โณ This skill is pending AI review.
Scores will appear once the review pipeline completes.
photo-abstract-editorial
Use when one or more photographs must become adaptive photo-plus-abstraction editorial compositions while source facts, spatial relationships, and strict photo preservation remain controlled.
Choose how to use this skill
You do not need every option. Choose the path your AI client supports. The stable page stays the same; versioned files are immutable.
1. Native installer
This listing has no registered native installer command. Use the complete package or source fallback below, depending on what your client supports.
Do not guess an installer command or replace an existing version without reviewing the diff.
2. Complete package recommended
Download the ZIP when available. It includes SKILL.md plus the references, security notes and version metadata.
No complete ProSkills package is published for this listing yet.3. Prompt-only
Copy the prompt above when the agent can read the stable page or when you want to adopt the workflow without installing a skill.
Need only the instruction file?
Download SKILL.md only if your client requires a single file. The complete ZIP is safer for a full installation because it preserves the references and release context.
No path installs or executes anything by itself. Your agent still needs access to the project files. Before updating, compare the installed version and review the diff.
// RATINGS
Not yet listed on ClawHub or SkillsMP
// README
๐จ Photo Abstract Editorial
A source-faithful Agent Skill that turns photographs into editorial artworks with adaptive layouts, controlled abstraction, and a deterministic composition path.
English ยท ็ฎไฝไธญๆ ยท ็น้ซไธญๆ
Photo Abstract Editorial keeps the source photograph factual, turns source-derived relationships into a restrained abstraction panel, and renders the editorial title locally where the verified path supports it.
Before โ After
Same source ยท Actual V3 output ยท Strict Fidelity / validator PASS
The source is shown unchanged. The result is the existing V3 same-source artifact in this repository; its manifest records a pixel-exact photographic region, and the published same-source check reports validator PASS.
| Before | After |
|---|---|
![]() | ![]() |
Jump to: Quick Start ยท Gallery ยท Choose an Edition ยท Compatibility ยท Validation ยท Releases ยท Environment Contract
โก 30-Second Quick Start
V3 Adaptive is the recommended starting point for most new users.
Install
Download the v3.0.0 release ZIP and install the included Skill, or use the existing installer method:
Use $skill-installer to install the root photo-abstract-editorial Skill from https://github.com/kwhi6693-web/photo-abstract-editorial.
Run
Use $photo-abstract-editorial to create an adaptive, source-faithful photo-and-abstract editorial from this image.
Optional Controls
Use $photo-abstract-editorial. Set Abstraction to 60, Creative Freedom to 45, Identity Preservation to 80, and Spatial Fidelity to 70.
๐ผ๏ธ Gallery
These five images are actual V3 outputs from the RC evaluation corpus. They are Codex Strict Fidelity results with validator PASS, not promotional reconstructions.
| Pure Portrait | Landscape | Architecture |
|---|---|---|
![]() | ![]() | ![]() |
| Still Life | Minimal / Light |
|---|---|
![]() | ![]() |
| Scene | Layout | Validation |
|---|---|---|
| Pure Portrait | Vertical Monument | Strict Fidelity ยท validator PASS |
| Landscape | Wide Horizon | Strict Fidelity ยท validator PASS |
| Architecture | Centered Archive | Strict Fidelity ยท validator PASS |
| Still Life | Sparse Object | Strict Fidelity ยท validator PASS |
| Minimal / Light | Sparse Object | Strict Fidelity ยท validator PASS |
The full corpus and manifests are in assets/evals/v3.0-rc1 and the RC evaluation report.
If this workflow is useful to you, a โญ helps more people discover the project.
Why Photo Abstract Editorial?
- Preserves the source photograph in the verified Strict Fidelity path.
- Chooses a scene-aware layout instead of forcing every source into one frame.
- Keeps abstraction controlled and traceable to the photograph.
- Renders exact local typography under Strict Fidelity.
- Uses structured QA before delivery.
- Provides a reproducible deterministic path where the required capabilities are available.
Generic image prompts may redraw the source photograph, lose subject identity, invent unsupported facts, produce inconsistent abstraction levels, or render unreliable typography. This project separates the factual photograph from the generated abstract motif, makes layout and creative decisions explicit, and adds a quality gate before delivery.
๐งญ Choose Your Edition
Recommended for most new users: V3 Adaptive.
| Original | V3 Adaptive | |
|---|---|---|
| Runtime | Codex only | Capability-based; validated on Codex and the tested DeepSeek Harness path |
| Layout | Fixed | Adaptive |
| Controls | Manual | Four creative controls |
| QA | Original validator | Structured V3 QA |
| Best for | Historical Codex workflow | Current recommended workflow |
Original remains supported for users who want the preserved fixed Codex workflow. V3 is the current formal stable release for scene adaptation, explicit controls, adaptive layouts, structured QA, and series work. The two editions can remain available side by side.
๐ Original vs V3 Comparison
| Dimension | Original Edition | V3 Adaptive Edition |
|---|---|---|
| Positioning | Preserved pre-V3 workflow | Adaptive photo-plus-abstraction Skill |
| Runtime environment | Codex only | Codex and the tested DeepSeek Harness capability path validated; capability-based compatibility designed |
| Complexity | Smaller, fixed workflow | More inputs, profiles, and verification boundaries |
| Art direction | Warm-ivory panel, muted source-derived motif, optical editorial spacing | Same editorial foundation plus scene-aware art direction and control resolution |
| Source fidelity | Pixel-exact photo region validated in the Original example | Strict Fidelity pixel-exact photo region validated in Codex and the tested DeepSeek Harness path |
| Creative controls | Manual title, panel, motif, alignment, and font overrides | Abstraction, Creative Freedom, Identity Preservation, Spatial Fidelity, each 0โ100 |
| Scene Profiles | None | 7: Pure Portrait, Environmental Portrait, Landscape, Architecture, Street/Crowd, Still Life, Minimal/Light |
| Portrait adaptation | Source-derived unequal vertical anchors when appropriate | Portrait-aware scene profile and identity-preservation resolution |
| Layout system | Lower Editorial predecessor: lower-left or bottom-center | 5 canonical profiles: Lower Editorial, Wide Horizon, Vertical Monument, Centered Archive, Sparse Object |
| QA | Machine validator plus visual QA; Original example has 9 validator checks | Structured 8-dimension QA, machine validator, and explicit HARD FAIL rules |
| Retry | One targeted motif correction; stop after two motif attempts | One targeted correction after a complete attempt; stop on a remaining critical HARD FAIL |
| Series | No Series Style Lock contract | Optional Series Style Lock process |
| Agent compatibility | Codex only | Codex and the tested DeepSeek Harness capability path are validated; other hosts remain capability-dependent |
| Installation | Historical Original package | V3 Adaptive package and current stable release |
| Best use cases | Stable, familiar, |
// HOW IT'S BUILT
// FILES
119 files shipped with this skill. Inline text is the catalog copy. The upstream link is the source repository.
SKILL.md3.4 KBce35c1fa4a29upstream โ
--- name: photo-abstract-editorial description: Use when one or more photographs must become adaptive photo-plus-abstraction editorial compositions while source facts, spatial relationships, and strict photo preservation remain controlled. --- # Photo Abstract Editorial Create a lightweight editorial composition from a supplied photograph. The photograph remains the sole factual source; generated content is an abstract motif, never a replacement for the final photo region or exact title. ## Capability routing Choose the highest-fidelity mode the host can actually execute: - **Strict Fidelity:** host visual-understanding capability, host image-generation capability, local filesystem access, local Python/script execution, Pillow-compatible image processing, and a usable serif font. This mode deterministically assembles photo, panel, motif, and typography, then runs machine validation. - **Native Image Edit:** host visual-understanding capability plus host image-editing/composition capability, without the complete local Strict toolchain. Use the source as the edit/reference image and disclose that pixel-exact, deterministic, machine-verified photo preservation is unavailable. - **Reference Generation:** host visual-understanding capability, host image-generation capability, and reference-image support. Use the source as the sole factual reference, perform best-effort fidelity, and disclose the same Strict guarantees are unavailable. If visual understanding or every image-generation/editing capability is absent, stop rather than fabricate a result. ## Workflow 1. Validate the required image input and inspect the source. 2. Classify the scene and resolve the four controls in [control-system.md](references/control-system.md) using [scene-profiles.md](references/scene-profiles.md). 3. Select a layout from [layout-profiles.md](references/layout-profiles.md) using scene facts, orientation, dominant axis, subject location, center of gravity, and negative space. Omitted layout falls back to Lower Editorial. 4. Read [art-direction.md](references/art-direction.md). Generate or edit only what the selected mode permits; keep the motif sparse, traceable, text-free, and source-derived. When the source supports a decisive vertical structure, use one primary and one subordinate irregular vertical stroke, unequal in height and width. 5. In Strict Fidelity, run [remove_chroma_key.py](scripts/remove_chroma_key.py), [compose_editorial.py](scripts/compose_editorial.py), and [validate_editorial.py](scripts/validate_editorial.py). Never overwrite source, motif, output, or manifest. 6. Run the structured eight-dimension QA in [quality-check.md](references/quality-check.md). Strict machine validation is necessary but not sufficient; visual/editorial QA is also required. 7. Apply at most one targeted correction. Re-run the relevant generation/composition, machine validation, and QA; stop after the second attempt if a critical HARD FAIL remains. 8. For an explicitly requested multi-image series only, load [series-style.md](references/series-style.md) and lock the visual family while re-analyzing each new source. The [validated example pair](references/example-pair.md) demonstrates package structure and regression expectations; do not copy its subject, palette, title, or coordinates. Return the final artifact, saved path, selected mode, and concise QA result. Do not claim Strict guarantees for Native Image Edit or Reference Generation.assets/evals/v3.0-rc1/README.md3.5 KBea2b66434589upstream โ
# V3.0 RC1 Evaluation Corpus This directory contains the reproducible local corpus and rendered evidence used for the V3.0 RC1 Codex-side evaluation. ## Provenance - Seven source images were generated with the built-in image-generation capability for this evaluation. They are fictional, ordinary-photography scenes created for testing; no user photographs, web downloads, stock assets, or copyrighted reference images were used. - Seven motif inputs were generated as transparent-background abstract marks from the same scene facts. Each motif was passed through the repository chroma-cleanup script before composition. - The generated-image originals remain outside the repository in the Codex generated-images store. The repository copies here are the exact inputs used by the recorded compositions. - All composition, cleanup, validation, and report generation was run locally by Codex on the RC1 evaluation branch. ## Scene corpus | Case | Scene profile | Source prompt intent | Motif prompt intent | Final layout | Result | |---|---|---|---|---|---| | E01 | pure-portrait | Quiet natural-light head-and-shoulders portrait of a fictional adult woman | Small charcoal and umber shoulder/axis mark; no face, text, or literal portrait | vertical-monument | result.png | | E02 | environmental-portrait | Fictional adult man in an olive jacket and helmet at a concrete harbor wall | Horizontal harbor-edge and figure-relation mark; no literal person or text | vertical-monument | result.png | | E03 | landscape | Coastal salt marsh at dusk with a long horizon and path | Broad lateral marsh, water, horizon, and path relationships | wide-horizon | result.png | | E04 | architecture | Generic vertical stone bell tower in quiet daylight | Narrow vertical stone/arch cue; no literal building rendering or text | centered-archive | result.png | | E05 | street-crowd | Anonymous rainy crosswalk crowd with umbrellas; no readable signage | Repeating pedestrian and umbrella rhythm; no faces, text, or logos | lower-editorial | result.png | | E06 | still-life | Ceramic cup and one pear on linen | Small planar cup/pear relation; no literal object drawing or text | sparse-object | result.png | | E07 | minimal-light | Empty plaster room with a chair and diagonal window light | Sparse diagonal light/room relationship; no literal furniture drawing or text | sparse-object | result.png | Each scene directory contains source.png, motif-generated.png, motif-clean.png, result.png, and the compositor manifest. E03 also retains result-attempt-1.png and its manifest to document the one targeted retry correction. ## Control corpus The control variants use the E03 landscape source and the same wide-horizon compositor geometry. They exercise the orchestration-level control prompts rather than a compositor command-line control schema: - abstraction-30, abstraction-60, abstraction-80: conservative, balanced, and strongly relational motif prompts. - freedom-low, freedom-high: conservative source-traceable and bolder mark-grouping prompts. Every control result has a machine-validated manifest. The control values are recorded in the evaluation report as applied prompt decisions; they are not claimed to be numeric values enforced by the runtime. ## Reproduction From the repository root, use the bundled Python dependency environment and run the cleanup, compositor, and validator scripts documented in the report. The generated result manifests record source hashes, rendered photo-region hashes, layout geometry, typography runs, and motif-cleanup details.README.md27 KB11167b32f894upstream โ
# ๐จ Photo Abstract Editorial A source-faithful Agent Skill that turns photographs into editorial artworks with adaptive layouts, controlled abstraction, and a deterministic composition path. [English](README.md) ยท [็ฎไฝไธญๆ](README.zh-CN.md) ยท [็น้ซไธญๆ](README.zh-TW.md) [](https://github.com/kwhi6693-web/photo-abstract-editorial/releases/tag/v1.0.0) [](https://github.com/kwhi6693-web/photo-abstract-editorial/releases/tag/v3.0.0) [](https://github.com/kwhi6693-web/photo-abstract-editorial) [](https://github.com/kwhi6693-web/photo-abstract-editorial#compatibility) Photo Abstract Editorial keeps the source photograph factual, turns source-derived relationships into a restrained abstraction panel, and renders the editorial title locally where the verified path supports it. <a id="before-after"></a> ## Before โ After **Same source ยท Actual V3 output ยท Strict Fidelity / validator PASS** The source is shown unchanged. The result is the existing V3 same-source artifact in this repository; its [manifest](assets/readme/comparisons/original-horizon/v3-result.png.manifest.json) records a pixel-exact photographic region, and the published same-source check reports validator PASS. | Before | After | |---|---| |  |  | <a id="navigation"></a> **Jump to:** [Quick Start](#quick-start) ยท [Gallery](#gallery) ยท [Choose an Edition](#choose-an-edition) ยท [Compatibility](#compatibility) ยท [Validation](#validation) ยท [Releases](#releases) ยท [Environment Contract](#environment-contract) <a id="quick-start"></a> ## โก 30-Second Quick Start V3 Adaptive is the recommended starting point for most new users. ### Install Download the [v3.0.0 release ZIP](https://github.com/kwhi6693-web/photo-abstract-editorial/releases/download/v3.0.0/photo-abstract-editorial-skill.zip) and install the included Skill, or use the existing installer method: ~~~text Use $skill-installer to install the root photo-abstract-editorial Skill from https://github.com/kwhi6693-web/photo-abstract-editorial. ~~~ ### Run ~~~text Use $photo-abstract-editorial to create an adaptive, source-faithful photo-and-abstract editorial from this image. ~~~ ### Optional Controls ~~~text Use $photo-abstract-editorial. Set Abstraction to 60, Creative Freedom to 45, Identity Preservation to 80, and Spatial Fidelity to 70. ~~~ <a id="gallery"></a> ## ๐ผ๏ธ Gallery These five images are actual V3 outputs from the RC evaluation corpus. They are Codex Strict Fidelity results with validator PASS, not promotional reconstructions. | Pure Portrait | Landscape | Architecture | |---|---|---| |  |  |  | | Still Life | Minimal / Light | |---|---| |  |  | | Scene | Layout | Validation | |---|---|---| | Pure Portrait | Vertical Monument | Strict Fidelity ยท validator PASS | | Landscape | Wide Horizon | Strict Fidelity ยท validator PASS | | Architecture | Centered Archive | Strict Fidelity ยท validator PASS | | Still Life | Sparse Object | Strict Fidelity ยท validator PASS | | Minimal / Light | Sparse Object | Strict Fidelity ยท validator PASS | The full corpus and manifests are in [assets/evals/v3.0-rc1](assets/evals/v3.0-rc1) and [the RC evaluation report](docs/evals/v3.0-rc1-real-world-evaluation.md). If this workflow is useful to you, a โญ helps more people discover the project. <a id="why"></a> ## Why Photo Abstract Editorial? - Preserves the source photograph in the verified Strict Fidelity path. - Chooses a scene-aware layout instead of forcing every source into one frame. - Keeps abstraction controlled and traceable to the photograph. - Renders exact local typography under Strict Fidelity. - Uses structured QA before delivery. - Provides a reproducible deterministic path where the required capabilities are available. Generic image prompts may redraw the source photograph, lose subject identity, invent unsupported facts, produce inconsistent abstraction levels, or render unreliable typography. This project separates the factual photograph from the generated abstract motif, makes layout and creative decisions explicit, and adds a quality gate before delivery. <a id="choose-an-edition"></a> ## ๐งญ Choose Your Edition **Recommended for most new users: V3 Adaptive.** | | Original | V3 Adaptive | |---|---|---| | Runtime | Codex only | Capability-based; validated on Codex and the tested DeepSeek Harness path | | Layout | Fixed | Adaptive | | Controls | Manual | Four creative controls | | QA | Original validator | Structured V3 QA | | Best for | Historical Codex workflow | Current recommended workflow | Original remains supported for users who want the preserved fixed Codex workflow. V3 is the current formal stable release for scene adaptation, explicit controls, adaptive layouts, structured QA, and series work. The two editions can remain available side by side. - [Original v1.0.0 release](https://github.com/kwhi6693-web/photo-abstract-editorial/releases/tag/v1.0.0) - [V3 Adaptive v3.0.0 release](https://github.com/kwhi6693-web/photo-abstract-editorial/releases/tag/v3.0.0) <a id="original-vs-v3"></a> ## ๐ Original vs V3 Comparison <details> <summary>Open the complete edition comparison</summary> | Dimension | Original Edition | V3 Adaptive Edition | |---|---|---| | Positioning | Preserved pre-V3 workflow | Adaptive photo-plus-abstraction Skill | | Runtime environment | Codex only | Codex and the tested DeepSeek Harness capability path validated; capability-based compatibility designed | | Complexity | Smaller, fixed workflow | More inputs, profiles, and verification boundaries | | Art direction | Warm-ivory panel, muted source-derived motif, optical editorial spacing | Same editorial foundation plus scene-aware art direction and control resolution | | Source fidelity | Pixel-exact photo region validated in the Original example | Strict Fidelity pixel-exact photo region validated in Codex and the tested DeepSeek Harness path | | Creative controls | Manual title, panel, motif, alignment, and font overrides | Abstraction, Creative Freedom, Identity Preservation, Spatial Fidelity, each 0โ100 | | Scene Profiles | None | 7: Pure Portrait, Environmental Portrait, Landscape, Architecture, Street/Crowd, Still Life, Minimal/Light | | Portrait adaptation | Source-derived unequal vertical anchors when appropriate | Portrait-aware scene profile and identity-preservation resolution | | Layout system | Lower Editorial predecessor: lower-left or bottom-center | 5 canonical profiles: Lower Editorial, Wide Horizon, Vertical Monument, Centered Archive, Sparse Object | | QA | Machine validator plus visual QA; Original example has 9 validator checks | Structured 8-dimension QA, machine validator, and explicit HARD FAIL rules | | Retry | One targeted motif correction; stop after two motif attempts | One targeted correction after a complete attempt; stop on a remaining critical HARD FAIL | | Series | No Series Style Lock contract | Optional Series Style Lock process | | Agent compatibility | Codex only | Codex and the tested DeepSeek Harness capability path are validated; other hosts remain capability-dependent | | Installation | Historical Original package | V3 Adaptive package and current stable release | | Best use cases | Stable, familiar, Codex-specific original behavior | Different scenes, explicit controls, layout adaptation, and series work | | Main trade-off | Less adaptive and not portable beyond Codex | More process and capability requirements; native tools still vary by host | </details> <a id="scope"></a> ## ๐ฏ Scope, Limitations & Guarantees <details> <summary>Open scope and guarantee details</summary> ### Original Edition Best suited for one photograph, one text-free motif, and one fixed editorial composition in Codex. It provides deterministic local assembly and the Original validator's source/panel checks. It does not provide the V3 control system, scene profiles, canonical auto-layout, Series Style Lock, or cross-Agent contract. It is not supported outside Codex. ### V3 Adaptive Edition Best suited for photographs that benefit from explicit scene reasoning, identity and spatial controls, adaptive layouts, or a series-level visual family. V3 does not invent unsupported people, buildings, objects, geography, text, logos, watermarks, swatches, or decorative facts. Image generation can vary across hosts and runs; current public evidence includes Codex Strict Fidelity and the independently tested DeepSeek Harness capability path. ### Guarantee Matrix | Capability | Original | V3 Strict Fidelity | V3 Native Image Edit | V3 Reference Generation | |---|---|---|---|---| | Source-aware behavior | Implemented and Codex-validated | Implemented and validated on Codex and the tested DeepSeek Harness path | Best effort | Best effort | | Scene adaptation | Fixed original behavior | Implemented and Codex-evaluated | Art-direction contract only | Art-direction contract only | | Creative controls | Manual overrides | 4 controls, 0โ100 | Host-dependent best effort | Host-dependent best effort | | Pixel-exact photo region | Validated for Original example | Machine-validated when Strict conditions hold | Not guaranteed | Not guaranteed | | Exact local typography | Original compositor path | Local typography path | Not guaranteed | Not guaranteed | | Deterministic composition | Original local compositor | V3 local compositor | Not guaranteed | Not guaranteed | | Machine validation | Original validator | V3 validator and manifest | Not a Strict output | Not a Strict output | | Host compatibility | Codex only | Codex and the tested DeepSeek Harness path validated; suitable capability-based hosts designed for | Capability-dependent | Capability-dependent | Native Image Edit and Reference Generation must never be described as machine-verified Strict output. </details> <a id="features"></a> ## โจ Key Features <details> <summary>Open the feature-level contract</summary> ### Original Edition - One source photograph and one sparse text-free motif. - Source-derived bands, gaps, offsets, and unequal vertical anchors when supported by the source. - Warm-ivory panel, muted palette, exact English title, and optical spacing. - Deterministic local composition and JSON manifest. - Original source/output collision guards and pixel-level photo-region validation. - Machine validation plus visual QA. ### V3 Adaptive Edition - Capability-based routing across Strict Fidelity, Native Image Edit, and Reference Generation. - Four Creative Controls: Abstraction, Creative Freedom, Identity Preservation, Spatial Fidelity. - Seven Scene Profiles and portrait-aware abstraction. - Five deterministic Layout Profiles. - Structured eight-dimension QA and bounded targeted retry. - Optional Series Style Lock process. - Portable chroma cleanup, exact local typography, manifest, machine validator, and reproducible package. </details> <a id="execution-modes"></a> ## ๐งฉ V3 Execution Modes <details> <summary>Open the mode boundaries</summary> ### Strict Fidelity When the host has visual understanding, image generation, local file access, Python, Pillow-compatible processing, and a usable serif font, V3 can provide deterministic composition, exact local typography, a manifest, a machine validator, and pixel-exact photographic-region verification when no resizing is requested. ### Native Image Edit Uses the source as the edit/reference image and follows the same art-direction and QA contract where the host supports it. It is best effort: no pixel-exact, deterministic, exact-local-typography, or machine-verified Strict guarantee is made. ### Reference Generation Uses the source as the sole factual reference for a model-dependent generation path. It is best effort and does not provide Strict Fidelity guarantees. </details> <a id="requirements"></a> ## ๐ Requirements <details> <summary>Open runtime requirements</summary> ### Original Edition **Codex only.** The Original workflow requires Codex visual inspection, the built-in image-generation path, the Codex workspace/runtime helper used by the historical workflow, Python 3.10โ3.13 (CI-tested), Pillow, a usable serif font, one photograph, and a transparent motif. ### V3 Strict Fidelity Visual understanding, image generation, local filesystem access, Python 3.10โ3.13 (CI-tested), Pillow-compatible processing, a usable serif font, one photograph, and a transparent motif. ### V3 Native Image Edit Visual understanding plus host image-editing/composition capability. Pixel-exact Strict guarantees are not available. ### V3 Reference Generation Visual understanding, image generation, and reference-image support. Results are model-dependent and best effort. </details> ### Development Tests The test runner is a development dependency and is not included in either runtime release archive. From a fresh checkout, create an isolated environment, install the repository development dependencies, and run `pytest` directly: ~~~powershell python -m venv .venv .venv\Scripts\Activate.ps1 python -m pip install -r requirements-dev.txt python scripts/doctor.py --development pytest python scripts/validate_editorial.py --source assets/examples/source-horizon.png --output assets/examples/result-horizon.png --manifest assets/examples/result-horizon.png.manifest.json ~~~ ~~~bash python3 -m venv .venv source .venv/bin/activate python -m pip install -r requirements-dev.txt python scripts/doctor.py --development pytest python scripts/validate_editorial.py --source assets/examples/source-horizon.png --output assets/examples/result-horizon.png --manifest assets/examples/result-horizon.png.manifest.json ~~~ <a id="environment-contract"></a> ## ๐ฉบ Environment Contract The local compositor and validator use `Pillow` as the only Python runtime dependency. Install it from [requirements.txt](requirements.txt) when running repository scripts; [requirements-dev.txt](requirements-dev.txt) reuses that runtime manifest and adds the development-only `pytest` runner. The read-only preflight is `python scripts/doctor.py`. It checks the tested Python versions (3.10โ3.13), Pillow and required imports, temporary storage, RGB/RGBA PNG round-trips, project files, the validator, serif-font discovery, and formal runtime archives. `python scripts/doctor.py --development --strict-packages` is the full local development check. Visual understanding, image generation, native image editing, and filesystem access are host capabilities, not pip dependencies. The CI contract covers Ubuntu/Linux on Python 3.10โ3.13, Windows on Python 3.12, and macOS on Python 3.13. Runtime release archives retain their existing Skill file contract and exclude tests, caches, and development-only files. The dependency/import check is `python tools/check_dependency_contract.py`; the package and repository hygiene gates are `python tools/check_package_parity.py` and `python tools/check_repository_hygiene.py`. <a id="installation"></a> ## ๐ฆ Installation <details> <summary>Open installation details for both editions</summary> ### Original Edition Download [photo-abstract-editorial-original.zip](https://github.com/kwhi6693-web/photo-abstract-editorial/releases/download/v1.0.0/photo-abstract-editorial-original.zip), extract the included photo-abstract-editorial folder, and install it in the Codex Skill directory. Original is Codex only. The repository copy is [dist/photo-abstract-editorial-original.zip](dist/photo-abstract-editorial-original.zip). ### V3 Adaptive Edition Download [photo-abstract-editorial-skill.zip](https://github.com/kwhi6693-web/photo-abstract-editorial/releases/download/v3.0.0/photo-abstract-editorial-skill.zip), or ask Codex to install the root Skill from the public repository. The repository copy is [dist/photo-abstract-editorial-skill.zip](dist/photo-abstract-editorial-skill.zip). The package contains only V3 runtime files. README assets, evaluation files, tests, caches, and temporary outputs are not part of the V3 runtime package. </details> <a id="usage"></a> ## ๐ Usage <details> <summary>Open additional usage patterns</summary> ### Original Edition ~~~text Use $photo-abstract-editorial to turn this photograph into a strict-fidelity photo-and-abstract editorial composition. ~~~ The Original prompt invokes the historical Codex-specific workflow. ### V3 Adaptive Edition โ Default The default V3 prompt is in [30-Second Quick Start](#quick-start). ### V3 Adaptive Edition โ Series ~~~text Use $photo-abstract-editorial for this three-image series. Keep a restrained warm-ivory editorial family, but re-analyze each photograph and choose its own motif, layout, and spatial relationships. ~~~ Controls are natural-language workflow inputs in the current Skill contract; they are not claimed to be a numeric compositor API. </details> <a id="input-output"></a> ## ๐ผ๏ธ Input โ Output Ordinary source photograph โ scene facts and source-derived relationships โ abstract motif, panel, and exact title โ machine validation and visual QA โ Photo Abstract Editorial result <a id="same-source-showcase"></a> ## โ๏ธ Same-Source Original vs V3 This comparison uses the historical Original README source. The Original result is reused unchanged; the V3 result was generated later from the same source with the current V3 workflow. | Same source | Original historical result | V3 Adaptive actual result | |---|---|---| |  |  |  | The Original keeps its fixed lower-left editorial behavior. V3 resolves the same long horizon as Wide Horizon and applies the current scene-aware composition. The comparison explains different decisions; it does not claim that one edition is universally better. <a id="control-showcase"></a> ## ๐๏ธ V3 Control Showcase These three outputs use the same E03 Landscape source and the same Wide Horizon layout. The variants are real Codex orchestration-level control prompts, and all three passed the machine validator. | Abstraction 30 | Abstraction 60 | Abstraction 80 | |---|---|---| |  |  |  | The current compositor does not expose these values as a numeric command-line argument; the report records them as applied orchestration decisions rather than machine-enforced scores. <a id="series-showcase"></a> ## ๐๏ธ V3 Series Showcase These three actual outputs were reviewed as a Codex-side series-style evaluation: same warm-ivory panel language, Bodoni typography family, whole-run kerning, restrained whitespace, and source-traceable mark vocabulary, while each image keeps its own scene logic. **Same visual family. Different source logic.** | Landscape | Street / Crowd | Minimal / Light | |---|---|---| |  |  |  | This is a real visual PASS review of the documented Series Style Lock contract on Codex outputs. The current runtime does not expose a separate machine flag named Series Style Lock; the series showcase itself remains Codex-side, while the independent DeepSeek validation covers the deterministic capability path described below. <a id="how-it-works"></a> ## โ๏ธ How It Works <details> <summary>Open the workflow sequence</summary> ### Original Edition Photograph โ Codex visual inspection โ text-free motif generation โ historical cleanup helper โ Original compositor โ Original validator โ visual QA. ### V3 Adaptive Edition Photograph โ Scene Analysis โ Creative Controls โ Layout Selection โ Art Direction โ Generation/Edit โ Quality Gate โ one targeted correction when needed โ Final Editorial. ### V3 Strict Fidelity Pipeline Transparent motif โ portable chroma cleanup โ deterministic composer โ exact local typography โ manifest โ machine validator. </details> <a id="validation"></a> ## โ Validation The numbers below are from the current public source and the v3.0.0 release, not historical estimates. - Original Edition: 32/32 tests passed in the isolated v1.0.0 worktree. - Original Edition: its own builder and --check passed; its historical example validator returned ok=true. - Original package: 9 runtime files; SHA-256 a1a44b1a9cec9ba04b379a7d3a14315701abb14bb93e953003337870772d0a6d. - V3 Adaptive Edition: 41/41 tests passed on Codex. - V3 package: 15 runtime files; package build and --check passed. - V3 RC evaluation: 7 scene results and 5 control variants passed the validator and photo-region pixel-exact checks. - V3 layouts: all 5 canonical layouts exercised on real source images. - V3 same-source comparison: validator PASS for the new V3 result. - V3 package excludes README assets, docs/evals, tests, caches, and temporary outputs. Built to be checked, not just generated. <a id="independent-validation"></a> ## ๐ฌ Independent Validation An independent DeepSeek Harness run completed the deterministic capability path. - DeepSeek Harness: PASS - Strict Fidelity: VERIFIED on the tested pipeline - Repository tests: 41/41 passed - Structured visual/editorial QA: 8/8 PASS - Machine validator: `ok: true`; validator errors: `[]` - Blocking issues: 0 - Cross-agent compatibility: VERIFIED for the tested capability path - Pipeline smoke checks: `remove_chroma_key.py`, `compose_editorial.py`, and `validate_editorial.py` all exited 0 See the concise [DeepSeek Harness validation summary](docs/evals/deepseek-harness-validation.md). DeepSeek Harness did not expose native neural image generation during this validation, so motif synthesis used deterministic procedural Pillow generation. This limitation did not change the verified photo-region pixel exactness, source hash, geometry, typography, deterministic composition, or machine-validation results. <a id="compatibility"></a> ## ๐ Compatibility ### Original Edition **CODEX ONLY.** Original is implemented and validated within its Codex-specific runtime contract. It is not supported on Claude, Gemini, Cursor, or other Agents. ### V3 Adaptive Edition - Validated runtime paths: Codex Strict Fidelity and an independent DeepSeek Harness Strict Fidelity run. - Designed for capability-based compatibility: suitable image-capable Agents/Harnesses that provide the required capabilities. - Cross-agent compatibility is verified for the tested capability path; this does not imply that every Agent exposes identical native tools. DESIGNED FOR COMPATIBILITY is not the same as VALIDATED. <a id="releases"></a> ## ๐ฅ Releases ### Photo Abstract Editorial โ Original Edition - Tag: v1.0.0 - Runtime: Codex only - Release: [Original Edition release](https://github.com/kwhi6693-web/photo-abstract-editorial/releases/tag/v1.0.0) - Artifact: [photo-abstract-editorial-original.zip](https://github.com/kwhi6693-web/photo-abstract-editorial/releases/download/v1.0.0/photo-abstract-editorial-original.zip) - Demo: preserved historical README source/result pair - Audit: [Original Edition feature and provenance audit](docs/releases/original-edition.md) ### Photo Abstract Editorial V3 Adaptive โ v3.0.0 - Tag: v3.0.0 - Release: [V3 Adaptive stable release](https://github.com/kwhi6693-web/photo-abstract-editorial/releases/tag/v3.0.0) - Artifact: [photo-abstract-editorial-skill.zip](https://github.com/kwhi6693-web/photo-abstract-editorial/releases/download/v3.0.0/photo-abstract-editorial-skill.zip) - Validation: Codex regression, build, package, and release-preflight checks PASS - Independent validation: DeepSeek Harness PASS for the tested deterministic capability path v3.0.0 is the current formal stable release. <a id="switching"></a> ## ๐ Switching / Upgrade Guide Stay on Original if you prefer the fixed historical visual behavior, use Codex, and want a smaller workflow. Try V3 if you need scene adaptation, four creative controls, automatic layouts, structured QA, or series work. No Original user is required to migrate. The two packages can remain available side by side. <a id="faq"></a> ## โ FAQ ### Can Original run outside Codex? No. Original is CODEX ONLY. ### Which edition should I download? Download Original for the historical fixed Codex workflow. Download V3 for adaptive scene logic and explicit controls. ### Does V3 require manually setting controls? No. The default workflow resolves them from the source. You may provide natural-language values when you want a deliberate bias. ### Why can Native Image Edit and Reference Generation not guarantee pixel-exact output? Those modes depend on host/model image operations rather than the local deterministic Strict compositor. ### Can both editions remain available? Yes. That is the purpose of the dual-edition release structure. ### Is V3 validated on Claude, Gemini, or Cursor? Not in this repository's published evidence. DeepSeek Harness is independently verified for the tested deterministic capability path; other Agents remain capability-dependent and untested. ### What does Series Style Lock preserve? It preserves a visual family such as panel language, palette, typography, whitespace, and mark vocabulary while re-analyzing each source instead of copying coordinates or motifs. <a id="technical-details"></a> ## ๐ ๏ธ Technical Details <details> <summary>V3 runtime structure</summary> ~~~text photo-abstract-editorial/ |- SKILL.md |- agents/openai.yaml |- references/ |- scripts/compose_editorial.py |- scripts/remove_chroma_key.py |- scripts/validate_editorial.py `- assets/examples/ ~~~ </details> <details> <summary>Reproducible packages</summary> Original uses the Original Edition builder from the historical v1.0.0 tree and contains 9 runtime files. V3 uses the current builder and contains 15 runtime files. Both builders exclude development-only material from their runtime archive and verify archive entries against their own source tree. </details> <details> <summary>Manifest and Strict checks</summary> V3 manifests record source/output hashes, rendered photo-region hash, panel geometry, layout profile, motif region, cleanup details, and typography runs. The validator checks the manifest, dimensions, geometry, panel corners, source hash, output hash, and photo-region pixels. </details> <a id="project-status"></a> ## ๐ Project Status - Original Edition: supported, preserved, and available as the historical v1.0.0 release. - V3 Adaptive Edition: current formal stable release with Codex validation and independent DeepSeek Harness validation for the tested capability path. - Documentation: three-language dual-edition README and real demo assets. - Cross-agent compatibility: verified for the tested DeepSeek Harness capability path; other Agents remain untested. - Formal v3.0.0: current stable release. <a id="license"></a> ## โ๏ธ License This project and its public releases are licensed under AGPL-3.0. See [LICENSE](LICENSE).scripts/compose_editorial.py33 KBa0d1d880f2d3upstream โ
Contents are not inlined in the catalog. Open the upstream file.
scripts/doctor.py18 KB9feb64774e52upstream โ
#!/usr/bin/env python3 """Read-only local environment preflight for Photo Abstract Editorial.""" from __future__ import annotations import argparse from dataclasses import dataclass import importlib import importlib.util from pathlib import Path import platform import subprocess import sys import tempfile REPO_ROOT = Path(__file__).resolve().parents[1] SUPPORTED_PYTHON = ((3, 10), (3, 11), (3, 12), (3, 13)) REQUIRED_IMPORTS = ( "PIL", "PIL.Image", "PIL.ImageColor", "PIL.ImageDraw", "PIL.ImageFont", "PIL.ImageOps", ) RUNTIME_PROJECT_FILES = ( "SKILL.md", "README.md", "README.zh-CN.md", "README.zh-TW.md", "requirements.txt", "agents/openai.yaml", "references/art-direction.md", "references/control-system.md", "references/example-pair.md", "references/layout-profiles.md", "references/quality-check.md", "references/scene-profiles.md", "references/series-style.md", "scripts/compose_editorial.py", "scripts/doctor.py", "scripts/remove_chroma_key.py", "scripts/validate_editorial.py", ) @dataclass(frozen=True) class CheckResult: name: str status: str detail: str what_failed: str | None = None why: str | None = None required: str | None = None detected: str | None = None how_to_fix: str | None = None def _pass(name: str, detail: str) -> CheckResult: return CheckResult(name=name, status="PASS", detail=detail) def _not_detectable(name: str, detail: str) -> CheckResult: return CheckResult(name=name, status="NOT DETECTABLE", detail=detail) def _fail( name: str, *, detail: str, what_failed: str, why: str, required: str, detected: str, how_to_fix: str, ) -> CheckResult: return CheckResult( name=name, status="FAIL", detail=detail, what_failed=what_failed, why=why, required=required, detected=detected, how_to_fix=how_to_fix, ) def render_check(result: CheckResult) -> str: line = f"{result.name}: {result.status}" if result.detail: line += f" โ {result.detail}" if result.status == "FAIL": fields = ( ("WHAT FAILED", result.what_failed), ("WHY", result.why), ("REQUIRED", result.required), ("DETECTED", result.detected), ("HOW TO FIX", result.how_to_fix), ) line += "\n" + "\n".join(f"{label}: {value}" for label, value in fields if value) return line def _check_platform() -> CheckResult: current = platform.system() supported = {"Windows": "Windows", "Linux": "Ubuntu/Linux", "Darwin": "macOS"} if current in supported: return _pass("OS/platform", f"{supported[current]} ยท {platform.machine()}") return _fail( "OS/platform", detail=f"unsupported platform {current!r}", what_failed="host platform support", why="the local filesystem and font contract is only verified on Windows, Linux, and macOS", required="Windows, Ubuntu/Linux, or macOS", detected=f"{current} ยท {platform.machine()}", how_to_fix="run this workflow on a supported platform or use a host capability mode", ) def _check_python() -> CheckResult: current = (sys.version_info.major, sys.version_info.minor) version = platform.python_version() if current in SUPPORTED_PYTHON: supported = ", ".join(f"{major}.{minor}" for major, minor in SUPPORTED_PYTHON) return _pass("Python", f"{version} ยท tested minors: {supported}") supported = ", ".join(f"{major}.{minor}" for major, minor in SUPPORTED_PYTHON) return _fail( "Python", detail=f"{version} is outside the tested support contract", what_failed="Python support status", why="the repository has not verified this Python minor version", required=f"one of: {supported}", detected=version, how_to_fix="install and activate one of the tested Python versions, then rerun python scripts/doctor.py", ) def _load_pillow() -> tuple[CheckResult, object | None]: try: pillow = importlib.import_module("PIL") except Exception as error: # pragma: no cover - depends on the caller's environment return ( _fail( "Pillow", detail="Pillow could not be imported", what_failed="Pillow import", why="the local compositor and validator require the runtime image dependency", required="Pillow==12.3.0 from requirements.txt", detected=f"{type(error).__name__}: {error}", how_to_fix="python -m pip install -r requirements.txt", ), None, ) return _pass("Pillow", str(getattr(pillow, "__version__", "version unavailable"))), pillow def _check_required_imports() -> CheckResult: missing: list[str] = [] for module_name in REQUIRED_IMPORTS: try: importlib.import_module(module_name) except Exception as error: # pragma: no cover - depends on the caller's environment missing.append(f"{module_name} ({type(error).__name__}: {error})") if not missing: return _pass("Required imports", ", ".join(REQUIRED_IMPORTS)) return _fail( "Required imports", detail="one or more required imports are unavailable", what_failed="required Python imports", why="a clean runtime environment cannot load the image pipeline", required=", ".join(REQUIRED_IMPORTS), detected="; ".join(missing), how_to_fix="python -m pip install -r requirements.txt", ) def _check_temp_directory() -> CheckResult: try: with tempfile.TemporaryDirectory(prefix="photo-abstract-editorial-doctor-") as temporary_directory: probe = Path(temporary_directory) / "write-probe.txt" probe.write_text("environment contract\n", encoding="utf-8") if probe.read_text(encoding="utf-8") != "environment contract\n": raise OSError("temporary file read-back mismatch") except Exception as error: # pragma: no cover - depends on the host filesystem return _fail( "Temporary directory", detail="temporary write/read failed", what_failed="temporary directory access", why="the compositor and tests need a writable, clean temporary workspace", required="a writable system temporary directory", detected=f"{type(error).__name__}: {error}", how_to_fix="check the system temporary directory permissions and rerun this doctor", ) return _pass("Temporary directory", "write/read/cleanup PASS") def _check_image_codecs() -> CheckResult: try: from PIL import Image with tempfile.TemporaryDirectory(prefix="photo-abstract-editorial-image-") as temporary_directory: root = Path(temporary_directory) rgb_path = root / "rgb.png" rgba_path = root / "rgba.png" Image.new("RGB", (7, 5), (24, 48, 72)).save(rgb_path, format="PNG") Image.new("RGBA", (7, 5), (24, 48, 72, 123)).save(rgba_path, format="PNG") with Image.open(rgb_path) as rgb: rgb.load() rgb_ok = rgb.mode == "RGB" and rgb.size == (7, 5) with Image.open(rgba_path) as rgba: rgba.load() alpha_ok = rgba.mode == "RGBA" and rgba.getpixel((0, 0))[3] == 123 except Exception as error: # pragma: no cover - depends on the caller's environment return _fail( "Image codecs", detail="PNG open/save or transparency probe failed", what_failed="Pillow image codec capability", why="the runtime must preserve ordinary RGB images and transparent motifs", required="RGB PNG and RGBA PNG read/write support", detected=f"{type(error).__name__}: {error}", how_to_fix="python -m pip install -r requirements.txt, then rerun python scripts/doctor.py", ) if not (rgb_ok and alpha_ok): return _fail( "Image codecs", detail="PNG read-back did not preserve expected mode, size, or alpha", what_failed="PNG codec round-trip", why="the image pipeline relies on deterministic RGB/RGBA behavior", required="RGB and RGBA PNG round-trip", detected=f"rgb_ok={rgb_ok}, alpha_ok={alpha_ok}", how_to_fix="reinstall the pinned Pillow runtime and rerun python scripts/doctor.py", ) return _pass("Image codecs", "RGB PNG and RGBA transparency round-trip PASS") def _check_project_files(repo_root: Path, development: bool) -> CheckResult: required = list(RUNTIME_PROJECT_FILES) if development: required.append("requirements-dev.txt") missing = [relative for relative in required if not (repo_root / relative).is_file()] if missing: return _fail( "Project files", detail="required repository files are missing", what_failed="project file contract", why="the workflow cannot run or be audited from this checkout", required="the runtime and documented project files", detected=", ".join(missing), how_to_fix="restore the missing files from the repository, then rerun python scripts/doctor.py", ) return _pass("Project files", f"{len(required)} required files present") def _run_read_only_command(command: list[str], cwd: Path, timeout: int) -> tuple[bool, str]: try: completed = subprocess.run( command, cwd=cwd, capture_output=True, text=True, check=False, timeout=timeout, ) except Exception as error: # pragma: no cover - depends on the host process return False, f"{type(error).__name__}: {error}" detail = (completed.stderr or completed.stdout).strip().replace("\n", " ") return completed.returncode == 0, detail def _check_validator(repo_root: Path) -> CheckResult: validator = repo_root / "scripts" / "validate_editorial.py" if not validator.is_file(): return _fail( "Validator", detail="validator script is missing", what_failed="validator executable", why="the repository cannot verify generated output", required="scripts/validate_editorial.py", detected=str(validator), how_to_fix="restore scripts/validate_editorial.py, then rerun python scripts/doctor.py", ) ok, detail = _run_read_only_command([sys.executable, str(validator), "--help"], repo_root, 20) if ok: return _pass("Validator", "scripts/validate_editorial.py --help PASS") return _fail( "Validator", detail="validator could not start", what_failed="validator executable", why="the machine validation entry point is not runnable in this environment", required="python scripts/validate_editorial.py --help returns exit code 0", detected=detail or "non-zero exit code", how_to_fix="python -m pip install -r requirements.txt, then rerun python scripts/doctor.py", ) def _check_serif_font(repo_root: Path) -> CheckResult: composer = repo_root / "scripts" / "compose_editorial.py" if not composer.is_file(): return _not_detectable("Serif font", "composer is missing") try: spec = importlib.util.spec_from_file_location("photo_abstract_editorial_doctor_composer", composer) if spec is None or spec.loader is None: raise ImportError("could not load compose_editorial.py") module = importlib.util.module_from_spec(spec) sys.modules[spec.name] = module spec.loader.exec_module(module) font_path = module.discover_font(None) if not Path(font_path).is_file(): raise FileNotFoundError(str(font_path)) from PIL import ImageFont ImageFont.truetype(str(font_path), size=16) except Exception as error: # pragma: no cover - depends on host fonts and Pillow return _fail( "Serif font", detail="no usable serif font was discovered", what_failed="serif font lookup", why="Strict Fidelity local typography requires a host-installed serif font", required="a usable serif font or an explicit --font path", detected=f"{type(error).__name__}: {error}", how_to_fix="install a system serif font or pass --font explicitly to the compositor", ) return _pass("Serif font", str(font_path)) def _check_pytest() -> CheckResult: try: pytest = importlib.import_module("pytest") return _pass("pytest", str(getattr(pytest, "__version__", "version unavailable"))) except Exception as error: # pragma: no cover - depends on the caller's environment return _fail( "pytest", detail="pytest could not be imported in development mode", what_failed="pytest development dependency", why="the repository test command cannot be reproduced from a fresh development environment", required="pytest==9.1.1 from requirements-dev.txt", detected=f"{type(error).__name__}: {error}", how_to_fix="python -m pip install -r requirements-dev.txt", ) def _check_dependency_contract(repo_root: Path) -> CheckResult: checker = repo_root / "tools" / "check_dependency_contract.py" if not checker.is_file(): return _not_detectable("Dependency contract", "checker is not present in this checkout") ok, detail = _run_read_only_command([sys.executable, str(checker)], repo_root, 20) if ok: return _pass("Dependency contract", "all Python imports map to declared manifests") return _fail( "Dependency contract", detail="undeclared import or manifest error detected", what_failed="dependency/import consistency", why="a fresh environment would not have a declared dependency needed by the repository", required="python tools/check_dependency_contract.py exits 0", detected=detail or "non-zero exit code", how_to_fix="update the correct requirements manifest, then rerun the dependency checker", ) def _check_package_structure(repo_root: Path, strict: bool) -> CheckResult: checker = repo_root / "tools" / "check_package_parity.py" dist = repo_root / "dist" if not checker.is_file() or not dist.is_dir(): if strict: return _fail( "Package structure", detail="package parity cannot be checked", what_failed="runtime artifact structure", why="the strict environment gate requires the formal runtime archives and parity checker", required="dist/ and tools/check_package_parity.py", detected=f"checker={checker.is_file()}, dist={dist.is_dir()}", how_to_fix="restore the formal dist archives and rerun python scripts/doctor.py --strict-packages", ) return _not_detectable("Package structure", "dist/ or package parity checker is not present") ok, detail = _run_read_only_command([sys.executable, str(checker)], repo_root, 90) if ok: return _pass("Package structure", "formal runtime archives pass parity and clean extraction smoke") return _fail( "Package structure", detail="runtime archive parity or extraction smoke failed", what_failed="runtime package structure", why="a release archive may be stale, incomplete, or not runnable after extraction", required="python tools/check_package_parity.py exits 0", detected=detail or "non-zero exit code", how_to_fix="rebuild the current package, inspect its exact entries, and rerun the package parity check", ) def run_checks( repo_root: Path = REPO_ROOT, *, development: bool = False, strict_packages: bool = False, ) -> list[CheckResult]: repo_root = Path(repo_root).resolve() pillow_result, _pillow = _load_pillow() results = [ _check_platform(), _check_python(), pillow_result, _check_required_imports(), _check_temp_directory(), _check_image_codecs(), _check_project_files(repo_root, development), _check_validator(repo_root), _check_serif_font(repo_root), _check_dependency_contract(repo_root), _check_package_structure(repo_root, strict_packages), ] if development: results.append(_check_pytest()) return results def overall_pass(results: list[CheckResult]) -> bool: return all(result.status != "FAIL" for result in results) def render_report(results: list[CheckResult]) -> str: lines = [ f"LOCAL ENVIRONMENT: {'PASS' if overall_pass(results) else 'FAIL'}", *[render_check(result) for result in results], "HOST CAPABILITIES:", "Visual understanding: NOT DETECTABLE โ host capability, not a pip dependency", "Image generation: REQUIRED FOR Strict Fidelity โ host capability, not a pip dependency", "Native image editing: OPTIONAL / REQUIRED BY MODE โ host capability, not a pip dependency", "Filesystem access: PASS โ local filesystem is available to this process", ] return "\n".join(lines) def main(argv: list[str] | None = None) -> int: parser = argparse.ArgumentParser(description=__doc__) parser.add_argument( "--development", action="store_true", help="also check requirements-dev.txt and import pytest", ) parser.add_argument( "--strict-packages", action="store_true", help="fail when formal runtime archives cannot be checked", ) args = parser.parse_args(argv) results = run_checks( development=args.development, strict_packages=args.strict_packages, ) print("Environment Contract: " + ("PASS" if overall_pass(results) else "FAIL")) print(render_report(results)) if not overall_pass(results): print("FAIL: see the actionable WHAT FAILED / WHY / REQUIRED / DETECTED / HOW TO FIX details above") return 0 if overall_pass(results) else 1 if __name__ == "__main__": raise SystemExit(main())scripts/remove_chroma_key.py8.7 KBe1404f9951b0upstream โ
#!/usr/bin/env python3 """Deterministically convert a configured chroma background into clean alpha.""" from __future__ import annotations import argparse import json import math import os import tempfile from collections import deque from pathlib import Path from typing import Any from PIL import Image, ImageColor def parse_color(value: str) -> tuple[int, int, int]: try: parsed = ImageColor.getrgb(value) except ValueError as exc: raise ValueError(f"invalid chroma color: {value!r}") from exc if len(parsed) != 3: raise ValueError("chroma color must be an opaque RGB color") return parsed def paths_refer_to_same_file(left: Path, right: Path) -> bool: if left.resolve() == right.resolve(): return True try: return left.exists() and right.exists() and left.samefile(right) except OSError as error: raise ValueError(f"cannot compare input and output paths: {error}") from error def remove_small_alpha_components( image: Image.Image, min_area: int, alpha_threshold: int = 32, ) -> tuple[Image.Image, int, int]: cleaned = image.convert("RGBA").copy() alpha = cleaned.getchannel("A") pixels = alpha.load() width, height = alpha.size visited = bytearray(width * height) removed_components = 0 removed_pixels = 0 for y in range(height): for x in range(width): index = y * width + x if visited[index]: continue visited[index] = 1 if pixels[x, y] < alpha_threshold: if pixels[x, y] != 0: pixels[x, y] = 0 removed_pixels += 1 continue component: list[tuple[int, int]] = [] queue: deque[tuple[int, int]] = deque([(x, y)]) while queue: current_x, current_y = queue.popleft() component.append((current_x, current_y)) for neighbor_y in range(max(0, current_y - 1), min(height, current_y + 2)): for neighbor_x in range(max(0, current_x - 1), min(width, current_x + 2)): neighbor_index = neighbor_y * width + neighbor_x if visited[neighbor_index]: continue visited[neighbor_index] = 1 if pixels[neighbor_x, neighbor_y] >= alpha_threshold: queue.append((neighbor_x, neighbor_y)) elif pixels[neighbor_x, neighbor_y] != 0: pixels[neighbor_x, neighbor_y] = 0 removed_pixels += 1 if len(component) < min_area: removed_components += 1 removed_pixels += len(component) for component_x, component_y in component: pixels[component_x, component_y] = 0 cleaned.putalpha(alpha) return cleaned, removed_components, removed_pixels def _neutralize_fringe( rgb: tuple[int, int, int], chroma: tuple[int, int, int], visibility: float, ) -> tuple[int, int, int]: """Reduce the chroma cast on semi-transparent edge pixels deterministically.""" luma = round((rgb[0] + rgb[1] + rgb[2]) / 3) blend = max(0.0, min(1.0, 1.0 - visibility)) return tuple(round(channel * (1.0 - blend) + luma * blend) for channel in rgb) def clean_rgba( image: Image.Image, chroma_color: tuple[int, int, int], tolerance: int, edge_softness: int, min_component_area: int, ) -> tuple[Image.Image, dict[str, int]]: if tolerance < 0: raise ValueError("tolerance must be non-negative") if edge_softness < 0: raise ValueError("edge_softness must be non-negative") if min_component_area < 1: raise ValueError("min_component_area must be at least 1") cleaned = image.convert("RGBA").copy() pixels = cleaned.load() removed_chroma_pixels = 0 fringe_pixels = 0 threshold = tolerance + edge_softness for y in range(cleaned.height): for x in range(cleaned.width): red, green, blue, alpha = pixels[x, y] distance = math.sqrt( (red - chroma_color[0]) ** 2 + (green - chroma_color[1]) ** 2 + (blue - chroma_color[2]) ** 2 ) if distance <= tolerance: pixels[x, y] = (0, 0, 0, 0) removed_chroma_pixels += 1 continue if edge_softness and distance <= threshold: visibility = (distance - tolerance) / edge_softness next_alpha = min(alpha, max(0, round(255 * visibility))) pixels[x, y] = (*_neutralize_fringe((red, green, blue), chroma_color, visibility), next_alpha) fringe_pixels += 1 cleaned, removed_components, removed_component_pixels = remove_small_alpha_components( cleaned, min_area=min_component_area, ) return cleaned, { "removed_chroma_pixels": removed_chroma_pixels, "fringe_pixels": fringe_pixels, "removed_components": removed_components, "removed_component_pixels": removed_component_pixels, "removed_pixels": removed_chroma_pixels + removed_component_pixels, } def atomic_save(image: Image.Image, destination: Path) -> None: destination.parent.mkdir(parents=True, exist_ok=True) handle = tempfile.NamedTemporaryFile( prefix=f".{destination.stem}.", suffix=destination.suffix, dir=destination.parent, delete=False, ) temporary_path = Path(handle.name) handle.close() try: image.save(temporary_path, format="PNG") os.replace(temporary_path, destination) finally: temporary_path.unlink(missing_ok=True) def clean_image( input_path: str | Path, output_path: str | Path, chroma_color: str = "#00FFFF", tolerance: int = 32, edge_softness: int = 16, min_component_area: int = 8, ) -> dict[str, Any]: input_path = Path(input_path) output_path = Path(output_path) if not input_path.is_file(): raise FileNotFoundError(f"input image not found: {input_path}") if output_path.suffix.lower() != ".png": raise ValueError("output must use the .png extension") if paths_refer_to_same_file(input_path, output_path): raise ValueError("input and output must be different files") if output_path.exists(): raise FileExistsError(f"refusing to overwrite existing output: {output_path}") chroma_rgb = parse_color(chroma_color) with Image.open(input_path) as opened: image = opened.convert("RGBA") cleaned, report = clean_rgba( image, chroma_rgb, tolerance=tolerance, edge_softness=edge_softness, min_component_area=min_component_area, ) atomic_save(cleaned, output_path) return { "ok": True, "input_file": input_path.name, "output_file": output_path.name, "output_mode": "RGBA", "chroma_color": chroma_color.upper(), "tolerance": tolerance, "edge_softness": edge_softness, "min_component_area": min_component_area, **report, } def build_parser() -> argparse.ArgumentParser: parser = argparse.ArgumentParser(description=__doc__) parser.add_argument("--input", required=True, type=Path) parser.add_argument("--output", required=True, type=Path) parser.add_argument("--color", default="#00FFFF", help="Configured chroma color") parser.add_argument("--tolerance", type=int, default=32) parser.add_argument("--edge-softness", type=int, default=16) parser.add_argument("--min-component-area", type=int, default=8) return parser def main() -> int: args = build_parser().parse_args() try: report = clean_image( args.input, args.output, chroma_color=args.color, tolerance=args.tolerance, edge_softness=args.edge_softness, min_component_area=args.min_component_area, ) except FileNotFoundError as error: code = "input_not_found" if str(error).startswith("input image not found") else "file_not_found" report = {"ok": False, "error": code, "detail": str(error)} except FileExistsError as error: report = {"ok": False, "error": "output_exists", "detail": str(error)} except ValueError as error: report = {"ok": False, "error": "invalid_input", "detail": str(error)} except Exception as error: # pragma: no cover - final machine-readable safety net report = {"ok": False, "error": "cleanup_exception", "detail": f"{type(error).__name__}: {error}"} print(json.dumps(report, ensure_ascii=False, sort_keys=True)) return 0 if report["ok"] else 1 if __name__ == "__main__": raise SystemExit(main())scripts/validate_editorial.py8.1 KBac65123c47efupstream โ
#!/usr/bin/env python3 """Validate a deterministic photo-abstract editorial composition.""" from __future__ import annotations import argparse import hashlib import json from pathlib import Path from typing import Any from PIL import Image, ImageColor, ImageOps EXPECTED_MANIFEST_VERSION = 1 KNOWN_LAYOUT_PROFILES = { "lower-editorial", "wide-horizon", "vertical-monument", "centered-archive", "sparse-object", } def sha256_file(path: Path) -> str: digest = hashlib.sha256() with path.open("rb") as handle: for chunk in iter(lambda: handle.read(1024 * 1024), b""): digest.update(chunk) return digest.hexdigest() def image_pixel_sha256(image: Image.Image) -> str: normalized = image.convert("RGB") digest = hashlib.sha256() digest.update(f"{normalized.width}x{normalized.height}:RGB\n".encode("ascii")) digest.update(normalized.tobytes()) return digest.hexdigest() def _load_manifest(path: Path) -> dict[str, Any]: if not path.is_file(): raise FileNotFoundError(f"manifest not found: {path}") payload = json.loads(path.read_text(encoding="utf-8")) if not isinstance(payload, dict): raise ValueError("manifest root must be an object") return payload def _expected_source_image(source_path: Path, manifest: dict[str, Any]) -> Image.Image: with Image.open(source_path) as opened: source = ImageOps.exif_transpose(opened).convert("RGB") target_width = manifest.get("options", {}).get("target_width") if target_width is not None: if not isinstance(target_width, int) or target_width < 1: raise ValueError("manifest target_width is invalid") target_height = max(1, round(source.height * target_width / source.width)) if (target_width, target_height) != source.size: source = source.resize((target_width, target_height), Image.Resampling.LANCZOS) return source def _region_inside_panel( region: Any, panel_y: int | None, panel_bottom: int | None, output_width: int, output_height: int, ) -> bool: if not isinstance(region, dict) or not isinstance(panel_y, int) or not isinstance(panel_bottom, int): return False values = (region.get("x"), region.get("y"), region.get("width"), region.get("height")) if not all(isinstance(value, int) for value in values): return False x, y, width, height = values return ( x >= 0 and y >= panel_y and width > 0 and height > 0 and x + width <= output_width and y + height <= min(panel_bottom, output_height) ) def validate( source_path: str | Path, output_path: str | Path, manifest_path: str | Path, ) -> dict[str, Any]: source_path = Path(source_path) output_path = Path(output_path) manifest_path = Path(manifest_path) if not source_path.is_file(): raise FileNotFoundError(f"source image not found: {source_path}") if not output_path.is_file(): raise FileNotFoundError(f"output image not found: {output_path}") manifest = _load_manifest(manifest_path) checks: dict[str, bool] = {} errors: list[str] = [] def record(name: str, passed: bool) -> None: checks[name] = bool(passed) if not passed: errors.append(name) record("manifest_version", manifest.get("manifest_version") == EXPECTED_MANIFEST_VERSION) record("source_sha256", manifest.get("source", {}).get("sha256") == sha256_file(source_path)) record("output_sha256", manifest.get("output", {}).get("sha256") == sha256_file(output_path)) declared_layout = manifest.get("layout_profile") or manifest.get("layout_grid", {}).get("profile") record("layout_profile", declared_layout is None or declared_layout in KNOWN_LAYOUT_PROFILES) expected_source = _expected_source_image(source_path, manifest) with Image.open(output_path) as opened: output = opened.convert("RGB") declared_output_size = manifest.get("output", {}).get("size") record("output_size", declared_output_size == [output.width, output.height]) photo = manifest.get("photo_region", {}) photo_geometry_ok = ( photo.get("x") == 0 and photo.get("y") == 0 and photo.get("width") == expected_source.width and photo.get("height") == expected_source.height and output.width == expected_source.width and output.height >= expected_source.height ) record("photo_region_geometry", photo_geometry_ok) if photo_geometry_ok: actual_photo = output.crop((0, 0, expected_source.width, expected_source.height)) record("photo_region_pixels", actual_photo.tobytes() == expected_source.tobytes()) record( "photo_region_hash", photo.get("pixel_sha256") == image_pixel_sha256(expected_source), ) else: record("photo_region_pixels", False) record("photo_region_hash", False) panel = manifest.get("panel", {}) panel_y = panel.get("y") panel_height = panel.get("height") panel_geometry_ok = ( isinstance(panel_y, int) and isinstance(panel_height, int) and panel_y == expected_source.height and panel_height > 0 and panel_y + panel_height == output.height ) record("panel_geometry", panel_geometry_ok) if panel_geometry_ok: try: panel_color = ImageColor.getrgb(str(panel.get("color"))) except ValueError: panel_color = None corner_pixels = [ output.getpixel((0, panel_y)), output.getpixel((output.width - 1, panel_y)), output.getpixel((0, output.height - 1)), output.getpixel((output.width - 1, output.height - 1)), ] record("panel_corner_color", panel_color is not None and corner_pixels == [panel_color] * 4) else: record("panel_corner_color", False) panel_bottom = panel_y + panel_height if isinstance(panel_y, int) and isinstance(panel_height, int) else None record( "motif_region_geometry", "motif_region" not in manifest or _region_inside_panel( manifest.get("motif_region"), panel_y if isinstance(panel_y, int) else None, panel_bottom, output.width, output.height, ), ) typography = manifest.get("typography", {}) record( "title_region_geometry", "title_region" not in typography or _region_inside_panel( typography.get("title_region"), panel_y if isinstance(panel_y, int) else None, panel_bottom, output.width, output.height, ), ) subtitle_region = typography.get("subtitle_region") record( "subtitle_region_geometry", subtitle_region is None or _region_inside_panel( subtitle_region, panel_y if isinstance(panel_y, int) else None, panel_bottom, output.width, output.height, ), ) return { "ok": not errors, "checks": checks, "errors": errors, "source_file": source_path.name, "output_file": output_path.name, "manifest_file": manifest_path.name, } def build_parser() -> argparse.ArgumentParser: parser = argparse.ArgumentParser(description=__doc__) parser.add_argument("--source", required=True, type=Path) parser.add_argument("--output", required=True, type=Path) parser.add_argument("--manifest", required=True, type=Path) return parser def main() -> int: args = build_parser().parse_args() try: verdict = validate(args.source, args.output, args.manifest) except Exception as error: verdict = { "ok": False, "checks": {}, "errors": ["validation_exception"], "detail": f"{type(error).__name__}: {error}", "source_file": args.source.name, "output_file": args.output.name, "manifest_file": args.manifest.name, } print(json.dumps(verdict, ensure_ascii=False, sort_keys=True)) return 0 if verdict["ok"] else 1 if __name__ == "__main__": raise SystemExit(main())references/art-direction.md2.3 KB73f1863dc18eupstream โ
# Photo Abstract Editorial Art Direction This reference defines the domain language shared by all execution modes. The supplied photograph is the sole factual source. Preserve relationships over surface detail and make every important mark traceable to a visible fact. ## Core principles - Identify three to six decisive facts: scale, position, axes, movement, intervals, overlap, depth, asymmetry, light/dark hierarchy, color roles, and negative space. - The motif should read as sparse abstraction before it recalls the source. Use one primary mark family and at most two supporting families. - Prefer flat or organic color masses, soft irregular circles, tapered strokes, short bars/bands, simplified architectural masses, structural axes, restrained rhythm, and negative-space cuts. - Preserve source-derived direction, scale, rhythm, center of gravity, and asymmetry. Do not invent symmetry, generic icons, surface ornament, or literal tracing. - Use a photo-derived muted palette: one dominant role, one dark structural role, one light/neutral role, and at most one or two small accents. - Keep a full readable photograph joined directly to a uniform warm-ivory panel. No frame, tape, shadow, torn edge, collage effect, gradient, texture, grain, or seam. - Use generous whitespace and a restrained editorial hierarchy. The panel contains one exact English title and only an optional exact subtitle when it adds a distinct idea. - Render typography locally and exactly when Strict Mode is available. Use native whole-run kerning, a high-contrast serif, and at most one smaller italic connector. ## Hard prohibitions No invented people, buildings, objects, geography, text, logos, watermarks, swatches, legends, dates, place labels, unsupported decoration, aggressive crop, outpainting, complete illustration, or recognizable vector tracing. Decisive vertical structures may become one primary and one subordinate vertical anchor, unequal in height and width, only when the source supports them. They remain subordinate to the main rhythm and omit crossarms, wires, latticework, and complete infrastructure outlines. The final work is accepted only when the source facts, abstraction level, spatial relationships, typography, panel, and editorial coherence pass the structured quality check. See the modular Scene, Control, Layout, Quality, and Series references for their specialized rules.references/control-system.md2.0 KB2cda23acdb6bupstream โ
# Creative Control System Use four internal control axes. Every value is clamped to `0โ100`; an explicit user value overrides a Scene Profile default. | Axis | Meaning | General default | |---|---|---:| | Abstraction | Distance from recognizable depiction toward relationships, rhythm, intervals, color roles, and negative space. `0` is highly concrete; `100` is strongly relational. | 60 | | Creative Freedom | Design latitude that remains inside the source facts. Low values stay conservative; high values allow bolder offsets, mark character, and regrouping without inventing content. | 35 | | Identity Preservation | Amount of identity cue retained for a meaningful person, landmark, or object. High values keep decisive silhouette, proportion, or cue; low values are suitable for anonymous rhythm or light. | Scene-dependent | | Spatial Fidelity | Strictness of left/right, upper/lower, scale, intervals, overlap, orientation, axes, gravity, perspective, and spatial rhythm. | 85 | ## Natural-language mapping Apply the following deltas to the active values, then clamp each result: | User wording | Change | |---|---| | โๆดๆฝ่ฑกไธ็นโ / โmore abstractโ | Abstraction `+15` | | โ็จๅพฎๆฝ่ฑกไธ็นโ / โslightly more abstractโ | Abstraction `+8` | | โๆดๅๅฎโ / โmore literalโ | Abstraction `-15` | | โๅๆฅ่ฟๅๅพไธ็นโ / โcloser to the sourceโ | Abstraction `-10`, Spatial Fidelity `+5` | | โๅคง่ไธ็นโ / โbolderโ | Creative Freedom `+15` | | โๆดๅ ๅถโ / โmore restrainedโ | Creative Freedom `-15` | | โๆดๅฟ ไบๅๅพโ / โmore faithfulโ | Spatial Fidelity `+10`, Creative Freedom `-10` | | โไบบ็ฉ่ฆๆๆพไธไบโ / โmake the person clearerโ | Identity Preservation `+20` | | โๅฐๆ ่ฆ็ๅพๅบๆฅโ / โkeep the landmark legibleโ | Identity Preservation `+15` | Do not force ordinary users to configure axes. Resolve Scene Profile defaults first, apply natural-language deltas second, and record the resolved values in the working decision or manifest when local composition is used.references/example-pair.md1.6 KBd35e40caaaa2upstream โ
# Validated Example Pair ## Source  This evaluation source was losslessly extracted from the photographic region of a legacy repository composite. It is included to demonstrate the strict workflow, not as a camera-original provenance claim. ## Result  The built-in image-generation path produced only the text-free band motif: a muted old-gold and oxblood palette, retained negative gaps, and two unequal vertical anchors derived from decisive structures in the source. The taller left anchor carries identity while the thinner right anchor acts as a quiet echo; both remain subordinate to the long bands and omit crossarms, wires, and literal infrastructure detail. `compose_editorial.py` removed isolated alpha components, placed the full source photograph, a uniform `#E8E1D5` panel, the transparent motif, and the exact title `Emberline at Dusk`. Bodoni whole-run shaping supplies natural kerning, while the single connector `at` uses a smaller matching italic and the title sits on an optical rather than mathematically equal margin. `validate_editorial.py` confirmed the source and output hashes, output geometry, panel corners, and exact equality of every decoded pixel in the photographic region. Do not reuse this subject, palette, title, or layout for another photograph. A new result must derive its decisions from its own source image.references/layout-profiles.md1.7 KBdb380548ff50upstream โ
# Layout Profiles Choose a profile from scene facts, source orientation, dominant axis, subject location, visual center of gravity, and negative space. Selection is deterministic; never choose randomly. If no decisive trigger exists, use **Lower Editorial**. | Canonical profile | Use when | Motif language | Typography | |---|---|---|---| | `lower-editorial` | Generic landscape, general street, or an ordinary scene without a stronger trigger | Lower-left optical grid, restrained horizontal relationship | Lower-left optical title | | `wide-horizon` | Sea, lake, road, long horizon, long horizontal architecture, or strong horizontal rhythm | Broad lateral motif, restrained vertical height, generous side spacing | Lower editorial title block | | `vertical-monument` | Pure Portrait, tower, tall building, or a decisive vertical subject/axis | Narrower vertical cue with editorial whitespace; never a poster icon | Calm centered title block | | `centered-archive` | Centered subject, near symmetry, monumental composition, or calm archival feeling | Centered but optically imperfect motif | Centered title block | | `sparse-object` | Still life, one object, minimal object scene, or generous negative space | Smaller motif, fewer marks, more silence | Restrained title with open margins | The current `lower-left` compositor value remains a backward-compatible alias for `lower-editorial`. The current `bottom-center` value remains accepted as a legacy centered alias. New callers may use the canonical names or `auto` with scene facts. The compositor maps each profile to explicit, bounded geometry. Motif and title must remain inside the panel, and omitted layout input must not alter the established Lower Editorial result.references/quality-check.md1.9 KB5872ea5597e1upstream โ
# Structured Visual Quality Check Run this check after composition and after any targeted correction. Record one status per dimension: `PASS`, `SOFT FAIL`, or `HARD FAIL`. | Dimension | PASS means | |---|---| | Source Fidelity | The source is the sole factual source; Strict Mode machine validation passes and no photo-region change is present. | | Abstraction Match | The motif is abstract before recollection and matches the resolved Abstraction/Freedom controls. | | Subject Identity | Required portrait, landmark, or object cues remain at the resolved Identity Preservation level. | | Spatial Traceability | Important positions, scale, overlap, axes, intervals, and gravity remain traceable to the source. | | Unsupported Content | No invented person, building, object, text, logo, watermark, swatch, or decorative fact appears. | | Composition | Photo, flat panel, motif, whitespace, and title form a coherent editorial hierarchy; layout geometry is legal. | | Typography | The exact title is present, locally rendered in Strict Mode, legible, restrained, and fully inside the panel. | | Editorial Coherence | Mark families, palette, whitespace, panel language, and tone feel intentional and consistent. | ## Failure policy `HARD FAIL` blocks delivery. In Strict Mode, any machine validator failure is a Source Fidelity `HARD FAIL`. A `SOFT FAIL` may be corrected only when a targeted change is clear and source-traceable. Apply at most one targeted correction after the first complete attempt. Re-run machine validation and the full visual checklist after that correction. If a critical `HARD FAIL` remains on the second attempt, stop and report the failure; never present it as a successful result. Native Image Edit and Reference Generation may use the same checklist, but must disclose that they cannot guarantee pixel-exact source preservation, deterministic photo preservation, or machine-verified Strict Mode output.references/scene-profiles.md2.3 KB4a52985005a2upstream โ
# Scene Profiles Classify the supplied photograph before choosing motif behavior. These are decision aids, not literal drawing recipes. Always preserve the supplied photograph as the sole factual source. | Profile | Abstraction | Creative Freedom | Identity Preservation | Spatial Fidelity | |---|---:|---:|---:|---:| | Pure Portrait | 30 | 25 | 90 | 80 | | Environmental Portrait | 45 | 35 | 75 | 85 | | Landscape | 65 | 50 | 30 | 90 | | Architecture | 55 | 35 | 65 | 95 | | Street / Crowd | 70 | 55 | 25 | 80 | | Still Life | 50 | 40 | 70 | 80 | | Minimal / Light | 80 | 60 | 20 | 70 | ## Profile behavior ### Pure Portrait Keep pose direction, head/shoulder relationship, body gravity, silhouette direction, clothing color role, subject/background proportion, light direction, and main negative space. Keep identity cues without facial reconstruction, realistic portrait illustration, literal vector tracing, or invented clothing details. Never reduce the principal subject to an anonymous short vertical mark. ### Environmental Portrait Balance person identity with environment structure, scale, body direction, horizon/architecture/ground relationship, clothing color role, and lighting relationship. The person and environment must remain traceable as a relationship. ### Landscape Prioritize horizon, depth, water, mountain, road, large color masses, light/dark hierarchy, dominant axis, atmospheric space, and spatial rhythm. Use abstraction to reduce surface detail, not to invent geography. ### Architecture Retain one to three identity cues such as a taper, arch, spire, roof/eave, characteristic negative space, or dominant proportion. Omit detailed window grids, facade tracing, and blueprint-like literalism. ### Street / Crowd Anonymous crowd and background people may become irregular short marks, restrained vertical ink marks, or rhythm clusters. This rule does not apply to a Pure Portrait or Environmental Portrait principal subject. ### Still Life Keep object scale, planar cues, relative position, dominant contour direction, object/background relationship, and material color role. Do not simulate realistic material or volume. ### Minimal / Light Prioritize shadow, light direction, small structural cues, negative space, color temperature, and minimal geometry. Keep the result quiet and avoid adding objects.references/series-style.md1.0 KB892d0cf0edbfupstream โ
# Optional Series Style Lock Only load this reference for multi-image series, an editorial set, or an explicit request for consistent visual style. Do not load it for an ordinary single-image task. ## First image After the first image passes QA, form an internal Style Profile containing: - abstraction baseline and Creative Freedom baseline; - typography character and panel language; - whitespace character and motif density; - overall editorial tone; - preferred layout-family tendency. ## Subsequent images Keep the overall abstraction language, Creative Freedom baseline, typography character, panel language, whitespace character, motif density, and editorial tone. Re-analyze every new source for scene, palette, decisive facts, identity, spatial relationships, motif geometry, and suitable layout. The lock defines a visual family, not a duplicated template. Do not copy exact motif coordinates, exact palette, exact layout, or exact spatial structure. A new photograph must remain the sole factual source for its own motif and composition.assets/evals/v3.0-rc1/controls/abstraction-30/motif-clean.png1.9 MBe4e92f9cd78bupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/controls/abstraction-30/motif-generated.png2.0 MBb22bae5feb1fupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/controls/abstraction-30/result.png2.4 MB38f7f99ced7eupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/controls/abstraction-30/result.png.manifest.json2.8 KBd053e4d2af86upstream โ
{ "layout_grid": { "inset": 123, "inset_ratio": 0.08, "profile": "wide-horizon" }, "layout_profile": "wide-horizon", "manifest_version": 1, "motif": { "cleanup": { "initial_min_component_area": 79, "removed_components": 19, "removed_pixels": 4369, "resized_min_component_area": 753 }, "file_name": "motif-clean.png", "sha256": "92f2419feeedf19f8b70770b2cf98d4fad9355eefd1703baf2598bb831292bae" }, "motif_region": { "height": 311, "width": 968, "x": 284, "y": 1170 }, "options": { "accent_font_path": null, "accent_scale": 0.7, "dominant_axis": "horizontal", "font_path": null, "layout": "wide-horizon", "margin_ratio": 0.07, "motif_center_x": null, "motif_top_ratio": 0.16, "motif_width_ratio": 0.54, "negative_space": "balanced", "panel_color": "#E8E1D5", "panel_height": null, "scene_profile": "landscape", "source_orientation": "landscape-wide", "subject_location": "left", "subtitle": null, "subtitle_color": null, "subtitle_font_path": null, "target_width": null, "title": "Marsh Study", "title_accent": null, "title_bottom_inset_ratio": 0.115, "title_color": "#34343A", "title_optical_offset_ratio": 0.012, "title_size_ratio": 0.041 }, "output": { "file_name": "result.png", "sha256": "d477a1f63f41e53d9d64f0ed8eed312241814d9e23d2759501fb4ced51b03406", "size": [ 1537, 1841 ] }, "panel": { "color": "#E8E1D5", "height": 818, "y": 1023 }, "photo_region": { "height": 1023, "pixel_exact": true, "pixel_sha256": "a2aa87818740e3bb93146982292cbf18aeb30dfeac6384f3d2eb80e122dd6220", "width": 1537, "x": 0, "y": 0 }, "source": { "file_name": "source.png", "rendered_pixel_sha256": "a2aa87818740e3bb93146982292cbf18aeb30dfeac6384f3d2eb80e122dd6220", "sha256": "888df6685fc45fa53c4ed04a5f9d820a2956ef6b2043161b44ca1a3ae2ea50ff", "visual_size": [ 1537, 1023 ] }, "typography": { "accent_font_file": null, "bottom_inset": 169, "font_file": "BOD_R.TTF", "kerning_mode": "whole-run", "layout": "wide-horizon", "layout_profile": "wide-horizon", "subtitle": null, "subtitle_color": "#34343A", "subtitle_font_file": null, "subtitle_font_size": null, "subtitle_region": null, "title": "Marsh Study", "title_accent": null, "title_anchor_x": 141, "title_color": "#34343A", "title_font_size": 63, "title_optical_offset": 18, "title_region": { "height": 58, "width": 341, "x": 141, "y": 1614 }, "title_runs": [ { "font_file": "BOD_R.TTF", "font_size": 63, "style": "primary", "text": "Marsh Study" } ] } }assets/evals/v3.0-rc1/controls/abstraction-60/motif-clean.png2.6 MBb1f11d49180fupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/controls/abstraction-60/motif-generated.png2.6 MB5d7a7a928198upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/controls/abstraction-60/result.png2.2 MB3b204c3656d2upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/controls/abstraction-60/result.png.manifest.json2.8 KBb2355b598647upstream โ
{ "layout_grid": { "inset": 123, "inset_ratio": 0.08, "profile": "wide-horizon" }, "layout_profile": "wide-horizon", "manifest_version": 1, "motif": { "cleanup": { "initial_min_component_area": 79, "removed_components": 27, "removed_pixels": 5911, "resized_min_component_area": 512 }, "file_name": "motif-clean.png", "sha256": "5d129db598465a12a513fbcad66de61ba357f9118e5748bee06a87c82c35b081" }, "motif_region": { "height": 311, "width": 658, "x": 440, "y": 1170 }, "options": { "accent_font_path": null, "accent_scale": 0.7, "dominant_axis": "horizontal", "font_path": null, "layout": "wide-horizon", "margin_ratio": 0.07, "motif_center_x": null, "motif_top_ratio": 0.16, "motif_width_ratio": 0.54, "negative_space": "balanced", "panel_color": "#E8E1D5", "panel_height": null, "scene_profile": "landscape", "source_orientation": "landscape-wide", "subject_location": "left", "subtitle": null, "subtitle_color": null, "subtitle_font_path": null, "target_width": null, "title": "Marsh Study", "title_accent": null, "title_bottom_inset_ratio": 0.115, "title_color": "#34343A", "title_optical_offset_ratio": 0.012, "title_size_ratio": 0.041 }, "output": { "file_name": "result.png", "sha256": "c3958199875eda96384db62becc16b1866926b3601ac9498c16d379821aa5452", "size": [ 1537, 1841 ] }, "panel": { "color": "#E8E1D5", "height": 818, "y": 1023 }, "photo_region": { "height": 1023, "pixel_exact": true, "pixel_sha256": "a2aa87818740e3bb93146982292cbf18aeb30dfeac6384f3d2eb80e122dd6220", "width": 1537, "x": 0, "y": 0 }, "source": { "file_name": "source.png", "rendered_pixel_sha256": "a2aa87818740e3bb93146982292cbf18aeb30dfeac6384f3d2eb80e122dd6220", "sha256": "888df6685fc45fa53c4ed04a5f9d820a2956ef6b2043161b44ca1a3ae2ea50ff", "visual_size": [ 1537, 1023 ] }, "typography": { "accent_font_file": null, "bottom_inset": 169, "font_file": "BOD_R.TTF", "kerning_mode": "whole-run", "layout": "wide-horizon", "layout_profile": "wide-horizon", "subtitle": null, "subtitle_color": "#34343A", "subtitle_font_file": null, "subtitle_font_size": null, "subtitle_region": null, "title": "Marsh Study", "title_accent": null, "title_anchor_x": 141, "title_color": "#34343A", "title_font_size": 63, "title_optical_offset": 18, "title_region": { "height": 58, "width": 341, "x": 141, "y": 1614 }, "title_runs": [ { "font_file": "BOD_R.TTF", "font_size": 63, "style": "primary", "text": "Marsh Study" } ] } }assets/evals/v3.0-rc1/controls/abstraction-80/motif-clean.png1.4 MB35f4a2ee0618upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/controls/abstraction-80/motif-generated.png1.5 MBd0e210ed3ff2upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/controls/abstraction-80/result.png2.0 MBc2940646d003upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/controls/abstraction-80/result.png.manifest.json2.8 KBa415cb4f68eeupstream โ
{ "layout_grid": { "inset": 123, "inset_ratio": 0.08, "profile": "wide-horizon" }, "layout_profile": "wide-horizon", "manifest_version": 1, "motif": { "cleanup": { "initial_min_component_area": 79, "removed_components": 148, "removed_pixels": 10059, "resized_min_component_area": 322 }, "file_name": "motif-clean.png", "sha256": "f17bf3ab50836b3fd8f9f4e8c014a0be40f34f7e377682d2e7fa56498dea66f5" }, "motif_region": { "height": 311, "width": 414, "x": 562, "y": 1170 }, "options": { "accent_font_path": null, "accent_scale": 0.7, "dominant_axis": "horizontal", "font_path": null, "layout": "wide-horizon", "margin_ratio": 0.07, "motif_center_x": null, "motif_top_ratio": 0.16, "motif_width_ratio": 0.54, "negative_space": "balanced", "panel_color": "#E8E1D5", "panel_height": null, "scene_profile": "landscape", "source_orientation": "landscape-wide", "subject_location": "left", "subtitle": null, "subtitle_color": null, "subtitle_font_path": null, "target_width": null, "title": "Marsh Study", "title_accent": null, "title_bottom_inset_ratio": 0.115, "title_color": "#34343A", "title_optical_offset_ratio": 0.012, "title_size_ratio": 0.041 }, "output": { "file_name": "result.png", "sha256": "492c13f53735912ff8990b8f9606d3d96f77066cc7895f04b031a8d89e6ebd25", "size": [ 1537, 1841 ] }, "panel": { "color": "#E8E1D5", "height": 818, "y": 1023 }, "photo_region": { "height": 1023, "pixel_exact": true, "pixel_sha256": "a2aa87818740e3bb93146982292cbf18aeb30dfeac6384f3d2eb80e122dd6220", "width": 1537, "x": 0, "y": 0 }, "source": { "file_name": "source.png", "rendered_pixel_sha256": "a2aa87818740e3bb93146982292cbf18aeb30dfeac6384f3d2eb80e122dd6220", "sha256": "888df6685fc45fa53c4ed04a5f9d820a2956ef6b2043161b44ca1a3ae2ea50ff", "visual_size": [ 1537, 1023 ] }, "typography": { "accent_font_file": null, "bottom_inset": 169, "font_file": "BOD_R.TTF", "kerning_mode": "whole-run", "layout": "wide-horizon", "layout_profile": "wide-horizon", "subtitle": null, "subtitle_color": "#34343A", "subtitle_font_file": null, "subtitle_font_size": null, "subtitle_region": null, "title": "Marsh Study", "title_accent": null, "title_anchor_x": 141, "title_color": "#34343A", "title_font_size": 63, "title_optical_offset": 18, "title_region": { "height": 58, "width": 341, "x": 141, "y": 1614 }, "title_runs": [ { "font_file": "BOD_R.TTF", "font_size": 63, "style": "primary", "text": "Marsh Study" } ] } }assets/evals/v3.0-rc1/controls/freedom-high/motif-clean.png2.7 MB5af02b17d86cupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/controls/freedom-high/motif-generated.png2.7 MB4fa9043f2c40upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/controls/freedom-high/result.png2.2 MB657b69297820upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/controls/freedom-high/result.png.manifest.json2.8 KB053e8deb0895upstream โ
{ "layout_grid": { "inset": 123, "inset_ratio": 0.08, "profile": "wide-horizon" }, "layout_profile": "wide-horizon", "manifest_version": 1, "motif": { "cleanup": { "initial_min_component_area": 79, "removed_components": 18, "removed_pixels": 5578, "resized_min_component_area": 435 }, "file_name": "motif-clean.png", "sha256": "52878ddf1492b4469e69acd8c8a4a5da1354ad9523926cb4cb2ba2469b673a20" }, "motif_region": { "height": 311, "width": 560, "x": 488, "y": 1170 }, "options": { "accent_font_path": null, "accent_scale": 0.7, "dominant_axis": "horizontal", "font_path": null, "layout": "wide-horizon", "margin_ratio": 0.07, "motif_center_x": null, "motif_top_ratio": 0.16, "motif_width_ratio": 0.54, "negative_space": "balanced", "panel_color": "#E8E1D5", "panel_height": null, "scene_profile": "landscape", "source_orientation": "landscape-wide", "subject_location": "left", "subtitle": null, "subtitle_color": null, "subtitle_font_path": null, "target_width": null, "title": "Marsh Study", "title_accent": null, "title_bottom_inset_ratio": 0.115, "title_color": "#34343A", "title_optical_offset_ratio": 0.012, "title_size_ratio": 0.041 }, "output": { "file_name": "result.png", "sha256": "dd86a8af53c50b461dc6580539d9cf41fabf80ac941c2015546d3ba02a9928ce", "size": [ 1537, 1841 ] }, "panel": { "color": "#E8E1D5", "height": 818, "y": 1023 }, "photo_region": { "height": 1023, "pixel_exact": true, "pixel_sha256": "a2aa87818740e3bb93146982292cbf18aeb30dfeac6384f3d2eb80e122dd6220", "width": 1537, "x": 0, "y": 0 }, "source": { "file_name": "source.png", "rendered_pixel_sha256": "a2aa87818740e3bb93146982292cbf18aeb30dfeac6384f3d2eb80e122dd6220", "sha256": "888df6685fc45fa53c4ed04a5f9d820a2956ef6b2043161b44ca1a3ae2ea50ff", "visual_size": [ 1537, 1023 ] }, "typography": { "accent_font_file": null, "bottom_inset": 169, "font_file": "BOD_R.TTF", "kerning_mode": "whole-run", "layout": "wide-horizon", "layout_profile": "wide-horizon", "subtitle": null, "subtitle_color": "#34343A", "subtitle_font_file": null, "subtitle_font_size": null, "subtitle_region": null, "title": "Marsh Study", "title_accent": null, "title_anchor_x": 141, "title_color": "#34343A", "title_font_size": 63, "title_optical_offset": 18, "title_region": { "height": 58, "width": 341, "x": 141, "y": 1614 }, "title_runs": [ { "font_file": "BOD_R.TTF", "font_size": 63, "style": "primary", "text": "Marsh Study" } ] } }assets/evals/v3.0-rc1/controls/freedom-low/motif-clean.png2.6 MBf78ce8ae6cf8upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/controls/freedom-low/motif-generated.png2.7 MB691a1a6cd425upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/controls/freedom-low/result.png2.2 MBaf6558dfe054upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/controls/freedom-low/result.png.manifest.json2.8 KBaf71e6305c51upstream โ
{ "layout_grid": { "inset": 123, "inset_ratio": 0.08, "profile": "wide-horizon" }, "layout_profile": "wide-horizon", "manifest_version": 1, "motif": { "cleanup": { "initial_min_component_area": 79, "removed_components": 2, "removed_pixels": 3695, "resized_min_component_area": 438 }, "file_name": "motif-clean.png", "sha256": "602b47db44af488bd2643b20620044072982742c2cc1d1f9e3cc81ed8d2fb6f8" }, "motif_region": { "height": 311, "width": 563, "x": 487, "y": 1170 }, "options": { "accent_font_path": null, "accent_scale": 0.7, "dominant_axis": "horizontal", "font_path": null, "layout": "wide-horizon", "margin_ratio": 0.07, "motif_center_x": null, "motif_top_ratio": 0.16, "motif_width_ratio": 0.54, "negative_space": "balanced", "panel_color": "#E8E1D5", "panel_height": null, "scene_profile": "landscape", "source_orientation": "landscape-wide", "subject_location": "left", "subtitle": null, "subtitle_color": null, "subtitle_font_path": null, "target_width": null, "title": "Marsh Study", "title_accent": null, "title_bottom_inset_ratio": 0.115, "title_color": "#34343A", "title_optical_offset_ratio": 0.012, "title_size_ratio": 0.041 }, "output": { "file_name": "result.png", "sha256": "a00bd73289c3f2fdccb587d67f62bfd9e468a1e6f2d94c32e22023c99d3d059c", "size": [ 1537, 1841 ] }, "panel": { "color": "#E8E1D5", "height": 818, "y": 1023 }, "photo_region": { "height": 1023, "pixel_exact": true, "pixel_sha256": "a2aa87818740e3bb93146982292cbf18aeb30dfeac6384f3d2eb80e122dd6220", "width": 1537, "x": 0, "y": 0 }, "source": { "file_name": "source.png", "rendered_pixel_sha256": "a2aa87818740e3bb93146982292cbf18aeb30dfeac6384f3d2eb80e122dd6220", "sha256": "888df6685fc45fa53c4ed04a5f9d820a2956ef6b2043161b44ca1a3ae2ea50ff", "visual_size": [ 1537, 1023 ] }, "typography": { "accent_font_file": null, "bottom_inset": 169, "font_file": "BOD_R.TTF", "kerning_mode": "whole-run", "layout": "wide-horizon", "layout_profile": "wide-horizon", "subtitle": null, "subtitle_color": "#34343A", "subtitle_font_file": null, "subtitle_font_size": null, "subtitle_region": null, "title": "Marsh Study", "title_accent": null, "title_anchor_x": 141, "title_color": "#34343A", "title_font_size": 63, "title_optical_offset": 18, "title_region": { "height": 58, "width": 341, "x": 141, "y": 1614 }, "title_runs": [ { "font_file": "BOD_R.TTF", "font_size": 63, "style": "primary", "text": "Marsh Study" } ] } }assets/evals/v3.0-rc1/E01-pure-portrait/motif-clean.png1.0 MBddeddf097c77upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E01-pure-portrait/motif-generated.png1.1 MBcf9057675ed8upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E01-pure-portrait/result.png1.8 MB35bd9ac19033upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E01-pure-portrait/result.png.manifest.json3.0 KB6ea1c04fc578upstream โ
{ "layout_grid": { "inset": 82, "inset_ratio": 0.08, "profile": "vertical-monument" }, "layout_profile": "vertical-monument", "manifest_version": 1, "motif": { "cleanup": { "initial_min_component_area": 79, "removed_components": 105, "removed_pixels": 5585, "resized_min_component_area": 151 }, "file_name": "motif-clean.png", "sha256": "493751d4d66bb5d770c931e5fa078071ecab2361f4c12621e0770a1292a41cea" }, "motif_region": { "height": 321, "width": 188, "x": 418, "y": 1637 }, "options": { "accent_font_path": null, "accent_scale": 0.7, "dominant_axis": "vertical", "font_path": null, "layout": "auto", "margin_ratio": 0.07, "motif_center_x": null, "motif_top_ratio": 0.16, "motif_width_ratio": 0.54, "negative_space": "generous", "panel_color": "#E8E1D5", "panel_height": null, "scene_profile": "pure-portrait", "source_orientation": "portrait", "subject_location": "right", "subtitle": null, "subtitle_color": null, "subtitle_font_path": null, "target_width": null, "title": "Held in Quiet", "title_accent": "in", "title_bottom_inset_ratio": 0.115, "title_color": "#34343A", "title_optical_offset_ratio": 0.012, "title_size_ratio": 0.041 }, "output": { "file_name": "result.png", "sha256": "816b9685e386690a4e07d1eb3a777b449c12f2c9d68d37c2068ce89766981b7b", "size": [ 1024, 2381 ] }, "panel": { "color": "#E8E1D5", "height": 845, "y": 1536 }, "photo_region": { "height": 1536, "pixel_exact": true, "pixel_sha256": "64d87e27ed0c676ef83c9cc03e1775a88320c15fcee10ee781fdab8dc53aff65", "width": 1024, "x": 0, "y": 0 }, "source": { "file_name": "source.png", "rendered_pixel_sha256": "64d87e27ed0c676ef83c9cc03e1775a88320c15fcee10ee781fdab8dc53aff65", "sha256": "51011f4183414b5f494e4a5ef7530faaf51a44430ccfbefdb9962c602891d1f4", "visual_size": [ 1024, 1536 ] }, "typography": { "accent_font_file": "BOD_I.TTF", "bottom_inset": 113, "font_file": "BOD_R.TTF", "kerning_mode": "whole-run", "layout": "auto", "layout_profile": "vertical-monument", "subtitle": null, "subtitle_color": "#34343A", "subtitle_font_file": null, "subtitle_font_size": null, "subtitle_region": null, "title": "Held in Quiet", "title_accent": "in", "title_anchor_x": 398, "title_color": "#34343A", "title_font_size": 42, "title_optical_offset": 0, "title_region": { "height": 38, "width": 228, "x": 398, "y": 2230 }, "title_runs": [ { "font_file": "BOD_R.TTF", "font_size": 42, "style": "primary", "text": "Held " }, { "font_file": "BOD_I.TTF", "font_size": 29, "style": "accent", "text": "in" }, { "font_file": "BOD_R.TTF", "font_size": 42, "style": "primary", "text": " Quiet" } ] } }assets/evals/v3.0-rc1/E01-pure-portrait/source.png1.9 MBd6c161be3f14upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E02-environmental-portrait/motif-clean.png955 KB377798fbb559upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E02-environmental-portrait/motif-generated.png1.0 MBc68b4fe2bb77upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E02-environmental-portrait/result.png2.0 MB64e54e95d46fupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E02-environmental-portrait/result.png.manifest.json3.1 KBb4b2a4a7af16upstream โ
{ "layout_grid": { "inset": 123, "inset_ratio": 0.08, "profile": "vertical-monument" }, "layout_profile": "vertical-monument", "manifest_version": 1, "motif": { "cleanup": { "initial_min_component_area": 79, "removed_components": 48, "removed_pixels": 4234, "resized_min_component_area": 299 }, "file_name": "motif-clean.png", "sha256": "7eb283afab63b402dd6206475a8104e140455cb085236bfa667ee9a8ac10b145" }, "motif_region": { "height": 311, "width": 384, "x": 576, "y": 1122 }, "options": { "accent_font_path": null, "accent_scale": 0.7, "dominant_axis": "horizontal", "font_path": null, "layout": "auto", "margin_ratio": 0.07, "motif_center_x": null, "motif_top_ratio": 0.16, "motif_width_ratio": 0.54, "negative_space": "balanced", "panel_color": "#E8E1D5", "panel_height": null, "scene_profile": "environmental-portrait", "source_orientation": "landscape", "subject_location": "left", "subtitle": null, "subtitle_color": null, "subtitle_font_path": null, "target_width": null, "title": "Across the Breakwater", "title_accent": "the", "title_bottom_inset_ratio": 0.115, "title_color": "#34343A", "title_optical_offset_ratio": 0.012, "title_size_ratio": 0.041 }, "output": { "file_name": "result.png", "sha256": "dd52f01b995b62e7d6d3864fe9c71ab9cfda51518b5c3b8b27d08df2cb9fc636", "size": [ 1536, 1843 ] }, "panel": { "color": "#E8E1D5", "height": 819, "y": 1024 }, "photo_region": { "height": 1024, "pixel_exact": true, "pixel_sha256": "a7bff1b9a8dcdb6a2bc0eaafc54cc154c3c8d3f1fad4eb3c64af91fd9f0a1c55", "width": 1536, "x": 0, "y": 0 }, "source": { "file_name": "source.png", "rendered_pixel_sha256": "a7bff1b9a8dcdb6a2bc0eaafc54cc154c3c8d3f1fad4eb3c64af91fd9f0a1c55", "sha256": "10349e818bb41ed44c2e9a71416c2d8825e78af268f7b9502f86870909afcd9d", "visual_size": [ 1536, 1024 ] }, "typography": { "accent_font_file": "BOD_I.TTF", "bottom_inset": 169, "font_file": "BOD_R.TTF", "kerning_mode": "whole-run", "layout": "auto", "layout_profile": "vertical-monument", "subtitle": null, "subtitle_color": "#34343A", "subtitle_font_file": null, "subtitle_font_size": null, "subtitle_region": null, "title": "Across the Breakwater", "title_accent": "the", "title_anchor_x": 485, "title_color": "#34343A", "title_font_size": 63, "title_optical_offset": 0, "title_region": { "height": 43, "width": 566, "x": 485, "y": 1631 }, "title_runs": [ { "font_file": "BOD_R.TTF", "font_size": 63, "style": "primary", "text": "Across " }, { "font_file": "BOD_I.TTF", "font_size": 44, "style": "accent", "text": "the" }, { "font_file": "BOD_R.TTF", "font_size": 63, "style": "primary", "text": " Breakwater" } ] } }assets/evals/v3.0-rc1/E02-environmental-portrait/source.png2.1 MB33e9b370e58aupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E03-landscape/motif-clean.png730 KB32e998fd0177upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E03-landscape/motif-generated.png764 KB5c7742a54ef8upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E03-landscape/result-attempt-1.png2.0 MBeb57631e0c22upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E03-landscape/result-attempt-1.png.manifest.json3.0 KBcc34b6a980e8upstream โ
{ "layout_grid": { "inset": 154, "inset_ratio": 0.1, "profile": "sparse-object" }, "layout_profile": "sparse-object", "manifest_version": 1, "motif": { "cleanup": { "initial_min_component_area": 79, "removed_components": 12, "removed_pixels": 4748, "resized_min_component_area": 242 }, "file_name": "motif-clean.png", "sha256": "de09edd049fbbedcbc7b17a581b34f1ce82c1085a370af787bd8547cc583ef8f" }, "motif_region": { "height": 210, "width": 461, "x": 538, "y": 1203 }, "options": { "accent_font_path": null, "accent_scale": 0.7, "dominant_axis": "horizontal", "font_path": null, "layout": "auto", "margin_ratio": 0.07, "motif_center_x": null, "motif_top_ratio": 0.16, "motif_width_ratio": 0.54, "negative_space": "generous", "panel_color": "#E8E1D5", "panel_height": null, "scene_profile": "landscape", "source_orientation": "landscape-wide", "subject_location": "left", "subtitle": null, "subtitle_color": null, "subtitle_font_path": null, "target_width": null, "title": "Where the Marsh Opens", "title_accent": "the", "title_bottom_inset_ratio": 0.115, "title_color": "#34343A", "title_optical_offset_ratio": 0.012, "title_size_ratio": 0.041 }, "output": { "file_name": "result.png", "sha256": "ea9538907fe043e1400fb74ac853771254fa89ba7d9af50777cc7ec96dac09ca", "size": [ 1537, 1841 ] }, "panel": { "color": "#E8E1D5", "height": 818, "y": 1023 }, "photo_region": { "height": 1023, "pixel_exact": true, "pixel_sha256": "a2aa87818740e3bb93146982292cbf18aeb30dfeac6384f3d2eb80e122dd6220", "width": 1537, "x": 0, "y": 0 }, "source": { "file_name": "source.png", "rendered_pixel_sha256": "a2aa87818740e3bb93146982292cbf18aeb30dfeac6384f3d2eb80e122dd6220", "sha256": "888df6685fc45fa53c4ed04a5f9d820a2956ef6b2043161b44ca1a3ae2ea50ff", "visual_size": [ 1537, 1023 ] }, "typography": { "accent_font_file": "BOD_I.TTF", "bottom_inset": 184, "font_file": "BOD_R.TTF", "kerning_mode": "whole-run", "layout": "auto", "layout_profile": "sparse-object", "subtitle": null, "subtitle_color": "#34343A", "subtitle_font_file": null, "subtitle_font_size": null, "subtitle_region": null, "title": "Where the Marsh Opens", "title_accent": "the", "title_anchor_x": 172, "title_color": "#34343A", "title_font_size": 63, "title_optical_offset": 18, "title_region": { "height": 58, "width": 601, "x": 172, "y": 1599 }, "title_runs": [ { "font_file": "BOD_R.TTF", "font_size": 63, "style": "primary", "text": "Where " }, { "font_file": "BOD_I.TTF", "font_size": 44, "style": "accent", "text": "the" }, { "font_file": "BOD_R.TTF", "font_size": 63, "style": "primary", "text": " Marsh Opens" } ] } }assets/evals/v3.0-rc1/E03-landscape/result.png2.1 MB713314f25708upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E03-landscape/result.png.manifest.json3.0 KBe561b1b60d10upstream โ
{ "layout_grid": { "inset": 123, "inset_ratio": 0.08, "profile": "wide-horizon" }, "layout_profile": "wide-horizon", "manifest_version": 1, "motif": { "cleanup": { "initial_min_component_area": 79, "removed_components": 17, "removed_pixels": 8089, "resized_min_component_area": 532 }, "file_name": "motif-clean.png", "sha256": "de09edd049fbbedcbc7b17a581b34f1ce82c1085a370af787bd8547cc583ef8f" }, "motif_region": { "height": 311, "width": 684, "x": 426, "y": 1170 }, "options": { "accent_font_path": null, "accent_scale": 0.7, "dominant_axis": "horizontal", "font_path": null, "layout": "auto", "margin_ratio": 0.07, "motif_center_x": null, "motif_top_ratio": 0.16, "motif_width_ratio": 0.54, "negative_space": "generous", "panel_color": "#E8E1D5", "panel_height": null, "scene_profile": "landscape", "source_orientation": "landscape-wide", "subject_location": "left", "subtitle": null, "subtitle_color": null, "subtitle_font_path": null, "target_width": null, "title": "Where the Marsh Opens", "title_accent": "the", "title_bottom_inset_ratio": 0.115, "title_color": "#34343A", "title_optical_offset_ratio": 0.012, "title_size_ratio": 0.041 }, "output": { "file_name": "result.png", "sha256": "2f7f0c91534b600b2ec97198f2eeeaea878948f3f20de1761a51489d8c08552a", "size": [ 1537, 1841 ] }, "panel": { "color": "#E8E1D5", "height": 818, "y": 1023 }, "photo_region": { "height": 1023, "pixel_exact": true, "pixel_sha256": "a2aa87818740e3bb93146982292cbf18aeb30dfeac6384f3d2eb80e122dd6220", "width": 1537, "x": 0, "y": 0 }, "source": { "file_name": "source.png", "rendered_pixel_sha256": "a2aa87818740e3bb93146982292cbf18aeb30dfeac6384f3d2eb80e122dd6220", "sha256": "888df6685fc45fa53c4ed04a5f9d820a2956ef6b2043161b44ca1a3ae2ea50ff", "visual_size": [ 1537, 1023 ] }, "typography": { "accent_font_file": "BOD_I.TTF", "bottom_inset": 169, "font_file": "BOD_R.TTF", "kerning_mode": "whole-run", "layout": "auto", "layout_profile": "wide-horizon", "subtitle": null, "subtitle_color": "#34343A", "subtitle_font_file": null, "subtitle_font_size": null, "subtitle_region": null, "title": "Where the Marsh Opens", "title_accent": "the", "title_anchor_x": 141, "title_color": "#34343A", "title_font_size": 63, "title_optical_offset": 18, "title_region": { "height": 58, "width": 601, "x": 141, "y": 1614 }, "title_runs": [ { "font_file": "BOD_R.TTF", "font_size": 63, "style": "primary", "text": "Where " }, { "font_file": "BOD_I.TTF", "font_size": 44, "style": "accent", "text": "the" }, { "font_file": "BOD_R.TTF", "font_size": 63, "style": "primary", "text": " Marsh Opens" } ] } }assets/evals/v3.0-rc1/E03-landscape/source.png1.9 MBe6b973982a9fupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E04-architecture/motif-clean.png938 KB56d12e0e1778upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E04-architecture/motif-generated.png962 KB37c63f9adcd0upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E04-architecture/result.png2.1 MB04c732a82d4eupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E04-architecture/result.png.manifest.json3.0 KBcd14220355e3upstream โ
{ "layout_grid": { "inset": 82, "inset_ratio": 0.08, "profile": "centered-archive" }, "layout_profile": "centered-archive", "manifest_version": 1, "motif": { "cleanup": { "initial_min_component_area": 79, "removed_components": 0, "removed_pixels": 1566, "resized_min_component_area": 121 }, "file_name": "motif-clean.png", "sha256": "67b1903fac6da5adeb42531adb883902d5a464dc0db47eed37f0f245c0111fbc" }, "motif_region": { "height": 321, "width": 151, "x": 436, "y": 1671 }, "options": { "accent_font_path": null, "accent_scale": 0.7, "dominant_axis": "vertical", "font_path": null, "layout": "auto", "margin_ratio": 0.07, "motif_center_x": null, "motif_top_ratio": 0.16, "motif_width_ratio": 0.54, "negative_space": "balanced", "panel_color": "#E8E1D5", "panel_height": null, "scene_profile": "architecture", "source_orientation": "portrait", "subject_location": "center", "subtitle": null, "subtitle_color": null, "subtitle_font_path": null, "target_width": null, "title": "Stone Over Light", "title_accent": "Over", "title_bottom_inset_ratio": 0.115, "title_color": "#34343A", "title_optical_offset_ratio": 0.012, "title_size_ratio": 0.041 }, "output": { "file_name": "result.png", "sha256": "3b025e98c2e39caf5481ab0431862eee3e4e2a1a034e04ab97d676c23141215d", "size": [ 1024, 2381 ] }, "panel": { "color": "#E8E1D5", "height": 845, "y": 1536 }, "photo_region": { "height": 1536, "pixel_exact": true, "pixel_sha256": "373aab1e027a10cc5f92ff3dbe5c0c7bf2aaf014e692843f816255cc5ed679be", "width": 1024, "x": 0, "y": 0 }, "source": { "file_name": "source.png", "rendered_pixel_sha256": "373aab1e027a10cc5f92ff3dbe5c0c7bf2aaf014e692843f816255cc5ed679be", "sha256": "e566cfe2f5f29e357434557ec36e33e48a88a3a291e7b4f7f03509fa2fc9a6fc", "visual_size": [ 1024, 1536 ] }, "typography": { "accent_font_file": "BOD_I.TTF", "bottom_inset": 123, "font_file": "BOD_R.TTF", "kerning_mode": "whole-run", "layout": "auto", "layout_profile": "centered-archive", "subtitle": null, "subtitle_color": "#34343A", "subtitle_font_file": null, "subtitle_font_size": null, "subtitle_region": null, "title": "Stone Over Light", "title_accent": "Over", "title_anchor_x": 376, "title_color": "#34343A", "title_font_size": 42, "title_optical_offset": 0, "title_region": { "height": 39, "width": 271, "x": 376, "y": 2219 }, "title_runs": [ { "font_file": "BOD_R.TTF", "font_size": 42, "style": "primary", "text": "Stone " }, { "font_file": "BOD_I.TTF", "font_size": 29, "style": "accent", "text": "Over" }, { "font_file": "BOD_R.TTF", "font_size": 42, "style": "primary", "text": " Light" } ] } }assets/evals/v3.0-rc1/E04-architecture/source.png2.1 MBef39a8a47713upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E05-street-crowd/motif-clean.png1.5 MBb499ec41c9dbupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E05-street-crowd/motif-generated.png1.6 MBf934cbc44b9eupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E05-street-crowd/result.png2.5 MB72ed5862e732upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E05-street-crowd/result.png.manifest.json3.0 KB70bf9653ece8upstream โ
{ "layout_grid": { "inset": 108, "inset_ratio": 0.07, "profile": "lower-editorial" }, "layout_profile": "lower-editorial", "manifest_version": 1, "motif": { "cleanup": { "initial_min_component_area": 79, "removed_components": 155, "removed_pixels": 14230, "resized_min_component_area": 379 }, "file_name": "motif-clean.png", "sha256": "ed34e8de48276fde691174ccf389f53c1794f56c1aefc00368ded5d34bdbffa5" }, "motif_region": { "height": 311, "width": 488, "x": 108, "y": 1155 }, "options": { "accent_font_path": null, "accent_scale": 0.7, "dominant_axis": "vertical", "font_path": null, "layout": "auto", "margin_ratio": 0.07, "motif_center_x": null, "motif_top_ratio": 0.16, "motif_width_ratio": 0.54, "negative_space": "balanced", "panel_color": "#E8E1D5", "panel_height": null, "scene_profile": "street-crowd", "source_orientation": "landscape", "subject_location": "center", "subtitle": null, "subtitle_color": null, "subtitle_font_path": null, "target_width": null, "title": "After the Rain", "title_accent": "the", "title_bottom_inset_ratio": 0.115, "title_color": "#34343A", "title_optical_offset_ratio": 0.012, "title_size_ratio": 0.041 }, "output": { "file_name": "result.png", "sha256": "1cb844fb5437a60245e111e3b85736a597492e87cd35a12fc162fd40d7bbbfd1", "size": [ 1536, 1843 ] }, "panel": { "color": "#E8E1D5", "height": 819, "y": 1024 }, "photo_region": { "height": 1024, "pixel_exact": true, "pixel_sha256": "4f002751d59c22a6c2d4c621c1806696736fb738e4b4feef2af434303d6b5304", "width": 1536, "x": 0, "y": 0 }, "source": { "file_name": "source.png", "rendered_pixel_sha256": "4f002751d59c22a6c2d4c621c1806696736fb738e4b4feef2af434303d6b5304", "sha256": "f2461086a619d730a4a8e1a00f7a7b52f947388d982514f7af4eeaf4941e01cc", "visual_size": [ 1536, 1024 ] }, "typography": { "accent_font_file": "BOD_I.TTF", "bottom_inset": 177, "font_file": "BOD_R.TTF", "kerning_mode": "whole-run", "layout": "auto", "layout_profile": "lower-editorial", "subtitle": null, "subtitle_color": "#34343A", "subtitle_font_file": null, "subtitle_font_size": null, "subtitle_region": null, "title": "After the Rain", "title_accent": "the", "title_anchor_x": 126, "title_color": "#34343A", "title_font_size": 63, "title_optical_offset": 18, "title_region": { "height": 43, "width": 351, "x": 126, "y": 1623 }, "title_runs": [ { "font_file": "BOD_R.TTF", "font_size": 63, "style": "primary", "text": "After " }, { "font_file": "BOD_I.TTF", "font_size": 44, "style": "accent", "text": "the" }, { "font_file": "BOD_R.TTF", "font_size": 63, "style": "primary", "text": " Rain" } ] } }assets/evals/v3.0-rc1/E05-street-crowd/source.png2.5 MB8c4506a4b909upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E06-still-life/motif-clean.png1.2 MB0ab444c37f71upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E06-still-life/motif-generated.png1.2 MB1593cd4b8da0upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E06-still-life/result.png1.8 MBc5c5950df50aupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E06-still-life/result.png.manifest.json3.0 KB096619146df9upstream โ
{ "layout_grid": { "inset": 125, "inset_ratio": 0.1, "profile": "sparse-object" }, "layout_profile": "sparse-object", "manifest_version": 1, "motif": { "cleanup": { "initial_min_component_area": 79, "removed_components": 0, "removed_pixels": 1997, "resized_min_component_area": 215 }, "file_name": "motif-clean.png", "sha256": "1ca7d05830b6e92ca17f8fab3475db0bef6e0315f393a84d08b7dc03dbbfef48" }, "motif_region": { "height": 229, "width": 376, "x": 439, "y": 1447 }, "options": { "accent_font_path": null, "accent_scale": 0.7, "dominant_axis": "horizontal", "font_path": null, "layout": "auto", "margin_ratio": 0.07, "motif_center_x": null, "motif_top_ratio": 0.16, "motif_width_ratio": 0.54, "negative_space": "generous", "panel_color": "#E8E1D5", "panel_height": null, "scene_profile": "still-life", "source_orientation": "square", "subject_location": "center", "subtitle": null, "subtitle_color": null, "subtitle_font_path": null, "target_width": null, "title": "Weight of Yellow", "title_accent": "of", "title_bottom_inset_ratio": 0.115, "title_color": "#34343A", "title_optical_offset_ratio": 0.012, "title_size_ratio": 0.041 }, "output": { "file_name": "result.png", "sha256": "4c3d109d4acbbd6e789ee6dd404c197775e3a16b266de14de682f4dc6a8af03b", "size": [ 1254, 2132 ] }, "panel": { "color": "#E8E1D5", "height": 878, "y": 1254 }, "photo_region": { "height": 1254, "pixel_exact": true, "pixel_sha256": "9a2a5194c569903ac292a19211f08a04028eae77d0018c4c73219b3c7ac806d6", "width": 1254, "x": 0, "y": 0 }, "source": { "file_name": "source.png", "rendered_pixel_sha256": "9a2a5194c569903ac292a19211f08a04028eae77d0018c4c73219b3c7ac806d6", "sha256": "2eb8a2e68c96ee72abb7b4879330ca009b219e49db5b3b1c36eb85376139ed6c", "visual_size": [ 1254, 1254 ] }, "typography": { "accent_font_file": "BOD_I.TTF", "bottom_inset": 150, "font_file": "BOD_R.TTF", "kerning_mode": "whole-run", "layout": "auto", "layout_profile": "sparse-object", "subtitle": null, "subtitle_color": "#34343A", "subtitle_font_file": null, "subtitle_font_size": null, "subtitle_region": null, "title": "Weight of Yellow", "title_accent": "of", "title_anchor_x": 140, "title_color": "#34343A", "title_font_size": 51, "title_optical_offset": 15, "title_region": { "height": 47, "width": 360, "x": 140, "y": 1935 }, "title_runs": [ { "font_file": "BOD_R.TTF", "font_size": 51, "style": "primary", "text": "Weight " }, { "font_file": "BOD_I.TTF", "font_size": 36, "style": "accent", "text": "of" }, { "font_file": "BOD_R.TTF", "font_size": 51, "style": "primary", "text": " Yellow" } ] } }assets/evals/v3.0-rc1/E06-still-life/source.png1.7 MB89c06eac23b9upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E07-minimal-light/motif-clean.png1.3 MB65acc7eb90dcupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E07-minimal-light/motif-generated.png1.3 MBfec6eddf067bupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E07-minimal-light/result.png1.8 MB27eaa9a3e6dcupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/evals/v3.0-rc1/E07-minimal-light/result.png.manifest.json3.0 KBb88d6ab774bfupstream โ
{ "layout_grid": { "inset": 154, "inset_ratio": 0.1, "profile": "sparse-object" }, "layout_profile": "sparse-object", "manifest_version": 1, "motif": { "cleanup": { "initial_min_component_area": 79, "removed_components": 0, "removed_pixels": 1867, "resized_min_component_area": 240 }, "file_name": "motif-clean.png", "sha256": "e448c52f370611e3123c8746b8aecd35f62066ec5b54995d2475bc1f7f4ba4ad" }, "motif_region": { "height": 311, "width": 309, "x": 614, "y": 1203 }, "options": { "accent_font_path": null, "accent_scale": 0.7, "dominant_axis": "diagonal", "font_path": null, "layout": "auto", "margin_ratio": 0.07, "motif_center_x": null, "motif_top_ratio": 0.16, "motif_width_ratio": 0.54, "negative_space": "large", "panel_color": "#E8E1D5", "panel_height": null, "scene_profile": "minimal-light", "source_orientation": "landscape", "subject_location": "right", "subtitle": null, "subtitle_color": null, "subtitle_font_path": null, "target_width": null, "title": "A Room for Light", "title_accent": "for", "title_bottom_inset_ratio": 0.115, "title_color": "#34343A", "title_optical_offset_ratio": 0.012, "title_size_ratio": 0.041 }, "output": { "file_name": "result.png", "sha256": "60b1e5f6df82f35a812662822f350e4f82dbaff11b8dd5291fb9825fbaf59d2f", "size": [ 1537, 1841 ] }, "panel": { "color": "#E8E1D5", "height": 818, "y": 1023 }, "photo_region": { "height": 1023, "pixel_exact": true, "pixel_sha256": "b4e9ef0fbbd0738202362ae9543a55009f21b8be04d136d51b9da92ba7eef0de", "width": 1537, "x": 0, "y": 0 }, "source": { "file_name": "source.png", "rendered_pixel_sha256": "b4e9ef0fbbd0738202362ae9543a55009f21b8be04d136d51b9da92ba7eef0de", "sha256": "85e9df9137f2c6ecf61fd238fcb64a85ea7cbcfb49791a3814f257d83772bf30", "visual_size": [ 1537, 1023 ] }, "typography": { "accent_font_file": "BOD_I.TTF", "bottom_inset": 184, "font_file": "BOD_R.TTF", "kerning_mode": "whole-run", "layout": "auto", "layout_profile": "sparse-object", "subtitle": null, "subtitle_color": "#34343A", "subtitle_font_file": null, "subtitle_font_size": null, "subtitle_region": null, "title": "A Room for Light", "title_accent": "for", "title_anchor_x": 172, "title_color": "#34343A", "title_font_size": 63, "title_optical_offset": 18, "title_region": { "height": 58, "width": 448, "x": 172, "y": 1599 }, "title_runs": [ { "font_file": "BOD_R.TTF", "font_size": 63, "style": "primary", "text": "A Room " }, { "font_file": "BOD_I.TTF", "font_size": 44, "style": "accent", "text": "for" }, { "font_file": "BOD_R.TTF", "font_size": 63, "style": "primary", "text": " Light" } ] } }assets/evals/v3.0-rc1/E07-minimal-light/source.png1.7 MBed3d4664fb27upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/examples/result-horizon.png332 KBe0a6e97cf0aaupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/examples/result-horizon.png.manifest.json2.8 KB92af7e22e057upstream โ
{ "layout_grid": { "inset": 72, "inset_ratio": 0.07 }, "manifest_version": 1, "motif": { "cleanup": { "initial_min_component_area": 79, "removed_components": 0, "removed_pixels": 5356, "resized_min_component_area": 138 }, "file_name": "motif-horizon-v4.png", "sha256": "6278db77d2db6800aac7aac61bcfb93f33f1ae6affa80917bcbec945d3a5d40e" }, "motif_region": { "height": 100, "width": 552, "x": 72, "y": 812 }, "options": { "accent_font_path": null, "accent_scale": 0.7, "font_path": null, "layout": "lower-left", "margin_ratio": 0.07, "motif_center_x": null, "motif_top_ratio": 0.16, "motif_width_ratio": 0.54, "panel_color": "#E8E1D5", "panel_height": null, "subtitle": null, "subtitle_color": null, "subtitle_font_path": null, "target_width": null, "title": "Emberline at Dusk", "title_accent": "at", "title_bottom_inset_ratio": 0.115, "title_color": "#4B3738", "title_optical_offset_ratio": 0.012, "title_size_ratio": 0.041 }, "output": { "file_name": "result-horizon.png", "sha256": "d4d7cd0d765c618bcac854768d566a7fa1780f3bcaa698c5f58366c6665a3e98", "size": [ 1023, 1296 ] }, "panel": { "color": "#E8E1D5", "height": 576, "y": 720 }, "photo_region": { "height": 720, "pixel_exact": true, "pixel_sha256": "69242c2e79928565d4da512222ee0f3da5f15b5242050151abb4be570f533856", "width": 1023, "x": 0, "y": 0 }, "source": { "file_name": "source-horizon.png", "rendered_pixel_sha256": "69242c2e79928565d4da512222ee0f3da5f15b5242050151abb4be570f533856", "sha256": "aea04d5b443a2c03de5e7842354ba76333c6ff94ec77421210ba93dee8d04a5e", "visual_size": [ 1023, 720 ] }, "typography": { "accent_font_file": "BOD_I.TTF", "bottom_inset": 118, "font_file": "BOD_R.TTF", "kerning_mode": "whole-run", "layout": "lower-left", "subtitle": null, "subtitle_color": "#4B3738", "subtitle_font_file": null, "subtitle_font_size": null, "subtitle_region": null, "title": "Emberline at Dusk", "title_accent": "at", "title_anchor_x": 84, "title_color": "#4B3738", "title_font_size": 42, "title_optical_offset": 12, "title_region": { "height": 29, "width": 318, "x": 84, "y": 1149 }, "title_runs": [ { "font_file": "BOD_R.TTF", "font_size": 42, "style": "primary", "text": "Emberline " }, { "font_file": "BOD_I.TTF", "font_size": 29, "style": "accent", "text": "at" }, { "font_file": "BOD_R.TTF", "font_size": 42, "style": "primary", "text": " Dusk" } ] } }assets/examples/source-horizon.png298 KBc7948a7195b9upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/readme/comparisons/original-horizon/motif-clean.png462 KB674ca1bf827dupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/readme/comparisons/original-horizon/motif-generated.png478 KBc849c586858cupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/readme/comparisons/original-horizon/original-result.png332 KBe0a6e97cf0aaupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/readme/comparisons/original-horizon/source.png298 KBc7948a7195b9upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/readme/comparisons/original-horizon/v3-result.png374 KBbcf86123870bupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/readme/comparisons/original-horizon/v3-result.png.manifest.json3.0 KBea300c7e3cc0upstream โ
{ "layout_grid": { "inset": 82, "inset_ratio": 0.08, "profile": "wide-horizon" }, "layout_profile": "wide-horizon", "manifest_version": 1, "motif": { "cleanup": { "initial_min_component_area": 79, "removed_components": 13, "removed_pixels": 3891, "resized_min_component_area": 214 }, "file_name": "motif-clean.png", "sha256": "e575b66cd1f5205ba8c6d045f25f831f4874a1330bb7cb64622729f5446183e3" }, "motif_region": { "height": 127, "width": 675, "x": 174, "y": 824 }, "options": { "accent_font_path": null, "accent_scale": 0.7, "dominant_axis": "horizontal", "font_path": null, "layout": "auto", "margin_ratio": 0.07, "motif_center_x": null, "motif_top_ratio": 0.16, "motif_width_ratio": 0.54, "negative_space": "generous", "panel_color": "#E8E1D5", "panel_height": null, "scene_profile": "landscape", "source_orientation": "landscape-wide", "subject_location": "center", "subtitle": null, "subtitle_color": null, "subtitle_font_path": null, "target_width": null, "title": "Emberline at Dusk", "title_accent": "at", "title_bottom_inset_ratio": 0.115, "title_color": "#34343A", "title_optical_offset_ratio": 0.012, "title_size_ratio": 0.041 }, "output": { "file_name": "v3-result.png", "sha256": "083237dfef0e4e1751ccd75c32f75c259f005e82b849a86f583800ae6d12ca4e", "size": [ 1023, 1296 ] }, "panel": { "color": "#E8E1D5", "height": 576, "y": 720 }, "photo_region": { "height": 720, "pixel_exact": true, "pixel_sha256": "69242c2e79928565d4da512222ee0f3da5f15b5242050151abb4be570f533856", "width": 1023, "x": 0, "y": 0 }, "source": { "file_name": "source.png", "rendered_pixel_sha256": "69242c2e79928565d4da512222ee0f3da5f15b5242050151abb4be570f533856", "sha256": "aea04d5b443a2c03de5e7842354ba76333c6ff94ec77421210ba93dee8d04a5e", "visual_size": [ 1023, 720 ] }, "typography": { "accent_font_file": "BOD_I.TTF", "bottom_inset": 113, "font_file": "BOD_R.TTF", "kerning_mode": "whole-run", "layout": "auto", "layout_profile": "wide-horizon", "subtitle": null, "subtitle_color": "#34343A", "subtitle_font_file": null, "subtitle_font_size": null, "subtitle_region": null, "title": "Emberline at Dusk", "title_accent": "at", "title_anchor_x": 94, "title_color": "#34343A", "title_font_size": 42, "title_optical_offset": 12, "title_region": { "height": 29, "width": 318, "x": 94, "y": 1154 }, "title_runs": [ { "font_file": "BOD_R.TTF", "font_size": 42, "style": "primary", "text": "Emberline " }, { "font_file": "BOD_I.TTF", "font_size": 29, "style": "accent", "text": "at" }, { "font_file": "BOD_R.TTF", "font_size": 42, "style": "primary", "text": " Dusk" } ] } }assets/readme/controls/abstraction-30.png2.4 MB38f7f99ced7eupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/readme/controls/abstraction-60.png2.2 MB3b204c3656d2upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/readme/controls/abstraction-80.png2.0 MBc2940646d003upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/readme/hero/v3-original-source.png374 KBbcf86123870bupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/readme/original/result-horizon.png332 KBe0a6e97cf0aaupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/readme/original/source-horizon.png298 KBc7948a7195b9upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/readme/series/landscape.png2.1 MB713314f25708upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/readme/series/minimal-light.png1.8 MB27eaa9a3e6dcupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/readme/series/street-crowd.png2.5 MB72ed5862e732upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/readme/v3/architecture.png2.1 MB04c732a82d4eupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/readme/v3/landscape.png2.1 MB713314f25708upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/readme/v3/minimal-light.png1.8 MB27eaa9a3e6dcupstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/readme/v3/pure-portrait.png1.8 MB35bd9ac19033upstream โ
Contents are not inlined in the catalog. Open the upstream file.
assets/readme/v3/still-life.png1.8 MBc5c5950df50aupstream โ
Contents are not inlined in the catalog. Open the upstream file.
.gitattributes73 B28ecd91899d9upstream โ
* text=auto eol=lf *.jpg binary *.jpeg binary *.png binary *.zip binary.github/dependabot.yml261 B7bbb62192f4aupstream โ
version: 2 updates: - package-ecosystem: pip directory: "/" schedule: interval: weekly open-pull-requests-limit: 5 - package-ecosystem: github-actions directory: "/" schedule: interval: weekly open-pull-requests-limit: 5.github/workflows/portability-pilot.yml5.1 KB35898f5278dcupstream โ
Contents are not inlined in the catalog. Open the upstream file.
.github/workflows/python-tests.yml5.9 KBfbe01fbdf338upstream โ
Contents are not inlined in the catalog. Open the upstream file.
.gitignore190 Bb3821a13ee24upstream โ
work/ docs/superpowers/ __pycache__/ *.pyc *.pyo .pytest_cache/ .mypy_cache/ .venv/ venv/ coverage/ htmlcov/ .coverage *.coverage tmp/ temp/ *.tmp *.temp .DS_Store Thumbs.db .idea/ .vscode/agents/openai.yaml291 Bf5893795aa20upstream โ
interface: display_name: "Adaptive Photo Abstract Editorial" short_description: "Create adaptive, QA-gated photo-and-abstract editorials." default_prompt: "Use $photo-abstract-editorial to turn this photograph into a source-faithful adaptive photo-and-abstract editorial composition."docs/evals/deepseek-harness-validation.md1.6 KB4e15696b4f4dupstream โ
Contents are not inlined in the catalog. Open the upstream file.
docs/evals/legacy-baseline.md1.7 KBc97f74c6f8a3upstream โ
Contents are not inlined in the catalog. Open the upstream file.
docs/evals/v3.0-rc1-real-world-evaluation.md12 KB9106a21297d2upstream โ
Contents are not inlined in the catalog. Open the upstream file.
docs/releases/original-edition.md2.9 KBf1584fb5f7cdupstream โ
Contents are not inlined in the catalog. Open the upstream file.
LICENSE34 KBbe3f7b28e564upstream โ
Contents are not inlined in the catalog. Open the upstream file.
README.zh-CN.md26 KB9b1a38838612upstream โ
Contents are not inlined in the catalog. Open the upstream file.
README.zh-TW.md26 KB8446fcdc8d4cupstream โ
Contents are not inlined in the catalog. Open the upstream file.
requirements-dev.txt161 Ba691e803e428upstream โ
# Install this dependency set for repository development and tests. # Runtime release archives exclude development-only files. -r requirements.txt pytest==9.1.1requirements.txt78 B54f89cda1cdcupstream โ
# Runtime dependencies for the local compositor and validator. Pillow==12.3.0tests/__init__.py61 B3ef253c7e365upstream โ
"""Repository test package for stable cross-test imports."""tests/test_compose_editorial.py16 KB60f07d186d1dupstream โ
Contents are not inlined in the catalog. Open the upstream file.
tests/test_environment_contract.py4.4 KB56fced0d5762upstream โ
Contents are not inlined in the catalog. Open the upstream file.
tests/test_remove_chroma_key.py3.7 KB99a9d3cd19f4upstream โ
Contents are not inlined in the catalog. Open the upstream file.
tests/test_skill_package.py11 KB0880503418b8upstream โ
Contents are not inlined in the catalog. Open the upstream file.
tests/test_validate_editorial.py6.7 KBd804f9cc1e90upstream โ
Contents are not inlined in the catalog. Open the upstream file.
tools/build_skill_package.py5.0 KB31b8ef227672upstream โ
Contents are not inlined in the catalog. Open the upstream file.
tools/check_dependency_contract.py6.8 KB4732baa688fbupstream โ
Contents are not inlined in the catalog. Open the upstream file.
tools/check_package_parity.py9.0 KBa662b8f39234upstream โ
Contents are not inlined in the catalog. Open the upstream file.
tools/check_repository_hygiene.py3.7 KBd9ea6e90f5aeupstream โ
Contents are not inlined in the catalog. Open the upstream file.
Plain-text copy, no login: /skill-bundles/kwhi6693-web-photo-abstract-editorial.md






