Spaces:
Sleeping
Sleeping
Commit History
-
Release v1.0.0: Official milestone with Alchemical Feedback, Thinking Model support, and Zoom-In Prompt editing.
c60a62f -
pref: Finalize Cloud deployment v0.1.5 with UI refinements
- Full engine isolation: ComfyUI & Ollama now correctly tag as '(Local Only)' when deployed on Hugging Face Spaces. - Smart engine prioritization: Hugging Face engines are now the hard-coded default when running in a Space environment. - Synthesis feedback: Integrated 'Loader2' spinners and processing states for 'Refine' and 'Synthesize' actions. - Lightbox UX: Refactored fullscreen viewer to maximize space for 9:16 portraits by moving the download button to a subtle top-right icon. - Version tracking: Bumped application version across the board to v0.1.5.ccc9c4e -
fix: Update Cloud UI constraint labels
- Changed `(Cloud Only)` to `(Local Only)` on the disabled dropdown options when the application is running in the cloud to correctly reflect their availability context.7973b76 -
fix: Overhaul dropdown logic for Cloud deployment contexts
- Exposed `.is_hf_space` state parameter from FastAPI to the frontend. - ComfyUI & Ollama options now gracefully grey out showing "(Cloud Only)" when they detect remote HF Spaces deployment constraint context instead of generic offline warnings. - Missing Token warnings on HF selection are now accurately repressed in the dropdowns if the user natively signs in through OAuth.33b76fc -
fix: Resolve HF Hub TS error and display app version
- Fixed an `@huggingface/hub` TypeScript type mismatch in `oauthHandleRedirectIfPresent` compilation. - Added `v0.1.0` dynamically below the Chronicle Portrait Studio logo for better version traceability in the UI.5ff1248 -
fix: Expose Oauth Client ID to static frontend
- Added `/api/config/oauth` GET endpoint to FastAPI to expose the Space's automatically injected `OAUTH_CLIENT_ID`. - Modified `ClientApp.tsx` to fetch the Client ID before attempting OAuth redirect validation or login URL generation. - Fixes "Missing clientId" errors thrown by `@huggingface/hub` caused by the static Next.js HTML being completely disconnected from the Docker container's environment variables.424b15d -
fix: Add OAuth scopes to README and improve Space HF Login error logging
- Added `hf_oauth_scopes: - inference-api` to the `README.md` metadata. Without this, the Hugging Face Space wrapper rejects login requests requesting inference scopes. - Improved the `catch` block in `handleHFLogin` inside `ClientApp.tsx` to display the actual underlying error to the UI status string if `oauthLoginUrl` fails.d8002cf -
chore: Restructure architecture for Hugging Face Docker Space deployment
- Updated `next.config.ts` to `output: 'export'` for highly-optimized static compilation. - Refactored `api.py` to securely mount the Next.js `out/` directory using FastAPI `StaticFiles`. - Updated `.dockerignore` to block local Node.js binaries and previous build steps, radically improving Docker context transfer times. - Wrote a new Multi-Stage `Dockerfile` that builds Next.js via Node, then serves it via Python, completely bypassing the legacy Gradio requirement. - Updated `README.md` default config from `sdk: gradio` to `sdk: docker`.8811f65 -
feat: Implement secure Hugging Face OAuth and patch Next.js vulnerability
- Removed manual HF token input in favor of official Hugging Face OAuth. - Implemented `oauthHandleRedirectIfPresent` in ClientApp state. - Refactored `api.py` request schemas and `modules/integrations.py` inference logic to support an `oauth_token`. - Added a Hybrid Token Priority system that falls back to local `.env` keys for seamless developer experience. - Designed a new "Hugging Face Identity" UI section providing status indicators (Authenticated, Developer Mode, Disconnected). - Bumbed `next` and `eslint-config-next` to `16.2.3` to patch a High Severity Denial of Service vulnerability (GHSA-q4gf-8mx6-v5v3). - Added `suppressHydrationWarning` to `layout.tsx` to fix mismatch errors caused by browser extensions during development.31bd72c
-
Phase 5.5: UI Model Selection & Enhanced Fault Diagnostics
- [UI] Converted Hugging Face model inputs to curated dropdowns with a 'Manual Entry' toggle. - [UI] Implemented high-visibility 'Synthesis Chamber Fault' error display in the central canvas. - [UI] Upgraded the action status bar to be reactive to failure states (pulsing red text). - [Backend] Restored SD-XL-base to the curated image model list in config.py. - [API] Exposed hf_text_models and hf_image_models in the config endpoint.e480cd9 -
Cleanup: remove temporary commit message file from tracking
2115291 -
feat: Add Advanced Engine Settings and native UI logo
- Implemented collapsible 'Advanced Engine Settings' allowing users to override Ollama and Hugging Face models directly from the UI. - Swapped the generic banner placeholder for the Chronicle Portrait Studio logo.svg. - Color-matched the SVG logo to the Next.js Obsidian & Gold palette (#0a0805, #DAB062).e1f9b50
-
Migrate to Next.js & FastAPI Architecture with Premium UI
- Decouples monolithic Gradio app into FastAPI backend (`api.py`) and Next.js frontend (`frontend/`). - Implements Stitch MCP-inspired 'Alchemist's Workshop' aesthetic with glassmorphism and Cinzel typography. - Replaces generic Gradio grids with precisely constrained, center-aligned HTML forms using Tailwind v4. - Preserves 100% functional parity including JSON Save/Load, Example Characters, and Image Generation. - Updates README documentation to reflect the new decoupled setup instructions.4eef090
-
fix: added some padding to textedit field so they look better in firefox.
cef8efc -
feat: refine AI backend options and UI layout
- Curated Hugging Face model lists (4 text, 4 image models). - Added manual Inference Provider support for HF backend (e.g., fal-ai). - Set default HF provider to 'auto' for robust model discovery. - Relocated 'Generate Image' button to the right column for better UX. - Improved error handling for prompt refinement to prevent UI प्रदूषण. - Added MIT License and updated README with new features and usage info.42bdcb2 -
fix: embed SVG logo in HTML to resolve Gradio TypeError and path issues
617d464 -
ui: hide interactive widgets on header logo
9d866c9 -
style: refine and center SVG logo layout
7c4f320 -
feat: add SVG logo and update UI layout
6a5d3c3 -
fix: enable hf_oauth in metadata
d8d0240 -
fix: use valid HF Space metadata color
a1165f2 -
docs: add HF Space metadata to README.md
5709e52 -
feat: integrate HF Router and enhance UI workflow
- Switched Hugging Face integration to the OpenAI-compatible Router. - Refactored `modules/integrations.py` to use `openai` SDK. - Updated `modules/config.py` with `HF_BASE_URL` and `Qwen/Qwen2.5-72B-Instruct` model. - Implemented dynamic backend availability checking for Ollama and ComfyUI at startup. - Fixed refined prompt persistence issue when loading characters. - Fixed aspect ratio handling for Hugging Face image generation. - Renamed "Settings & Generation" section to "AI Backend Configuration" and relocated "Refine Prompt" button for better workflow. - Updated `requirements.txt` with `openai` and `python-dotenv`.54d620e
-
UX Polish, Metadata Embedding & Local Name Generation
- Character Naming: Added Character Name field to Identity tab. - Procedural Naming: Integrated 'fictional-names' library for instant, thematic fantasy name generation (🎲 button). - Metadata Embedding: Prompts and character names are now embedded into PNG 'tEXt' chunks (Comment, CharacterName). - Smart Filenaming: Portraits and character data are saved using the character's name. - Clipboard: Added native 'Copy' buttons to prompt textboxes. - UI Stability: Fixed multiple runtime exceptions (NameErrors, AttributeErrors) and implemented a UI build test script. - Documentation: Updated design.md and created walkthrough.md to document new features.de132df
-
feat: stabilize Hugging Face integration and update Gemini models
- Refactored Hugging Face integration to use `huggingface_hub` SDK for improved stability. - Updated Gemini models to `gemini-3-pro-preview` (text) and `imagen-4.0-generate-001` (images). - Fixed 410/404 errors by using the correct router endpoints. - Added support for `share=True` remote access with individual token override. - Enhanced UI branding and added remote access guidance notes. - Updated `requirements.txt` with `huggingface_hub`. - Included example scripts for Hugging Face API testing.be1294b
-
feat: integrate Hugging Face Inference API for text and images
3ff0143 -
feat: add Docker configuration for Hugging Face Spaces deployment
fd1d3cf -
refactor: modularize codebase and reorganize utilities
- Split monolithic app.py into modules/config.py, modules/integrations.py, modules/core_logic.py, and modules/ui_layout.py. - Consolited ComfyUI utilities into a dedicated comfy/ directory. - Updated imports to use the new package structure. - Enhanced .gitignore to include __pycache__ and temporary character saves. - Documented new architecture and safer port management in design.md.467c85e -
style: reorganize UI into tabs and refine button hierarchy
- Implemented gr.Tabs for character attributes - Aligned randomization dice with dropdowns on the same line - Reordered output column (prompts above image) - Grouped Save/Load buttons with example characters - Moved 'Randomize' button below technical prompt - Updated design docs with process management lessonsdb5af08 -
feat: integrate Ollama for local prompt refinement
- Implemented dynamic Ollama model discovery and status checking. - Externalized LLM system prompts into prompts.yaml. - Updated Gradio UI with refinement backend selector and model dropdown. - Updated documentation (README.md, design.md) and environment examples.bd05868
-
docs: Update design and readme to reflect ComfyUI integration and .env config
0eb4e19 -
feat: ComfyUI integration, .env config system, and character load persistence fixes
439c579
-
feat: Adjust supported aspect ratios and add Alfred & Frida examples
df5838f -
feat: Add example character loader to UI
db8982c -
feat: Add character examples folder
ae55ba3 -
fix: Clear refined prompt when randomizing features
647b5c0 -
docs: Update README and design doc with Gemini and Save/Load features
1f2c958 -
feat: Integrate Gemini AI for prompt refinement and image generation
- Migrated to google-genai SDK. - Implemented prompt refinement with gemini-3-pro-preview. - Integrated image generation using imagen-4.0-generate-001. - Fixed image display by converting Gemini response to PIL Image. - Added PNG conversion and friendly filenames (rpg_portrait.png). - Expanded exotic weapon, armor, and accessory variants in features.yaml. - Added support for dual accessory selection. - Implemented "Save Character" and "Load Character" functionality (JSON). - Added detailed traceback logging for debugging. - Fixed Gradio theme and CopyButton compatibility issues.0493f14 -
Initial commit: RPGPortrait Prompt Builder
- Implemented Gradio UI with 24 character parameters. - Dynamic prompt generation with detailed mappings for advanced AI models. - Randomization feature with checkboxes and a regenerate button. - YAML-based data storage for features and templates. - Comprehensive documentation (README, design, walkthrough).5b0247a