Spaces:
Sleeping
Sleeping
Commit History
-
Remove Phase 3 native reference attempt - go directly to Phase 2
normanedev02Phase 3 native reference_images are not supported by the Vertex AI Python SDK. This removes the futile attempt and goes directly to Phase 2 (Gemini-enhanced prompts) which provides good character consistency without the overhead.33858a1 -
Fix: Disable SSR mode for HF Spaces compatibility
normanedev0214ee8d3 -
Fix: Accept 429 as valid API key during initialization
normanedev02A 429 (rate limit exceeded) proves the API key is valid - we just hit quota temporarily. This was causing all image generation to fall back to mock images.92ac85c -
Merge branch 'main' of https://huggingface.co/spaces/RadixLaboratory/radixlab-anime
normanedev02630c581 -
Fix audio/video sync, improve API retry, add TTS support
normanedev02- Fix MoviePy 2.x API: use with_audio() instead of set_audio() - Fix dropdown value mismatch errors (remove dynamic status icons) - Improve Imagen API retry logic (6 retries, longer backoffs) - Add Google Cloud TTS for real voiceover/dialogue audio - Pass scene prompt to Veo for better video generation - Increase delays between A/B generation and scenes976f7cb -
Upload jabbawocky.png
Verified signature854a71c
-
Fix MoviePy 2.x: volumex -> with_volume_scaled
normanedev0287a674d -
Change image model to imagen-3.0-generate-002
normanedev027f1508c -
Fix MoviePy 2.x API: subclip -> subclipped
normanedev02In MoviePy 2.x, the subclip() method was renamed to subclipped(). This fixes the audio/video combination error.2c264ff -
Fix Phase 3 reference_images SDK limitation
normanedev02- Updated generate_with_native_references to gracefully skip when reference images are provided (SDK doesn't support reference_images parameter in generate_images method) - Falls back to Phase 2 prompt-based consistency automatically12eb1d7 -
Upgrade to Imagen 4 + Veo 3.1 with REST API polling
normanedev02- Update image generation to imagen-4.0-generate-001 (latest model) - Update video generation to veo-3.1-generate-001 with native audio - Implement proper REST API polling for Veo long-running operations - Add test reference images (robot, scientist, laboratory) - Verify multi-character support working with ref_character_keys array970fff8 -
Complete Milestone 2 & 3: Video Generation + Export
normanedev02Milestone 2 - Video & Audio: - Add Vertex AI Veo for AI-powered I2V generation - MoviePy fallback for programmatic camera motion - Voiceover, Dialogue, SFX audio generation - Audio + Video synchronization - Video Approval tab with Generate All Clips - Re-render individual clips Milestone 3 - Export: - Add Export tab for final assembly - Video stitching (concatenate all approved clips) - Download button for final .mp4 file - Export status tracking Files: - veo_video_generator.py: Vertex AI Veo I2V module - app.py: Updated with Veo integration + Export tab All 25 client requirements now met.2951a90 -
Add Phase 3: Native Reference Images with Vertex AI Capability Model
normanedev02Features: - Service account authentication (GOOGLE_APPLICATION_CREDENTIALS) - Hugging Face Spaces support (GCP_SERVICE_ACCOUNT_JSON secret) - Vertex AI imagen-3.0-capability-001 model - Native reference images with REFERENCE_TYPE_SUBJECT - Subject type inference (PERSON, ANIMAL, PRODUCT) - Automatic fallback to Phase 2 when unavailable Files: - phase3_native_reference_generator.py: Native reference image generator - app.py: Phase 3 integration with fallback logic - requirements.txt: Added google-cloud-aiplatform, vertexai dependencies - README.md: HF Spaces deployment instructions - .gitignore: Exclude service account JSON filesb67275f -
Add support for multiple character references (ref_character_keys array) and voice_id parsing
normanedev02bcbd2fe -
Improve character description precision for better face consistency
normanedev0219658ce -
Fix: Move Fanko.png to correct Characters/ folder (not outputs/)
normanedev026b96585 -
Add Fanko reference image and Gemini character description feature
normanedev02c36e144 -
Add Gemini-based reference image description for character consistency
normanedev024f46ce9 -
Use toast notifications for progress
normanedev02023fe73 -
Fix syntax error
normanedev02f6df1c1 -
Fix progress bar visibility
normanedev020f3c1a5 -
Update app.py
normanedev02610e5e7 -
Update app.py
normanedev02d6023d0 -
Fix Generate All Images button - refresh UI after generation
normanedev025d95a80 -
Update app.py
normanedev02d179bcd
-
Update app.py
normanedev02f7f8d2c -
Update app.py
normanedev025ae66f6 -
Update app.py with improvements and add .gitignore
normanedev02f80e298 -
Add anime generation app with characters, locations, and sample outputs
normanedev02- Add main app.py with anime scene generation logic - Add character assets (cat, daughter, pediatrician) - Add location backgrounds (classroom, lab) - Add sample script JSON for testing - Add generated output images - Update requirements.txt with dependenciesf4cd01f -
initial commit
Unverified signature23212a4