empty
90ceb76296
feat: Optimize VLM prompt for storyboard consistency, focus on constant features
2026-01-07 09:44:27 +08:00
empty
1b54552fec
fix: Handle nested JSON structures in VLM response parsing
2026-01-07 09:40:21 +08:00
empty
f19804facb
feat: Add VLMConfig to schema for proper config.yaml VLM support
2026-01-07 09:38:45 +08:00
empty
bc077475c6
fix: Remove duplicate import os causing UnboundLocalError
2026-01-07 09:35:45 +08:00
empty
9675b9c23b
feat: Add VLM config to config.example.yaml with config.yaml support
2026-01-07 09:32:16 +08:00
empty
92183b083b
feat: Add Qwen VL support for character analysis, configurable via VLM_PROVIDER
2026-01-07 09:29:43 +08:00
empty
be216eacad
fix: Increase VLM max_tokens to 2000 to avoid response truncation
2026-01-07 03:37:55 +08:00
empty
8d82cf91d5
fix: Auto-detect and use GLM-4V vision model for character analysis
2026-01-07 03:33:56 +08:00
empty
8c35b0066f
fix: Enhance VLM response parsing to handle markdown code blocks
2026-01-07 03:31:42 +08:00
empty
b3cf9e64e5
feat: Implement Character Memory V1 - VLM analysis and prompt injection
2026-01-07 03:08:29 +08:00
empty
2be9256c48
fix: Use OpenAI multimodal message format for VLM style extraction
2026-01-07 00:22:33 +08:00
empty
297f3ccda4
feat: Enhance StyleGuard with VLM-based style extraction for specific style_prefix
2026-01-07 00:16:57 +08:00
empty
56db9bf9d2
feat: Add hybrid quality evaluation system with CLIP and VLM support
...
- Add FeatureExtractor for CLIP-based image/text feature extraction
- Add ObjectiveMetricsCalculator for technical quality metrics
- Add VLMEvaluator for vision language model evaluation
- Add HybridQualityGate combining objective + VLM evaluation
- Enhance CharacterMemory with visual feature support
- Add quality optional dependency (torch, ftfy, regex)
- Add unit tests for new modules
🤖 Generated with [Claude Code](https://claude.com/claude-code )
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com >
2026-01-05 15:56:44 +08:00
empty
ca018a9b1f
feat: Add comprehensive timeline editor with frame editing and regeneration capabilities
2026-01-05 14:48:43 +08:00
