键盘 let5see
0 Followers · 0 Following
  • Joined on 2025-11-29
Repositories
18
 Projects Packages Public Activity Starred Repositories
let5see pushed to main at let5see/AI-Video 2026-01-07 09:56:06 +08:00
c854bd80e0 feat: Switch to flux2 workflow for image regeneration
let5see pushed to main at let5see/AI-Video 2026-01-07 09:44:31 +08:00
90ceb76296 feat: Optimize VLM prompt for storyboard consistency, focus on constant features
let5see pushed to main at let5see/AI-Video 2026-01-07 09:40:23 +08:00
1b54552fec fix: Handle nested JSON structures in VLM response parsing
let5see pushed to main at let5see/AI-Video 2026-01-07 09:38:48 +08:00
f19804facb feat: Add VLMConfig to schema for proper config.yaml VLM support
let5see pushed to main at let5see/AI-Video 2026-01-07 09:35:47 +08:00
bc077475c6 fix: Remove duplicate import os causing UnboundLocalError
let5see pushed to main at let5see/AI-Video 2026-01-07 09:32:20 +08:00
9675b9c23b feat: Add VLM config to config.example.yaml with config.yaml support
let5see pushed to main at let5see/AI-Video 2026-01-07 09:29:46 +08:00
92183b083b feat: Add Qwen VL support for character analysis, configurable via VLM_PROVIDER
let5see pushed to main at let5see/AI-Video 2026-01-07 03:37:58 +08:00
be216eacad fix: Increase VLM max_tokens to 2000 to avoid response truncation
let5see pushed to main at let5see/AI-Video 2026-01-07 03:33:59 +08:00
8d82cf91d5 fix: Auto-detect and use GLM-4V vision model for character analysis
let5see pushed to main at let5see/AI-Video 2026-01-07 03:31:44 +08:00
8c35b0066f fix: Enhance VLM response parsing to handle markdown code blocks
let5see pushed to main at let5see/AI-Video 2026-01-07 03:18:00 +08:00
44249889df fix: Add file upload endpoint and fix frontend upload path
let5see pushed to main at let5see/AI-Video 2026-01-07 03:11:45 +08:00
49e667cc94 feat: Enhance CharacterPanel with image upload and VLM analysis
let5see pushed to main at let5see/AI-Video 2026-01-07 03:08:32 +08:00
b3cf9e64e5 feat: Implement Character Memory V1 - VLM analysis and prompt injection
let5see pushed to main at let5see/AI-Video 2026-01-07 00:34:41 +08:00
da98d0842a feat: Persist regenerated image/audio paths to storyboard.json
let5see pushed to main at let5see/AI-Video 2026-01-07 00:31:52 +08:00
c0eb4ed320 fix: Add onClick handlers to top Settings and Save buttons with alerts
let5see pushed to main at let5see/AI-Video 2026-01-07 00:22:35 +08:00
2be9256c48 fix: Use OpenAI multimodal message format for VLM style extraction
let5see pushed to main at let5see/AI-Video 2026-01-07 00:16:59 +08:00
297f3ccda4 feat: Enhance StyleGuard with VLM-based style extraction for specific style_prefix
let5see pushed to main at let5see/AI-Video 2026-01-07 00:05:58 +08:00
a3ab12e87c feat: Add detailed debug logging to editor functions
let5see pushed to main at let5see/AI-Video 2026-01-06 23:54:39 +08:00
4d3c89a8f6 fix: Await async extract_style_anchor call
let5see pushed to main at let5see/AI-Video 2026-01-06 23:50:29 +08:00
7da6ed6a74 feat: Auto-apply style anchor prefix when regenerating frame images