IXIWORKS-KIMJUNGHO-ComfyUI-VideoDescription

72 Commits 1 Branch 0 Tags

2 Commits

Include renames

Author	SHA1	Message	Date
KIMJUNGHOandClaude Sonnet 4.6	94929c6bd0	feat: v2.0.0 — Major refactor and node reorganization ## Nodes - Unified node key prefix to Util* (UtilSwitch, UtilStringToList, etc.) - Added UtilSaveFile: combined Save Text & Image node with batch/list support, shared counter, subfolder, INPUT_IS_LIST - Added UtilJsonToList: JSON file/string to STRING list via dot-notation key - Added UtilEmptyLatent channels param (16=Flux/Lumina/SD3, 4=SD/SDXL) - Updated UtilLoadImageList: count-based dropdown selectors from Input folder - Added CNStepControl, CNPreprocessor (canny/depth/lineart/openpose/mlsd) - Added ImageCropResize, LoRAStepLoader - Added SBPromptFilter (Claude API style/controlnet filtering) - Removed layer_nodes.py, sampler_nodes.py (layer separation experiments) - Removed experiments/ directory ## Dependencies - Removed controlnet_aux — direct preprocessor implementations in preprocessors/ Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-03-05 10:56:17 +09:00
KIMJUNGHOandClaude	eb8669890d	feat: add ComfyUI video description nodes with Qwen3-VL and DAM support Implement custom nodes for video description using advanced vision-language models. Features: - Video Description (Qwen3-VL): Full video analysis node - Works on CUDA, Apple Silicon (MPS), and CPU - 3 analysis types: detailed, summary, keywords - Smart path resolution for easy video loading - Optimized inference with model caching - Video Description (DAM Region): Region-based video analysis node - CUDA-only (requires NVIDIA GPU) - Region specification via JSON coordinates (point/box/polygon) - 3 analysis types: detailed, summary, action - Localized description generation Architecture: - Singleton model cache for efficient memory usage - ComfyUI standard directory structure - Comprehensive error handling and validation - Independent node operation without interference 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>	2025-10-22 11:31:18 +09:00

Author

SHA1

Message

Date

KIMJUNGHOandClaude Sonnet 4.6

94929c6bd0

feat: v2.0.0 — Major refactor and node reorganization

## Nodes
- Unified node key prefix to Util* (UtilSwitch, UtilStringToList, etc.)
- Added UtilSaveFile: combined Save Text & Image node with batch/list support,
  shared counter, subfolder, INPUT_IS_LIST
- Added UtilJsonToList: JSON file/string to STRING list via dot-notation key
- Added UtilEmptyLatent channels param (16=Flux/Lumina/SD3, 4=SD/SDXL)
- Updated UtilLoadImageList: count-based dropdown selectors from Input folder
- Added CNStepControl, CNPreprocessor (canny/depth/lineart/openpose/mlsd)
- Added ImageCropResize, LoRAStepLoader
- Added SBPromptFilter (Claude API style/controlnet filtering)
- Removed layer_nodes.py, sampler_nodes.py (layer separation experiments)
- Removed experiments/ directory

## Dependencies
- Removed controlnet_aux — direct preprocessor implementations in preprocessors/

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

2026-03-05 10:56:17 +09:00

KIMJUNGHOandClaude

eb8669890d

feat: add ComfyUI video description nodes with Qwen3-VL and DAM support

Implement custom nodes for video description using advanced vision-language models.

Features:
- Video Description (Qwen3-VL): Full video analysis node
  - Works on CUDA, Apple Silicon (MPS), and CPU
  - 3 analysis types: detailed, summary, keywords
  - Smart path resolution for easy video loading
  - Optimized inference with model caching

- Video Description (DAM Region): Region-based video analysis node
  - CUDA-only (requires NVIDIA GPU)
  - Region specification via JSON coordinates (point/box/polygon)
  - 3 analysis types: detailed, summary, action
  - Localized description generation

Architecture:
- Singleton model cache for efficient memory usage
- ComfyUI standard directory structure
- Comprehensive error handling and validation
- Independent node operation without interference

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

2025-10-22 11:31:18 +09:00