A collection of custom nodes for ComfyUI.

ComfyUI-DeZoomer-Nodes

Generates detailed captions for videos using the Qwen2.5-VL model.

Input Parameters:
- images: Input video frames to process (ComfyUI's IMAGE type)
- user_prompt: Detailed instructions for what aspects to analyze (default provided)
- system_prompt: Instructions for the model's behavior and output style
- model_name: Qwen2.5-VL model to use (7B, 1.5B, or 72B variants)
- temperature: Controls randomness in generation (0.1-1.0)
- use_flash_attention: Enables faster attention implementation
- low_cpu_mem_usage: Optimizes for low CPU memory usage
- quantization_type: Memory optimization (4-bit or 8-bit)
- keep_model_loaded: Whether to keep the model in memory after processing
- seed: Random seed for reproducible generation

The node processes video frames and generates comprehensive descriptions covering:
- Main content and characters
- Object and character details
- Actions and movements
- Background elements
- Visual style
- Camera work
- Scene transitions

ComfyUI Node: Video Captioning

Category

Inputs

Outputs

Extension: ComfyUI-DeZoomer-Nodes

More nodes in ComfyUI-DeZoomer-Nodes

Run ComfyUI workflows without the setup