CVT - Cinematic Video Tools

Cinematic video prompt tools for ComfyUI.

Nodes

ThreeShot Sequence

Builds 1-3 shot video prompts with per-shot controls for camera motion, framing, color grading, mood, and lighting. Supports image-driven parameter extraction via CLIP, Qwen 3 VL, Gemini, ChatGPT, Claude, or the ComfyUI API proxy.

SceneCamera

Takes a scene image and a camera movement selection, outputs an optimized video prompt with 25 camera move presets (push in, orbit, crane, dolly zoom, drone, steadicam, etc.).

Storyboard Assembler

Parses 9 shot descriptions from Gemini API output and assembles them into a prompt for a 3x3 cinematic contact sheet.

Extract First Frame

Extracts a single frame from a VIDEO input as an IMAGE.

Gate Nodes

Simple pass-through gates (Any, String, Image) that enable/disable signal flow with a boolean toggle.

Image Concatenate Multi

Concatenates multiple images in any direction with optional size matching.

List To Batch

Converts a list of images into a batched IMAGE tensor.

Installation

Install via the Comfy Registry:

comfy node install comfyui-cvt

Or clone into your custom_nodes directory:

cd ComfyUI/custom_nodes
git clone https://github.com/f1tzcarrald0/comfyui-cvt.git
pip install -r comfyui-cvt/requirements.txt

Example Workflows

See the example_workflows/ folder for ready-to-use workflow files.

Vision Analysis Models

The ThreeShot node can analyze input images to automatically determine color grading, mood, and lighting parameters. Supported backends:

| Model | Type | Requirements | |-------|------|-------------| | CLIP | Local | transformers (included in dependencies) | | Qwen 3 VL 8B | Local | GPU with 16GB+ VRAM, auto-downloads model | | Gemini | API | Google AI API key | | ChatGPT | API | OpenAI API key | | Claude | API | Anthropic API key | | ChatGPT (ComfyUI) | API proxy | ComfyUI account (uses account credits) |