toyxyz/ComfyUI_toyxyz_test_nodes

ComfyUI_toyxyz_test_nodes – a toolbox of custom ComfyUI nodes

What it is – A collection of user‑contributed nodes for ComfyUI, the visual‑flow interface for Stable Diffusion‑style image generation. The repo bundles a variety of utilities that the author (toyxyz) frequently uses, ranging from webcam capture and region‑mask creation to advanced prompt‑generation helpers for MiniMax‑H3 video workflows.

Why it matters – ComfyUI’s power comes from its modular nodes. By installing this package you instantly gain dozens of ready‑made nodes that would otherwise require you to write Python scripts or manually wire together multiple primitive nodes. The nodes cover:

• Real‑time webcam and screen capture (including region‑specific windows)
• Mask generation (draw‑area, visual‑area, region‑based masks) for regional prompting
• OpenPose editing and pose interpolation
• Utilities for handling video (trim, concatenate, export to .glb, depth‑to‑normal conversion, noise removal)
• Text helpers (load random lines from files, Lora‑hook support)
• A MiniMax‑H3‑prompter node that builds audiovisual prompts for the MiniMax H3 video model, handling shot/move timelines, camera‑style metadata and reference analysis via Qwen‑3.8‑27B or MiniMax‑H3‑Rewriter.

Installation

1. Clone the repo into your ComfyUI custom_nodes directory:

   git clone https://github.com/toyxyz/ComfyUI_toyxyz_test_nodes
   

2. Run the provided setup.bat inside CaptureCam to install the small webcam helper application.
3. Restart ComfyUI – the new nodes appear under a “toyxyz” category.

Typical workflow – The README ships three example graphs (default, render‑preview, direct webcam capture). A common pattern is:

1. Capture an image or video frame via the Capture Webcam or Direct_screenCap node.
2. Mask the region you want to treat specially using Draw area mask or Visual area mask.
3. Prompt the region with a base prompt (e.g., ComfyCouple Region) and optionally add LoRA hooks.
4. Generate the image, then Save image to path or Export video.
5. For video‑centric projects, feed the generated prompt from MiniMax‑H3‑prompter into a downstream H3 pipeline.

Key nodes (selected)

Node	Purpose
MiniMax‑H3‑prompter	Constructs timed audiovisual prompts (shots, moves, camera style) for MiniMax H3 video generation.
Capture Webcam / Load Webcam Image	Grab frames from a webcam (real‑time) or load pre‑saved captures.
Direct_screenCap	Capture a screen region or a specific window.
Draw area mask / Visual area mask	Interactive creation of binary masks for regional prompting.
Openpose Editor / Pose Interpolation	Edit body‑part keypoints or generate smooth transitions between poses.
ComfyCouple Region / Mask	Regional prompting compatible with SD‑1.5, SDXL, Flux and Anima; supports LoRA hooks.
Cut Video / Connect Video	Trim or concatenate VIDEO tensors while preserving audio and FPS.
Depth to normal / Remove noise	Post‑process depth maps or normal maps for better 3D‑aware generation.
Export glb	Packages a flat .glb file containing color, normal and alpha maps.
Load Random Text From File	Pull random or indexed lines from a text file to use as prompts.

Who should use it – Anyone already working with ComfyUI who wants quick access to webcam‑based capture, region‑mask workflows, or the MiniMax‑H3 prompt builder without writing custom code. The nodes are especially handy for creators building AI‑augmented video pipelines or interactive image‑to‑video setups.

License & contribution – The repository follows the standard open‑source model used by most ComfyUI custom nodes (MIT/Apache‑like). Contributions are accepted via pull‑requests; the author maintains a changelog with dates for each added node.