Custom ComfyUI nodes for music video workflows and high-quality video/image enhancement.
Includes HUMO-based multi-scene workflows, audio splitting, video combining, and realtime-ready enhancement tools.
Turn a reference image and an audio track into a fully AI-generated music video.
The workflow automatically splits lyrics, generates scene prompts, and syncs everything into a cinematic final video.
- Upload a reference image (your main character).
- Import your audio file (with lyric syncing).
- Set a folder name for outputs.
- Customize the Prompt Creator to define style, mood, and scenes.
✨ Everything else — transcription, scene building, video chunks, and final rendering — runs automatically.
The result: a stylized, synced AI-driven music video.
-
🎞️ Fast Film Grain (
FastFilmGrain)
Add controllable, grayscale or color grain for cinematic texture.
➕ Now includes abatch_sizesetting to reduce out-of-memory (OOM) issues on long or high-res videos. If you have a strong GPU, trybatch_size = 8. If you hit OOM, lower the value (e.g., to 4 or 2). -
🎨 Color Match to Reference (
ColorMatchToReference)
Align image tones to a reference image using LAB color matching.
➕ Also includes abatch_sizesetting to help manage VRAM usage. Start at 8 if your GPU allows — lower it if you get OOM errors. -
🎯 Fast Unsharp Sharpen (
FastUnsharpSharpen)
Simple and efficient sharpening using unsharp masking. -
🌀 Fast Laplacian Sharpen (
FastLaplacianSharpen)
Edge-based sharpening via Laplacian kernel for crisp detail. -
📏 Fast Sobel Sharpen (
FastSobelSharpen)
Gradient-based edge enhancement using Sobel filters. -
⚡ Optimized for image or video workflows and per-frame processing in ComfyUI.
-
💻 Designed to be efficient on modern GPUs.
- Open ComfyUI.
- Go to the Manager tab → Install Custom Nodes.
- Search:
vrgamedevor use this Git URL:https://github.com/vrgamegirl19/comfyui-vrgamedevgirl
- Clone or download this repo to your
ComfyUI/custom_nodesdirectory. - Restart ComfyUI.
A requirements.txt file is now needed.
To install the required packages, run this from inside the python_embeded folder:
python.exe -m pip install -r ..\custom_nodes\comfyui-vrgamedevgirl\requirements.txt
requirements.txt (in the repo) includes:
kornia
librosa
imageio
Adds customizable film grain to each frame.
Grain can be grayscale or saturated RGB, and sized to match your output resolution.
Inputs:
images: Frame tensor input.grain_intensity: Blend amount (0 to 1).saturation_mix: 0 = grayscale grain, 1 = full RGB noise.
🟢 Video-safe and very fast.
Matches image color distribution to a reference image using LAB space normalization.
Inputs:
images: Your video frames.reference_image: A single image to match tone and color against.match_strength: Blend between original and matched (0.0–1.0).
Applies unsharp masking to enhance edges with a fast, low-cost blur pass.
Inputs:
images: Input image tensor.strength: Sharpening amount (0.0 to 2.0)
🚀 Lightweight and ideal for subtle sharpening.
Enhances edges by applying a Laplacian kernel to bring out high-frequency detail.
Inputs:
images: Input image tensor.strength: Sharpening amount (0.0 to 2.0)
🧪 Gives a more "punchy" sharpen effect great for detail recovery.
Uses Sobel filters to detect image gradients and amplify edge contrast.
Inputs:
images: Input image tensor.strength: Sharpening amount (0.0 to 2.0)
🧠 Sharpens by boosting directional edge response — great for outlines and detail clarity.
- 🎞️ Fast Film Grain (
FastFilmGrain) - 🎨 Color Match To Reference (
ColorMatchToReference) - 📏 Fast Sobel Sharpen (
FastSobelSharpen) - 🌀 Fast Laplacian Sharpen (
FastLaplacianSharpen) - 🎯 Fast Unsharp Sharpen (
FastUnsharpSharpen) - 🌫️ Local Contrast / Dehaze
- 🎛️ LUT Loader or Approximate Match
comfyui-vrgamedevgirl/
│
├── init.py
├── nodes.py
├── README.md
├── LICENSE
├── requirements.txt
└── .gitignore
VRGameDevGirl
✨ Custom tools for cinematic AI workflows
💌 Questions or collabs? Reach out via GitHub
This project is licensed under the MIT License.