ComfyUI Extension: Vega Flow V8.9 — Temporal Stabilisation

Authored by vegaflowltd

Created

Updated

0 stars

Run ComfyUI workflows without the setup

No installs, no CUDA version roulette, no GPU sitting idle on your bill. Bring a workflow and run it in the browser.

Professional temporal stabilisation engine. Eliminates flicker, luminance drift, and frame-to-frame inconsistency in AI-generated and traditional video footage.

Looking for a different extension?

Custom Nodes (0)

    README

    Vega Flow V8.9 — ComfyUI Node

    Vega Flow is a professional temporal stabilisation engine built by an independent deep tech studio based in Edinburgh, Scotland. We develop enterprise OpenFX plugins for professional VFX pipelines, and are releasing our V8.9 engine to the ComfyUI community as a free public beta.

    Benchmarked against DaVinci Resolve's native deflicker on RED 4K footage — Vega Flow outperforms across all measured metrics.

    This node applies advanced mathematical stabilisation to fix temporal inconsistency, flickering, and color/luminance shifts in video generation.

    🎯 What V8.9 Excels At (Current Architecture)

    V8.9 is a global temporal processor. It is highly optimized for workflows where the entire frame requires stabilization. It performs exceptionally well on:

    • Older-Gen AI Video: AnimateDiff, Stable Video Diffusion (SVD), Pika 1.0, and Gen-2.
    • Archival/Traditional Footage: Correcting extreme camera flicker, drone prop-shadow, and analog film degradation.
    • Stylized Renders: Deflickering Deforum or batch Image-to-Image stylization.

    🚧 Known Limitations (The V9.0 Roadmap)

    Please Note: V8.9 is not optimized for next-generation DiT models like LTX-2.3 or Sora.

    Newer models often suffer from "localized micro-boil" (e.g., the foreground subject is perfectly stable, but the brick wall in the background is melting). Because V8.9 applies stabilization globally, running it on LTX-2.3 may inadvertently soften sharp foregrounds while trying to lock the background.

    We are actively building Vega Flow V9.0 (Semantic Spatial Masking) to solve this exact problem by routing the math only to unstable pixels. Until then, use V8.9 for AnimateDiff, SVD, and global deflickering workflows!

    🔒 Why is the core engine compiled?

    The Vega Flow engine contains proprietary mathematics protected by a pending patent. The core is distributed as a compiled binary to protect this IP.

    This node runs entirely locally on your machine:

    • No internet connection required.
    • No API keys.
    • No data leaves your machine.
    • Totally free to use.

    ⚙️ Installation

    1. Navigate to your ComfyUI/custom_nodes folder.
    2. Clone this repository: git clone https://github.com/vegaflowltd/ComfyUI-VegaFlow
    3. Restart ComfyUI.
    4. Search for "Vega Flow V8.9" in the node menu.

    🎛️ Recommended Parameters for AI Video (AnimateDiff/SVD)

    • Media Profile: AI-Generated
    • Anchor Mode: Aggressive
    • Structure Preset: Architectural (for environments) or Organic (for characters)
    • Flicker Reduction: 5.0 - 7.0

    Run ComfyUI workflows without the setup

    No installs, no CUDA version roulette, no GPU sitting idle on your bill. Bring a workflow and run it in the browser.

    Learn more