ComfyUI-PrunaVAED loads the PrunaVAED pruned video VAE for LTX-2.3 into ComfyUI as a standard VAE, providing faster decoding and lower VRAM usage.

Stars

8

7-day growth

No data

Forks

1

Open issues

1

License

GPL-3.0

Last updated

2026-07-28

AI repository intelligence
FR-AI / ANALYSIS

Why it is worth attention

It delivers up to ~1.97× faster video decoding with about 50% lower peak VRAM compared to the stock LTX-2.3 VAE, requires no extra Python dependencies, and integrates seamlessly with ComfyUI's existing VAE interface.

Who it is for

  • ComfyUI users working with LTX-2.3 video generation
  • Developers optimizing video workflow performance
  • Researchers evaluating pruned VAE models
  • Content creators seeking faster video output

Use cases

  • Accelerating text-to-video decoding in LTX-2.3 workflows
  • Enabling faster image-to-video encoding and decoding
  • Running reproducible VAE decode benchmarks
  • Replacing the stock LTX-2.3 video VAE in existing ComfyUI pipelines

Strengths

  • No additional Python packages required
  • Standard ComfyUI VAE output compatible with VAE Decode and tiled decode nodes
  • Built-in benchmark node for reproducible performance measurement
  • Explicit format validation and error messages for unsupported checkpoints

Considerations

  • Only accelerates the LTX-2.3 video VAE; audio VAE and transformer remain unchanged
  • Performance gains depend on GPU architecture, resolution, tiling, and dtype selection
  • Manual model download and placement required; no automatic download

README quick start

Installation

Description

Native ComfyUI loader and benchmark nodes for the accelerated PrunaAI PrunaVAED LTX-2.3 video VAE.

Related repositories

Similar projects matched by category, topics, and programming language.

xiejunjie524
Featured
xiejunjie524 GitHub avatar

handdraw-story-video

A tool that converts 7–9 hand-drawn story keyframes into a 35–45 second vertical video with progressive line art and color animation, configurable via JSON and renders with HyperFrames and GSAP.

Design & Creative
692
bytec-ai
Featured
bytec-ai GitHub avatar

book-video-factory

A portable multi-account short video production skill for books that manages accounts, scripts, images, voiceovers, subtitles, and final exports through a unified workspace.

Design & Creative
136
Krishnagangwal
Featured
Krishnagangwal GitHub avatar

CS-Fundamentals

A curated collection of Computer Science fundamentals (PDFs, notes, cheatsheets, interview question banks) for placement preparation, covering seven core subjects plus general resources.

Data & DatabasesDatabases & Storage
2,326