Sign In
homemodelsimagesvideospostsarticlesbountieschallengeseventsupdatesshop

Torsten's Wan2.2 i2v GGUF Workflow VERSION 3.0!

10

TorstenTheNord's Avatar

TorstenTheNord

Oct 27, 2025

(Updated: 3 days ago)

workflows
Torsten's Wan2.2 i2v GGUF Workflow VERSION 3.0!

Version 3.0 of Torsten's Wan2.2 i2v 14B Low-VRAM Workflow is released!

If this is your first time seeing my content, or if you're already a follower who was eagerly anticipating this update, hello and thank you for checking out my work! I'm a hobbyist AI tinkerer and I was on a hiatus to focus on other important life events. Now I'm back with an exciting update of VERSION 3.0 of my highly-rated Low-VRAM Wan2.2 workflow that even a basic computer with 8GB of VRAM can generate high-quality videos with!

Download, review, and share your videos on the model page:

https://civitai.com/models/1824962?modelVersionId=2350988

What's new in Version 3.0?

  • The workflow now includes grouped sections to break the process down into steps.

  • There is a very detailed instruction manual at the far left. This provides model links and directions for how to generate high-quality videos from a reference image.

  • Re-worked a few node in/out links to prevent video generation errors.

  • Added a node pack called RES4LYF. This new pack adds more options for samplers and schedulers within the default KSampler nodes, and allows users to preview the de-noise sigmas prior to running a full generation, fine-tuning user control over the quality.

Here's an example of RES4LYF de-noise sigma value preview:

RES4LYF_Preview.png

This is an example of De-Noise Sigma Value with 20 steps WITHOUT USING LightX2V LORA
*** If using LightX2V speed lora, I only recommend using Simple or Normal scheduler with a ModelShift value of 1-2 ***

Updated Recommendations for Choosing Quantization Model:

K_S = Small | K_M = Medium | K = Full/Large | Less VRAM = Smaller Quant Number & Size

12-16+ GB VRAM - Q8_0 Model

10-12 GB VRAM - Q6_K Model, or Q5_K_M if you get frequent OOM errors

8GB VRAM (or Less) - Q4_K_M Model

Here is a video generated with the new V3.0 Workflow (upscaled using an external workflow)

If you enjoy using my workflows, please share your generated videos on the model page and leave likes, comments, or even Buzz if you feel like it! Thank you for taking the time to read this post, and I hope you have fun using this simple and effective workflow for your AI videos.

10