# looking for i2v minimax h3 — retained record

> Retained reddit record from reddit:StableDiffusion.

Canonical: https://highsignal.app/data/records/cf8fcbdae7cca1fc

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:StableDiffusion)  published 2026-09-30 16:20 ingested 2026-10-01 02:41

[original source ↗](https://www.reddit.com/r/StableDiffusion/comments/1wu9arj/looking_for_i2v_minimax_h3/)

## retained text



```
i know there are a lot of posts about this but i  just can't find what i need.

  
I'm looking for I2v workflow which incorportae kitchen attention (and any other speedups/optimizations) + easy way to load loras into it. hopefilly workflow which keeps face consitancy.

a video upscaler at the and of workflow will be a nice bonus.

32gb systme ram + 4070rtx 12gb vram.

the regular comfyui template works fine, but very slow and unintuitive for loras. 

  
any suggestions?

Edit - any local llm for creating prompts for it? 
```



document fetched 2026-10-01 00:35 · canonical https://reddit.com/r/StableDiffusion/comments/1wu9arj/looking_for_i2v_minimax_h3
