# Minimax H3  - What Happens When You Train With Stills Alone? — retained record

> Retained reddit record from reddit:StableDiffusion.

Canonical: https://highsignal.app/data/records/5f0c2b427489cc02

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


source: [reddit](https://highsignal.app/data/reddit) (reddit:StableDiffusion)  published 2026-09-18 16:49 ingested 2026-09-19 02:40

[original source ↗](https://www.reddit.com/r/StableDiffusion/comments/1wjvje2/minimax_h3_what_happens_when_you_train_with/)

## retained text



```
I wanted to make this demo as there is a lot of old info out there (old by ai standards) about how training with images breaks motion ability etc with Minimax.   This is a very short before and after video with the same prompt with and without a lora trained on 341 stills, with same seed, settings, fully T2V.

[https://github.com/shootthesound/Fizgig/releases/tag/v6.2.0](https://github.com/shootthesound/Fizgig/releases/tag/v6.2.0)
```



document fetched 2026-09-19 00:31 · canonical https://reddit.com/r/StableDiffusion/comments/1wjvje2/minimax_h3_what_happens_when_you_train_with
