# workflow minimax 6 steps basic and 4 steps taomate lora example — retained record

> Retained reddit record from reddit:StableDiffusion.

Canonical: https://highsignal.app/data/records/97de9aaf588c9ee8

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


source: [reddit](https://highsignal.app/data/reddit) (reddit:StableDiffusion)  published 2026-09-14 16:14 ingested 2026-09-15 02:38

[original source ↗](https://www.reddit.com/r/StableDiffusion/comments/1wg7vfq/workflow_minimax_6_steps_basic_and_4_steps/)

## retained text



```
i made a new post since the one used a suspicious site for the wf. here is the pastebin link 

[https://pastebin.com/DefpKa7d](https://pastebin.com/DefpKa7d)

Since everybody is talking about this lora that apparently is great as a refiner. i tried to implement it in my wf with the help of Chatgpt. 

I HAVE LIMITED EXPERIENCE WITH WORKFLOW SO IS NOT PERFECT 

But maybe together we can find the best way to make it work.



 the video is not perfect, but maybe is a start? 

2 clownshark samplers euler/beta57. 

6 steps base without speed lora and 4 with taomate 3 steps lora 0.65 strenght. 

vs 

10 steps with 8step lora 1 strenght. 

0.8 mp, 7s, same seed 

the quality is not equally good but is more dynamic. total rendering time around 308s for both 



let me know if you find best combination
```



document fetched 2026-09-15 00:33 · canonical https://reddit.com/r/StableDiffusion/comments/1wg7vfq/workflow_minimax_6_steps_basic_and_4_steps
