← sources

Qwen 2.1 styles comprehension.

source: reddit (reddit:StableDiffusion)published 2026-09-30 15:42ingested 2026-10-01 02:41

retained text

I'm far from being an expert, but for me Qwen is not a Turbo Model so it cannot be as good as Krea 2 Turbo.

Also their dataset seems to be not good, it's hard to make doodles, cartoon styles, and its character knowledge is poorer.

And because it is not a Turbo model, the prompt needs to be very long and you need to specify everything, from lighting, to every small details or it come as a messy noisy image,  very visible on hair for exemple in photographic portraits.

Even negatives don't help a lot at CFG=3.

I'll go back to Krea 2,and waiting for a good Turbo lora (or a real Turbo version).

If you found a solution to make as good as a tubro model, prove me wrong.

document fetched 2026-10-01 00:35 · canonical https://reddit.com/r/StableDiffusion/comments/1wu8auh/qwen_21_styles_comprehension