Qwen Image 2.1 Turbo - extend 'magic' 8 steps sigmas further
So, I've decided to try Qwen Image 2.1 Turbo with the following 'magic' sigmas, taken directly from their Diffusers implementation:
Proof1 community thread
Key points
- This works okay, but for fine details we need slightly more steps (for better skin, better composition etc.) And to my knowledge, there's no working way to interpolate the resulting curve to any N steps (e.g. 10, 12, 14).
- Here's the original 'magic' 8-step curve:
- If we use RES4LYF 'Custom Sigmas' node and try to interpolate to, say, 14 steps, we get this:
- Meet a 'Magic AI Slop Scheduler for Qwen Image 2.1 Turbo'😆
Sources (1)
- [1]Qwen Image 2.1 Turbo - extend 'magic' 8 steps sigmas furtherr/StableDiffusion (top, day) · Oct 11, 01:01 AM
So, I've decided to try Qwen Image 2.1 Turbo with the following 'magic' sigmas, taken directly from their Diffusers implementation:
This works okay, but for fine details we need slightly more steps (for better skin, better composition etc.) And to my knowledge, there's no working way to interpolate the resulting curve to any N steps (e.g. 10, 12, 14).
Extractive summary: sentences quoted from the sources.
Before this
- Oct 11, 2026TensorSharp now supports Qwen Image 2.1 Turbo + LoRA — Here's a quick image editing demo
- Oct 10, 2026Krea2 Turbo Distill 2 step LoRA - FINAL checkpoint released (chk51195)
- Oct 10, 2026How do you guys upscale your images? Is anything even close to Magnific?
- Oct 9, 2026Qwen/Qwen-Image-2.1-Turbo
- Oct 8, 2026unslothai/unsloth v0.1.905-beta: Sandboxing is here!
- Oct 7, 2026unslothai/unsloth v0.1.904-beta: Train your own Decision model
