Y
Prompt Workshop
Article · SD parameters

Stable Diffusion Sampling Steps Explained

Steps are how many denoising iterations Stable Diffusion runs. More is not always better. Most samplers converge by 25–35 steps.

Per-sampler sweet spots

SamplerSweet spotAbove is
Euler a25–30Style keeps drifting (non-converging)
DPM++ 2M Karras25–35Essentially unchanged
UniPC15–20Essentially unchanged
DDIM30–50Tiny detail gains
Flux Dev20–25No change
Flux Schnell2–4Collapses

When to push higher

  • Hires fix: 50–80 steps
  • Detail-critical product shots: 40 steps
  • Complex multi-subject scenes: 35–40 steps

When to go lower

  • Batch drafts: UniPC 15 steps
  • Schnell: 4 steps is enough
  • img2img with low denoise: 15 steps

Frequently asked questions

Does more steps = more detail?

Past the sweet spot, no. 30 vs 80 steps on DPM++ 2M is nearly indistinguishable.

Why does Schnell only need 4 steps?

Schnell is a distilled variant trained to converge in very few steps. Pushing it higher collapses.

Try this parameter live in the editor

The local editor includes Midjourney / SDXL / Flux templates with a visible, editable parameter panel.

Open the editor →
Yan · AI Prompt Workshop editorial team|Last updated on 2026-06-12。This site does not call any cloud model. Every prompt and parameter in this article was tested and refined locally by the editorial team.