Symptom

  • You specify a single person or a single dog, but several people or objects appear.
  • Only certain parts, like hands or faces, seem to multiply.

When it happens

  • When generating with Stable Diffusion 1.5 at too high a resolution, such as 1024px or more.
  • When generating at an extremely tall or wide resolution.

Cause

  • SD1.5 was trained on square images around 512px, so its compositions tend to fall apart at larger resolutions.

Solution

  • Generate at a size close to the model's recommended resolution
    • For SD1.5, try around 512-768px.