ThereforeGames/unprompted

zoom_enhance with bypass_adaptive_hires bug

ComplexRobot opened this issue · 4 comments

ComplexRobot commented

Due diligence

  • I checked for similar issues and couldn't find any.
  • My WebUI and Unprompted are both up-to-date.
  • I disabled my other extensions but the problem persists.

Describe the bug

If I use the bypass_adaptive_hires argument in zoom_enhance, it gives the error 'target_multiplier' referenced before assignment. Then, unprompted breaks until I reload the entire Stable Diffusion UI.

Prompt

No response

Log output

stable-diffusion-webui\extensions\_unprompted/shortcodes\stable_diffusion\zoom_enhance.py", line 305, in run_atomic
        target_size = (w * h) / (upscale_width * upscale_height * target_multiplier)
    UnboundLocalError: local variable 'target_multiplier' referenced before assignment

Unprompted version

9.16.1

WebUI version

1.6.0

Other comments

No response

ThereforeGames commented

Hi ComplexRobot,

Thank you for bringing this to my attention. The bypass_adaptive_hires parg has been fixed in Unprompted v10.1.3. Let me know if you have any questions. 🙂

ComplexRobot commented

Hi ComplexRobot,

Thank you for bringing this to my attention. The bypass_adaptive_hires parg has been fixed in Unprompted v10.1.3. Let me know if you have any questions. 🙂

Thank you! It doesn't produce an error anymore, but I don't think the setting is working as intended. I have logging level set to debug and it says it's still changing the CFG dynamically. Also, I was experiencing an error message about sharpness, so I fixed it by using sharpen_amount=0. Therefore, I don't know if that is working as intended.

I'm going off of this from the manual:

Supports the bypass_adaptive_hires positional argument. By default, the shortcode will scale up some inference settings such as CFG scale and sharpness depending on the resolution of the init image. Include this argument to disable that behavior.

In the meantime, it seems I can use cfg_scale, denoising_strength and sharpen_amount manually to disable the behavior, but it makes the bypass_adaptive_hires argument pointless. 😅

ThereforeGames commented

Ah, looks like it was broken in more ways than one. I just released another small patch for this - I think it's working correctly now. Let me know if it still gives you any trouble.

Also, what was the error message regarding sharpen_amount? I haven't been able to reproduce that one. Thanks!

ComplexRobot commented

Also, what was the error message regarding sharpen_amount? I haven't been able to reproduce that one. Thanks!

I don't think it was a bug. I think that was a misunderstanding on my part.

  • 👍1