[Bug]: Move to Inpaint Sketch Slowdown Crash (Repro vid)

[scpedicini]

Is there an existing issue for this?

• [X] I have searched the existing issues and checked the recent builds/commits

What happened?

Using the inpaint-sketch tab directly and loading an image seems to work correctly with no slowdown issues.

Using Copy to inpaint sketch results in some kind of infinite GET loop that eventually causes the inframed window to crash.

From the developer console, it looks like its doing recursive HTTP GET requests on the Base64 image, each time increasing the size of the fetch - 1.8mb - 3.6mb -> 7.2mb, etc.

https://user-images.githubusercontent.com/2040540/233862830-4b124ab5-0686-46e2-8401-57132572d80c.mp4

Note: There is an existing issue that may be related filed back in January, but I'm not sure it is the same issue since I do not encounter an automatic1111 log errors related to IndexError. https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/7083

Steps to reproduce the problem

1. Load an image into img2img
2. Copy image to: inpaint sketch

What should have happened?

Browser should not have crashed.

Commit where the problem happens

22bcc7be

What platforms do you use to access the UI ?

MacOS

What browsers do you use to access the UI ?

Google Chrome, Brave, Apple Safari

Command Line Arguments

`--listen --xformers`

List of extensions

No

Console logs

venv "C:\dev\automatic1111\venv\Scripts\Python.exe"
Python 3.10.9 (tags/v3.10.9:1dd9be6, Dec  6 2022, 20:01:21) [MSC v.1934 64 bit (AMD64)]
Commit hash: 22bcc7be428c94e9408f589966c2040187245d81
Installing requirements for Web UI


Launching Web UI with arguments: --listen --xformers --ckpt-dir C:\generative\models\Stable-Diffusion
Loading weights [eafc19923b] from C:\generative\models\Stable-diffusion\526mixV14_v14-inpainting.safetensors
Creating model from config: C:\generative\models\Stable-diffusion\526mixV14_v14-inpainting.yaml
LatentInpaintDiffusion: Running in eps-prediction mode
DiffusionWrapper has 859.54 M params.
Applying xformers cross attention optimization.
Textual inversion embeddings loaded(3): bad-artist, EasyNegative, ng_deepnegative_v1_75t
Textual inversion embeddings skipped(3): nartfixer, nfixer, nrealfixer
Model loaded in 2.2s (create model: 0.4s, apply weights to model: 0.5s, apply half(): 0.3s, move model to device: 0.3s, load textual inversion embeddings: 0.6s).
Running on local URL:  http://0.0.0.0:7860

To create a public link, set `share=True` in `launch()`.
Startup time: 9.6s (import torch: 0.7s, import gradio: 0.5s, import ldm: 0.3s, other imports: 0.5s, load scripts: 0.5s, load SD checkpoint: 2.2s, create ui: 0.5s, gradio launch: 4.3s).

### Additional information

_No response_

[JonyTianming]

My problem is the same as yours, but my problem is to generate a picture from tex2img, then send to img2img, and then click the sketch below, and then automatically infinitely enlarge until the picture explodes, and a small sad face appears in the upper left corner. There will be an error prompt before, indicating that the picture has exceeded 280 million pixels, so it is cracked. After I updated it again, there was no error, but the problem is still there. Then many plug-ins that need to use the sketch functionality are affected and cannot be used.

---

Error message before reinstallation:：

After I reinstalled it all, the error message no longer appears, but the problem is still there As can be seen from the video (brush size adjustment lever), the resolution of the picture is automatically magnified until it finally explodes. I didn't adjust the size of the brush, but it changes automatically, indicating that the resolution of the picture has been automatically enlarged in the background.

https://user-images.githubusercontent.com/128905039/235833824-079788d6-83a7-482b-9e34-e1c5734cdfe4.mp4

https://user-images.githubusercontent.com/128905039/235833826-2e033f52-405e-440b-82ec-f0330b92da22.mp4

[karry12138]

My problem is the same as yours, but my problem is to generate a picture from tex2img, then send to img2img, and then click the sketch below, and then automatically infinitely enlarge until the picture explodes, and a small sad face appears in the upper left corner. There will be an error prompt before, indicating that the picture has exceeded 280 million pixels, so it is cracked. After I updated it again, there was no error, but the problem is still there. Then many plug-ins that need to use the sketch functionality are affected and cannot be used.

Error message before reinstallation:：

After I reinstalled it all, the error message no longer appears, but the problem is still there As can be seen from the video (brush size adjustment lever), the resolution of the picture is automatically magnified until it finally explodes. I didn't adjust the size of the brush, but it changes automatically, indicating that the resolution of the picture has been automatically enlarged in the background.

bandicam.2023-05-03.12-35-54-726.mp4 bandicam.2023-05-03.12-34-53-513.mp4

I got the problem similar to you in skectch/inpaint skectch.When sending img to skectch or inpaint skectch,the img will automaticly enlarged by the north-west corner and I find no method to stop this. This is really disappointing.

https://github.com/AUTOMATIC1111/stable-diffusion-webui/assets/66634899/5a440379-ded8-46ff-9d26-732d4001464b

[cain666-4u]

yep, same issue here on windows