Principles of controlnet workflow (linux)

[fractal-fumbler]

i am using webui on archlinux

• cuda 11.8.0
• cudnn 8.7.0.84
• python-pytorch-opt-cuda 1.13.1
• xformers 0.0.17+57edbb7.d20230226
• gradio 13.6.2

sd-webui-controlnet version is c16489c
command to start automatic

--xformers --xformers-flash-attention --listen --no-half-vae --enable-insecure-extension-access

i've installed extension via webui and downloaded pruned models from https://huggingface.co/webui/ControlNet-modules-safetensors
then i
a) selecting preprocessor none
b) choosing model control_canny-fp16.safetensors
c) uploading image to image section
and didn't get any result

picture when not working

but when i am clicking several times on image (image is getting green) i am getting result
same behaviour with newly released controlnet model for SD-2.x

picture when partly working

so what i am doing wrong or what additional steps i need to make?

[fractal-fumbler]

on one of videos about controlnet i've noticed that annotators didn't automactically downloaded. can this be an issue?

[fractal-fumbler]

it seems that browser addons screwed up controlnet behaviour.