stable-diffusion-webui

stable-diffusion-webui copied to clipboard

Is there an existing issue for this?

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

What would your feature do ?

grab an image from a webcam and use it for processing then grabbing a new image when done.

Proposed workflow

1. Go to .... the img2img tab
2. Press .... an not yet excisting "use camera" button
3. Press .... an not yet excisting "loop" toggle ui element if desired
4. Press ... Generate

Additional information

https://www.youtube.com/watch?v=g75ipNzWnbo the video here gives an impression of the process im sure lots of us would enjoy

Yes this would be a great experience

This feature world be usefull for me.

+1

You can easily make the UI take a webcam as input by modifying this one line

https://github.com/AUTOMATIC1111/stable-diffusion-webui/blob/5ab7f213bec2f816f9c5644becb32eb72c8ffb89/modules/ui.py#L707

change your local stable-diffusion-webui/modules/ui.py to say webcam i.e.

init_img = gr.Image(label="Image for img2img", elem_id="img2img_image", show_label=False, source="webcam", interactive=True, type="pil", tool="editor", image_mode="RGBA").style(height=480) 

then restart ./webui.sh. Tested and works on MacOS.

You can easily make the UI take a webcam as input by modifying this one line

https://github.com/AUTOMATIC1111/stable-diffusion-webui/blob/5ab7f213bec2f816f9c5644becb32eb72c8ffb89/modules/ui.py#L707

change your local stable-diffusion-webui/modules/ui.py to say webcam i.e.

init_img = gr.Image(label="Image for img2img", elem_id="img2img_image", show_label=False, source="webcam", interactive=True, type="pil", tool="editor", image_mode="RGBA").style(height=480) 

then restart ./webui.sh. Tested and works on MacOS.

thanks, I can see the webcam but when I hit 'generate' I get an error "AttributeError: 'NoneType' object has no attribute 'split' "

+1

You can easily make the UI take a webcam as input by modifying this one line

https://github.com/AUTOMATIC1111/stable-diffusion-webui/blob/5ab7f213bec2f816f9c5644becb32eb72c8ffb89/modules/ui.py#L707

change your local stable-diffusion-webui/modules/ui.py to say webcam i.e.

init_img = gr.Image(label="Image for img2img", elem_id="img2img_image", show_label=False, source="webcam", interactive=True, type="pil", tool="editor", image_mode="RGBA").style(height=480) 

then restart ./webui.sh. Tested and works on MacOS.

sounds like OP means they'd like a live webcam input that loops like a video, rather than taking a simple image and generating based off of that

You get "AttributeError: 'NoneType' object has no attribute 'split' " because it doesn't return an image unless you press the camera icon.

Worked for me with the following change. Which combined with "generate forever" option gives a constantly updating render.

init_img = gr.Image(label="Image for img2img", elem_id="img2img_image", streaming=True, show_label=False, mirror_webcam=False, source="webcam", interactive=True, type="pil", tool="editor", image_mode="RGBA", height=opts.img2img_editor_height)

You get "AttributeError: 'NoneType' object has no attribute 'split' " because it doesn't return an image unless you press the camera icon.

Worked for me with the following change. Which combined with "generate forever" option gives a constantly updating render.

init_img = gr.Image(label="Image for img2img", elem_id="img2img_image", streaming=True, show_label=False, mirror_webcam=False, source="webcam", interactive=True, type="pil", tool="editor", image_mode="RGBA", height=opts.img2img_editor_height)

thanks that worked, how are you toggling between normal image upload and webcam? it'll be great if there was a new tab for the input type to be webcam

You get "AttributeError: 'NoneType' object has no attribute 'split' " because it doesn't return an image unless you press the camera icon. Worked for me with the following change. Which combined with "generate forever" option gives a constantly updating render.

init_img = gr.Image(label="Image for img2img", elem_id="img2img_image", streaming=True, show_label=False, mirror_webcam=False, source="webcam", interactive=True, type="pil", tool="editor", image_mode="RGBA", height=opts.img2img_editor_height)

thanks that worked, how are you toggling between normal image upload and webcam? it'll be great if there was a new tab for the input type to be webcam

I'm not. Yeah to do that would require a new tab. Which shouldn't be too hard to implement.

thanks, I tried mucking around with it, the easiest way is just to replace the input in the sketch tab since I dont use the sketch function anyway, works great

Many helps! But here is a new question, if I want to change the source to virtual camera such as OBS, how can I set it?