Or I should perhaps say specifically when using demo_web with an initial image. It's very difficult and confusing to use.
I can't find any pattern to when I'm supposed to click "send to Streamlit". Is it before or after I create a mask? Because the generator starts up pretty randomly when you mess around in the gui. Also it's unclear to me when the prompt is supposed to be entered, so many times I've gotten the initial image back from the generator. Sometimes when a mask has been entered, the prompt makes no diffierence and the image is returned with extremely subtle and minor changes.
I think this part of the project is in dire need of some usability testing and improvements.
Also it would be great to be able to generate multiple images for a prompt and have them returned as a series of images, in stead of having to click and wait for each individual image.
Or I should perhaps say specifically when using demo_web with an initial image. It's very difficult and confusing to use.
I can't find any pattern to when I'm supposed to click "send to Streamlit". Is it before or after I create a mask? Because the generator starts up pretty randomly when you mess around in the gui. Also it's unclear to me when the prompt is supposed to be entered, so many times I've gotten the initial image back from the generator. Sometimes when a mask has been entered, the prompt makes no diffierence and the image is returned with extremely subtle and minor changes.
I think this part of the project is in dire need of some usability testing and improvements.
Also it would be great to be able to generate multiple images for a prompt and have them returned as a series of images, in stead of having to click and wait for each individual image.