First few attemps about ai-friends

[lhucklen]

After subsequent attempts was able to get the app to download the info from hugging. I enjoy inpainting my creations and am looking for a way to achieve this with mobile. Guidance would help in this endeavor.

[Skquark]

That error sounds like you forgot to run the first setup code block in Colab, which initializes the sdd-settings.json file with those prefs. Sounds like you got past that though... I assume you got the web app to launch and you were in the settings.
So Inpainting with a mask image does work, and it's using the new HuggingFace Inpaint model which does better, but you do have to accept the new Inpainting model card that's in the instructions, as well as the new 1.5 checkpoint model file if you haven't already. However, I don't currently have a painting gui built in to draw on top of the init image like I had in the Enhanced Stable Diffusion notebook that used Gradio for that part. Also the file picker to select and upload the init and mask image is still not working, you just have to use the File Manager in the side of Colab to upload the image or select from your mounted google drive, then right click and copy path, and paste it into the image text field.
Hope that helps, I haven't played with this on a mobile, I've been using it on desktop/laptop with nice big screen, but should work on the smaller size with a little more doing. As a side note, I find Colab runs better on a mobile phone with a Firefox or Edge browser rather than Chrome, which is strange since it's a Google product.
Let me know if you run into any other questions, still a work in progress but I believe all the kinks have been worked out and it's ready for usage. Enjoy.

[Capandallama]

Hey, love the ui but how do I download my generations? It won’t let me, all it does is download the website link. Also I have a issue where when i’m using my Ipad 7th generation (15.1), and whenever I touch my screen it updates lists, primarily my prompt list. Which leaves me with a whole lot of blanks that I’d rather not see.

[Skquark]

Your generations are automatically saved into your Google Drive folder, so you don't need to individually download them. I suppose I could add the feature to download the image locally in the browser if you right click or long press on the generated image, but didn't think that was needed since most are happy storing it all in the GDrive folder and downloading locally from there.

Regarding it adding the prompt text to the list when you tap anywhere on your iPad, that's more unexpected. It adds the text to the prompts when you press Enter/Submit, which is equivalent to pressing the Add Prompt button, but it might be sending Submit on the iPad when you tap elsewhere.. I don't know, I've only tested on desktop/laptop browser, and not sure how to fix that bug you experienced without sacrificing the feature.

I also didn't expect the "Run Diffusion on Prompts" FAB button to show up on the top left on iPad instead of floating on the bottom right where it's supposed to be. I'm wondering if this is just another Safari browser incompatibility issue, maybe you can try it on your tablet with Chrome or Firefox browser and let me know if it runs any different..

I'm not sure you meant by "a whole lot of blanks".. You mean those extra prompt lines that say red panda? Those aren't blanks, they're just the same prompt duplicated. The first one on your list shows the parameter overrides on the subtext line, which is the difference to the set default Image Parameters. When you change any of the parameters in that tab, there's a button that comes up to Apply to Prompt List if you want to. Hope that makes sense... Lemme know if that covers it, or if you think there's something I can code in to fix what you ran into.. Right now I don't know what to change since it appears to be an iPad Safari issue. Thanks.