this post was submitted on 19 Aug 2023
9 points (90.9% liked)

Stable Diffusion

4324 readers
16 users here now

Discuss matters related to our favourite AI Art generation technology

Also see

Other communities

founded 1 year ago
MODERATORS
 

I'm new to SD. Not exactly my forte (art or AI), but I'm having fun playing with it. I had an idea the other day, I think SD is the solution, but I don't know how to do it.

I wanted to try putting my son into a pic. He's into space right now and I'd love to make some art of him as an astronaut. Something better than a cut and paste of his head onto an astronaut cutout. How do I do that with SD? I can't figure it out.

top 8 comments
sorted by: hot top controversial new old
[–] CloverSi@lemmy.comfysnug.space 8 points 1 year ago* (last edited 1 year ago) (1 children)

I would train a LoRA, here's a guide (also has links to other guides at the bottom)

[–] loopedcandle@lemmynsfw.com 3 points 1 year ago (1 children)

Ok . . . so I have a Mac M2. And have been using Diffusion Bee to avoid the terminal. But I'll see if I can figure it out.

[–] Stampela@startrek.website 4 points 1 year ago

Much simpler way, even if not as cool: try https://apps.apple.com/app/id6444050820 that is free and great. First download a model, if you already have one you like just use the Civitai.com link, otherwise there's a selection of them and I'd try... Realistic Visions? The 8 bit one saves you a little space. Then use the photo icon in the bottom to import a photo, you want something with the face clearly visible. Use the eraser to get rid of everything other than the face. Now try prompting for an astronaut! Should work already but you can try adding Controls: Tile, Inpainting and Pose specifically.

Performance wise you can go at the bottom of the left side, click on the cpu looking icon and select (if it's not selected already) Use CoreML: Yes and for the CoreML compute units go with All.

My M2 Mini takes a dozen seconds for a normal generation like that. With the Control modules and Inpainting is going to be a little slower obviously but hey, it's a fast app.

[–] strawman@lemmy.dbzer0.com 4 points 1 year ago

The easiest way is take a picture of your son and "inpaint" the parts that are not him with the style you want. The more difficult way would be to train a LORA. Here is a blog by someone who did exactly that. https://www.shruggingface.com/blog/self-portraits-with-stable-diffusion-and-lora

[–] FredericChopin_@feddit.uk 3 points 1 year ago

I found this repo that may help here

If you’re still stuck though, I’ll happily figure out how to do and give you the steps to take. I’m not home at the moment so will be sometime over the weekend.

[–] JackbyDev@programming.dev 2 points 1 year ago

In-painting. You can selectively generate some parts of an image. So you can take pics of your son and generate the parts he isn't in.

[–] IvanOverdrive@lemm.ee 2 points 1 year ago (1 children)

Roop is what you are looking for. MUCH simpler than creating a LORA. https://youtu.be/8SWon8erHZw?si=XKFDvXh0sJhF6I1D

And if you haven't subscribed to Olivio Sarikas's YouTube channel, you should

[–] PipedLinkBot@feddit.rocks 2 points 1 year ago

Here is an alternative Piped link(s): https://piped.video/8SWon8erHZw?si=XKFDvXh0sJhF6I1D

Piped is a privacy-respecting open-source alternative frontend to YouTube.

I'm open-source, check me out at GitHub.