AI model To edit pics
- Thread starter Anon2334
- Start date
tanstaafl
Engaged Member
- Oct 29, 2018
- 2,263
- 2,856
- 311
You must be registered to see the links
You must be registered to see the links
Go nuts.
Anon2334
Member
- Jun 1, 2024
- 154
- 214
- 121
Civit AI changes the whole image. Is there free easy to use way to just add armpit hair on pics
tanstaafl
Engaged Member
- Oct 29, 2018
- 2,263
- 2,856
- 311
tanstaafl
Engaged Member
- Oct 29, 2018
- 2,263
- 2,856
- 311
Inpaint is the best option if he wants to keep the original picture unchanged. A Lora would redo the original, though I've never tried a Lora while inpainting, lol.
- Jul 4, 2017
- 3,420
- 6,577
- 707
It works just the same as LORA on a whole image. source: me, i use it all the time.
Dark Lord Nova
Active Member
- Nov 7, 2017
- 587
- 1,232
- 405
Funny enough, I totally agree with you about AI—except when it comes to inZOI. That's where we split. Lol. I'm always on Hugging Face and CivitAI hunting for checkpoints and LoRAs. Still, I run into issues all the time. Getting hands and fingers to look right is a pain, and faces can get weird too. Sometimes I just want a full body shot, head to toe, but it only gives me the upper half no matter how clear I am in the prompt. I'm still pretty much a beginner to all this stuff, so I'm figuring it out. Lol.
tanstaafl
Engaged Member
- Oct 29, 2018
- 2,263
- 2,856
- 311
Like I said in my last post on that thread, I don't want Inzoi to fail, I'm just pointing out that it currently isn't doing well. As for AI, I'm still on that upward curve of learning...it's fun.
MissCougar
Active Member
- Feb 20, 2025
- 865
- 3,701
- 262
- Jan 21, 2022
- 11,975
- 23,379
- 913
I don't think that means what you think it meansconstant failures.
@tanstaafl is a pro at it.
MissCougar
Active Member
- Feb 20, 2025
- 865
- 3,701
- 262
Oh no. Did I do something I don't know what I did again? Let me know so I can be in the know and not screw up in the future!I don't think that means what you think it means
MissCougar
Active Member
- Feb 20, 2025
- 865
- 3,701
- 262
tanstaafl
Engaged Member
- Oct 29, 2018
- 2,263
- 2,856
- 311
I'm good at annoying people, writing code, and a few other unmentionables. I'll leave it at that.
- Jul 4, 2017
- 3,420
- 6,577
- 707
- Jul 4, 2017
- 3,420
- 6,577
- 707
It is a built-in functionality in the web clients for diffusion image generators, such as Automatic1111.
You drag in an image, use mouse to draw an area over the top to select what areas should be modified, then run the generator with whatever prompt you want. Only the areas you marked on the original image will be modified.
There's various settings which affect how much of the underlying original image is retained or not, or how carefully the join between the marked area and the surroundings are blended.
Check a youtube video about it, it's easier to show than to explain.
- Jul 4, 2017
- 3,420
- 6,577
- 707
I'm not certain that inpainting does have lower vram requirements - the big memory hog is a model itself and that doesn't change.
I've not needed to investigate that myself as I am GPU rich, but a quick searchengine run found lots of discussion and "guides" on how to run on smaller (<12gb) VRAM cards. I suggest you research and play around.
Alternatively, using one of the online services, which does cost $, is potentially a lot cheaper than buying a new GPU for $$$$. Some of those online vendors have pretty good support for all the useful extensions.