CREATE YOUR AI CUM SLUT ON CANDY.AI TRY FOR FREE
x

Daseson

Well-Known Member
Modder
Dec 30, 2020
1,154
8,371
View attachment 2412008 View attachment 2412009 View attachment 2412010 View attachment 2412011 View attachment 2412012

ControlNet was released for Stable Diffusion, so I gave it a spin. Better than the img2img stuff, but still misses the mark when it comes to this.

They do have another model for coloring anime line drawings, but they want to "evaluate the risks" before releasing it. Here's a sample:

View attachment 2412013
As always beautiful... although there are still deformities :(
 

Shadesishere

Well-Known Member
Modder
Dec 5, 2020
1,638
12,487
View attachment 2412008 View attachment 2412009 View attachment 2412010 View attachment 2412011 View attachment 2412012

ControlNet was released for Stable Diffusion, so I gave it a spin. Better than the img2img stuff, but still misses the mark when it comes to this.

They do have another model for coloring anime line drawings, but they want to "evaluate the risks" before releasing it. Here's a sample:

View attachment 2412013
It's great... until you spend more then 10 seconds looking for errors... then you see that the hands 90% of the time are horrifying... which is strange considering it can do ears a-okay 95% of the time.
 

mario6714

Member
Jun 22, 2020
392
231
View attachment 2412008 View attachment 2412009 View attachment 2412010 View attachment 2412011 View attachment 2412012

ControlNet was released for Stable Diffusion, so I gave it a spin. Better than the img2img stuff, but still misses the mark when it comes to this.

They do have another model for coloring anime line drawings, but they want to "evaluate the risks" before releasing it. Here's a sample:

View attachment 2412013
Can you teach me how to use "ControlNet was released for Stable Diffusion"? thanks
 

kengador

Newbie
Aug 6, 2020
63
165
I hate AI drawn images now that my brain knows what to look for. My brain sees the deformations first. Fucked up limbs, genitalia , sometimes eyes and mouths,ears. The more crowded a picture is the more deformation you see. Sometimes even non human objects get deformed. Everything melting into each other. Like the blob genetic mutation at the end of Akira. Nah i'll just wait 10 more years before i bother with ai stuff.
 
  • Like
Reactions: Eterai and Mczipper

hurrdurrderpderp

Active Member
Apr 4, 2020
901
3,368
I hate AI drawn images now that my brain knows what to look for. My brain sees the deformations first. Fucked up limbs, genitalia , sometimes eyes and mouths,ears. The more crowded a picture is the more deformation you see. Sometimes even non human objects get deformed. Everything melting into each other. Like the blob genetic mutation at the end of Akira. Nah i'll just wait 10 more years before i bother with ai stuff.
That's because most of people posting these don't bother with inpainting, they're just dumping to general public what machine spew out and that's it. Some artists are already using AI generated images as base and fiddle with inpainting until it looks better - just then fixing composition errors, adding detals, decorations and particles, creating fitting background and making image really shine with blending layers together.
Before that AI stuff artists were using models in blender to get poses right, some other prefered wooden models (these are quite great and fun to play as well) or even notorious and dishonorable act of tracing photographs (literally every artist in existence done that).

Tl;dr you still need to put some elbow grease and in meantime, in year or two, AI will learn what 90% of self proclaimed artists can't wrap their heads around - why exactly anatomy is important.
 
  • Thinking Face
Reactions: akapp

FishermansFriend

New Member
Jan 9, 2022
6
42
As always beautiful... although there are still deformities :(
Yeah, it's still not there I'm afraid, which is why your coloring work is even more impressive.

It's great... until you spend more then 10 seconds looking for errors... then you see that the hands 90% of the time are horrifying... which is strange considering it can do ears a-okay 95% of the time.
Ears are much simpler. They rarely bend or twist and are always relatively in the same place.

Can you teach me how to use "ControlNet was released for Stable Diffusion"? thanks
There's plenty of tutorials on YouTube :)

Tl;dr you still need to put some elbow grease and in meantime, in year or two, AI will learn what 90% of self proclaimed artists can't wrap their heads around - why exactly anatomy is important.
Yep! It's not a replacement for artists. It's a tool like any other tool, like photoshop or blender. You need to build up skill to use it and traditional artists skills help there. Sadly, my skills are not artistic, so I treat AI more like gacha. Pull the lever, see what comes out. If it's good, it's good. If it's bad, pull again.

But I'm still taken back by the speed of improvement for open source AI. If you compare these to my earlier img2img post, they are much better already in a span of a month.
 

JakeTapper

Member
May 17, 2020
211
582
As someone who uses stable diffusion locally I can say if you're trying to do img2img and it starts transforming things too much you just want to put the bias more towards the original image. Otherwise you're gonna sit there looking for that 1/500 output that doesn't have some obvious error. That said, if you get something that's almost perfect with a small issue somewhere you can always inpaint that specific spot until it's fixed.

Anyways, for a project like this the consistency of coloring by hand is going to be much better than stable diffusion. Not to mention SD throws off the entire 'vibe' (for better or worse but it just feels different). It's certainly interesting seeing what HD re-renders of scenes can look like though.
 

ーはかせ

Member
Sep 25, 2021
111
304
Hello,

Testing my own mod I've seen a lot of interesting bugs.

You don't have permission to view the spoiler content. Log in or register now.

If you have found any additional errors please let me know so I can correct them.
It has been some time since I checked the forums and I almost missed it because it was not linked in the opening post 'Extras', but I'm loving your color mod! I usually like to play the game without color mods, but I really liked the way you colored this. Looks like a new playthrough for me :p

As someone commented earlier, her hair was supposed to regain color after the 'adventure', but my opinion is it makes sense to keep the hair color that you used. If someone really wanted to have her blonde hair after the 'adventure', maybe have a 'version B' of the mod to manually replace, but that would be just too much effort for a small detail.
 
  • Like
  • Heart
Reactions: akapp and Daseson

Daseson

Well-Known Member
Modder
Dec 30, 2020
1,154
8,371
It has been some time since I checked the forums and I almost missed it because it was not linked in the opening post 'Extras', but I'm loving your color mod! I usually like to play the game without color mods, but I really liked the way you colored this. Looks like a new playthrough for me :p

As someone commented earlier, her hair was supposed to regain color after the 'adventure', but my opinion is it makes sense to keep the hair color that you used. If someone really wanted to have her blonde hair after the 'adventure', maybe have a 'version B' of the mod to manually replace, but that would be just too much effort for a small detail.
Thank you very much for your beautiful words.

The hair at the moment is in a process of change since for reasons that I could not control I lost all my work, so I am redoing it.

And regarding the fact that my mod is not referenced in the initial post, I would think it is because as it is constantly changing, Shade is waiting for me to finish it to reference it.

Note: Around here testing my mod I find more and more errors ಥ_ಥ
 

Settum

Well-Known Member
Jan 10, 2019
1,010
922
Yeah, can't really post a screen without context. In the post the image just above shows two other sister games with the commentary that "imouto" games took the ranking spots... thus, it seems humanity is looking for a younger sister, after all.
 
4.70 star(s) 97 Votes