(Poll) Should I make an easy to follow Stable Diffusion guide?

[applecider212]

Stable Diffusion is a new piece of software that can allow many more people to easily make artwork their games and/or improve their artworks through AI. All you have to do is write a prompt of what you want and you'll have an image as your output.

I've been a lurker and this is my first post here, so I unfortunately can't add images for now. You can generate things from scenery to character art, and in the right hands, there are limitless capabilities.

As of right now, the process is confusing and it's hard for a newcomer to find the right resources. Some guides feel too convoluted and everything is everywhere. However, I've been compiling a compendium of resources that I think will let anyone set up and use Stable Diffusion. This is especially directed toward game developers that lack the technical expertise or funding to get good art for their games and people that just want good art (and smut) but can't find any online.

So, would the F95 community like an easy to follow guide on how to setup this software?

 

[FAXz]

AI is great to fool around. more appreciated guide to generate same characters, envs, props in different ways (like camera angle, position, etc)

Main issue to use in in gamedev - draw same object in different env, pose, etc.

 

[Bandi]

I have been struggling so much with the Textual Inversion section of it, trying to train a model on specific characters or artists, and it just garbling the output into something not even objects. Any assistance on that front would be fantastic.

 

[baloneysammich]

It pretty much already exists:

You must be registered to see the links

[to installing

You must be registered to see the links

]

 

[applecider212]

AI is great to fool around. more appreciated guide to generate same characters, envs, props in different ways (like camera angle, position, etc)

Main issue to use in in gamedev - draw same object in different env, pose, etc.

There have been a lot of things that came out since this post, some of which are methods to do what you're asking

I have been struggling so much with the Textual Inversion section of it, trying to train a model on specific characters or artists, and it just garbling the output into something not even objects. Any assistance on that front would be fantastic.

There can be many things causing this to happen, like too high/low learning rates, vectors, etc. I'd need more info to help you

I'm working on the guide right now, so stay tuned for when that comes out