• Sign up or login, and you'll have full access to opportunities of forum.

Fallenmystic's AI-assisted works

Go to CruxDreams.com
Thanks! And if you like blond subs I suppose you may also like this, which is an alternate version of the render I posted above:

View attachment 1317884
Excellent work fallenmystic. This image (ans also the alternate one) brings out the humiliation of the chained naked girl so vividly. Her face is filled with shame and looks so stunning.
Maybe she is just looking for a little bit of privacy to pee or shit. But she is having to do it under the full glare of the spectators who are enjoying every bit of it.
Or maybe she has just been gangraped. And she is fearful as to how soon they will gangrape her again.
This is a picture that brings out many possibilities. And that is the beauty of it !
 
I just did a quick render from a random doodle I drew. I use a few different workflows but starting from a crude sketch is among the most straightforward and easiest. I'll post additional images to share the workflow.

This was my initial doodle (made with Krita):

sketch.png

And I developed it using the txt2img tool of Stable Diffusion:

00276-20230726005058.png 00438-20230726012830.png

This is the control map image (lineart - realistic) I made from the above intermediary renders:

lineart.png

And here's the final result after upscaling (img2img) and postprocessing (using Darkroom):

00478-20230726023100.jpg

(I know the dragan effect is rather extreme but I don’t have time to fix it at the moment. After all, it’s a quick render ;))
 
Last edited:
I just did a quick render from a random doodle I drew. I use a few different workflows but starting from a crude sketch is among the most straightforward and easy. I'll post additional images to share the workflow.

This was my initial doodle (made with Krita):

View attachment 1327722

And I developed it using the txt2img tool of Stable Diffusion:

View attachment 1327723 View attachment 1327724

This is the control map image (lineart - realistic) I made from the above intermediary renders:

View attachment 1327726

And here's the final result after upscaling (img2img) and postprocessing (using Darkroom):

View attachment 1327742
Thanks for the info.
I used to doodle myself and this is an interesting approach to me.
 
Getting more realistic!
I hate to admit it, but the main reason is I didn't try to do much with the composition this time. The more complex the scene becomes and the more precise I try to direct it, it tends to be less realistic. They look photoreal because they're simple portrait renders with minimal guidance from me (i.e. just a prompt and some LORAs).

It's something I'm still trying to overcome. :)
 
Inspired by this post, I tried to make a render with a similar theme this weekend. Unfortunately, my installation of SD.next got broken after an update so I can't render photorealistic images with it anymore.

I had spent more than half a day before I managed to find a workaround to produce this:

View attachment 1324721

I won't make another render until the problem gets resolved. The developers of the client have been very active so I expect I won't have to wait very long.
Great! And it´s great to have you back here. Please more artworks with hard labour. I loved your story about the prison.
 
I hate to admit it, but the main reason is I didn't try to do much with the composition this time. The more complex the scene becomes and the more precise I try to direct it, it tends to be less realistic. They look photoreal because they're simple portrait renders with minimal guidance from me (i.e. just a prompt and some LORAs).

It's something I'm still trying to overcome. :)
Interesting
 
Another remaster of my early AI render. I wanted to do a quick test of a new LORA that provides an easy way to make dark/high contrast scenes, for which task Stable Diffusion is known to struggle.

I liked the result enough to make a 2k render version, which is shown below:

View attachment 1319891

P.S.: I really like the combination of Krita + Stable Diffusion + Darktable. I fixed mangled toes in Krita and applied a few post-processing effects (e.g. sharpen, vignette, colour correction, etc.) in Darktable.
That's one scary-looking pussy! Damn! There's a sarlacc in there!!!:eek::eek::eek:
 
As I mentioned in another thread, I'm still waiting for ControlNet support to arrive in SD.next, which is essential for my workflow. Until then, I can't make anything serious with SDXL, so I've been generating test renders with SDXL to find an optimal setting for my future works instead.

Tonight, however, I happened to create renders that I thought were good enough to be included in this thread. They are still test images, so I didn't apply any post-processing or detailing/error-fixing steps:

00313.jpg 00307.jpg 00306.jpg 00298.jpg 00276.jpg 00275.jpg
 
Last edited:
On a side note, I just realised that the daily limit of the number of images can be posted in a single thread is 5, not 6 as I thought it was. From my previous posts, it looks like I got confused about the matter recently so I’ve been consistently making posts with 6 images as of late.

Sorry, it wasn’t intentional and I will adhere to the rules the next time I post images.
 
On a side note, I just realised that the daily limit of the number of images can be posted in a single thread is 5, not 6 as I thought it was. From my previous posts, it looks like I got confused about the matter recently so I’ve been consistently making posts with 6 images as of late.

Sorry, it wasn’t intentional and I will adhere to the rules the next time I post images.
I’ve been browsing old threads and it seems that there are quite many duplicates about same images in different or even same threads.

I don’t know if the disk space is a cost or just traffic, but wouldn’t be bad idea to run some image similarity analysis and replace duplicates with links.
 
I’ve been browsing old threads and it seems that there are quite many duplicates about same images in different or even same threads.

I don’t know if the disk space is a cost or just traffic, but wouldn’t be bad idea to run some image similarity analysis and replace duplicates with links.
I don’t think I posted duplicate images except for a couple of exceptions (like those included in the first post of this thread.) So I assume you mentioned of such a tendency shown in other threads.

I agree that it’d be nice if one of those who have administrative privileges on the server could have sufficient time and knowledge to do that. But since this is a thread I made for sharing my renders and getting feedback about them, I’d refrain from discussing the matter here at length.
 
I don’t think I posted duplicate images except for a couple of exceptions (like those included in the first post of this thread.) So I assume you mentioned of such a tendency shown in other threads.

I agree that it’d be nice if one of those who have administrative privileges on the server could have sufficient time and knowledge to do that. But since this is a thread I made for sharing my renders and getting feedback about them, I’d refrain from discussing the matter here at length.
Sorry for misunderstanding, I didn’t mean you. Mostly duplicates seems to be pictures from the net.
 
After learning that SDXL can generate legible texts, unlike its predecessor, I tried to depict realistic brands. And here's my result:

View attachment 1363986

I'm glad that now I don't have to rely on an external program for the task as it was the case before.
Where else would you place her identification so it's perfectly visible? Logical placement. Easily seen.

So, she's a Delta ("D-class"). Only fit for harsh outdoor servitude.

What about Alphas or Betas?

Immediately made me think of a harsher version of Brave New World.
 
Back
Top Bottom