Your AI-enhanced images (tutorial with your pictures)

[Prof Sai]

It looks like you can create some interesting backgrounds with it (to insert either rendered characters or real people).

It's not as good as just using Google photos. I know how to use negative prompts (they go in a separate window) but haven't gotten it to lower the viewpoint or make any SWs yet.

[who did this ?]

floor level view of a dollsized tiny woman standing on the floor of an a out-of-scale giant kitchen,Fujifilm XT3,studio quality. insanely detailed, lush detail, perfectionism, max detail, 4k uhd, HQ

negative prompt:
doll,toy,puppet,insect,bugs,ugly,duplicate, morbid, mutilated, out of frame, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, ugly, blurry, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, out of frame, ugly, extra limbs, bad anatomy, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, mutated hands, fused fingers, too many fingers, long neck



fullbody wide angle floor level view of a doll sized woman standing on the floor of an giant kitchen,Fujifilm XT3,studio quality. insanely detailed, lush detail, perfectionism, max detail, 4k uhd, HQ,no toy,no doll,real women,ultra-wide angle

negative:
doll,toy,puppet,insect,bugs,ugly,duplicate, morbid, mutilated, out of frame, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, ugly, blurry, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, out of frame, ugly, extra limbs, bad anatomy, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, mutated hands, fused fingers, too many fingers, long neck,doll,puppet,toy.duplication

[Granelex]

Wow these AI generated images you make are getting better and better Who. You are really refining the process of making these images. Great work.

[who did this ?]

Wow these AI generated images you make are getting better and better Who. You are really refining the process of making these images. Great work.

Thanks. This is the first time it worked without any source image.And if the little ladies would turn around, i would be 100% happy.

[Prof Sai]

:: Jealousy ::

[bobvan]

Cool

[bobvan]

:: Jealousy ::

I agree

[Dr.Minimizer]

Sigh, all that time learning to use Photoshop to hack my way through photocomics and now this newfangled technology is going to make me obsolete. Now I know how the Pony Express riders felt.

[Sumguy14]

Sigh, all that time learning to use Photoshop to hack my way through photocomics and now this newfangled technology is going to make me obsolete. Now I know how the Pony Express riders felt.

You... will never be obsolete.

[Prof Sai]

floor level view of a dollsized tiny woman standing on the floor of an a out-of-scale giant kitchen,Fujifilm XT3,studio quality. insanely detailed, lush detail, perfectionism, max detail, 4k uhd, HQ

negative prompt:
doll,toy,puppet,insect,bugs,ugly,duplicate, morbid, mutilated, out of frame, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, ugly, blurry, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, out of frame, ugly, extra limbs, bad anatomy, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, mutated hands, fused fingers, too many fingers, long neck

DefusionBee actually rejects this as being too long to work with, so I guess the problem isn't just me. That is a relief, looking for different tools is next then. It is also strange that you need to continually remind the AI that you don't want all those negative things - you'd think it would have some common sense defaults.

It's hard to think verbally about visuals like this. I hope someone makes an AI that generates 3d models that can then be adjusted as such. It would be so much easier to click and drag a limb with the mouse then explain to the AI what the problem is.

[who did this ?]

floor level view of a dollsized tiny woman standing on the floor of an a out-of-scale giant kitchen,Fujifilm XT3,studio quality. insanely detailed, lush detail, perfectionism, max detail, 4k uhd, HQ

negative prompt:
doll,toy,puppet,insect,bugs,ugly,duplicate, morbid, mutilated, out of frame, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, ugly, blurry, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, out of frame, ugly, extra limbs, bad anatomy, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, mutated hands, fused fingers, too many fingers, long neck

DefusionBee actually rejects this as being too long to work with, so I guess the problem isn't just me. That is a relief, looking for different tools is next then. It is also strange that you need to continually remind the AI that you don't want all those negative things - you'd think it would have some common sense defaults.

It's hard to think verbally about visuals like this. I hope someone makes an AI that generates 3d models that can then be adjusted as such. It would be so much easier to click and drag a limb with the mouse then explain to the AI what the problem is.

You have to see it like this: You can create virtually anything that makes no sense at all and the AI is like a chef who always has all the ingredients in stock and could prepare anything, because he has basically seen all the recipes and knows in principle what belongs in which dish. Unfortunately, he has no sense of taste, can't smell and you have to tell him that everything that comes out of an oven or a pan must not be charred. And if you want a cake, he doesn't have to use all five ovens at the same time.
And that's the current problem with stable diffusion.
I once wanted to see a submarine in a lake. I puzzled for a long time why SD didn't show me a submarine. Because submarines dive. It wasn't until I typed in surfaced submarine....

[bobvan]

Without a doubt a skill he has put hours into and has got the hang of it which gives him a unique perspective. Having said that some images look stunning, I still think rendering still does a better job in some cases. BUT! this is early days for AI perhaps one day you will not need to counteract with what one does not want. I do have to get off my lazy @ss and give this a try @ some point. Who is kind enough to do his best to help others

[who did this ?]

Sigh, all that time learning to use Photoshop to hack my way through photocomics and now this newfangled technology is going to make me obsolete. Now I know how the Pony Express riders felt.

As a photoshop user you could see it as another tool to complement your workflow. There is already a plugin for stable diffusion that allows you to add or change things within photoshop. The photoshop own AI in the latest versions is the same thing as all the other AI-solutions (changing emotions or age of a face). So it is already there, they call it "neural filter".But stable diffusion gives you more control.
If one of your models wears a superhero costume, you had only one chance to shoot a certain pose and that costume looks not like you planned it, you could draw a mask over it and the AI could add more wrinkles, smooth it .. whatever.
It will take some years before any AI can create anything like your fantastic Photoshop work.
(And .. just look at the hands and feet. After creating virtually thousands of AI-images, there is not one of them with perfect AI-generated fingers. So the Pony Express riders are safe!)

[bobvan]

Sigh, all that time learning to use Photoshop to hack my way through photocomics and now this newfangled technology is going to make me obsolete. Now I know how the Pony Express riders felt.

o the Pony Express riders are safe!)

[coolguy69]

"..."

Wow!!! These are fantastic! I wonder how they would render a cartoon

[who did this ?]

Water turns into beer ..

[Granelex]

Thanks for showing us this tool Who and fantastic work bobvan. Though my work cant compare to either of yours the tutorial on how to use this tool ha sbeen handy. It's really allowed some pictures of little ladies to get a little biit of refinement. Though it does seem to warp them a bit. I do think poser and daz can make for better shots but it has been fun to play with some pictures.




and when put through NMKD Stable Diffusion GUI





It did more Vanessa's shoes from wedges to heels but it has given her a more realistic look.

[who did this ?]

Thanks for showing us this tool Who and fantastic work bobvan. Though my work cant compare to either of yours the tutorial on how to use this tool ha sbeen handy. It's really allowed some pictures of little ladies to get a little biit of refinement. Though it does seem to warp them a bit. I do think poser and daz can make for better shots but it has been fun to play with some pictures.

and when put through NMKD Stable Diffusion GUI

It did more Vanessa's shoes from wedges to heels but it has given her a more realistic look.

Thanks, Granelex, thats what i wanted to see.Now there is the time to play with all the different samplers, keywords, cameras, resolutions for the output image and so on ..
(try to add seductive as a keyword, the AI is very sexist and adds breasts )

[bobvan]

Thanks for showing us this tool Who and fantastic work bobvan. Though my work cant compare to either of yours the tutorial on how to use this tool ha sbeen handy. It's really allowed some pictures of little ladies to get a little biit of refinement. Though it does seem to warp them a bit. I do think poser and daz can make for better shots but it has been fun to play with some pictures.


and when put through NMKD Stable Diffusion GUI



It did more Vanessa's shoes from wedges to heels but it has given her a more realistic look.

Nice to see others starting to get it

[coolguy69]

Shrunk message down

Wow!! What a fascinating result. I love the second one because the height difference is most clear

Would love to see the AI interpret her not able to reach it, of course

Thanks for testing!

[coolguy69]

I wonder what the AI would do here - the shading/lighting being perfected would be pretty cool

[bugmaker]

It's fun to play with Stable Diffusion. The whole thing just needs some more time to mature. Especially the hands and feet are a nightmare. Pretty impressed with what it can achieve so far.

Here's two SW experiments of mine I did a couple of weeks ago. I used Automatic1111 UI with a mix of the sd 1.5 model and F111.



Prompts:
naked (((45yr woman))) ((busty)) ((bbw)) ((round boobs)), ((((standing on top of giant office desk)))), in front of (((giant office supplies and giant computer ))), ((high angle)), (((facing camera))) (((scared))) ((wide open eyes)), ((long blonde hair))

Negative:
(((deformed face))), disfigured, ((deformed)), deformations, [blurry], (out of frame), long neck, long arms,, (bad hands), zoomed in, (deformed arms), ((extra limbs)), extra nipples, (extra legs), (extra arms), (extra fingers), amputee, pants, bad anatomy, bad proportions, disfigured, mutation, furry, mutated, (extra_limb), extra limbs, penis, text, watermark, shoes, socks



Prompts:
a confused worried tiny shortstack blonde 45 year old woman wearing baggy oversized bathrobe, clothes way too big, giant kitchen_floor, giant kitchen, giant high towering kitchen counters, ((highly detailed)), facing camera, full body shot, huge natural breasts, (((high angle))), photo 4k HQ, very detailed, nikon d850

Negative:
monochrome, blurry. lowres. lowrez. low resolution, shoes, monochrome, penis, mutant. mutation. deformed. Signature. Watermark. Text. Contortionist. Missing Limbs. Extra Limbs. Missing Fingers. Extra Fingers

Note that these are the results of doing a whole bunch of img2img. So no chance of you getting the same results, even if I shared the seed and the settings. Still, the prompts might come in handy.

[bobvan]

Ha ha neat

[xrattletrapx]

Thanks a LOT, Who!

I feel like we're learning a lot as a community thanks to your initiative.
We're a...smaller...community, and I'm sure the others are deep in AI now.
This is our...little...way of standing out!

[who did this ?]

It's fun to play with Stable Diffusion. The whole thing just needs some more time to mature. Especially the hands and feet are a nightmare. Pretty impressed with what it can achieve so far.

Here's two SW experiments of mine I did a couple of weeks ago. I used Automatic1111 UI with a mix of the sd 1.5 model and F111.

...
Note that these are the results of doing a whole bunch of img2img. So no chance of you getting the same results, even if I shared the seed and the settings. Still, the prompts might come in handy.

I love it. AI can be scary at times and you will see things you can not make unseen.Heads growing out of arms, Bodys melting into other bodys, this is an horror authors dream tool. I like the look of the first sw

Thanks a LOT, Who!

I feel like we're learning a lot as a community thanks to your initiative.
We're a...smaller...community, and I'm sure the others are deep in AI now.
This is our...little...way of standing out!

In a few months we will see so many incredible sw pictures, it will be crazy. And posting them on DA will be fun, there are already so many "artists" posting this stupid "NoAI" sign...