Give me a prompt and i'll make an image

[Bohandas]

Oh..you should never look at hands in AI portraits..and you should certainly avoid counting them or the associated digits.

To be fair most human artists qn't draw hands right either. That's why cartoon characters are generally drawn with less than five fingers, it's simpler to draw them that way

 

[Gammadoodler]

To be fair most human artists qn't draw hands right either. That's why cartoon characters are generally drawn with less than five fingers, it's simpler to draw them that way.

Yeah that's fair.

I think the difference is that the human artists who can draw photorealistic or near photorealistic portraits of people are mostly able to draw hands that don't bear close resemblance to the creature from "The Thing".

I suspect there is some aspect of model training with hands that is significantly more difficult than with faces and other body parts. Possibly something related to the diversity of postures and angles that hands can be drwn or photographed in.

In either case it's, one of the interesting and pretty common uncanny valley aspects of AI Art.

It can make some beautiful faces, but don't count the fingers.

 

[Bagpuss]

I've used Perchance that can now link to Stable Diffusion to knock up a very quick character portrait generator.

Character Generator

Create Characters with Stable Diffusion.

perchance.org

If you scroll down you can see some examples using the same prompt, but different genre and style modifiers.

"a university professor in his late forties, overweight and wearing glasses, with wild hair and a beard, in a library" (at least until people add their own to the public gallery)

 

[Bagpuss]

New slightly improved version.

Character Generator

Create Characters with Stable Diffusion.

perchance.org

You can now specify a range, close-up, middle or full body. It's not 100% effective, it does prefer middle to close up if you mention lots of facial details and if you keep mentioning shows and boots, it will go for full even if you have said close up.

Also added a field to put in negative prompts, and added Old Photograph which works great with the CoC genre. If anyone wants to suggest different styles, genres of games or prompts they have found that are good at getting better refined pictures in the styles and genres already mentioned let me know.

 

[trappedslider]

Female cyberpunk assassin

 

[trappedslider]

It gave her the lightsaber with no prompting

work flow:
Female cyberpunk assassin

Negative prompt: deformed, bad anatomy, disfigured, poorly drawn face, mutation, mutated, extra limb, ugly, disgusting, poorly drawn hands, missing limb, floating limbs, disconnected limbs, malformed hands, blurry, ((((mutated hands and fingers)))), watermark, watermarked, oversaturated, censored, distorted hands, amputation, missing hands, obese, doubled face, double hands,canvas frame, cartoon, 3d, ((disfigured)), ((bad art)), ((deformed)),((extra limbs)),((close up)),((b&w)), wierd colors, blurry, (((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))), Photoshop, video game, ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, mutation, mutated, extra limbs, extra legs, extra arms, disfigured, deformed, cross-eye, body out of frame, blurry, bad art, bad anatomy, 3d render

Steps: 128, Sampler: DPM++ 2M SDE Karras, CFG scale: 7, Seed: 4185545525, Face restoration: CodeFormer, Size: 768x768, Model hash: 5dd1c2014e, Model: cyberrealistic_v21, Version: ## 1.4.0

 

[Bagpuss]

The "cyberpunk" prompt seem to love purple hair for some reason.

 

[Paul Farquhar]

I think the difference is that the human artists who can draw photorealistic or near photorealistic portraits of people are mostly able to draw hands that don't bear close resemblance to the creature from "The Thing".

Human portrait artists frequently avoid showing the hands at all. That's why muffs are so popular in portraiture.

But AI learns from copying humans. So if human artists avoid showing hands, or draw them badly, the AI will learn to do likewise.

 

[trappedslider]

The "cyberpunk" prompt seem to love purple hair for some reason.

I took the image and uses img2img to do green hair and this happened

then went back to txt2img and went with Female cyberpunk assassin with blue hair

 

[trappedslider]

I like the Jacket on this one