AI [AI Art] - Show Us Your AI Skill *NO TEENS*

4.80 star(s) 6 Votes

FoolsBeGone

Well-Known Member
May 21, 2023
1,039
3,128
so I tried one of those free AI conversion things online.....I was very disappointed but it might just be that I am not using the right words or terminology to give the AI what it needs to make the picture work.

Here are the two pictures I got
prompt1.png prompt.png
here are the prompts I used:
side view of a five foot tall white female between 17 and 20 years old with 35-24-36 measurements, green eyes, freckles, full lips, cute nose, long wavy auburn hair reaching the small of her back, looking at the viewer, wearing a tight fitting teal sleeveless silk shirt with a leather sash, midriff exposed, firm abs, tight leather pants and thigh high boots, carrying a five foot long, 2 1/2"diameter oak staff the staff touching the ground and her hand placed 3/4 of the way to the top of the staff, holding the staff with her right hand, casting a fireball from her left hand, standing near a stream with water running over a large boulder in the middle of the stream, an oak forest in the background. trout jumping out of the stream, owl landing in a tree in the back ground, muscular black panther sitting beside her......

Yeah it got SOME of that but nowhere close to all of it
obviously I had to keep it pg because those sites will go nuts if you go anywhere near nsfw. Optimally I want her to be showing more cleavage and I want a frigging FIREBALL damnit!!!!
 
  • Like
Reactions: mams3425

FoolsBeGone

Well-Known Member
May 21, 2023
1,039
3,128
ok so I gave up on the fireball and chose a few different things and I actually like the way this one came out
prompt2.png
they didn't get the leather shorts and the shirt isn't silky as I asked and I see no blue flowers. Also I think the sash is confusing it cause its got sash everywhere.

I would definitely love to see her in a VN though. I think she is beautiful
 

Artiour

Member
Sep 24, 2017
314
1,318
so I tried one of those free AI conversion things online.....I was very disappointed but it might just be that I am not using the right words or terminology to give the AI what it needs to make the picture work.

Here are the two pictures I got
View attachment 2864004 View attachment 2864005
here are the prompts I used:
side view of a five foot tall white female between 17 and 20 years old with 35-24-36 measurements, green eyes, freckles, full lips, cute nose, long wavy auburn hair reaching the small of her back, looking at the viewer, wearing a tight fitting teal sleeveless silk shirt with a leather sash, midriff exposed, firm abs, tight leather pants and thigh high boots, carrying a five foot long, 2 1/2"diameter oak staff the staff touching the ground and her hand placed 3/4 of the way to the top of the staff, holding the staff with her right hand, casting a fireball from her left hand, standing near a stream with water running over a large boulder in the middle of the stream, an oak forest in the background. trout jumping out of the stream, owl landing in a tree in the back ground, muscular black panther sitting beside her......

Yeah it got SOME of that but nowhere close to all of it
obviously I had to keep it pg because those sites will go nuts if you go anywhere near nsfw. Optimally I want her to be showing more cleavage and I want a frigging FIREBALL damnit!!!!
that's stable diffusion 1.5, it's a pain in the ass but a necessary step for learning, keep looking around, you'll sure find websites with other models
 

FoolsBeGone

Well-Known Member
May 21, 2023
1,039
3,128
that's stable diffusion 1.5, it's a pain in the ass but a necessary step for learning, keep looking around, you'll sure find websites with other models
are there certain phrases I should use to fine tune it or is that basically about as good as 1.5 gets?
 

Artiour

Member
Sep 24, 2017
314
1,318
are there certain phrases I should use to fine tune it or is that basically about as good as 1.5 gets?
you could use artists styles from this
there is fine tuning to be done but needs a lot of knowledge in photography field ,art field, movie making, comics, or even art softwares. (this part is a never ending journey)
but you're doing good so far, the full body view usually mess up the face badly
here is something to try and learn something from if you're up for experimenting
You don't have permission to view the spoiler content. Log in or register now.
 
Last edited:
  • Wow
Reactions: FoolsBeGone

FoolsBeGone

Well-Known Member
May 21, 2023
1,039
3,128
you could use artists styles from this
there is fine tuning to be done but needs a lot of knowledge in photography field ,art field, movie making, comics, or even art softwares. (this part is a never ending journey)
but you're doing good so far, the full body view usually mess up the face badly
here is something to try and learn something from if you're up for experimenting
You don't have permission to view the spoiler content. Log in or register now.
Holy SHITE....thats a lot. I thought I had been detailed but I see now that I wasn't even close-gonna give what you said a try and see what turns up
 

FoolsBeGone

Well-Known Member
May 21, 2023
1,039
3,128
I basically copied and pasted some of what you wrote just to see what would happen and added it to what I had written earlier and this is what it gave me
Prompt(wouldn't allow negative without purchase)
side view of a five foot tall white female between 17 and 20 years old with 35-24-36 measurements, green eyes, freckles, full lips, cute nose, long wavy auburn hair reaching the small of her back, looking at the viewer, wearing a tight fitting teal sleeveless silk shirt with a leather sash, midriff exposed, firm abs, tight leather pants and thigh high boots, carrying a five foot long, 2 1/2"diameter oak staff the staff touching the ground and her hand placed 3/4 of the way to the top of the staff, holding the staff with her right hand, casting a fireball from her left hand, standing near a stream with water running over a large boulder in the middle of the stream, an oak forest in the background. hyperrealistic, hyper detailed, photorealistic :: a masterpiece, incredible composition, amazing depth, imposing, meticulously composed, 16k resolution concept art, Insanely Detailed, Intricate dress, gorgeous Beautiful perfect Face, Artgerm, WLOP, tom bagshaw, magali villenueve; Photograph Taken on Nikon D750, Intricate, Digital Illustration, Scenic, Hyper-Realistic, Hyper-Detailed, 16k, cinema 4D, ethereal sun rays iridescent beautiful colorful ethereal
here is the pic it popped out. Its late now:sleep: and I am a bit frustrated being my very first day of trying this so fresh start tomorrow I think - Thanks for the help its MUCH appreciated:)
50e4d1b614734b179360b9fdea5d599a.png
actually I tried on last on and it came out pretty good although it still didn't do everything I asked it, I think it was a good start
image3.png
 
Last edited:

Dagg0th

Member
Jan 20, 2022
303
3,005
are there certain phrases I should use to fine tune it or is that basically about as good as 1.5 gets?
There is a lot you can do, add Loras and embeddings for example; use different models. If you use SD1.5 you definitely need to add to the prompt:
"(masterpiece, best quality, high quality), hyperdetailed, highres, high detail, intricate detail, HDR, 8k wallpaper, sharp focus, detailed background, realistic, photorealistic, perfect hands"

Negative prompts:
"(worst quality, low quality, normal quality:1.2), lowres, monochrome, grayscale"

that prompt alone will greatly increase the quality of your image

00007-1061861824.jpg
 

FoolsBeGone

Well-Known Member
May 21, 2023
1,039
3,128
There is a lot you can do, add Loras and embeddings for example; use different models. If you use SD1.5 you definitely need to add to the prompt:
"(masterpiece, best quality, high quality), hyperdetailed, highres, high detail, intricate detail, HDR, 8k wallpaper, sharp focus, detailed background, realistic, photorealistic, perfect hands"

Negative prompts:
"(worst quality, low quality, normal quality:1.2), lowres, monochrome, grayscale"

that prompt alone will greatly increase the quality of your image

View attachment 2864304
She is gorgeous!
 
  • Like
Reactions: Kass69

Synalon

Member
Jan 31, 2022
228
668
Holy SHITE....thats a lot. I thought I had been detailed but I see now that I wasn't even close-gonna give what you said a try and see what turns up
Kiss Test.jpg

This is from a test I did so its still in the early stages, hence the washed out look, and dead eyes, but I find for photos theres kind of a prompt workflow that works best for me.

1) The style of photo, such as glamour, documentary, candid etc.
2) What the photo is of, such as a woman, man, motorbike, landscape etc.
3) Add some details that you want in for certain.
4) Add details you would like but don't mind if they are missing.
5) The type of photo, such as full body, portrait etc.
6) The background and details you want in it.
7) The type of lighting, shadows you want. To start with you can use cinematic lighting.
8) The type of camera you want it shot on. ( Helps if you know cameras )
9) The type of camera film you want to use. ( Also helps to know the difference )
10) The photographer you want it in the style of.

I've found with some models, and SDXL for certain you can reduce a lot of it to something like directed by (directors name) and cinematography by (cinematographers name). That will cut out a lot of messing about with cameras, vignettes, lighting.

Negatives prompts are sometimes more important than positive prompts, but also look up embeddings, some of the models like to use negative embeddings and their own Vae.

Also watch some youtube videos on inpainting, quite often you will get an image you really like apart from a minor flaw and inpainting can help fix that.
 

Sharinel

Active Member
Dec 23, 2018
613
2,586
so I tried one of those free AI conversion things online.....I was very disappointed but it might just be that I am not using the right words or terminology to give the AI what it needs to make the picture work.

Here are the two pictures I got
View attachment 2864004 View attachment 2864005
here are the prompts I used:
side view of a five foot tall white female between 17 and 20 years old with 35-24-36 measurements, green eyes, freckles, full lips, cute nose, long wavy auburn hair reaching the small of her back, looking at the viewer, wearing a tight fitting teal sleeveless silk shirt with a leather sash, midriff exposed, firm abs, tight leather pants and thigh high boots, carrying a five foot long, 2 1/2"diameter oak staff the staff touching the ground and her hand placed 3/4 of the way to the top of the staff, holding the staff with her right hand, casting a fireball from her left hand, standing near a stream with water running over a large boulder in the middle of the stream, an oak forest in the background. trout jumping out of the stream, owl landing in a tree in the back ground, muscular black panther sitting beside her......

Yeah it got SOME of that but nowhere close to all of it
obviously I had to keep it pg because those sites will go nuts if you go anywhere near nsfw. Optimally I want her to be showing more cleavage and I want a frigging FIREBALL damnit!!!!
You're putting to much shite into the prompt :) Remember that stable diffusion only has so many 'tokens' to employ to generate the image, you are overloading it with rubbish and then wondering why it's missing out on stuff. An owl and a panther because why not? may as well add an autobot and Luke Skywalker in the background as well.
Quick glance over the prompt -
5 foot tall woman. Why? It's not as if you are putting her next to a 6 foot 9 inch warrior so you can see the height differences. I don't stare at the left picture above and say achtually with the size of the rocks in the stream I can calculate this woman is 5'2" - what an outrage. Bin it, not needed
between 17 and 20 - if you can tell the difference between a generated 19 year old and one that is 20, you're a better man than I Gunga Din. Just say 18 year old white female.
five foot long 2 1/2" diameter oak staff - Let me fix that for you - "staff". there you go. You'll be lucky if you don't get the fishing rod you got in the left hand image or that staff that doesn't match up because of her 6 fingered hand in the right (AI is notoriously bad at hands, you want to hide them in as many pics as you can, i'd take out the bit for 3/4 way up the staff)

I'm starting to get myself worked up into a Gordon Ramsey style rant here, but that's unfair as it's your first day doing it. Set expectations though, you won't get all of what you want in the pic
 

Synalon

Member
Jan 31, 2022
228
668
Ok here is my next try (it generated an extra hand that wasn't even connected to the body so I just cropped it to show the good part. The Robe was supposed to be black silk and ankle length but that didn't take either.
View attachment 2867063
She is beautiful though if I do say so myself.
Are you using an online version of Stable Diffusion, or are you using a version you downloaded?
 

FoolsBeGone

Well-Known Member
May 21, 2023
1,039
3,128
it didn't take long to figure out that whatever I create is going to be sorely limited since I am unwilling to pay the cost for all the so called "free" ai art generators. You make your picture then it starts telling you too many words or cant use that prompt you need to sign up for an account first etc.. etc.. etc..... But I was never going to be someone that tried to do this for anything other than enjoyment of making something beautiful so I wont pay. Anyway here are todays attempts lol
View attachment 2867231 View attachment 2867235
Are you using an online version of Stable Diffusion, or are you using a version you downloaded?
I am using
 
4.80 star(s) 6 Votes