- Aug 21, 2018
- 2,193
- 3,591
- Aug 21, 2018
- 2,193
- 3,591
No Wednesday, but at least b/w and colors ^^
That flag in her hair sucks - but so close to what I wanted ^^
The other one is cool, but really creepy
But I got there ...
PS: I used the F222 model for all of them.
That flag in her hair sucks - but so close to what I wanted ^^
The other one is cool, but really creepy
But I got there ...
PS: I used the F222 model for all of them.
Last edited:
- Feb 13, 2019
- 2,694
- 18,287
I wanted to give this a try so I followed
I may have to reinstall some stuff, which isn't an issue since I have almost no time invested in this. I have tried sd1.5, sd2.0 and sd2.1.
I would like to continue because this seems very interesting, although I am not sure if I would call it art, my analytical self sees it more like writing code and running it through a compiler that just happens to output an image.
You must be registered to see the links
video on installing stable-diffusion and automatic1111 and had it running for a few minutes, now all I get is an error message and am looking for a solution. I seem to always get this error
Code:
AttributeError: 'NoneType' object has no attribute 'process_text'I would like to continue because this seems very interesting, although I am not sure if I would call it art, my analytical self sees it more like writing code and running it through a compiler that just happens to output an image.
- May 4, 2019
- 456
- 1,634
Did you install any extensions? You mentioned you had it running, so it worked and then the error appeared after restarting? I found that installing extensions is not that straightforward and can cause errors like that on startup, but then only the extension was broken, not the entire webui.
- Feb 13, 2019
- 2,694
- 18,287
After exercising my best Google MOJO I found that some people fixed their implementations by deleting the venv subdirectory and then allowing webui-user.bat to reinstall everything.
This certainly is more intensive that installing and running DAZ for the first time.
- Aug 21, 2018
- 2,193
- 3,591
Yes, but good coding can be art too
It is just fun to finde the right combination of prompts, weighting, negatives and even the right model or merger of models to get a result you have in mind.
Just getting something is quite easy and fun, but to get a specific result is some kind of "art".
The post frome me above yours took quite some effort to get that result - but I had fun doing it and in the end that is the most important thing for me with this
PS: I run SD 1.5, haven't tried 2.0.
- Oct 29, 2017
- 812
- 2,197
Yes, but good coding can be art too
It is just fun to finde the right combination of prompts, weighting, negatives and even the right model or merger of models to get a result you have in mind.
Just getting something is quite easy and fun, but to get a specific result is some kind of "art".
Guys, guys, just don't go there, there is already 1 or more threads talking about that in the forum. It doesn't make any sense to destroy this productive thread with that topic.
- Oct 29, 2017
- 812
- 2,197
I agree, any traditional artist with an open mind will definitely kill it with a tool like SD.
- Aug 21, 2018
- 2,193
- 3,591
Not what I wanted, but some really nice looking women. ^^
I have to use inpainting more and try to get good with that - currently I mainly use text2img
I have to use inpainting more and try to get good with that - currently I mainly use text2img
- Feb 13, 2019
- 2,694
- 18,287
This is basically my first attempt after using "cat" as the testing prompt. I downloaded Hentai-Diffusion 1.7 and followed their directions on
You must be registered to see the links
. This was img2img. Setup is Automatic1111, HD-17, RTX3090, R9-5950X. I experimented with the sampling method and I recall this image used DPM++.
You don't have permission to view the spoiler content.
Log in or register now.
- Aug 21, 2018
- 2,193
- 3,591
My "ice phase" - tried some variations with women and ice
"Failed" but nice looking ...
Now we are talking ...
Then the next idea - ice statue with human hair ...
"Failed" but nice looking ...
Now we are talking ...
And then to end this - a first try on "diamond"
"Failed" but nice looking ...
Now we are talking ...
Then the next idea - ice statue with human hair ...
"Failed" but nice looking ...
Now we are talking ...
And then to end this - a first try on "diamond"
- Oct 9, 2019
- 713
- 3,700
This is really fucking addictive.
Everything was img2img of my old renders, I like 6 and .6 as the weights but I'll play around with them.
Everything was img2img of my old renders, I like 6 and .6 as the weights but I'll play around with them.
You don't have permission to view the spoiler content.
Log in or register now.
- Oct 29, 2017
- 812
- 2,197
Trying to work with some of those hot details that you always see on japanese hentai...
Last edited:
- Oct 9, 2019
- 713
- 3,700
WORLD CUP OF WRESTLING
I decided to create the world cup of wrestling. For no reason other than I was having a great time algorithming muscle girls with big breasts. I hope you enjoy. This took way longer and was slightly more difficult than I thought. All renders are on 40 steps in euler a because I can render each image in 10 seconds. I start by doing small batches of test renders, lock in my prompts and then generally do 2-5 batches of 20. Then I pick the best ones. Generally, I'll find between 2-10 good ones. Usually around 4 or 5 so I have a ton of them still saved that are really good, but I thought these were better.
Bear in mind, most of these utilize tropes and stereotypes. I also had to keep the art style consistent, so I did use an artist's name (said artist has also been lambasted for using the same face on every female he draws in the past... and he kind of does.) And no inpainting so if the hands look weird, yeah, I didn't fix that on any.
Enjoy the 24 nations.
No USA because this model apparently doesn't have the "props" I wanted to use.... but honorable mentions:
I decided to create the world cup of wrestling. For no reason other than I was having a great time algorithming muscle girls with big breasts. I hope you enjoy. This took way longer and was slightly more difficult than I thought. All renders are on 40 steps in euler a because I can render each image in 10 seconds. I start by doing small batches of test renders, lock in my prompts and then generally do 2-5 batches of 20. Then I pick the best ones. Generally, I'll find between 2-10 good ones. Usually around 4 or 5 so I have a ton of them still saved that are really good, but I thought these were better.
Bear in mind, most of these utilize tropes and stereotypes. I also had to keep the art style consistent, so I did use an artist's name (said artist has also been lambasted for using the same face on every female he draws in the past... and he kind of does.) And no inpainting so if the hands look weird, yeah, I didn't fix that on any.
Enjoy the 24 nations.
You don't have permission to view the spoiler content.
Log in or register now.
No USA because this model apparently doesn't have the "props" I wanted to use.... but honorable mentions:
You don't have permission to view the spoiler content.
Log in or register now.
- Aug 24, 2022
- 309
- 734
As a huge fan of
I've played around with mimicking their style.
I am getting pretty happy with the results.
(There's a few small imperfections I need to clean up still, namely the right strap of her panties denoised into a sort of strap on her back that merged with her dark hair, should be pretty simple to fix though)
You must be registered to see the links
..I've played around with mimicking their style.
I am getting pretty happy with the results.
(There's a few small imperfections I need to clean up still, namely the right strap of her panties denoised into a sort of strap on her back that merged with her dark hair, should be pretty simple to fix though)
- Aug 17, 2019
- 5,847
- 9,315
Not a dig, to be absolutely clear, but I've gotta ask...
Where's the skill in AI art, exactly?
Where's the skill in AI art, exactly?
- Aug 24, 2022
- 309
- 734
As a general user of something like SD - personally I don't think there's much "hard skill" required at all.
Anyone can fairly quickly set it up and begin training checkpoints, it's only a factor of cost / hardware.
I would rate the iteration process on generating AI images at about the same complexity as a DAZ render with kitbashed pre-rigged assets / poses / blendshapes / lighting & render options.
I'd argue that's kind of the point..