Modified AI art thread - I'll share the thread if anyone wants to add theirs

amo

Well-known member
Messages
125
I can't think of AI art without thinking of commercial art, something to sell. So these are all products. Frankly, I try to "recycle" almost all the art I make for sale somewhere, just because I need the income. (And sales help me justify the time I spend working on art to my family.) Anyway, I don't want to come across as spamming or anything. It's just the nature of the thing I am making.

DALL-E images are almost too small to be usable for anything when they're generated (1024x1024 pixels). I experimented with some online services that enlarge one way or another, and some seem okay, but I'm happier right now making collages of smaller images. That adds at least a little more of my artistic input, too. (Sometimes there's plenty of artistic input already, depending on what DALL-E gave me to start with.)

So now I have this "inventory" of images, so far in two categories, steampunk fish and amigurumi things found in space. (Why didn't I think of spaceships before now?) I'm sure I will make more products with the images now that I have them, but I don't plan to post any more of these two sets to this thread.

I accidentally hit "save" before I was finished, so edited to add the images. Guess that's all I wanted to say for now. Comments, criticism, whatever welcome.
 

Attachments

  • Right Eight Steampunk Goldfish Redbubble Generic with subtle bubbles (Custom).png
    Right Eight Steampunk Goldfish Redbubble Generic with subtle bubbles (Custom).png
    835.3 KB · Views: 101
  • AOS for Twitter.png
    AOS for Twitter.png
    233.5 KB · Views: 114
The fish are very cool! (y) The repetition of the patterns reminds me of wall paper for a kids room, wrapping paper, or cover design for cell phones. No idea how you break into those markets, but somebody does. :)
 
The fish are very cool! (y) The repetition of the patterns reminds me of wall paper for a kids room, wrapping paper, or cover design for cell phones. No idea how you break into those markets, but somebody does. :)
That's a great idea! I've got them on Redbubble designs, but I know there's a market for wallpapers and fabrics . . . Have to see if I can find an "in"
 
That's a great idea! I've got them on Redbubble designs, but I know there's a market for wallpapers and fabrics . . . Have to see if I can find an "in"
I am guessing, on further reflection, that DALL-E's involvement will eliminate these designs from consideration, but I guess I won't know unless I ask.

Edited to add: No, looks like WeaveUp will take my designs. If they comply in all the other ways; haven't checked that yet.
 
Last edited:
I like both of those. Amigurumis in space was a great idea! I've never tried to sell anything, but I have been playing with the AI image generators. So we can post AI generated images in this thread?

Furry Night, in collaboration with artbreeder:
furrynight.png
 
  • Like
Reactions: amo
I like both of those. Amigurumis in space was a great idea! I've never tried to sell anything, but I have been playing with the AI image generators. So we can post AI generated images in this thread?
Sure! I plan to modify all mine (although I may show the originals sometimes; I have in other threads), but the more the merrier, modified or not!
 
How do you mean "modify?" Like performing some Gimpcraft on them, maybe? I've been doing a bit of that.
Mine are digitally manipulated in all kinds of ways, crop, cut/paste, dodge/burn, etc. The steampunk goldfish were 8 DALL-E-generated fish images plus two background images. I have no idea how many images there are in the amigurumi space one. A lot. The ones I've posted in this thread are essentially collages. The other one I have on the site here somewhere is not a collage, but it still underwent significant revision.

The reason I say "modified AI" is, one, because I have modified them, and two, I don't want people (on Redbubble for example) to steal my image because DALL-E images are in the public domain. I wouldn't give AI credit at all, except it's part of the terms of use for DALL-E. I may eventually move to another AI because of that. But I'm just trying to call them the same thing everywhere I share them.
 
The reason I say "modified AI" is, one, because I have modified them, and two, I don't want people (on Redbubble for example) to steal my image because DALL-E images are in the public domain. I wouldn't give AI credit at all, except it's part of the terms of use for DALL-E. I may eventually move to another AI because of that.
Ahhh... I understand now. Smart! I assume I should mention artbreeder for some legal reason, so that's why I keep mentioning the site. I'm doing some Gimp stuff for sure on many of the ones I keep.
 
Ahhh... I understand now. Smart! I assume I should mention artbreeder for some legal reason, so that's why I keep mentioning the site. I'm doing some Gimp stuff for sure on many of the ones I keep.
Each AI seems to have different terms of use. I believe a United States court case declared AI images to be public domain, so I don't know how much weight the terms of use hold in the U.S., but I don't want OpenAI to cut me off :)
 
Thanks for sharing your thoughts. You've got me thinking that it might be fun to make postcards if the AI and I come up with something I particularly like. There's a place nearby where I could leave some to sell.

Apropos of nothing at all... Warhol meets van Gogh, furry render:
vanandy.png

artbreeder.com
 
  • Like
Reactions: amo
Thanks for sharing your thoughts. You've got me thinking that it might be fun to make postcards if the AI and I come up with something I particularly like. There's a place nearby where I could leave some to sell.

Apropos of nothing at all... Warhol meets van Gogh, furry render:
View attachment 23858
artbreeder.com
DALL-E images are pretty small, but they might be big enough for postcards. I paid for bigjpg after trying a few different image enlargers. So far so good. DALL-E spits out 1024x1024. What size are Artbreeder's images?
 
Thought I'd share some "swing and a miss" AI results. I had heard that, for some reason, AI is good at putting things in glass, like ships in a bottle or a snowman in a snowglobe, etc. And I had an image come to mind of a hollow robot filled with fireflies. Well, I never got it :)

Maybe I'll keep trying. I have more failures than I posted here. I think I failed to save some, because I didn't know I was going to share.
 

Attachments

  • NightcafeStudio.jpg
    NightcafeStudio.jpg
    37.1 KB · Views: 84
  • glass robot with a jar full of fireflies for its torso, digital art.jpg
    glass robot with a jar full of fireflies for its torso, digital art.jpg
    13.2 KB · Views: 91
  • glass robot with a jar full of fireflies for its body, digital art.jpg
    glass robot with a jar full of fireflies for its body, digital art.jpg
    9.8 KB · Views: 84
  • glass robot with a jar full of fireflies for its body, 3d render 2.jpg
    glass robot with a jar full of fireflies for its body, 3d render 2.jpg
    8 KB · Views: 90
  • glass robot jar full of fireflies, digital art.jpg
    glass robot jar full of fireflies, digital art.jpg
    11.2 KB · Views: 91
  • glass robot jar full of fireflies, digital art 2.jpg
    glass robot jar full of fireflies, digital art 2.jpg
    12.3 KB · Views: 86
Other "swing and a miss" incidents when I hoped to improve on my own unsatisfactory art. DALL-E only made it worse :)
 

Attachments

  • Screenshot 2022-08-12 205030.jpg
    Screenshot 2022-08-12 205030.jpg
    57.2 KB · Views: 79
  • Screenshot 2022-08-12 205116.jpg
    Screenshot 2022-08-12 205116.jpg
    66.3 KB · Views: 75
Regarding my previous post, maybe I'm biased. Do you think DALL-E made my art better?
Here's another one I put on Redbubble. The original colors were a little different, and one of the eyes was really funky. The image here is the finished product.
Happy Bunny with Heart Redbubble (Large).png
 
The time is coming when I'm going to have to get a "real job" if I don't get more freelance work (some art related, some not) and art/photography/design sales, so I'm really hitting this hard between other projects. If I had worked as hard on traditional art as I did yesterday on digital/AI art, I would also likely have three images. It was at least a 12-hour day, and I usually work pretty quickly.

Anyway, yesterday, I did the image above and two more (all with the help of OpenAI's DALL-E). Don't really have too much to say about Grumpy Girl:
Grumpy Girl (Large).png
But I did think it's worth sharing a little more about the technique I used for I'm Just Here for the Cookies:
I'm just here for the cookies (Large).png
Because I digitally combined these images to arrive at my final image, using elements of each:
DALL·E 2022-08-13 09.24.19 - a bear with a cookie jar.png
Screenshot 2022-08-13 124501.jpg

Screenshot 2022-08-13 124949.jpg

I originally started by opening them all in layers and erasing what I didn't want, but it turned out to be simpler to rectangle select the parts I wanted to keep and paste them together, with a little digital clean-up at the end.

I've never been much of a digital artist, but I'm learning an awful lot from these kinds of projects. Skills I'm sure will have broader application.
 
I originally started by opening them all in layers and erasing what I didn't want, but it turned out to be simpler to rectangle select the parts I wanted to keep and paste them together, with a little digital clean-up at the end.
I like the images you're coming up with, and it's generous of you to share technique. I'll definitely try the above.

Relatedly* perhaps, yesterday I started using the paths tool in the Gimp to roughly select large areas of unwanted background: enclose a big area with the paths tool, choose "select from path," and then give it the old Ctrl-x. That left little areas that were easy to clean up and sped up the process considerably for me.

*Not a real word 😦
 
I like the images you're coming up with, and it's generous of you to share technique. I'll definitely try the above.

Relatedly* perhaps, yesterday I started using the paths tool in the Gimp to roughly select large areas of unwanted background: enclose a big area with the paths tool, choose "select from path," and then give it the old Ctrl-x. That left little areas that were easy to clean up and sped up the process considerably for me.

*Not a real word 😦

Happy to share any techniques. I'm not an experienced digital artist, though. Just fumbling my way through.

I used to understand paths a little better than I do now, but I guess I never had a good grasp on how to use them because I've forgotten everything I ever knew. I need to take the time and re-learn, though, because I know they can be very useful.
 
Back
Top