Thoughts on Artificial Intelligence Enhanced Images?
Gogger
Posts: 2,485
I recently discovered Google's Artificial Intelligence painting algorithm, Deep Dreams (er, uh, Google it and you will find it), and have been mesmerized by it. I've been throwing images at it non-stop and am just fascinated by it. When it comes to a finished piece of art, I am a "by whatever means necessary" kinda guy so I mix photography and 3D rendering with graphic design to get something that pleases me (hopefully). Here are a couple examples of what I have come up with. The one image shows the source along with the AI image applied for effect and the result. The rest are just the results. You get the idea. Curious what people's thoughts are on USING Artifical Intelligence to augment "ART"?
[Edited for non-AI assisted spelling - HA HA! ]
Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2025 Daz Productions Inc. All Rights Reserved.
Comments
Does a good job looking like it was painted by whim
the phone video's freak me out more
one of mine with warping, sadly stills only feasable with online filters so short gifs the best I can do otherwise
My take on Artificial Intelligence; well it's artificial, (no one home)
I find it fun and fascinating, and I can't wait to see what happens as the AI's get better. I would probably use it all the time if it were able to handle larger images, which I'm sure they will be at some point...
3D augmented by AI deepfakes as art? I'm sure this would catch on.
It you think that's fascinating, try the demo for this: https://www.mediachance.com/dap/
Creates the painting from scratch, in the style of the master.
Well, it's interesting. Just know this kind of thing is great for experimentation, but since it is difficult to assert copyright on the algorithmic output of a machine (at least for images significantly modified from the original) it has limited use commercially. Copyrights generally require human input during the process beyond simply pushing a button.
Totally cool. Signed up... got my first one baking!
I've definitely played around with the demo! It's fun. Not perfect, by any means, but fun.
Re Deep Dreams, even if you don't use the actual results, the mashups can inspire an actual original piece. It's great for finding inspiration.
I signed up & tried to. I like the results in the one render I used and they were quite nice. I liked the results better than the PS actions Novica has been demonstraiting in another thread.
The results were closing to what a person with talent would do as compared to the typical image filters solutions we see alot of. I guess that shouldn't be surprising since the AI software looked at individual artists art & create AI filters to convert images to that artist's art style. One can even upload their own art with their own style & train the AI to apply their art style to other images.
It's pretty neat.
Oh, I KNOW this one! I use it REGULARLY and am constantly AMAZED! Digital Art Painter is AMAZING.
I was stunned when this is what came of a recent Unicorn 3D Render I did and a photo of some craft feathers fed into the Google Deep Dreams AI painting program - far exceeds anything I could have possibly come up with on my own!
Unicorn Feathers - 3D and Human/AI Collaboration - Erik Pedersen
Where does one find the Deep Dream Generator? I love it.
I'm sort of underwhelmed by this. It looks like a software package wtih some filter themes that it runs... Its not displaying any intelligence that I can see... But maybe I've just had bad luck.
I think it's allowed since it's both not a competitor to DAZ, you asked for a link, and it's 'donate if you want': deepdreamgenerator.com
It's shown me that is can detect different sorts of patterns, eg colors, shapes, lines, ...and then substitute those patterns from one or more pictures into the target image as sub-patterns. There is a lot of parameterization of how much strength to use in overriding an image's original patterns with the patterns detected in the source images(s). Unfortunately there is not a lot of control we as end-users got on altering those patterns we must get get to alter influence strength.
The main way the end user can influence the AI substitutions to create the filters in by supplying a source image(s). eg, I tried different pages of text and music notes and then ran the Deep Dream generation and it did an amusing job. I will probably do it again but slightly change the strengths of the various patterns detected by the AI fin the sourses or filtering through to be applied to the target image to make it a bit more clearly musical notation.
It can do quite decent imitation of watercolors and oil painting and so on, eg, Starry Night style, in my tests.
Digital Art Painter renditions: Sunflower in Van Gogh Starry night style, Crocus on Gruppe style. Moonbridge Monet style. Original photos by me. The program is fascinating. Won't go OT again.
those are super!
Not each image generated is a masterpiece, by any stretch of the imagination, but MAN it sure has given a second life to a lot of my 'test' renders that otherwise would not see the light of day. I imagine it will not appeal to everybody, but my wife and I are both having a lot of fun with it. It is online, it is free, and it claims to be using artificial intelligence. Sure, merely saying it doesn't make it so, but if anybody is going to do it I would imagine Google would be the one. I remember they played with some algorithms where you could upload say, an image of a blueberry and it would identify it and help you find more things like it. Not sure if that flopped, or morphed, but this is one likely off-shoot of that.
Google Image Search is integrated into Google, and this AI is part of how it identifies what is in a picture you may upload. MS Bing also has this, it even has an app where you can use your mobile camera. It doesn't need to take a picture, you just hold the camera over the object in question. Google translate works with AI as well, the mobile app has a feature where you can use your camera and point it at text you wish to translate. Sometimes it works amazingly well. Amazon's app features a camera scanner where you can point your camera at an item and it will try to identify and of course sell it to you. All of these are using some form of AI to help accomplish this. If you watch Nvidia recent conference, the one where they revealed the newest and biggest Quadro and DGX-2, they dedicated a long segment to AI. They showed their GPUs scanning and processing images, and producing real time ray tracing using AI. The AI is the secret sauce of how real time ray tracing is possible.
Its amazing to see what Google is doing with this imaging AI.
God I miss LSD.
That looks amazing!
This is a little OT, but a while ago I used a free program call FotoSketcher to make a funky DeepDream image... (I hated the results)...
But FotoSketcher is kinda useful... It's a image editor that applies a couple of oil color, pastel, watercolor, brushstroke etc. filters to images...
It's pretty good for a free program...
https://fotosketcher.com
Sorry to interrupt.
Have to say, I like Fotosketcher much more. Allot more control too.
Here is a quickie I did using the colored pencil option!
I have used FotoSketcher for years now and even contributed to the author! I have liked it over the years, but pretty much stopped using it once I got ahold of DAP (Digital Art Painter) which is just amazing! But I find each of these programs gives me something the others don't so I am happy to have several arrows in my quiver of paint programs.