Create a collage from anything, then fine tune the AI to give it the right amount freedom to make something you are satisfied with.
Yesterday I wrote an "X" File about using a Visual Prompt from a collage that I made. It described how A.I. models can create original artwork, even when the artist uses 'borrowed' images from the web. It also described how, when an artist makes the A.I. follow their image closely, it can still retain the artist's original concept artwork and message from the image that it worked off of.
Today's post will go in the totally opposite direction! It will show you how your simple collages can take you to far away places when you feed them into the A.I., and then give it a LOT of 'artistic freedom' to work with, instead of following your image too closely, as described in PART ONE of this series.
There's a fine line when you direct the A.I. with your Visual Prompts and then tell it how closely you want it to follow your image, but when you reach that sweet spot, you'll recognize it! Also, as the creator of the image, only YOU can determine where eXactly that sweet spot is!
Today I'll discuss the following image and tell you how I created it...
What a weird image! I can honestly say that I've NEVER seen anything quite like it, (in AI-land), and that's a GOOD THING!
Trying to make 'original' artwork using Artificial Intelligence these days is increasingly more difficult as time ticks on. Sure, the different models and styles that A.I. offers us to create with can pump out some great looking images, but what good are they when they're 'generic' and look like everyone else's images do?
I may be wrong, but it seems to me that as A.I. advances and gets 'better', many of the images it is generating are starting to look the same, regardless of who the artist was that created the image, or the subject matter the image portrays.
Where's the uniqueness from the artists themselves? Has the A.I. really reduced us all to people who just think up ideas, write the prompts, and have it illustrate it for us?
There are ways for artists to circumvent the algorithms that developers have invented to make pretty pictures for the masses, while somehow limiting the artist's personal talents and styles in the process.
Here is my story on how I made an 'original' piece of art that is different and out of the ordinary, even though I used a previous generic piece to make it...
This is the image I started with, isn't it awful? It is generic and boring and lacks any points of interest. It is totally 'pedestrian', eXcept for that eye! The eye sitting there all by itself interested me, so I thought I might like to play with it a bit and see if anything could be done with it.
So I downloaded the image and made a quick cut 'n' paste collage from it. I just cut out the area that I found interesting and pasted it over top of the original image from the center out to the right edge, then I made a mirror image of that on the left side.
OK, that's already more interesting than the original image was!
So I continued on my merry way, changing my written prompt as I went along. I had an issue with noses. The A.I. insisted that this image needed several of them! At one time I think I had 4 noses going along the center of the picture, but I finally got down to 2.
Try as I might, I just couldn't get the A.I. to create an image with just one nose. I tweaked my written prompt, I tweaked my negative prompts, but I think the image was just too weird for the A.I. to comprehend it.
Then it occurred to me that maybe I could just add a mouth over one of the noses and that would take care of it for me, so I went to Google Images and found one.
I got a good laugh from my image, but it caused a new problem for me. Even after several tries, I couldn't get the mouth to look like it belonged. Those ruby red lips were here to stay. I needed a new mouth that had colours that matched more closely, and after working on the above image for a while, I decided its mouth was far too happy anyway.
So back to Google Images I went, this time looking for a frowning mouth that had colours more suitable for my work in progress. I had to play with the colour of the lips a bit, but finally got them 'close enough' and then plastered them on top of a different version of my picture that I thought they would blend in better with.
I had sort of lost my nose on this version, but I thought it was more interesting, especially for the new mouth I had found.
So now I had something I could really work with. I played with the colours of the image, sharpened it several times, changed my prompt to add "brains" and have "no buildings", and did several evolutions/renderings on it.
Obviously, I had to give the A.I. a LOT of freedom while doing my various renderings. Most were done around 30% and I remember at least once going down to only 20%.
I somehow managed to get my nose back, but I also ended up with an eXtra set of eyes, but I thought that was interesting. Also, I saw an entity growing out of the brains section, (at least that's what it looked like to me), so I thought I'd change my prompt once again to work on that.
I finally arrived at a place I liked with my picture. I had a nose, I had a mouth that suited it, and I now had the brain creature too! I downloaded the image, fine tuned it manually, and added my watermark.
It was done!
The name for this piece came quickly for me. First, it's a near endless supply of stimuli for the viewer. Hey, there's a LOT to take in here!
Second, it's an endless supply of stimuli for me! Creating quick collages to work with in A.I. is just wonderful artistically, and this collage was so easy to make, it took me less than a minute!
As weird and strange as this image is, it's really grown on me. I put a bit of work into it, for sure, but its originality from other A.I. artwork out there made it all worth while.
I could have continued working on this image, fine tuning it and making it better and more realistic, but I was tired of it by the time I got to this spot and I thought it looked pretty good as is. Of course the beauty of A.I. is that I can always pick up and continue from where I left off some time in the future.
One thing is for sure though:
This image is certainly NOT 'pedestrian'!
I hope you enjoyed this little creative journey I took you on today,
February 5th, 2024