Fun with AI and no politics

Ok, so how is it, there is a Hobbits from the 1978 Lord of the Rings cartoon dressed for the holidays?  Well, that would be me playing with OpenAI's new integrated photo editor.  This is how it ended, but where did it starts?

The other day, I was watching nerdy YouTube (this would be my go to as a tech junkie) and came across a discussion about OpenAI's new text to image and image manipulation capability.  Prior to this change, editing services were part of a few offerings, but for the most part you could only get one iteration of an image and couldn't continue to edit the same photo.  

Well, I had a little fun after work and figured I would give it a try.  I quickly scanned some pictures and decided to go with a picture of my wife and I from Disney World in 2021.  It was our 20th anniversary and we were heading out to a Disney Halloween Party (note: these are more for the kids, but the short lines were worth the admission.  It was also the International Food/Wine Festival at Epcot - banging).  This was before I knew I had cancer, but to be honest, I knew something was wrong.  However, back to the fun stuff.

The original picture is below.  It has bad shadows, etc...  Since there was so much going on with the hats, glasses, Mickey Mouse ears, and Godzilla shirts.  Hell, in my case I have two Godzilla shirts.


Since, the Ghibli style requests seemed to be all the rage, I figured I would give it a shot.  I hadn't seen them all over as... I'm not on social media any longer and that has been wonderful.  So of course, I went with that style first and HOLY SHIT, THIS IS AWESOME.  The model, pulled the glasses, the facial features, the clothing, including adding dinosaurs (Godzilla is copyrighted), and of course didn't put any Mickey or Minnie's on the ears as also, copyrighted.  However, I think it did a damn good job with all that was going on in that picture and how washed out the lighting was.  Notice, it even put the tattoo from my arm in.


I then told it to make us look like a Hanna Barbera cartoon from the 80's (note, it is using the prior image so I expect every generation to lose details, but it did  a pretty good job)


Then, to really test, I asked to make us like Rankin Bass Christmas specials, slim us down (the Hanna Barbera went a bit overboard.  Now here is the kicker, I told the model to remove the dinosaurs from the shirts and to add red nosed reindeer and Santa.  Seriously, this is pretty damn amazing.


Finally, I had to South Park the picture - seriously well done.


Then I took a walled sized picture (that I scanned or something) into the model and told it to take out the subjects and put them into an oil painting that is reminiscent of the portraits of royalty.  Considering the low quality of the original picture, I am pretty impressed with the results.  Except that painting guy is way more attractive than me.




Honestly, if we don't destroy ourselves first, AI is going to be doing a lot of the heavy lifting of our work.

For now, I'm good (MMGA)
- JJ

 

Comments

Popular posts from this blog

No more social media

A break from the nostalgia

There are some things you can't use your teeth for