Sunday, 23 November 2025

Reconnecting with AI ART

I've gotten behind on AI ART while absorbed by printmaking, and a lot of innovations have come out recently:

PROFILES
in Midjourney

My Profile on Midjourney is -- krrrl:



This week (November 19, 2025Midjourney came out with "PROFILES."  Subscribers can highlight their best work on the front page for others to see.  However you have to be a subscriber to see other people's profiles -- and you can also see their archive of all the other images they have generated.  In short, this is an easy way to see what everyone else is doing.  I like that, but a lot of other people want to keep their images private.

UNIQUE STYLE?

Is my Midjourney style unique?  I used OBS Studio to film my Midjourney Profile as I scrolled down, and posted that video online as seen above.  I wanted the overview to see how my style compared to that of what other subscribers were generating. My images lean towards Moebius' style -- but some of my drawing style still manages to surface in the generations.  Or are they all in the Midjourney style?

 

I have noted that style is critical in past blog posts: 

A good piece of art is when the style complements the content.




2D to 3D
and into PARTS

Now the 2D to 3D programs will also break the 3D model down into parts:

(YouTube): AI Generate 3D in parts - worth it? Honest review:
PARTCRAFTER

PartCrafter is probably the best program for breaking the model down.  It will convert a 2D image into a 3D file -- and then separate the parts so that they can be printed with a 3D printer.  



I first evolved one of my drawings into a more mechanical and articulated figure with AI in KREA:

I then animated that figure with AI in Midjourney to imagine it moving:


Finally I uploaded the 2D image to PartCrafter to make a 3D image out of it:

My AI figure
separated into different colored parts


META

Meta would also transform a 2D into into 3D, but it would not let me download the result:

(YouTube): Turn ANY Photo in 3D Models with Meta's Sam 3D - 100% FREE


***

I was not sure how to separate the parts for 3D printing. So I tried a more mechanical image and figured it out.  A while back I converted one of my figure drawings into a machine:

prompt: "3D model created in Solidworks"



Then I exploded the machine with AI in by animating it in Midjourney:

prompt: "Exploding 3D model into parts
 that was created in Solidworks"


I then uploaded the 2D image above to PartCrafter, to make a 3D image that separated into individual parts.  The result did break the 3D model into individual colored parts, but was huge -- 70 MB.  It would not open in Paint3D:



I had to upload the result into Blender to separate some of the parts. However Blender did not show any colors.  The exploding model was now 370 MB. I rendered a single image after I moved some of the parts:


Then I colored that exploding image in Nano Banana, and animated it with SORA 2:



COMBO ANIMATION

The Mayan and Machine images above
combined into an animation in Midjourney


DEEP DREAM GENERATOR
MUSIC

Deep Dream Generator came out with a Music Generator a while back (Q3) apparently, using Google's Lyria.  I tried it for free, but got censored for asking for Chopin. It worked the second time:




In a blog post last December 2024 -- Future of AI ART -- I discussed some of the drawbacks of AI ART programs last year.   However things have really improved in the last year.

However it is still touch-and-go.  I uploaded last year's Google generated image of flying javelinas, and  asked Google's new Nano Banana Pro to transform it into a photorealistic scene -- however it hardly changed anything:

prompt: "Can you convert this image into a photorealistic scene"

Later on I got a lot better results from Google's new Nano Banana Pro however.



No comments:

Post a Comment

Note: only a member of this blog may post a comment.