.png)
-effectapp.png)
KREA Realtime prompt:
"sculpture of a man in style of Bernini"
KREA Realtime prompt:
"sculpture of a man in wire"
KREA Realtime prompt:
"paper sculpture of a man in style of Bocioni"
FREESTYLE
Tuesday, 14 April 2026
More RIGGING and MOTION
I found that MESHY was really good for converting 2D images into 3D objects -- and also rigging and animating them too. They do have a free tier:
I then turned the T-POSE image into a 3D model in MESHY -- but using the MESHY 5 model, which allowed me to download the GLB file:
NOTE: MESHY did a lot better job converting 2D-to-3D than Huanyuan 3D-2.1 in KREA:
NOTE: MESHY also did a better job converting 2D-to-3D than Hunyuan 2.1Image-to-3D Simplified -- which we did with the jaguar when we ran the program locally at Quelab on a computer with 12 GBs of VRAM:
With MESHY I am all that much closer to placing my FAKE Augmented Reality sculptures in the Albuquerque Museum Sculpture Garden:
I converted and animated one of my figure drawings from April 13, 2026 -- that had first been transformed into a more Bernini-like figure with KREA.
The key is to reposition your 2D image into a T-Pose. I did this with FLOW/NanoBanana2 (but could not get it done in Midjourney):
Converted into 3D with the MESHY 5 model
The MESHY 6 model gave better, more detailed, 2D-to-3D conversion results, but one needs to have a subscription to download the MESHY 6 GLB files:
.gif)
Huanyuan 3D-2.1 in KREA
was good, but not as good as MESHY
when converted into a GLB file
as the human figure when converted in MESHY
*****
With MESHY I am all that much closer to placing my FAKE Augmented Reality sculptures in the Albuquerque Museum Sculpture Garden:
My Augmented Reality Sculpture
Monday, 13 April 2026
April 13, 2026
Drawn at Art Buddies in Albuquerque:
.jpeg)
.jpeg)
Odd observation: The AI models all think in light and shadow, and shoe horns the results into more classical images. Imagine an AI model that thinks in motion instead, say by first breaking the image down into pieces and anticipating the physics of each part.
.jpeg)
"Sculptures in Zocalo Mexico City"
My FAKE sculptures in Mexico City
***
***
.jpeg)
"Wire sculptures in the city"
Skateboarding over
my FAKE wire sculptures in Mexico City
DRAWINGS
KREA Realtime prompt:
"exploding sculpture of a man broken into separate pieces"
KREA Realtime prompt:
"paper sculpture of a man broken into separate pieces"
KREA (Hunyuan 3D-2.1)
Download: "260413B.GLB" (587.19 KB)
KREA Realtime prompt:
"pieces of sculpture of a man floating in the air"
KREA Realtime prompt:
"paper sculpture of a man broken into separate pieces"
KREA Realtime prompt:
"paper sculpture of a man broken into separate pieces"
KREA Realtime prompt:
"sculpture of wire"
Odd observation: The AI models all think in light and shadow, and shoe horns the results into more classical images. Imagine an AI model that thinks in motion instead, say by first breaking the image down into pieces and anticipating the physics of each part.
They tore down the house across the street
on April 12, 2026
- (YouTube): Soviet Jetsons: Escape from Paradise -- AI cartoon about Orbitgrad
Subscribe to:
Comments (Atom)