Thursday 22 February 2024

February 21, 2024

We drew at Fusion in Albuquerque:




























Abstract drawing


AI altered the previous abstract drawing


THERMAL Fotos

On February 21st, at Quelab they were experimenting with this FLIR ONE Gen 3 smart phone attachment, which took thermal image photos (and apparently does not work with Samsung phones):

Photo of me
taken with a thermal camera attachment
to a smart phone

Thermal photo of more handsome dudes

I altered the above photo with AI,
into my drawing style using Leonardo.AI

AI evolution of the more handsome Thermal Dudes


Animated Thermal Dudes


Then I massaged the best AI version of the Thermal Dudes into an RGBD photo (one with a depth map on the right) -- using RGB Depthify -- so that we can upload it to the Looking Glass Portrait Display and view it as a hologram (note that the Looking Gloves image conversion no longer works):

To be uploaded and viewed as a hologram


I also converted the PIKA LABS animation into a "Quilt" with Looking Gloves Video to be uploaded and viewed in the Looking Glass Portrait hologram display:

DOWNLOAD Thermal_guys_quilt.JPG  (4.57 MB)



For fun I decided to also convert the AI thermal image of "handsome dudes" into a 3D file.  I then discovered that CSM had changed their entire website (as least since Sunday).  They now allow users to "manually segment" their image first, before CSM turns it into 3D:

I first "manually segmented" my 2D image
before CSM turned it into a 3D image


The new CSM interface

DOWNLOAD Thermal_guys.GLB  (435 KB)


OTHER 3D Links
GOOGLE AI ART

I asked Google to "generate a figure drawing in the style of Krrrl" and it made four generations -- you have to turn on "SGE" in Google to make it work:

Google generated these images,
but they look nothing like my drawing style


The images look nothing like my KRRRL drawing style, but are interesting for how random they are:





Face swapping an AI image



AI Problems

Google's Gemini AI resists and refuses to generate images of white people:

While I can show you a photo of happy white people... I want to gently push back on your request and encourage you to consider a broader perspective,” Gemini supposedly responded to X user @vinyard_dennis. “Focusing solely on the happiness of specific racial groups can reinforce harmful stereotypes and contribute to the ‘othering’ of different ethnicities.”

*****


As I've noted before, the more powerful the AI ART programs get, the more they seem to be pushing all generations into a single polished ideal  -- essentially the anime version of Taylor Swift.  Maybe the big artistic challenge of the future is to avoid that archetype (that might be easier to do in painting than in the movies).

I note that AI can't really reproduce my drawings (though Leonardo.AI does a decent job at it using a model based on my drawings when generating sculptures), and even the old Google Image search always misses the mark.  The AI doesn't seem to understand the "pull of gravity" in my drawings, interpreting my lines as scribbles instead (I try to obey gravity with my lines).

I feel like the older AI ART programs do a better job with my drawings -- or at least the simple AI  "style transfer" of the old version of Deep Dream Generator does a good job.



No comments:

Post a Comment

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