I transformed some of the figures I drew last year into Fake Robert Williams using the "Text 2 Dream" option of Deep Dream Generator:
- Previously I transformed my drawings into Fake Michelangelos and Fake Egon Schieles.
The base image is a collaboration
between John Tollett and I
- I took more liberties by editing the results a bit in Photoshop Elements, creating transparent backgrounds and outlining the figures (Layer>Layer Style>Style Settings...>stroke)
The image below was based on the first image posted on this blog, outpainted in DALL-E 2:
prompt: "Background in the style of Robert Williams"
After outpainting with DALL-E 2, I uploaded that whole scene into the "Text 2 Dream" option in Deep Dream Generator and generated the Fake Robert Williams scenario below:
I think the colored images transformed into better Fake Robert Williams, but I wanted to see what would happen starting with a regular pencil drawing:
Lexica (Robert Williams search)
Similar, but...
NOTES
- I saw the Robert Williams exhibition in Santa Fe in 2016 at CCA (Center for Contemporary Arts)
- The Robert Williams documentary -- "Mr Bitchin" -- is on YouTube.
- Robert Williams co-founded Juxtapoz Magazine
- He grew up in Albuquerque
- RK Sloane is similar and also from Albuquerque
- Joe Turner is a young artist from Albuquerque with a similar style
THOUGHTS
While these Fake Robert Williams images obeyed the composition of the drawing I uploaded as a template, the results bore little resemblance to my drawings. Less resemble than the Michelangelo and Egon Schiele transformations.
Moreover I think the colored images yielded more successful results, more than the monochrome pencil drawings.
I'm craving a more powerful AI. Actually I really want more control over the AI, which is what Deep Dream Generator is giving me. Their "Text 2 Dream" option obeys my bad figure drawings the best, whereas Midjourney pretty much ignores any image I upload.
However I don't mind a few surprises. I believe the job of art is to show me something that does not exist, rather than just render something I already know, like taking a photograph. Nevertheless, I would like to steer my AI generations into something more meaningful.
No comments:
Post a Comment
Note: only a member of this blog may post a comment.