Paper Pixels is open for public mint now! Many thanks for your support from @Digitaldoes and me
@genekogan
-
AI-Powered Animation: AnimateDiff and Runway Gen-2 Video Collection
By
–
Finally, we rounded out the collection with some animations. Through a combination of AnimateDiff and @runwayml Gen-2 to induce animation in a controllable way, 12 videos were produced.
— Gene Kogan (@genekogan) 6 avril 2024
Credit to the DOES team for the ambient soundtrack backing these (sound on!) pic.twitter.com/EeIq6Uj3k3Finally, we rounded out the collection with some animations. Through a combination of AnimateDiff and @runwayml Gen-2 to induce animation in a controllable way, 12 videos were produced. Credit to the DOES team for the ambient soundtrack backing these (sound on!)
-
Interpolation Videos: Pre-AnimateDiff Creative AI Experiments
By
–
As an aside, during this whole process, I would occasionally take a detour to try making interpolation videos (the old style, pre-AnimateDiff).
— Gene Kogan (@genekogan) 6 avril 2024
It never really took off, although some of it was nice. I sometimes think I was a setting or two off from these being more impactful.… pic.twitter.com/c0PCnxyrVuAs an aside, during this whole process, I would occasionally take a detour to try making interpolation videos (the old style, pre-AnimateDiff). It never really took off, although some of it was nice. I sometimes think I was a setting or two off from these being more impactful.
-
Genetic Algorithm Image Optimization Through Ruthless DOES Curation
By
–
Once we locked in on a theme, we entered a feedback loop where we used relentless curation from the DOES team to optimize the input settings. It was ruthless — 97% of the images were deleted. The surviving ones would mutate and multiply in the next generation, like in a genetic
-
Training LoRAs to Control Stable Diffusion Style Consistency
By
–
It was clear that training needed to be part of the process. SD has way too much entropy to consistently and narrowly represent a single style. Over a few weeks, I trained over a hundred LoRAs on subsets of the source images. LoRA preserves the diversity of base SD while
-
Delayed AI Model Release: Pursuing Advanced Finetuning Results
By
–
We had originally targeted October 2023 to release (lol) but neither DOES nor I were satisfied yet, so we shifted towards full finetunes. We started getting more abstract results that looked like DOES's paper collages up close but in another dimension. Some of them even
-
Iterative AI Prompt Development with Team Feedback Loop
By
–
At first, we alternated between me trying out technical ideas and sending them to DOES team to give feedback and modify the approach, then rinse and repeat. Although they were new to AI, they could intuit how the prompts and settings would affect the outputs. Over time, our
-
Remix Techniques and Stable Diffusion Latent Space Exploration
By
–
Early on, the first thing I tried was using some of the "remix" techniques we've been developing at @eden_art_
. I'd project any and all of the source images into the latent space of Stable Diffusion and try making variations. It was promising, but we were caught between making -
DOES: From Graffiti Virtuoso to AI-Driven Paper Collage Artist
By
–
DOES is originally a virtuoso Graffiti artist who also has been making intricate paper collages over the last couple years and has a very handmade style. His art is best appreciated up close. The last 10 months have been an exercise in what I think is the essential problem of
-
Paper Pixels: AI Reimagines Handmade Artistic Style
By
–
Very excited to be releasing Paper Pixels with the amazing @Digitaldoes!
— Gene Kogan (@genekogan) 6 avril 2024
This project was made possible by @Streetlab_io who brought us together all the way back in June 2023 with a simple idea for me to "reimagine" DOES's eclectic handmade style with AI.
No problem I thought,… pic.twitter.com/m53kvvGk7KVery excited to be releasing Paper Pixels with the amazing @Digitaldoes
! This project was made possible by @Streetlab_io who brought us together all the way back in June 2023 with a simple idea for me to "reimagine" DOES's eclectic handmade style with AI. No problem I thought,