i've been following the idea of living portraits for several years now. last week, i came across a live portrait node for comfyui. after some quick testing, i now know i can implement this using large displays and spatial tracking to create portraits that react to the space around them. lets see

I don’t want to seem out of touch but I don’t actually understand the economy anymore.

!

2024 – not designed by rams / flux lora strength test with a flux lora finetuned on 200 dieter rams products. this was part of an art/technology exploration from 2022 - 2024

flashback to 2023. seedtravel with a stable diffusion 1.5 model finetuned on 200 dieter rams products.

i am starting to miss this kind of artificial style.

I made a new mouth mechanism. I'm calling it Selectramatronics.

#fuck

finally i can do microdata refinement with my outie. thanks claude-sonnet-3.7 moritzpongratz.com/mdr/

#severance #claude-sonnet

with the realease of github.com/Microsoft/TR... and my old dieter rams models, i got much closer to my prompt2rams pipline

vimeo.com/1038011260

#trellis #3dprint #prompt2object #ai

found my 10 year old transparent screen and had to mix it with some physics and positional tracking for an installation i am currently experimenting with.

vimeo.com/944120079

it has now been over a year since i made notdesignedbyrams.com , a RaaS (rams as a service) prototype for my talk at the ADC digital congress. it's still running strong and some people still seem to use it for actual inspiration or research. still around 5-20 generations a day.

quick test with FLUX.1-Redux-dev chaining three inputs into one style. no prompts, nearly uncontrolable but a lot of fun. one step closer to my infinte moodboard #flux #comfyui