ComfyUI
Is anyone else experimenting with this free open-source AI image generation tool for AECO?
I’m looking for good workflows for archviz, both text-to-image and image-to-image. Here are some of my first generated images using this tool following PAACADEMY!
text->image
image->image
input image made with Blender
outputs image+prompt
Comments
@elschilling cool pics, I used lokal "easy diffusion" and the architecture models (for outdoor, indoor) from Civitai.green. For sure from outlines, or Render+enviroment. The generativeAI from speckle.systems is also fun to play. But I like the option to change partial things with masks. (Render in blender or outlines, mask the rest... write a prompt... do some atempts e voila.) Your first pics are of the sort of "lets dream of vacation", and the quality is "Wow" at one 2 suns are a little bit freaky. :D
Not as ultrarealistic as the pics from elschilling, but fast and 1024x512 (original). thx for the hint! Comfy looks great, have to play around some nights :-)
hey, slow down, do you know how fast you were scrolling?
after my first steps with comfyui, I can say it´s super flexible. Used blenders "lineart modifier" for controlnet, tried some easydiffusion archmodels,... but that are only the first steps... maybe I´m trying to make a short vids out of the pics.
