Workflows with Stable Diffusion

Comments

  • edited April 24

    Hi Theoryshaw, I don´t know the goal/usecase of the viz. My way to do viz is... using AI.
    Software: stable diffusion (local! nvidia rtx-gpu works best) (not as blender addon). Blender. (AI for Archviz is a little bit trial and error)
    1. duplicate the ifc-file. (or use an elevationview then step 2-4 you don´t need to do)
    2.remove the ifc stuff (make it to a "simple" blenderfile")
    3. config freestyle lineart
    4. Create Cam, take a shot from a nice angle or just use a elevationview. (The outlines are needed)
    5. Take the generated pic and use it as controlnet-pic in (easy diffusion) (therefore u have to download a model for architectual exterior. One place to get such models is civitai.com)
    6. Enter Prompts - Play around with config (tip: set picture size only on sizes that have a *)
    7. Resize the pic by AI.
    You will find some Tuts on YT for the first steps of config and the general use of easy diffusion. U can also do the render of the building in Blender,... and only create the exterior (stick it together in Gimp or Photoshop or use masks in Easydiffusion)
    Greetings

    theoryshaw
  • Awesome. I separated this out into its own thread, as Stable Diffusion workflows are a big and exciting topic.

  • result AI-generated content :-P

    BedsonMassimoemiliotassoNigelsteverugi
Sign In or Register to comment.