We are inspired by the CN model of semantic segmentation and hope to train a designer specific controllet model. Through this model, we can solve this pain point once and for all, so that we can obtain the results we want more accurately when using AI for drawing. Finally. After spending a long time manually annotating a lot of data and trained this controlnet, this problem has finally been solved! Today we have also opened up our CN material control model Material_control V0.7, and we will open up a better version depending on the situation in the future.
At present, the open version supports precise control of # 76380b (wood), # cd5c5c (brick), # 4c4c4c (concrete), # 00bfff (glass), and # c0c0c0 (metal). After multiple tests, it was found that the model can not only control the position and texture of materials, but also achieve excellent results with very simple prompts without Lora, which is something we did not expect.
Recommended large models: realistic models for architecture, all other types of large models are available
Recommended weight: 1
Recommended tip word: You can write the desired materials or not, starting with "a modern building" and writing freely afterwards
This model is prohibited for commercial use. If commercial use is required, please confirm with me
Description
FAQ
Comments (10)
Do you have a preprocessor to generate the image with colors or would I need to do this coloring manually?
u are right, no preprocessor
yeah，maybe later i would do a preprocessor model, now we can only get that input image manually
try segment anything to gen pics and then photoshop to replace the color?
tryin by sketchup modeling, add code to material library.
amazing work
This is amazing!
what was the checkpoint model you have used in the examples? will it work when I pair it with multiple controlnets?
niubee
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.