3D ML WOrkflow

One of the most tedious parts of building out my 3D scenes, for me at least, is always the modelling.

Discovering the power of some of the newer models on #huggingface spaces has allowed me to develop a new workflow for creating some quick assets

💭 A stable diffusion model allows you to develop a text to image reference that forms the basis for your character or object

🖼️ Cleanup your elements in the image in Photoshop or Figma and quickly make your reference ready for the next model

🎮 Using DreamGaussian to take your image to 3D, probably the most back and forth portion of this as you try to fine tune and play with settings to get it just right

The only note is that the models themselves are highly primitive and working with their details can be difficult but overall a fun quick way to simulate