ArtificialGuyBR

Home / Projects / 3DMDT1

Other · Text-to-Image

3DMDT1

Generalist Stable Diffusion model focused on 3D rendering style (Blender/Unreal), also handles landscapes, portraits, photorealism, and illustrations. Trigge…

About this model

3DMDT1 is a custom text-to-image model developed by ArtificialGuyBR in collaboration with Shadow (creator of the PRMJ model). The model was trained with a focus on 3D rendering — producing images that resemble Blender or Unreal Engine output — but also delivers strong results across landscapes, portraits, photorealism, and illustrations.

The model uses the trigger word "3DMDT1" in prompts. For photorealistic results, include photography keywords like "raw photo"; for 3D renders, include "3d render". Without style keywords, the model defaults toward 3D-style output.

Recommended generation settings (from the author): DPM++ 2M Karras sampler, 30 steps, no embeddings, standard negative prompts. The model is distributed under the OpenRAIL license and available as SafeTensor, CKPT, and YAML config files.

Project signals

  • 6 Hugging Face likes

Topics

art · text-to-image

Explore the source