AI Image Gen · Checkpoint

3DMDT1 — Generalist Model for 3D Renders (SD 2.1 768)

SD 2.1 768 checkpoint tuned for 3D render style with strong generalist capability across photorealism, illustration, and fantasy.

About this model

3DMDT1 is a Stable Diffusion 2.1 768 checkpoint trained by ArtificialGuyBR in collaboration with Shadow_Shinigami (PRMJ model creator). The model started as a 3D-oriented fine-tune aiming to produce images that look like Blender or Unreal Engine renders, then evolved into a generalist checkpoint that handles 3D renders, photorealism, landscapes, portraits, and illustration styles.

Base model: SD 2.1 768. Trigger word: `3dmdt1` — include it in prompts to activate the 3D render style. For photorealistic results, add photography keywords ("raw photo", "photograph", "8k"); for 3D renders, include "3d render", "octane render", "blender", "unreal engine". Without style keywords the model leans toward 3D output.

Recommended settings (from author): DPM++ 2M Karras, 30 steps, CFG 7, no embeddings, standard negative prompt, no face restore. Model files: fp16 SafeTensor (2.4 GB) and fp16 .ckpt (2.4 GB), both with a companion YAML config. License: not explicitly stated; Civitai defaults apply.

Downloads: ~2,500 on Civitai, 485 likes, published February 2023.