Originally posted to HuggingFace by TheLastBen

V2.1-768 Model by TheLastBen

This model is trained on 11 Midjourney images 768x768, 1300 steps and 300 steps text_encoder (30% because the total steps is low, normally 15%)

Prompts to start with :

ttdddd , __________, movie, ultra high quality render, high quality graphical details, 8k, volumetric lighting, micro details, (cinematic)

Negative : bad, low-quality, 3d, game

The prompt also can be as simple as the instance name : ttdddd and you will still get great results.

You can also train your own concepts and upload them to the library by using fast-DreamBooth.