Instructions to use dream-textures/texture-diffusion with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use dream-textures/texture-diffusion with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("dream-textures/texture-diffusion", dtype=torch.bfloat16, device_map="cuda") prompt = "pbr brick wall" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
- Local Apps
- Draw Things
- DiffusionBee
How to prompt engineer or re-train to make it do well with weirder textures?
#4
by MaxLohMusic - opened
It has no issues with common things like dirt, stone, and cobblestone. As soon as you try anything rarer, it has some trouble, e.g. the floor textures for "ocean floor", "treacherous terrain covered with green slime" or even "snowy ground" which never looks snow-covered. I would like a way to prompt it to be more "creative" and allow for a fantasy color palette that might not be prevalent in the training set, yet still follow tiling rules.