Instructions to use dataautogpt3/TempestV0.1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use dataautogpt3/TempestV0.1 with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("dataautogpt3/TempestV0.1", dtype=torch.bfloat16, device_map="cuda") prompt = "Food photography style photo RAW,piece of fried grilled meat, splashes of ketchup and mustard sauce, (rosemary), spices, exceptional shallow depth-of-field capabilities, atmospheric haze blur,vivid colors,high quality textures of materials, volumetric textures, coating textures, metal textures . Appetizing, professional, culinary, high-resolution, commercial, highly detailed" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
- Local Apps
- Draw Things
- DiffusionBee
Are there (will there be) more aspect ratios / dimensions?
It's amazing what you have achieved here! I'm loving it! For me, the Artistic version is giving more realistic, nicer, aesthetically more pleasing results.
However, as I'm doing characters designs, I'm having a bit of trouble with these aspect ratios, as characters get stretched quite a bit, they have elongated bodies. Anatomy-wise, they are fine, have only 2 legs and 2 arms, but they look like they have marfan syndrome.
Are there any other aspect ratios, that work? Maybe 4:3 or 3:2?
I tried using some random W:H numbers, but then the composition or anatomy got really messy, I couldn't quite figure out decent dimensions.
Thank you for your hard work nonetheless! I will soon be uploading some images to Twitter to spread the word.