import torch
from diffusers import DiffusionPipeline
# switch to "mps" for apple devices
pipe = DiffusionPipeline.from_pretrained("StarkEater/NyteTyde150626-Anima-fullfinetune-noflip", dtype=torch.bfloat16, device_map="cuda")
prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
image = pipe(prompt).images[0]Yes, this is the most accurate outcome. If interested, then use only TD Russel's Diffusion pipe, which is a better tool than Kohya's SD-scripts specifically for Anima.
Don't worry, it can handle pure NL, hybridised NL, and pure tags easily. The proof lies in actually using the model.
- Downloads last month
- -
Model tree for StarkEater/NyteTyde150626-Anima-fullfinetune-noflip
Base model
nvidia/Cosmos-Predict2-2B-Text2Image Finetuned
circlestone-labs/Anima