import torch
from diffusers import DiffusionPipeline
# switch to "mps" for apple devices
pipe = DiffusionPipeline.from_pretrained("undefined", torch_dtype=torch.bfloat16, device_map="cuda")
pipe.load_lora_weights("tm-hf-repo/tmnby")
prompt = "tmnbny"
image = pipe(prompt).images[0]tmnby
Model description
3000 step prompt
Model was Nanobanana, prompt was:
- convert this to a very loose and abstract pencil drawing cartoon illustration style, minimalist art, keep the original colors, in the style of a "The New Yorker" cartoon, empty background, maintain the same colors, composition, and pose
Some Thoughts
- There was extremely high variability in output, filtered out from about 700 generations to 144 final, tried to have a loose colored cartoon sketch style (also it looks nothing like a New Yorker Cartoon)
- It should be a lot better at naturally maintaining skin tone
Trigger words
You should use tmnbny to trigger the image generation.
EXAMPLES
INPUT
DEFAULT PROMPT
WITH ADDITIONAL "maintain the same skin tone"
Download model
Weights for this model are available in Safetensors format.
Download them in the Files & versions tab.
- Downloads last month
- 7