|
|
--- |
|
|
tags: |
|
|
- text-to-image |
|
|
- lora |
|
|
- diffusers |
|
|
- template:diffusion-lora |
|
|
widget: |
|
|
- text: '-' |
|
|
output: |
|
|
url: images/custom2.png |
|
|
base_model: black-forest-labs/FLUX.1-dev |
|
|
instance_prompt: animanflux |
|
|
license: cc-by-nc-nd-4.0 |
|
|
--- |
|
|
# Anthropomorphic |
|
|
|
|
|
<Gallery /> |
|
|
|
|
|
## Model description |
|
|
|
|
|
Super Realistic Anthropomorphic Animal Photography |
|
|
Recommended parameters: |
|
|
|
|
|
CFG: 1-4, related to anime and realism |
|
|
|
|
|
Weight: 0.8-1.2 |
|
|
|
|
|
Trigger word: animanflux |
|
|
|
|
|
Animal type: fly in the sky, run on the ground, swim in the water, whatever comes out. |
|
|
|
|
|
Manga style, realistic style can be switched very well, using manga and realistic type of prompt word to distinguish. |
|
|
|
|
|
There is no problem with compatibility with other models in the current test. |
|
|
|
|
|
By the way, the model is lora trained by the new dev model, so the limb degradation is almost negligible. |
|
|
|
|
|
And the anthropomorphic hands and feet of the animal retain the animal's features very well, except in individual cases. For example, if an anthropomorphic horse rides a horse, there is a probability that the hands and feet will be human. |
|
|
|
|
|
The model is trained by hundreds of anime-style and hyper-realistic pictures of anthropomorphic animals, which are labeled with a palace grid splicing. In order to ensure the generalization and fitting effect of the model, rank128, A100 trained. |
|
|
|
|
|
## Trigger words |
|
|
|
|
|
You should use `animanflux` to trigger the image generation. |
|
|
|
|
|
|
|
|
## Download model |
|
|
|
|
|
Weights for this model are available in Safetensors format. |
|
|
|
|
|
[Download](/Keltezaa/animan_V2.0/tree/main) them in the Files & versions tab. |
|
|
|
