|
--- |
|
license: openrail |
|
language: |
|
- en |
|
--- |
|
|
|
 |
|
# Model Card |
|
|
|
CyberRealistic v8 is PONY based model from author Cyberdelia, its used for image inference. |
|
|
|
## Model Details |
|
|
|
GGUF version of original CyberRealistic v8 model |
|
|
|
CLIP G and CLIP L from CyberRealistic v8 |
|
|
|
VAE from SDXL |
|
|
|
simple workflow samples in .json and .png for ComfyUI use |
|
|
|
### Model Description |
|
|
|
Its just split model obtained from CyberRealistic v8 nothing fancy, reason for making it was https://github.com/pollockjj/ComfyUI-MultiGPU, |
|
which allows running models (in GGUF format) even on very low VRAM graphic cards. Provided you have enough RAM, it could probably run even on 3GB VRAM. |
|
|
|
Its made with https://github.com/calcuis/gguf which along GGUF loaders and other stuff also has tool that can turn basically anything into GGUF (their ZERO converter). |
|
|
|
Ive split it into individual files, cause you can already force via ComfyUI VAE and CLIP where you want it, but main issue is model ofc. |
|
|
|
I think its probably possible to make even smaller fp8 (q8?) model, if needed/wanted. |
|
|
|
### Model Source |
|
|
|
- **Repository:** [https://civitai.com/models/443821?modelVersionId=1346181] |
|
|
|
### Model Author |
|
|
|
[https://huggingface.co/cyberdelia] |
|
|
|
## Uses |
|
|
|
Use within ComfyUI, simple workflow is with other needed files. It probably wont work outside ComfyUI for now at least. |
|
|
|
### Out-of-Scope Use |
|
|
|
Whatever you generate with this is your responsibility. Dont share anything that can end with you in jail, it wont help you or community. |
|
|
|
## Bias, Risks, and Limitations |
|
|
|
It can do very much NSFW. |
|
|
|
### Recommendations |
|
|
|
Dont make and then share illegal stuff. |
|
|
|
### Model Architecture and Objective |
|
|
|
[PONY mode which is based on SDXL] |