|
--- |
|
|
|
license: creativeml-openrail-m |
|
|
|
--- |
|
|
|
### Model info |
|
|
|
--- |
|
|
|
This is a dreambooth model trained with the data set of [FloralMarble](https://huggingface.co/datasets/spaablauw/FloralMarble_dataset) on top of stable diffusion 1.5, all creadits to [spaablauw](https://huggingface.co/spaablauw) for original images. |
|
|
|
I left several models uploaded, all the intermediate steps + two anime models that I merged into. |
|
|
|
I would recomend try [the 4000 steps model](https://huggingface.co/N75242/FloralMarbles_Model/resolve/main/FloralMarble_step_4000.ckpt) or the [7000 steps one](https://huggingface.co/N75242/FloralMarbles_Model/resolve/main/FloralMarble_step_7000.ckpt), it depends a bit in what you want, I had relly good result in booth. |
|
For img2img 7000 step version is better. |
|
|
|
[Download Eimis Merge](https://huggingface.co/N75242/FloralMarbles_Model/resolve/main/EimisAnimeDiffusion_1-0v_0-FloralMarble_step_3000.safetensors) |
|
|
|
[Download Anything Merge](https://huggingface.co/N75242/FloralMarbles_Model/resolve/main/Anything-V3.0_0-FloralMarble_step_3000_1.safetensors) |
|
|
|
Use whatever VAE you want. |
|
|
|
--- |
|
|
|
### Examples, download images to get prompts from exif data |
|
|
|
 |
|
 |
|
 |
|
 |
|
|
|
 |
|
 |
|
 |
|
 |
|
 |
|
 |
|
|
|
 |
|
 |
|
 |
|
 |
|
|
|
--- |
|
|
|
### Tag list |
|
|
|
[Get the tag list images had here](https://huggingface.co/N75242/FloralMarbles_Model/resolve/main/tags.txt) |
|
|
|
I used "flrmrbl" as an unique token, so it should activate the model traing data, also "floral marble" is present in all images, but its more generic si probably less powerfull. |
|
But as an alternative use "in the style of flrmrbl" or "flrmrbl style". |
|
|
|
Have fun! |
|
|
|
|
