Update README.md
Browse files
README.md
CHANGED
|
@@ -21,12 +21,28 @@ library_name: diffusers
|
|
| 21 |
|
| 22 |
# BPModel (and other models)'s UNET merged into SD 1.4 #
|
| 23 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 24 |
## Self explained. ##
|
| 25 |
|
| 26 |
- [BPModel](https://huggingface.co/Crosstyan/BPModel/tree/main) is a model intentionally preserved its UNET, *and dropped all the others, especially Text Encoder and VAE*. [PR has been made](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/7523), and it never get entertained. Workaround (Switch to SD 1.4 and then BPModel) works fine, [until I have some other use for it.](https://github.com/6DammK9/nai-anime-pure-negative-prompt/blob/main/ch05/README.MD)
|
| 27 |
|
| 28 |
- I need to align all the models I want to merge before I start merging them, to ensure *preserving the original Text Encoder will maximize the variance of the generation.*
|
| 29 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 30 |
## Models to be merged (download links) ##
|
| 31 |
|
| 32 |
- [BPModel]
|
|
|
|
| 21 |
|
| 22 |
# BPModel (and other models)'s UNET merged into SD 1.4 #
|
| 23 |
|
| 24 |
+
|
| 25 |
+
|
| 26 |
+
```
|
| 27 |
+
parameters
|
| 28 |
+
|
| 29 |
+
(ushanka:0.98), [[braid]], [[astolfo]], [[[[moscow, russia]]]], [[[[[[snow]]]]]]
|
| 30 |
+
Negative prompt: (bad:0), (comic:0), (cropped:0), (error:0), (extra:0), (low:0), (lowres:0), (speech:0), (worst:0)
|
| 31 |
+
Steps: 48, Sampler: Euler, CFG scale: 11, Seed: 1370168022, Size: 768x768, Model hash: 4a15b47ed1, Model: bp_nman_e29-sd_14, VAE hash: 551eac7037, VAE: vae-ft-mse-840000-ema-pruned.ckpt, Clip skip: 2, Dynamic thresholding enabled: True, Mimic scale: 1, Separate Feature Channels: False, Scaling Startpoint: MEAN, Variability Measure: AD, Interpolate Phi: 0.3, Threshold percentile: 100, Version: v1.6.0
|
| 32 |
+
```
|
| 33 |
+
|
| 34 |
## Self explained. ##
|
| 35 |
|
| 36 |
- [BPModel](https://huggingface.co/Crosstyan/BPModel/tree/main) is a model intentionally preserved its UNET, *and dropped all the others, especially Text Encoder and VAE*. [PR has been made](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/7523), and it never get entertained. Workaround (Switch to SD 1.4 and then BPModel) works fine, [until I have some other use for it.](https://github.com/6DammK9/nai-anime-pure-negative-prompt/blob/main/ch05/README.MD)
|
| 37 |
|
| 38 |
- I need to align all the models I want to merge before I start merging them, to ensure *preserving the original Text Encoder will maximize the variance of the generation.*
|
| 39 |
|
| 40 |
+
## Why SD 1.4 ##
|
| 41 |
+
|
| 42 |
+
- "*It is by design."* - Not author, quote lost long ago
|
| 43 |
+
|
| 44 |
+
- To verify, [head to my artwork long ago and try to replicate.](https://www.pixiv.net/en/artworks/104582308) *I made a more fancy version of him because dynamic CFG OP*
|
| 45 |
+
|
| 46 |
## Models to be merged (download links) ##
|
| 47 |
|
| 48 |
- [BPModel]
|