Updated sampler grids
Browse files
README.md
CHANGED
|
@@ -4,7 +4,7 @@ license: unknown
|
|
| 4 |
|
| 5 |
# StableMondAI-SDG
|
| 6 |
|
| 7 |
-
A model to bring back SDXL
|
| 8 |
|
| 9 |
### Model Description
|
| 10 |
|
|
@@ -14,7 +14,7 @@ A model to bring back SDXL base styles to noobAI XL
|
|
| 14 |
|
| 15 |
## Uses
|
| 16 |
|
| 17 |
-
This model is an attempt to make the recent noobAI XL based merges again understand the
|
| 18 |
|
| 19 |
This model is in V-PREDICTION mode.
|
| 20 |
|
|
@@ -36,44 +36,37 @@ Starter Negative prompt:
|
|
| 36 |
worst quality
|
| 37 |
```
|
| 38 |
|
| 39 |
-
If you are using artist tags some additional quality tags might come in handy. This is only an example and what quality tags might work best in your specific circumstances requires some experimentation.
|
| 40 |
-
|
| 41 |
-
Advanced Positive prompt:
|
| 42 |
-
```
|
| 43 |
-
masterpiece, best quality, newest, highres
|
| 44 |
-
```
|
| 45 |
-
|
| 46 |
-
Advanced Negative prompt:
|
| 47 |
-
```
|
| 48 |
-
worst quality, low quality, sketches
|
| 49 |
-
```
|
| 50 |
-
|
| 51 |
-
|
| 52 |
### Generation Recommendations
|
| 53 |
|
| 54 |
The following settings are recommended starting points:
|
| 55 |
|
| 56 |
-
* Sampler: Euler
|
| 57 |
* Scheduler: SGM Uniform
|
| 58 |
* Steps: 24-32
|
| 59 |
-
* CFG:
|
| 60 |
|
| 61 |
-
RescaleCFG is not required.
|
| 62 |
|
| 63 |
Initial tests show that other samplers work as well but need appropiate settings. See examples below.
|
| 64 |
|
| 65 |
## More Information
|
| 66 |
|
| 67 |
-
###
|
| 68 |
|
| 69 |
-
|
| 70 |
-
<
|
|
|
|
| 71 |
</a>
|
| 72 |
|
| 73 |
-
|
|
|
|
|
|
|
|
|
|
| 74 |
|
| 75 |
-
|
| 76 |
-
|
|
|
|
|
|
|
| 77 |
</a>
|
| 78 |
|
| 79 |
### Example Generations (including metadata)
|
|
|
|
| 4 |
|
| 5 |
# StableMondAI-SDG
|
| 6 |
|
| 7 |
+
A model to bring back SDXL art styles to noobAI XL
|
| 8 |
|
| 9 |
### Model Description
|
| 10 |
|
|
|
|
| 14 |
|
| 15 |
## Uses
|
| 16 |
|
| 17 |
+
This model is an attempt to make the recent noobAI XL based merges again understand the art styles that were present in the SDXL 1.0 base model but got lost in the lineage of style merges. It has a stable builtin style and follows artist tags well. Photorealism is not supported well, use one of the other models for that.
|
| 18 |
|
| 19 |
This model is in V-PREDICTION mode.
|
| 20 |
|
|
|
|
| 36 |
worst quality
|
| 37 |
```
|
| 38 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 39 |
### Generation Recommendations
|
| 40 |
|
| 41 |
The following settings are recommended starting points:
|
| 42 |
|
| 43 |
+
* Sampler: Euler
|
| 44 |
* Scheduler: SGM Uniform
|
| 45 |
* Steps: 24-32
|
| 46 |
+
* CFG: 4-5
|
| 47 |
|
| 48 |
+
RescaleCFG is **not** required.
|
| 49 |
|
| 50 |
Initial tests show that other samplers work as well but need appropiate settings. See examples below.
|
| 51 |
|
| 52 |
## More Information
|
| 53 |
|
| 54 |
+
### Sampler / Scheduler Comparison Grid
|
| 55 |
|
| 56 |
+
Base style
|
| 57 |
+
<a href="https://cdn-uploads.huggingface.co/production/uploads/674081772cb82e06227eee49/PZjC-GHVo6EJVCKE-R01M.jpeg">
|
| 58 |
+
<img src="https://cdn-uploads.huggingface.co/production/uploads/674081772cb82e06227eee49/PZjC-GHVo6EJVCKE-R01M.jpeg" width="1024" height="auto">
|
| 59 |
</a>
|
| 60 |
|
| 61 |
+
Realism
|
| 62 |
+
<a href="https://cdn-uploads.huggingface.co/production/uploads/674081772cb82e06227eee49/vpKjB7fU_zb1XVzt6aUOP.jpeg">
|
| 63 |
+
<img src="https://cdn-uploads.huggingface.co/production/uploads/674081772cb82e06227eee49/vpKjB7fU_zb1XVzt6aUOP.jpeg" width="1024" height="auto">
|
| 64 |
+
</a>
|
| 65 |
|
| 66 |
+
### Style Examples Grid
|
| 67 |
+
|
| 68 |
+
<a href="https://cdn-uploads.huggingface.co/production/uploads/674081772cb82e06227eee49/FTUVdkceWohrV7QLDteT8.jpeg">
|
| 69 |
+
<img src="https://cdn-uploads.huggingface.co/production/uploads/674081772cb82e06227eee49/FTUVdkceWohrV7QLDteT8.jpeg" width="1024" height="auto">
|
| 70 |
</a>
|
| 71 |
|
| 72 |
### Example Generations (including metadata)
|