Disty0
commited on
Commit
•
17ece57
1
Parent(s):
888d1e8
Update README
Browse files
README.md
CHANGED
@@ -1,3 +1,105 @@
|
|
1 |
---
|
2 |
license: creativeml-openrail-m
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: creativeml-openrail-m
|
3 |
+
language:
|
4 |
+
- en
|
5 |
+
library_name: diffusers
|
6 |
+
pipeline_tag: text-to-image
|
7 |
+
tags:
|
8 |
+
- art
|
9 |
+
- anime
|
10 |
+
- stable diffusion
|
11 |
+
- openvino
|
12 |
+
- onnx
|
13 |
---
|
14 |
+
|
15 |
+
|
16 |
+
SoteMix V3 is trained at 1024x1536 for high resolution image generations.
|
17 |
+
|
18 |
+
This model is tested on SD.Next with Diffusers backend and HyperTile size 384.
|
19 |
+
|
20 |
+
Positive Prompts:
|
21 |
+
```
|
22 |
+
masterpiece, best quality, highres, 1girl,
|
23 |
+
```
|
24 |
+
|
25 |
+
Negative Prompts:
|
26 |
+
```
|
27 |
+
(worst quality, low quality, lowres), zombie, interlocked fingers,
|
28 |
+
```
|
29 |
+
Do not use any negative embeddings.
|
30 |
+
|
31 |
+
Sampler: `Euler a`
|
32 |
+
Steps: `30-40`
|
33 |
+
Clip Skip: `1`
|
34 |
+
CFG: `4-7`
|
35 |
+
Base Resolution: `512x` / `768x` / `1024x` / `768x1280` / `960x1280` / `1024x1536` / `1920x1080`
|
36 |
+
Model can still be chaotic at `1024x1536` and `1920x1080`.
|
37 |
+
|
38 |
+
Second Pass / Hires:
|
39 |
+
Sampler: `Euler` / `Euler a`
|
40 |
+
Steps: `10` with `Euler` / `20` with `Euler a`
|
41 |
+
Upscaler: `RealESRGAN 4x+ Anime6B` / `ESRGAN 4x-AnimeSharp` with `0.2`-`0.3` denoise strength.
|
42 |
+
CFG: `6-9`
|
43 |
+
Resolution: `2x` of the base resolution.
|
44 |
+
|
45 |
+
|
46 |
+
|
47 |
+
Training:
|
48 |
+
My GPU couldn't handle full model training at these resolutions so i trained it as a `512` layer Lora with SoteMix V1 as the base.
|
49 |
+
Used highres as the trigger word. Also used raifu trigger word with my OC character.
|
50 |
+
|
51 |
+
Resolution: `1024x1536 with Bucketing`
|
52 |
+
Batch Size: `1`
|
53 |
+
Steps: `40000`
|
54 |
+
GPU: `Intel ARC A770 16GB`
|
55 |
+
|
56 |
+
Bucket:
|
57 |
+
```
|
58 |
+
bucket 0: resolution (832, 1664), count: 49
|
59 |
+
bucket 1: resolution (896, 1280), count: 1
|
60 |
+
bucket 2: resolution (896, 1536), count: 2
|
61 |
+
bucket 3: resolution (960, 1408), count: 8
|
62 |
+
bucket 4: resolution (960, 1472), count: 57
|
63 |
+
bucket 5: resolution (960, 1536), count: 12
|
64 |
+
bucket 6: resolution (960, 1600), count: 537
|
65 |
+
bucket 7: resolution (1024, 1344), count: 266
|
66 |
+
bucket 8: resolution (1024, 1408), count: 349
|
67 |
+
bucket 9: resolution (1024, 1472), count: 1535
|
68 |
+
bucket 10: resolution (1024, 1536), count: 950
|
69 |
+
bucket 11: resolution (1088, 1280), count: 63
|
70 |
+
bucket 12: resolution (1152, 1216), count: 62
|
71 |
+
bucket 13: resolution (1152, 1280), count: 147
|
72 |
+
bucket 14: resolution (1152, 1344), count: 114
|
73 |
+
bucket 15: resolution (1216, 1152), count: 44
|
74 |
+
bucket 16: resolution (1216, 1216), count: 409
|
75 |
+
bucket 17: resolution (1216, 1280), count: 53
|
76 |
+
bucket 18: resolution (1280, 576), count: 20
|
77 |
+
bucket 19: resolution (1280, 640), count: 94
|
78 |
+
bucket 20: resolution (1280, 704), count: 217
|
79 |
+
bucket 21: resolution (1280, 768), count: 102
|
80 |
+
bucket 22: resolution (1280, 832), count: 118
|
81 |
+
bucket 23: resolution (1280, 896), count: 280
|
82 |
+
bucket 24: resolution (1280, 960), count: 137
|
83 |
+
bucket 25: resolution (1280, 1024), count: 32
|
84 |
+
bucket 26: resolution (1280, 1088), count: 27
|
85 |
+
bucket 27: resolution (1280, 1152), count: 61
|
86 |
+
bucket 28: resolution (1280, 1216), count: 24
|
87 |
+
bucket 29: resolution (1344, 1024), count: 17
|
88 |
+
bucket 30: resolution (1344, 1152), count: 38
|
89 |
+
bucket 31: resolution (1536, 896), count: 94
|
90 |
+
bucket 32: resolution (1536, 1024), count: 34
|
91 |
+
bucket 33: resolution (1600, 960), count: 196
|
92 |
+
bucket 34: resolution (1664, 832), count: 21
|
93 |
+
bucket 35: resolution (2048, 768), count: 3
|
94 |
+
bucket 36: resolution (2304, 576), count: 1
|
95 |
+
mean ar error (without repeats): 0.01257769833438255
|
96 |
+
```
|
97 |
+
|
98 |
+
Merge:
|
99 |
+
Merged SoteMix V1 with Lunar Radiance Light and then merged the Hires Lora i trained on top of it.
|
100 |
+
Merge ratio: `(0.6 SoteMix V1 + 0.4 Lunar Radiance Light) + 0.7 Hires Lora`
|
101 |
+
|
102 |
+
|
103 |
+
|
104 |
+
IMAGE
|
105 |
+
|