NorbyQWEN commited on
Commit
d334e04
·
verified ·
1 Parent(s): 70d259d

Upload 3 files

Browse files
.gitattributes CHANGED
@@ -50,3 +50,4 @@ KRAST_v2/9e2b24f3-e352-4db4-b745-8fab3bc5f9bc.jpeg filter=lfs diff=lfs merge=lfs
50
  KRAST_v2/comfyui_00076.png filter=lfs diff=lfs merge=lfs -text
51
  Equinox_BSS_Semi-real_Int8Convrot.jpg filter=lfs diff=lfs merge=lfs -text
52
  godago_v3_fp8.jpg filter=lfs diff=lfs merge=lfs -text
 
 
50
  KRAST_v2/comfyui_00076.png filter=lfs diff=lfs merge=lfs -text
51
  Equinox_BSS_Semi-real_Int8Convrot.jpg filter=lfs diff=lfs merge=lfs -text
52
  godago_v3_fp8.jpg filter=lfs diff=lfs merge=lfs -text
53
+ fubuwa_MIX_3D_REAL_v2.jpg filter=lfs diff=lfs merge=lfs -text
fubuwa_MIX_3D_REAL_v2.jpg ADDED

Git LFS Details

  • SHA256: 09d7a749cebe6b4fff2bda66a83a5f12cc8c6252da0010a8f0be7f91cd00a298
  • Pointer size: 131 Bytes
  • Size of remote file: 250 kB
fubuwa_MIX_3D_REAL_v2.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5d322dd6f219be4ac6520b98669f27859624ec4d69410fb5d23c76bbb6ef74c5
3
+ size 13785330624
fubuwa_MIX_3D_REAL_v2/info.txt ADDED
@@ -0,0 +1,73 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ Version notes
2
+ v2
3
+ Increased colour saturation, brightness.
4
+
5
+ Increased head size/reduced broad shoulders.
6
+
7
+ Less dreamy, lesser artifact.
8
+
9
+ Prompting should be more responsive.
10
+
11
+
12
+ -----------
13
+
14
+
15
+ v2 - Bright, natural style
16
+
17
+ -----------
18
+
19
+ 3DCG/Semi realistic face; sharper/mature facial features with subtle skin texture.
20
+
21
+ -----------
22
+
23
+ Weakness
24
+ Noise pattern on image. Recommend using WAN 2.1 fp32 VAE and film grain postprocessing node.
25
+
26
+ Hands can be funky sometimes.
27
+
28
+ Anime style hairstyle can be hard to prompt
29
+
30
+ -----------
31
+
32
+ Recommended settings
33
+ (can see workflow via uploaded images)
34
+ euler + simple
35
+
36
+ steps 8-10, cfg 1
37
+
38
+ standard, simple, stable, works most of the time
39
+
40
+ downside - lacks texture, uniform texture, softer looking
41
+
42
+ euler + beta/(beta57-more tex)
43
+
44
+ steps 8-12, cfg 1
45
+
46
+ more texture than simple but can cause more artifacts
47
+
48
+ euler_a + beta57
49
+
50
+ steps 8-15, cfg 1
51
+
52
+ even more texture and details but also can get the image wrong
53
+
54
+ can cause the image to look more illustrative/less realistic
55
+
56
+ deis_2m / deis_3m + beta57
57
+
58
+ steps 3-4, cfg 1
59
+
60
+ similar to euler+beta57 but can have more stable structure while having as much texture
61
+
62
+ might have more realistic proportions compared to euler
63
+
64
+ downside - can be soft?
65
+
66
+ deis_3m is longer but might have better structure
67
+
68
+ Remarks
69
+ Personally use 1024x1536 resolution most of the time. Lower resolutions makes the noise pattern more noticeable when upscaled.
70
+
71
+ Upscale model recommendation: 2xLiveActionV1_SPAN. Keeps the image similar to the original without extra sharpening or lost of details while 2x resolution. Also fast.
72
+
73
+ WAN 2.1 fp32 VAE and film grain postprocessing node for noise pattern reduction.