Add Readme
Browse files- .gitattributes +1 -0
- README.md +94 -0
- statics/00171-3779625862.png +3 -0
- statics/00182-3779625873.png +3 -0
- statics/00267-3274412811.png +3 -0
- statics/banner.png +3 -0
- statics/file.png +3 -0
.gitattributes
CHANGED
@@ -32,3 +32,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
32 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
33 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
34 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
32 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
33 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
34 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
35 |
+
*.png filter=lfs diff=lfs merge=lfs -text
|
README.md
CHANGED
@@ -1,3 +1,97 @@
|
|
1 |
---
|
2 |
license: creativeml-openrail-m
|
|
|
|
|
|
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: creativeml-openrail-m
|
3 |
+
pipeline_tag: text-to-image
|
4 |
+
tags:
|
5 |
+
- ' stable-diffusion'
|
6 |
+
- ' stable-diffusion-diffusers '
|
7 |
---
|
8 |
+
|
9 |
+
# MechaDream
|
10 |
+
### A Stable Diffusion model for Mecha
|
11 |
+
![banner](./statics/banner.png)
|
12 |
+
---
|
13 |
+
## Available Models
|
14 |
+
### MechaDream-V1_lora
|
15 |
+
|
16 |
+
**Base model:** Counterfeit-V2.5
|
17 |
+
|
18 |
+
**The Road to this Lora**
|
19 |
+
|
20 |
+
Step 1: Fine-tuned on a dataset consisting of 130,000+ images of mecha/robots from various sources
|
21 |
+
|
22 |
+
Step 2: Mix the Lora models trained using different parameters
|
23 |
+
|
24 |
+
Step 3: Test different mixing ratios to find the best model
|
25 |
+
|
26 |
+
**Examples**
|
27 |
+
|
28 |
+
![sample1](./statics/00182-3779625873.png)
|
29 |
+
```
|
30 |
+
((masterpiece)), ((chibi)), a ((mecha)) with metallic hues and neon lights,symmetric body, full body, sky background
|
31 |
+
<lora:mechaDream-v1_lora:1>
|
32 |
+
Negative prompt: EasyNegative, bad-hands-5, human
|
33 |
+
Steps: 20, Sampler: UniPC, CFG scale: 15, Seed: 3779625873, Size: 680x456, Model hash: a074b8864e, Model: Counterfeit-V2.5_pruned, Denoising strength: 0.7, Clip skip: 2, Hires upscale: 2, Hires upscaler: R-ESRGAN 4x+ Anime6B
|
34 |
+
```
|
35 |
+
|
36 |
+
![sample2](./statics/00171-3779625862.png)
|
37 |
+
```
|
38 |
+
((masterpiece)), ((chibi)), a ((mecha)) with metallic hues and neon lights,symmetric body, full body, sky background
|
39 |
+
<lora:mecha_v2-1_u04t10_mecha_v1-2_u08t03:1>
|
40 |
+
Negative prompt: EasyNegative, bad-hands-5, human
|
41 |
+
Steps: 20, Sampler: UniPC, CFG scale: 15, Seed: 3779625858, Size: 680x456, Model hash: a074b8864e, Model: Counterfeit-V2.5_pruned, Denoising strength: 0.7, Clip skip: 2, Hires upscale: 2, Hires upscaler: R-ESRGAN 4x+ Anime6B
|
42 |
+
```
|
43 |
+
|
44 |
+
![sample3](./statics/00267-3274412811.png)
|
45 |
+
```
|
46 |
+
((masterpiece)), a (((robot))) with sleek and menacing design, glowing eyes, full body, highly detailed
|
47 |
+
<lora:mecha_v2-1_u04t10_mecha_v1-2_u08t03:0.8>
|
48 |
+
Negative prompt: EasyNegative, bad-hands-5, signature, boarder, word
|
49 |
+
Steps: 20, Sampler: UniPC, CFG scale: 15, Seed: 3274412811, Size: 720x408, Model hash: a074b8864e, Model: Counterfeit-V2.5_pruned, Denoising strength: 0.7, Clip skip: 2, Hires upscale: 2, Hires steps: 30, Hires upscaler: Latent (nearest)
|
50 |
+
```
|
51 |
+
|
52 |
+
![sample5](./statics/file.png)
|
53 |
+
```
|
54 |
+
((masterpiece)), a metallic crimson (((robot))) with sleek and menacing design, glowing eyes, full body, highly detailed
|
55 |
+
<lora:mecha_v2-1_u04t10_mecha_v1-2_u08t03:0.8>
|
56 |
+
Negative prompt: EasyNegative, bad-hands-5, signature, boarder
|
57 |
+
Steps: 20, Sampler: UniPC, CFG scale: 15, Seed: 1562808839, Size: 720x408, Model hash: a074b8864e, Model: Counterfeit-V2.5_pruned, Denoising strength: 0.7, Clip skip: 2, Hires upscale: 2, Hires steps: 30, Hires upscaler: Latent (nearest)
|
58 |
+
```
|
59 |
+
|
60 |
+
|
61 |
+
### MechaDream-V1
|
62 |
+
|
63 |
+
🚧 Pre-merged model is comming soon 🚧
|
64 |
+
|
65 |
+
---
|
66 |
+
## Recommended Setting
|
67 |
+
|
68 |
+
**Checkpoint:** Counterfeit-V2.5
|
69 |
+
|
70 |
+
**VAE:** mecha_v2_e3-pruned.ckpt
|
71 |
+
|
72 |
+
**Sampler:** UniPC (Personal preference)
|
73 |
+
|
74 |
+
**CFG Scale:** 15
|
75 |
+
|
76 |
+
**Negative Embeddings:** EasyNegative, bad-hands-5
|
77 |
+
|
78 |
+
**Hires.fix with scale >= 2 is highly recommend**
|
79 |
+
|
80 |
+
---
|
81 |
+
## My workflow
|
82 |
+
Step 1: Batch generate images
|
83 |
+
|
84 |
+
Step 2: Select the best image and inpaint any flaws
|
85 |
+
|
86 |
+
Step 3: Upscale using MultiDiffusion
|
87 |
+
|
88 |
+
*🡳 Some comparison images between original image and after the above workflow 🡳*
|
89 |
+
|
90 |
+
https://imgsli.com/MTcyNDM2
|
91 |
+
|
92 |
+
https://imgsli.com/MTcyNDM3
|
93 |
+
|
94 |
+
https://imgsli.com/MTcyNDM5
|
95 |
+
|
96 |
+
|
97 |
+
|
statics/00171-3779625862.png
ADDED
Git LFS Details
|
statics/00182-3779625873.png
ADDED
Git LFS Details
|
statics/00267-3274412811.png
ADDED
Git LFS Details
|
statics/banner.png
ADDED
Git LFS Details
|
statics/file.png
ADDED
Git LFS Details
|