Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
Please consider visiting my [patreon](https://www.patreon.com/ChainGirlAI/membership), the first post is a full wildcard of my **1,240 prompts for this model. It's a fast track way to use my model for creativity and see the full potential.**
|
2 |
+
|
3 |
+
See 1330 txt2img unprocessed samples from this model in an interactive mosaic here
|
4 |
+
[https://mosaically.com/photomosaic/chaingirl-surreality](https://mosaically.com/photomosaic/chaingirl-surreality)
|
5 |
+
|
6 |
+
This is my surreal realism model. **Surreality.** This is my first release publicly but I've created models amongst friends and small community, but if people like this I will either extend it or create new models on other styles.
|
7 |
+
|
8 |
+
It's mostly SFW as it's trained directly on SD1.5 and has no merges.
|
9 |
+
|
10 |
+
**Prompting**
|
11 |
+
|
12 |
+
All my examples here are basic prompts. I left my stable diffusion webui generating 1k images with a wildcard of the prompts I trained. No cherry picking to upload, just grabbed any random images from the outputs. Use **chaingirldark style** when prompting
|
13 |
+
|
14 |
+
**Sample prompts**
|
15 |
+
|
16 |
+
1. _a path through a forest with lots of trees and lights on it's sides and a trail leading to the trees , chaingirldark style lightning storm, speedpainting, matte painting, concept art, 4k_
|
17 |
+
|
18 |
+
2. two horses running in a field in the foggy day time, with one horse looking at the camera , chaingirldark style ultra detailed, 8k, 35mm lens, beautiful, classical painting, smoke tendrils, clouds, foggy, volumetric lighting, cloudy, trending on Artstation, concept art, hyperdetailed, hyperrealistic
|
19 |
+
|
20 |
+
3. a black and white photo of a church on a rocky hill with a stream running between it and a tree , chaingirldark style digital art, photorealistoc, art by greg rutkowski, hyperdetailed, western comic style, comic, comic style, sharp lineart, professional lighting, deviantart, artstation, trevor henderson, rossdtaws, cinematic, dramatic
|
21 |
+
|
22 |
+
4. a desk with a computer and a monitor on it in a room with a lot of windows and plants , chaingirldark style cyberpunk 2077, blade runner, neon volumetric lights, 35mm lens, beautiful lighting, hyper detailed, digital art, trending in artstation
|
23 |
+
|
24 |
+
5. a city street with a lot of flowers and buildings in the background at night time with pink and purple lights , chaingirldark style cyberpunk, futuristic, ultra detailed, concept art
|
25 |
+
|
26 |
+
6. check example images metadata or my patreon wildcard for more prompts
|
27 |
+
|
28 |
+
|
29 |
+
**Negative prompt**: boring, 3d render, long neck, out of frame, extra fingers, mutated hands, , ((poorly drawn hands)), 3DCG, cgstation, extra heads, close up, text ,watermarks, logo
|
30 |
+
|
31 |
+
|
32 |
+
**Training:**
|
33 |
+
|
34 |
+
Finetuned SD1.5 with 1200 surreal and sci-fi images. No merges.
|
35 |
+
|
36 |
+
- trained chaining resolutions from low 300px up to 768px in increments
|
37 |
+
|
38 |
+
- overall 36 epochs 29,970 steps when combining epoch's on lower res training with the 768 training
|