Update README.md
Browse files
README.md
CHANGED
|
@@ -38,4 +38,24 @@ It is not based on any existing models, the UNet is trained from scratch.
|
|
| 38 |
- **GPU**: ~6 GB VRAM minimum (recommend 8+ GB)
|
| 39 |
- **CPU**: ~7 GB RAM. Image generation is extremely slow on CPU. (recommend 10+ GB)
|
| 40 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 41 |
|
|
|
|
| 38 |
- **GPU**: ~6 GB VRAM minimum (recommend 8+ GB)
|
| 39 |
- **CPU**: ~7 GB RAM. Image generation is extremely slow on CPU. (recommend 10+ GB)
|
| 40 |
|
| 41 |
+
## Prompting
|
| 42 |
+
|
| 43 |
+
Aniimage uses plain text captions as well as Danbooru tags.
|
| 44 |
+
|
| 45 |
+
An example of acceptable captions:
|
| 46 |
+
|
| 47 |
+
"A smiling anime girl with red hair and a school uniform"
|
| 48 |
+
|
| 49 |
+
"1girl, solo, smile, red_hair, school_uniform, anime_coloring"
|
| 50 |
+
|
| 51 |
+
Long prompts and enhanced prompts usually result in worse overall results. Try to keep your prompts short.
|
| 52 |
+
|
| 53 |
+
Adding a negative prompt is required for a good output. The one used during inference in the preview image is
|
| 54 |
+
"low quality, ugly, blurry, distorted, deformed, bad anatomy, bad proportions, extra limbs, missing limbs, watermark,
|
| 55 |
+
text, signature, washed out, flat colors, manga panel, disfigured, poorly drawn, jpeg artifacts, cropped, out of frame"
|
| 56 |
+
|
| 57 |
+
for the best quality, use DPM++ SDE at 50 steps.
|
| 58 |
+
|
| 59 |
+
**important** If you do not put 'NSFW' in the negative prompt, it will most likely make nsfw images.
|
| 60 |
+
|
| 61 |
|