Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
HOW TO RUN THIS MODEL
|
| 2 |
|
| 3 |
- This is a terminal-snr-v-prediction model and you will need an accompanying configuration file to load the checkpoint in v-prediction mode. Relevant configuration files are available in this repository. Place them in the same folder as the checkpoint. ComfyUI users will need to place this configuration file in models/configs and use the Load Checkpoint (With Config) node.
|
|
|
|
| 1 |
+
V10-FunnerEdition?
|
| 2 |
+
|
| 3 |
+
- Tweaked UNET with supermerger adjust to dialback noise/detail that can resolve eye sclera bleed in some cases.
|
| 4 |
+
- Adjusted contrast and color temperature. (Less orange/brown by default)
|
| 5 |
+
- CLIP should theoretically respond more to natural language. (Don't conflate this with tags not working or having to use natural language. Also it is not magic, so don't expect extremely nuanced prompts to work better.)
|
| 6 |
+
- FunEdition and FunEditionAlt are earlier versions before adjusting the UNET further to fix color temperature and color bleed. CLIP on these versions may less predictable as well.
|
| 7 |
+
|
| 8 |
HOW TO RUN THIS MODEL
|
| 9 |
|
| 10 |
- This is a terminal-snr-v-prediction model and you will need an accompanying configuration file to load the checkpoint in v-prediction mode. Relevant configuration files are available in this repository. Place them in the same folder as the checkpoint. ComfyUI users will need to place this configuration file in models/configs and use the Load Checkpoint (With Config) node.
|