Update README.md
Browse files
README.md
CHANGED
|
@@ -3,6 +3,24 @@ license: creativeml-openrail-m
|
|
| 3 |
base_model: runwayml/stable-diffusion-v1-5
|
| 4 |
---
|
| 5 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6 |
# STORYTIME -- **teaching Stable Diffusion to use its words**
|
| 7 |
|
| 8 |
|
|
|
|
| 3 |
base_model: runwayml/stable-diffusion-v1-5
|
| 4 |
---
|
| 5 |
|
| 6 |
+
# UPDATE: Introducing the ConceptSmasher
|
| 7 |
+
|
| 8 |
+
So... While randomly messing around with ComfyUI (this will only work in ComfyUI), I stumbled upon this neat little
|
| 9 |
+
configuration of nodes. It is particularly interesting when used with Storytime, which is not a thing I planned at
|
| 10 |
+
all. Being a person who enjoys random/strange/weird AND someone who likes to use models in ways that may not
|
| 11 |
+
reflect intended use, this appeals to my personal interest in exploring what ML models can do, artistically speaking.
|
| 12 |
+
|
| 13 |
+
The ConceptSmasher is designed to work with some rather wild settings -- no prompt and a CFG of 100. You can, of course,
|
| 14 |
+
explore other uses/ideas/etc, but this is my "fire and forget" configuration. Inasmuch as ComfyUI has nodes for randomly
|
| 15 |
+
loading images from folders, that would make this 100% more awesome. ALSO inasmuch as ComfyUI could have the ability
|
| 16 |
+
to randomly modify _any_ node values at all (selectable per slider/value/etc), THAT would be EXTREMELY AWESOME. I might
|
| 17 |
+
make this, but until then... What this monstrosity looks like:
|
| 18 |
+
|
| 19 |
+
|
| 20 |
+
And some sample images:
|
| 21 |
+
|
| 22 |
+
|
| 23 |
+
|
| 24 |
# STORYTIME -- **teaching Stable Diffusion to use its words**
|
| 25 |
|
| 26 |
|