Hassanim
An attempt to improve over HassamBlend with the help of ReVAnimated. This was a prompt that created a composition I loved with SD1.5, but the face always looked bad. I tried dozens of models and they didn't improve the input much, except for ReVAnim that improved on the compossition, and HassanBlend provided the best face. At the end the model remains 95% HassanBlend.
Comparison:
(click for larger)
Prompt:
A ultradetailed beautiful painting of a stylish Pretty CUTE girl wearing streetwear standing in a convenience store, oil painting, by ilya kuvshinov, greg rutkowski and makoto shinkai in the style of ross tran
Aniblend
The only known model that created a blue puff jacket with this prompt.
RevAnimHassan & RevHassanimated
The first was a classic blend of the models required to create the other ones, the latter creates the best possible faces that ReVAnimated can create, at the cost of image's compositions.
Recipes:
- Add Difference 1.0
Primary model:
ReVAnimated
Secondary model:
ReVAnimated
Tertiary model:
v1-5-pruned-fp16-no-ema (https://huggingface.co/Yntec/DreamLikeRemix/resolve/main/v1-5-pruned-fp16-no-ema.safetensors)
Output Model:
ReVAnimatedEssense
- Super Merger Weight sum Train Difference 0.70
Model A:
ReVAnimatedEssense
Model B:
HassanBlend1.2
Output:
ReVAnimHassan
- Super Merger Weight sum Train Difference use MBW 0,0,0,0,0,0,0,0,0,0,0,0,0,0.5,1,1,1,1,1,1,1,1,1,1,1,1
Model A:
ReVAnimHassan
Model B:
HassanBlend1.2
Output:
RevHassanimated
- Super Merger Weight sum Train Difference use MBW 1,1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,1,1,1,1,1,1,0,0,0
Model A:
RevHassanimated
Model B:
ReVAnimated
Output:
AniBlend
- Super Merger Weight sum Train Difference use MBW 0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0
Model A:
HassanBlend1.2
Model B:
Unknown (I didn't record what model was used here, but it was one of ReVAnimHassan, RevHassanimated, or AniBlend. Probably AniBlend.)
Output:
Hassanim
