Home

A: GhostMix civitai

B: Lyriel civitai

C: Realistic Vision v2.0 civitai

alpha: 0.5

beta: 0.5

M: resulting merged model with weights_clip = False

wc: resulting merged model with weights_clip = True

a cat as a DJ at the turntables
Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 1897848000, Size: 512x716

weighted_sum

$$M = (1 - \alpha)\times A + \alpha B$$

A	B	M	wc

add_difference

Note, in this case C is SD 1.5 base model hf

$$M = A + \alpha (B - C)$$

A	B	C	M	wc

weighted_subtraction

$$ M = \frac{(A - \alpha \times \beta \times B)}{(1 - \alpha \times \beta)} $$

A	B	M	wc

sum_twice

$$M = (1-\beta)\left[(1 - \alpha) \times A + \alpha B \right]+\beta C$$

A	B	C	M	wc

triple_sum

$$M = (1 - \alpha - \beta) \times A + \alpha B + \beta C$$

A	B	C	M	wc

similarity_add_difference

$$ M=M(\alpha, \beta, A, B, C) $$

A	B	C	M	wc

transmogrify_distribution

$$ M = M(A, B) $$

A	B	M	wc

tensor_sum

$$ M = M(\alpha, \beta, A, B) $$

A	B	M	wc

multiply_difference

$$ M = C + S(|(A - C)^{1 - \alpha} (B - C)^\alpha|, (1 - \beta)\times A + \beta B - C) $$

Where $S(v, u)$ copies the signs of $u$ and applies them to $v$ element-wise.

More information here.

A	B	C	M	wc