Yes, I am having the same issue. Masked control net is a great feature!
At the moment it fails with an error
The size of tensor a (12555) must match the size of tensor b (12420) at non-singleton dimension 1
if I try to specify mask_optional. Without mask_optional works still great! Would be perfect to have such option available with flux also.
Yes, I am having the same issue. Masked control net is a great feature! At the moment it fails with an error
The size of tensor a (12555) must match the size of tensor b (12420) at non-singleton dimension 1if I try to specifymask_optional. Withoutmask_optionalworks still great! Would be perfect to have such option available with flux also.