v1 vs. v2 models

You should be able to use both v1.x and v2.x models. With a small warning, if you (for example) use sd.textencode() the size of the tensor will be different depending on which version of model you use. [1, 77, 768] for v1 and [1, 77, 1024] for v2.

If you get an error similar to this you probably have a tensor of the wrong size somewhere. But don't worry, just change the size and try again. (But you might have to reset or clear the gallery if you've added a broken image.)

Error: 'ERROR: Sizes of tensors must match except in dimension 0. Expected size 768 but got size 1024 for tensor number 1 in the list.'