Exactly same dataset and settings for both models: v1a, v1b.

As a result: different images with same seed.

Thats all i wanted to know)

Generation of tusks still bad, so ADetailer in help.