Morning.

>>66333
Wan does seem retarded in following prompts. But probably due to lack of good data to make the textual inversion.
There are better prompt methods, like using annotations on the picture, but I think only the other ones use it.
I haven't tried the other video models yet to know if they are better.