>>17249
it's absurd how much more powerful grok is. for the WAN stuff i'd need to load in different LORAs depending on what i wanted to happen on the portion of the video, generate starting/ending pictures, and then use a ton of keywords during prompting to try and summon up some violence.

for grok you can just type "shoot the catgirl in the head" and get something cohesive - it's wild tech and i think it's spoiled me from going back to less powerful models. (yes that's literally all i wrote in the prompt after generating the image)