Why has everyone been making versions of Dyno_High recently?
#2
by
pymo
- opened
To be honest, there have really been too many new models and technologies lately. For someone like me who wants to conduct slightly in-depth research, it's a bit overwhelming.
May I ask why everyone (For example"Kijai") has started launching the "Dyno_High" version of the model recently?
Is it mainly to speed up the composition generation of T2V? And then use the ordinary LOW model for refinement? I wonder if my understanding is correct.
Basically they released the speed lora v2, but it had issues so they released the full model too, and as you said you are supposed to use this high and normal + lora low, that way you can generate stuff in 4-8 steps with higher quality (;