PLEASE SUPPORT ME BY POSTING A VIDEO BELOW!
This is my most ambitious lora by far with a dataset of 2700 video clips which took a very long time to prepare and to train the model so I very much appreciate your support.
No Trigger Needed
Minimax H3 is a very capable model and this lora is not teaching it brand new concepts but is training a style based on 2700 cinematic fantasy video clips ranging from 1 to 3 seconds long. Check out all the samples that include the basic prompts for an idea of the changes to expect. The audio layer was not trained, video only.
