About Frame control

[YuehengLuo]

Hi！
How can I control the number of frames of model input and output?
1.I found that using the model weights you provided for text to motion to get a video length of 6 seconds the results_len.txt that I get shows 120 frames per video. This means that there are 20 frames per second, can I control this parameter like 5 frames per second or that the output is only 1 frame.
2.Can the input to the model be just one frame? My goal is to generate a single frame of the image that best fits the text.

[GuyTevet]

1. The model was trained with a 20fps dataset. No flexibility here (except training with different data)
2. Naiivle, no. Try our recent https://setarehc.github.io/CondMDI/