ComfyUI Node: ⏱️ Latent Motion Retime

⏱️ Latent Motion Retime:

LatentMotionRetime is a specialized node designed to correct the pacing of an already-upsampled latent sequence. Its primary function is to adjust the velocity of each frame in a sequence to achieve a natural motion pace, particularly when converting from a lower frame rate to a higher one. This node is particularly useful for AI artists who work with video sequences that have been upsampled from a lower frame rate, such as 24fps, to a higher frame rate, like 30fps. By attenuating the per-frame velocity based on the ratio of the source and target frame rates, LatentMotionRetime ensures that the motion in the video appears smooth and natural, avoiding the unnatural speed-up that can occur when frame rates are mismatched. This correction is crucial for maintaining the intended artistic expression and ensuring that the motion in the video aligns with the viewer's expectations.

⏱️ Latent Motion Retime Input Parameters:

latent

The latent parameter represents the input latent sequence that has been upsampled and requires pacing correction. It is a multi-dimensional tensor containing the video frames that need to be adjusted for natural motion pacing. This parameter is crucial as it provides the raw data that the node will process to achieve the desired frame rate correction.

source_fps

The source_fps parameter specifies the original frame rate of the latent sequence before upsampling. It is a critical value that determines how much the velocity of each frame needs to be adjusted to match the target frame rate. The default value is typically set to the frame rate at which the original content was created, such as 24fps.

target_fps

The target_fps parameter indicates the desired frame rate for the output sequence after pacing correction. This value is essential for calculating the attenuation factor that will be applied to the frame velocities. The default value is often set to a standard playback frame rate, such as 30fps, to ensure smooth motion.

⏱️ Latent Motion Retime Output Parameters:

latent

The latent output parameter is the corrected latent sequence with adjusted frame pacing. This output retains the original structure of the input latent but with modified frame velocities to ensure natural motion at the target frame rate. The corrected sequence is ready for further processing or direct playback, providing a seamless viewing experience.

⏱️ Latent Motion Retime Usage Tips:

• Ensure that the source_fps and target_fps parameters are set accurately to reflect the original and desired frame rates. This will ensure that the pacing correction is applied correctly.
• Use LatentMotionRetime in conjunction with other nodes like LatentMotionSharpener to enhance both the temporal and spatial quality of your video sequences.

⏱️ Latent Motion Retime Common Errors and Solutions:

"Expected 5D, got {samples.shape} - passthrough"

• Explanation: This error occurs when the input latent does not have the expected five-dimensional shape, which is necessary for processing.
• Solution: Verify that the input latent is correctly formatted as a 5D tensor with dimensions corresponding to batch size, channels, frames, height, and width. Adjust the input data structure if necessary.

"MotionRetime: {samples.shape} | scale={scale:.4f} atten={atten:.4f}"

• Explanation: This message indicates that the attenuation factor is too low, resulting in minimal or no pacing correction.
• Solution: Check the source_fps and target_fps values to ensure they are set correctly. If the attenuation factor is still too low, consider adjusting the frame rates or reviewing the input data for inconsistencies.