For MiniMax H3 videos LONGER than about 15 seconds: 30 seconds, a minute or more, made in one run with no joins. For 15 seconds or less you do not need it (with it, nothing changes). Put it on the model wire before H3 Audio Sync and the sampler, and set the video length as long as you want.
H3 was trained on clips of up to about 15 seconds. This node still makes the whole video as one piece, but on every step the model looks at it in overlapping windows of about 10 seconds, each with its own part of your song, and the windows are blended where they overlap. Because every window is worked on at every step, they agree with each other and there is no seam, flash or quality drop anywhere.
It also needs far less graphics memory than one long run and is faster: a minute at 864 took 4.5 minutes instead of 9.4 on a 4090, and with only 12 GB of graphics memory free it ran at the same speed (a minute at 864, and 30 seconds at 1024), where one long run could not finish a step.
This node only works with MiniMax H3. Find it by searching for h3, minimax, long video, windows.
By pixaroma·Created 7 months ago·Updated a day ago· 449
H3 Long Video Pixaroma
model
model
◄window_seconds10.0►
◄overlap_seconds1.5►
Category👑 Pixaroma/🎵 Audio
Inputs (3)
Name
Type
Default
Description
model
MODEL
Your MiniMax H3 model (after its LoRAs). Wire the output on to H3 Audio Sync and the sampler.
window_seconds
FLOAT
10.04–15
How much of the video the model looks at at once. 10 seconds is the tested default; H3 was trained on 4 to 15 seconds. Shorter windows use less memory but hold together a little less well. Snapped to H3's frame grid (10 becomes 10.12). A video no longer than one window is made in one piece as usual.
overlap_seconds
FLOAT
1.50.5–7
How much two neighbouring windows share; they are blended across it. 1.5 seconds is the tested default. More overlap = smoother but slower. Snapped to H3's frame grid (1.5 becomes 1.62).
Outputs (1)
Name
Type
Description
model
MODEL
Your model, set to make long videos in windows. Wire it to H3 Audio Sync (or straight to the sampler).