EzFlex-TimePlan
- timeplan
- frames
◄modelminimax_h3►
◄total30.0►
◄segment5.0►
◄fps24►
◄overlap22►
◄tolerance0►
◄alignalign►
◄config{}►
◄trim_tail0►
◄audio_context_length24►
◄grid_b5►
◄grid_s17►
◄max_frames362►
◄index0►
CategoryEzFlex
Inputs (14)
| Name | Type | Default | Description |
|---|---|---|---|
| model | COMBO | minimax_h3 | 3 options: minimax_h3, wan, ltx |
| total | FLOAT | 30.0 | Total length in seconds. Split into ceil(total / segment) segments. |
| segment | FLOAT | 5.0 | Target length of each segment in seconds (before overlap). |
| fps | INT | 241–120 | — |
| overlap | INT | 220–200 | Frames re-used from the previous segment's tail. Snapped DOWN to the model grid (MiniMax H3: 0 or 5+17k). |
| tolerance | INT | 00–5 | Extra HEAD frames trimmed after generation for the transition (total head trim = overlap + head). Generation is NOT increased for it; total length gets shorter. |
| align | COMBO | align | align = total length wins; strict = every segment (but the first) uses the SAME grid step. |
| config | STRING | {} | Plan panel state (segOverride = per-segment manual durations, seconds). |
| trim_tail | INT | 00–24 | Frames dropped from the END of each non-final segment (also shortens the plan, so the timeline stays in sync). |
| audio_context_length | INT | 240–400 | Transition audio length in video frames (official H3 example: video 22 / audio 24). MotionContext uses this when it follows the plan. |
| grid_b | INT | 50–1000 | Frame-grid base B. Frames must be B + S*k (H3 default 5 + 17k). |
| grid_s | INT | 171–1000 | Frame-grid step S. Frames must be B + S*k (H3 default 5 + 17k). |
| max_frames | INT | 3621–100000 | Maximum generated frames per segment. |
| indexopt | INT | 0 | Segment index (0-based). Wire EzFlex-LoopStart.index here. Out of range = 0 frames (lets the loop finish). |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| timeplan | EZFLEX_TIMEPLAN | — |
| frames | INT | — |