ComfyUI Node
LTX Latent Resizer (Divisible by 32)
Resizes an IMAGE by scale_factor and rounds both dimensions up to a multiple of `multiple_of`. Run this before VAEEncode -- it operates on pixels, not latents, despite the node's name. Optionally also trims the batch/frame dimension down to a valid count for LTX-Video, Hunyuan Video, WAN, or a custom multiple-of-N+remainder formula via `video_preset`.
LTX Latent Resizer (Divisible by 32)
- image
- IMAGE
- width
- height
- frame_count
◄scale_factor1.00►
◄multiple_of32►
◄resize_methodbilinear►
◄video_presetnone (spatial resize only)►
◄custom_frame_multiple8►
◄custom_frame_remainder1►
CategoryLTXVideo/Utils
Inputs (7)
| Name | Type | Default | Description |
|---|---|---|---|
| image | IMAGE | — | |
| scale_factor | FLOAT | 1.000.1–8 | — |
| multiple_ofopt | INT | 321–256 | 32 fits LTX and most video VAEs; change it if yours needs otherwise. |
| resize_methodopt | COMBO | bilinear | 4 options: bilinear, bicubic, nearest, area |
| video_presetopt | COMBO | none (spatial resize only) | Trims the batch (frame) dimension down to the nearest valid count for the chosen model's VAE, e.g. LTX needs 8k+1 frames. 'none' leaves the frame count untouched (spatial resize only, the original behavior of this node). |
| custom_frame_multipleopt | INT | 81–64 | Only used when video_preset is 'Custom'. |
| custom_frame_remainderopt | INT | 10–63 | Only used when video_preset is 'Custom'. Must be less than custom_frame_multiple. |
Outputs (4)
| Name | Type | Description |
|---|---|---|
| IMAGE | IMAGE | — |
| width | INT | — |
| height | INT | — |
| frame_count | INT | — |