Animate Diff Lora Loader
Camera-motion LoRAs (zoom, pan, tilt, roll) — v2 module only
- lora_stack
- MOTION_LORA_STACK
Don't confuse this with a regular LoRA loader - it's not for character or style LoRAs. AnimateDiffLoraLoader loads Motion LoRAs, a specific small set trained by the AnimateDiff team to control camera movement: zoom in, zoom out, pan left/right, tilt up/down, roll clockwise/counterclockwise. Stack one in and instead of your subject just moving on the spot, the whole "camera" pushes, pans, or rotates around the scene - a much more directed effect than hoping the base motion module does something similar on its own.
They're small (tens of megabytes, not full checkpoints) because they're not teaching the model new content - they're steering the motion module's existing temporal behavior in a specific direction. That's also why this node's output type is a stack rather than a single loaded LoRA: you can chain more than one AnimateDiffLoraLoader together (via the optional lora_stack input) to combine effects - a slow zoom-in stacked with a gentle pan, for example - rather than being limited to picking one.
The inputs and outputs that matter
lora_name- a dropdown populated from whatever motion LoRA files you've placed incomfyui-animatediff/loras/. If the dropdown is empty, that's not a bug, it means the folder is empty - download at least one motion LoRA and restart ComfyUI to see it appear.alpha(default 1, range 0–10) - strength of the effect. 1 is the sane starting point; push it up for a more exaggerated camera move, though at the extremes you'll likely see it fight with or distort your subject rather than just moving the "camera" more.lora_stack(optional, MOTION_LORA_STACK) - feed in the output of a previousAnimateDiffLoraLoaderto chain multiple motion LoRAs together instead of replacing one with another.
Output: MOTION_LORA_STACK - a chainable stack format meant to be wired downstream to wherever this pack's loader consumes motion LoRAs.
How to install it
Comes with the pack, no separate node install needed.
Via ComfyUI Manager: search "comfyui-animatediff" and install.
Manually:
cd ComfyUI/custom_nodes
git clone https://github.com/ArtVentureX/comfyui-animatediff
Restart ComfyUI. Then, separately, download the motion LoRA files themselves - the README points to huggingface.co/guoyww/animatediff - and place them in comfyui-animatediff/loras/. Restart again (or refresh the node) so the lora_name dropdown picks them up.
Common issues & troubleshooting
The camera effect doesn't seem to be doing anything. This is the one gotcha worth knowing before you spend time debugging: the README is explicit that motion LoRAs only work with the AnimateDiff v2 module (mm_sd_v15_v2.ckpt). If your AnimateDiffLoader has a different motion module loaded - v1, v3, or a community-finetuned one - the motion LoRA effect simply isn't going to show up the way you expect, no matter how high you push alpha. Swap your motion module to mm_sd_v15_v2.ckpt first, then troubleshoot further if it's still not working.
lora_name dropdown is empty. You haven't put anything in comfyui-animatediff/loras/ yet, or you added a file without restarting ComfyUI afterward. Both are easy to fix - download a motion LoRA, drop it in the folder, restart.
Combining several motion LoRAs makes the result chaotic instead of a smooth combined move. Camera-motion LoRAs are additive effects on top of an already-moving scene - stacking two strong ones (say, a full-strength zoom and a full-strength pan) can genuinely fight each other. Try lowering alpha on each one when stacking rather than running every LoRA at full strength.
Inputs (3)
| Name | Type | Default | Description |
|---|---|---|---|
| lora_name | COMBO | 0 options: | |
| alpha | FLOAT | 1.0000–10 | — |
| lora_stackopt | MOTION_LORA_STACK | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| MOTION_LORA_STACK | MOTION_LORA_STACK | — |