πΌοΈ Load Image Temporarily
Load an image without cluttering your input folder
- image
- mask
- width
- height
Stock LoadImage copies whatever you pick into ComfyUI's /input folder, and it just sits there forever - every test render, every reference image you've ever tried, permanently cluttering the file picker. Load Image Temporarily does the same upload-and-load job, but drops the file into /temp instead, so it doesn't pile up.
How it works
Same file-picker widget as stock LoadImage, same alpha-mask handling - the node's own description notes the mask is "inverted like ComfyUI LoadImage," so it matches the behavior you're already used to. The only real difference is where the file physically lands on disk: /temp instead of /input.
That distinction matters more than it sounds like on paper. ComfyUI's /input folder is also what populates the file dropdown on every LoadImage-style node in every workflow you open - so every image you've ever tested with is sitting in that same dropdown, forever, getting harder to scroll through the longer you use ComfyUI. Routing test uploads through /temp instead keeps that list to the images you actually meant to keep around.
The inputs and outputs that matter
image- the file picker/upload widget. That's the only input.
Outputs: image (the loaded tensor), mask (the alpha channel), width, height.
How to install it
Via ComfyUI Manager: search ComfyUI-mnemic-nodes, install, restart. Manually:
cd ComfyUI/custom_nodes
git clone https://github.com/MNeMoNiCuZ/ComfyUI-mnemic-nodes
Restart ComfyUI. No model downloads or extra dependencies.
Common issues & troubleshooting
The image disappears after a restart, or after a while. That's /temp doing exactly what it's supposed to - ComfyUI periodically clears its temp folder, including on restart, so nothing you load through this node is meant to stick around long-term. If you need the file to persist for reuse across sessions, use stock LoadImage or this pack's own Load Image Advanced instead, both of which write to /input.
When to actually reach for this one. Anything you're testing once and don't need to keep - a quick reference image, a one-off comparison, images coming in through the API where you'd rather not have every upload permanently listed in ComfyUI's file browser afterward. It's a small quality-of-life swap, not a different feature: same loading behavior as LoadImage, just without the long-term footprint on disk.
When not to. If you're building a dataset you'll revisit, or a reference image other team members or other workflows need to find later, /temp is the wrong destination by definition - reach for stock LoadImage or this pack's own Load Image Advanced instead, both of which keep the file in /input where it'll actually still be there next week.
Inputs (1)
| Name | Type | Default | Description |
|---|---|---|---|
| image | COMBO | 1 options: |
Outputs (4)
| Name | Type | Description |
|---|---|---|
| image | IMAGE | The loaded image tensor. |
| mask | MASK | The image alpha mask (inverted like ComfyUI LoadImage). |
| width | INT | Image width. |
| height | INT | Image height. |