Emiewn Img Paste

The name is the whole pitch: click the node, hit Ctrl+V, and whatever's in your clipboard is now an image in your graph. No "save screenshot to a file, find it in the ComfyUI input folder, wire up a Load Image" dance. Emiewn Img Paste turns the node itself into a paste target, which is one of those tiny quality-of-life things you stop noticing until you're on a machine that doesn't have it.

How it works

The node is half Python, half frontend. A JavaScript extension draws a custom box on the node - click it to focus, then Ctrl+V a screenshot, or just drag and drop an image file onto it. Either way the file gets uploaded to ComfyUI's /upload/image endpoint, and the returned filename is written into the node's hidden string input. The backend then loads that path, applies EXIF orientation (so phone photos don't come in sideways), and hands you back a tensor.

Two things make it nicer than the stock Load Image. First, it shows an instant preview with the resolution in the corner, and the node auto-resizes to match your image's aspect ratio. Second, and this is the part most paste nodes skip, it also returns a mask built from the image's alpha channel.

Inputs and outputs

Only one input, and it's not what it looks like: image is a STRING holding the ComfyUI input filename that the frontend wrote for you. You basically never type here - it's a pipe the JS fills in.

• IMAGE - your pasted image, ready to wire into anything that takes an image.
• MASK - where the transparency is. If you paste a PNG with an alpha channel, transparent pixels become a white mask; opaque areas are black. Paste a screenshot or JPEG with no alpha and you get an all-zero mask.

That mask output is the reason to reach for this over a plain load node: paste a transparent sticker, logo, or cutout and you can feed the mask straight into compositing or inpainting without building one yourself.

Installing it

This is one of five nodes in the Emiewn pack, so you install the whole thing:

cd ComfyUI/custom_nodes
git clone https://github.com/emiewnn/ComfyUI-Emiewn-Nodes.git

Then restart ComfyUI. Or search "ComfyUI-Emiewn-Nodes" in ComfyUI Manager and install from there. The pack's requirements (huggingface-hub, opencv-python, pyyaml, omegaconf) install alongside it; the paste node itself is dependency-light, so none of the heavy GIMM-VFI stuff in this pack matters for it.

Notes and gotchas

• The paste box needs focus first. If you paste while the canvas has focus, ComfyUI will happily drop the image on the graph instead of into the node. Click the box, then paste.
• If you run headless or queue up workflows without the UI, this is not your node - the frontend does the upload, so it won't work in pure API mode.
• It's a thin convenience wrapper, not a clever image tool. If you need resize, crop, or EXIF stripping, put nodes after it. For "grab a screenshot and get it into the graph in three seconds," it's hard to beat.