Svg Image ♾️Mixlab

Svg Image takes vector artwork and hands it to the rest of ComfyUI, which otherwise only speaks raster. Upload an SVG through the node's own widget, and it does two things at once: gives you back a flat IMAGE you can treat like any raster load, and - the more interesting half - decomposes the SVG into its constituent shapes and hands those back as a LAYER list, Mixlab's own layer type. That second output is what makes this node genuinely useful rather than just a fancy image loader: if your SVG has separate paths or groups (icons, illustrations, poster elements exported from Illustrator or Figma), each one can arrive as its own positionable layer instead of one flattened blob.

That layers output plugs directly into Mixlab's own layer-compositing system - the same one built around New Layer and Merge Layers. Feed the layers this node extracts into that chain (or straight into Merge Layers, if you don't need to reposition anything first) and you get a composited IMAGE plus MASK back out, ready to sample on top of, feed into ControlNet, or save as-is. In effect, Svg Image is the on-ramp that lets a vector design become individually-editable layers inside a pipeline that otherwise only ever works with flat pixels.

There's exactly one input, upload, typed SVG - the file you drop into the node's widget. There's no width/height/scale field exposed on this node itself; if you need a specific raster size out of the flat IMAGE output, resize it downstream the normal way. The two outputs are IMAGE (the flattened raster, straightforward to use anywhere a plain image goes) and layers (a LAYER list, one entry per decomposed shape or group, meant for New Layer / Merge Layers).

Install is the pack-wide standard:

cd ComfyUI/custom_nodes
git clone https://github.com/shadowcz007/comfyui-mixlab-nodes.git
cd comfyui-mixlab-nodes
install.bat

or via ComfyUI Manager (search "comfyui-mixlab-nodes"), or pip3 install -r requirements.txt in a venv. No model download needed - this is file parsing and rasterization, not inference.

If you only ever see one layer come back regardless of how your SVG was structured, the likely cause is upstream of this node: an SVG exported as a single flattened path (common if the source file was "outlined" or exported without preserving groups) genuinely has nothing more granular to decompose - check the source file in a vector editor first to confirm it actually has separate paths or groups before assuming the node dropped something. If text elements or gradients look wrong or missing after import, that's a common gap in SVG parsers generally - fonts and certain gradient/filter definitions don't always survive a conversion cleanly, so converting text to outlined paths and flattening unusual effects in your vector editor before export is the safer bet if fidelity matters. And if the layers output won't connect where you expect, remember it's Mixlab's own LAYER type - it's built to feed New Layer or Merge Layers specifically, not a generic image-list input from another pack.