Recraft V4 Styles Pro Vector
When the deliverable is an SVG, not pixels
- referenceImages
- file_path
Every other image node in this pack hands you pixels. This one hands you an SVG. Recraft V4 Styles Pro Vector is the vectorization lane of Recraft's design model - you describe a graphic, apply a style, and instead of an IMAGE you get an actual vector file that scales to a billboard without ever getting blurry. If you've ever generated a logo at 1024px and then watched it pixelate on a poster, you already know why this node exists. It's the deliverable-shaped output that raster models can't give you.
The way it feels different in ComfyUI starts with the output. This node's return type isn't an image tensor - it's file_path, a string pointing at a file saved into your ComfyUI output folder. That's the pack's mechanism for non-raster results: the README is explicit that files save to your output folder and the node returns the path. So the mental model shifts from "wire this into Preview Image" to "run it, then find the .svg in your output directory and bring it into your design tool." The outputFormat default says it plainly: SVG.
Inputs are the Recraft family's style controls in their leanest form. positivePrompt is required, sentence-style. referenceImages supplies a visual reference. styleId applies a Recraft style - the same style-ID system as the raster V4 nodes - and settings.styleMatch (default precise) sets how strictly it's followed. size starts at 2K (1:1), and advanced_json remains the route to background-color and palette overrides. That's the whole list - notably no numberResults for variations and no watermark group, because vectorization is a different pipeline and this node stays out of its way. Output is the file_path string.
Install
The shared pack setup, once: ComfyUI Manager → search Runware → install → restart, or clone https://github.com/Runware/ComfyUI-Runware into custom_nodes and pip install -r requirements.txt. Runware API key from the dashboard (Settings → Runware API key, RUNWARE_API_KEY, or runware auth login), then you're set.
The honest caveats
Recraft V4 is a closed hosted model, and this node is the most niche use of it - you're paying cloud inference rates for vector output that's genuinely hard to generate well. Keep expectations realistic: AI vectorization produces clean, editable shapes for the graphic-design material Recraft is good at (logos, icons, flat illustration), and it is not magic for dense photographic scenes. And note the workflow difference: since the result is a saved file, there's no live IMAGE to preview in the graph - check your output folder, and remember the per-run cost on the node's title bar is the price of a file that would otherwise have cost you an hour in a vector editor.
Inputs (9)
| Name | Type | Default | Description |
|---|---|---|---|
| positivePrompt | STRING | Text prompt describing elements to include in the generated output. | |
| referenceImagesopt | IMAGE | — | |
| sizeopt | COMBO | 2K (1:1) | Output size. Only one of resolution or dimensions is sent. |
| styleIdopt | STRING | Identifier of an existing style to apply. | |
| settings.styleMatchopt | COMBO | precise | How closely the style is followed. Overrides the match value stored on the style for this request. |
| ttlopt | BOOLEAN | false | Enable to set ttl. Off uses the model's default. |
| ttl_valueopt | INT | 60 | Time-to-live (TTL) in seconds for generated content. Only applies when `outputType` is `URL`. |
| outputFormatopt | COMBO | SVG | File format for the vectorized output. |
| advanced_jsonopt | STRING | Optional JSON merged into the request. For: settings.backgroundColor.rgb, settings.colors |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| file_path | STRING | — |