☁️BizyAir Binary Lines
BizyAir Binary Lines — ComfyUI Node Guide
- image
- IMAGE
This is the bluntest tool in BizyAir's line-extraction lineup, and that's not a criticism - it's genuinely the right choice for a specific, narrow job. Where Canny does actual edge detection with gradient analysis and Lineart uses a trained model to produce natural-looking outlines, Binary just picks a brightness cutoff and slams every pixel to pure black or pure white on either side of it. No cleverness, no learned features, just a threshold.
When that's actually the right call
Reach for this specifically when your source image is already mostly line art or high-contrast content - a manga page, an already-inked sketch, a screenshot of flat-colored comic art. In that situation, a smart edge detector or a trained lineart model can actually work against you: it'll try to interpret shading, gradients, or anti-aliasing artifacts as meaningful structure and produce noisier output than the source deserves. A hard threshold, by contrast, just cleanly separates "this is line" from "this is not," which is exactly what a genuinely binary source image wants.
For photographs or anything with real tonal gradation, this is the wrong preprocessor - you'll get a harsh, blotchy black-and-white result that throws away far more information than Canny or a softedge preprocessor would. Save this one for source material that's already close to line art.
Inputs and outputs
- image - required.
- bin_threshold - optional, 0 to 255, default 100. This is the actual cutoff: pixels brighter than the threshold go one way, darker go the other. Push it up if too much of your image is being classified as "line" (over-inclusive), push it down if fine details are getting lost into solid black or white regions.
- resolution - optional, default 512, range 64–1024 in steps of 64. Processing resolution, separate from your output size.
Output: a single IMAGE, a pure black-and-white line map, ready for a ControlNet Apply node paired with a line/canny-trained ControlNet checkpoint.
How to install it
Ships with the full BizyAir pack.
ComfyUI Manager: search "BizyAir," Install, restart.
git clone:
cd ComfyUI/custom_nodes
git clone https://github.com/siliconflow/BizyAir.git
API key required, same as every node in the pack.
Common issues
Result looks like harsh, blotchy patches instead of clean lines. This almost always means the source image had real tonal gradation (a photo, a shaded illustration) that a hard threshold isn't equipped to handle. Switch to BizyAirCannyEdgePreprocessor or a lineart-style preprocessor instead - Binary is for content that's already close to pure line art.
Threshold tuning feels like trial and error. It kind of is, by design - there's no "correct" threshold in the abstract, only what separates line from non-line for your specific source image's brightness distribution. If the default of 100 isn't working, nudge it in increments of 20–30 and check the output visually rather than trying to reason your way to the right number.
Fine detail disappearing. At certain threshold values, thin lines or small details can get absorbed entirely into a solid region on one side of the cutoff. If detail matters, try a slightly higher resolution setting alongside threshold adjustments - that gives the thresholding operation more pixels to work with before it collapses detail.
Inputs (3)
| Name | Type | Default | Description |
|---|---|---|---|
| image | IMAGE | — | |
| bin_thresholdopt | INT | 1000–255 | — |
| resolutionopt | INT | 51264–1024 | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| IMAGE | IMAGE | — |