<aside> 🛠

Classical Canny Edge Maps - Ready for ControlNet-Canny

Turn any image into a clean white-edges-on-black edge map using OpenCV's classical Canny algorithm. CPU-only, no neural network, fully deterministic. The output is exactly the format ControlNet-Canny expects, so this is the standard preprocessor for any image-generation flow that uses edges as structural guidance. Two intuitive sliders (min / max edge strength) control how aggressively edges are detected..

</aside>

What it does

Canny Edge Annotator runs OpenCV's classical Canny edge detector on the input image and outputs a binary edge map - white edges on a black background. Internally the algorithm computes image gradients, suppresses non-maxima, and applies double-threshold hysteresis: pixels above the Maximum edge strength are accepted as "strong" edges; pixels between the two thresholds are kept only if they're connected to a strong edge; pixels below the Minimum edge strength are discarded. The result is the canonical input format for ControlNet-Canny.

Problem it solves

• ControlNet-Canny conditioning – Generate the exact white-edges-on-black control image that ControlNet-Canny expects, in one node
• Structural guidance for generation – Drive Image Generation / Image Edit models to preserve the silhouette and layout of a reference image
• Deterministic, CPU-only edges – Classical algorithm; same input always produces the same output (no neural-network variance)
• Fast preprocessing – No GPU required, runs instantly even on large batches
• Reference for sketch / line-art workflows – Convert photos into clean line drawings for downstream stylization
• Composition control – Lock in the geometry of a scene before letting a generative model fill in the style
• Diagnostic / inspection tool – Visualize edge strength and detail level in source images before deciding on resolution / cropping
• Stable across runs – Unlike learned edge detectors, output is fully reproducible from the same parameters

Input/Output

<aside>

• Input: Single image
  • Format: PNG, JPEG, or standard image formats
  • Resolution: Any size (CPU-only - runtime scales linearly with pixel count)
  • Content: Any image; works best on images with clear shapes / contours </aside>

<aside>

• Output: Binary edge map

  • Format: PNG (grayscale)
  • Content: White pixels (255) on edges, black pixels (0) elsewhere
  • Resolution: Same as input
  • Use: Drop directly into ControlNet-Canny or any downstream tool that consumes edge maps

  

</aside>

Configuration Options

• Minimum edge strength (lower threshold): Pixels below this gradient strength are discarded outright.
  • Default: 100
  • Lower value = more edges detected (including weaker / noisier ones)
  • Higher value = fewer edges; cleaner but less detail
  • Role in the algorithm: Used during hysteresis — borderline pixels (between min and max) are kept only if they touch a strong edge

• Maximum edge strength (upper threshold): Pixels above this gradient strength are accepted as definite ("strong") edges.
  • Default: 200
  • Lower value = more pixels qualify as strong edges → busier output
  • Higher value = only very strong gradients qualify → only the most prominent edges kept