DEV Community

keeper
keeper

Posted on

Printsight v0.2 — Now Shows Exactly Where Your 3D Print Defects Are

#python #3dprinting #opensource #cv

Last week I released Printsight — an open-source CLI tool that detects 3D print defects from a photo using pure OpenCV.

The feedback was consistent: "The text report is useful, but show me where on the print the problems are."

So that's exactly what v0.2 does:

printsight my_print.jpg --annotate
# → my_print_annotated.png with color-coded defect markings
Enter fullscreen mode Exit fullscreen mode

What You Get

Run with the --annotate flag and Printsight saves a marked-up version of your photo:

printsight failed_print.jpg -a
Enter fullscreen mode Exit fullscreen mode

The output image uses four colors:

Color What it marks
🔴 Red Stringing — thin polylines over detected wisps
🟡 Yellow Layer issues — horizontal bands where regularity is poor
🔵 Blue Warping — circles on lifted corners
🟢 Green Reference line — where the bottom edge should be

Plus a score overlay in the top-right corner.

How the Annotations Work

Stringing (Red)

The stringing detector uses two methods in parallel:

  1. Hough Line Transform on Canny edges — finds thin, non-horizontal line segments
  2. Morphological erosion — erodes the print body to isolate thin wisps

Both sets of detected features are drawn as red polylines directly on the image. More red = more stringing.

Layer Issues (Yellow)

The layer analyzer computes horizontal gradient magnitude → projects onto vertical axis → measures peak regularity.

When peak spacing deviates significantly from the mean, that horizontal band is highlighted in yellow. The more bands, the more irregular the extrusion.

Warping (Blue + Green)

The warping detector finds the print's bottom edge (lower 20% of bounding box) and fits a straight line to it.

  • Green line: the ideal straight bottom edge
  • Blue circles: points that deviate significantly from the line (lifted corners)
  • Green dots: normal, well-adhered bottom points

Full Workflow with v0.2 

# Before tuning — get a baseline
printsight bridge_test.jpg -a
# Shows red stringing strands + warped corners on the annotated output

# After tuning retraction + bed temp
printsight bridge_test.jpg -a
# Compare annotated images side by side
Enter fullscreen mode Exit fullscreen mode

This is the feedback loop that makes iterative tuning actually work. No more squinting and guessing.

Companion Tools

Printsight is part of a three-tool open-source workflow:

SupportSage ──► FilamentDB ──► Printsight
(optimize)      (settings)      (inspect)
Enter fullscreen mode Exit fullscreen mode
  • SupportSage — AI-optimized support structures (33% material savings)
  • FilamentDB — 25+ filament profiles with recommended settings
  • Printsight v0.2 — print quality inspection with visual annotations

All MIT licensed, pure Python, zero ML dependencies.

Install:

pip install https://github.com/bossman-lab/printsight/releases/download/v0.2.0/printsight-0.2.0-py3-none-any.whl
Enter fullscreen mode Exit fullscreen mode

Usage:

printsight photo.jpg -a
Enter fullscreen mode Exit fullscreen mode

Links:

Top comments (0)