Camera mapping fails

Check the required point count

Camera Mapping requires at least 3 non-collinear point pairs. Camera Alignment requires at least 4 well-spread pairs for perspective correction. Its recommended 9-point layout also estimates radial lens distortion; the solver needs at least 6 pairs to estimate that distortion.

If the points are on one line

Both sides of the point pairs must span a non-zero area. Three points along one bed edge are not enough. Use points that form a large triangle, or 4 or more points spread around the useful image area.

If an image point is outside the frame

Image X must be between 0 and the frame width. Image Y must be between 0 and the frame height. Capture a current frame and check its dimensions at the top of the dialog.

If RMSE is high

Recheck every pair. Each image point must refer to the same physical feature as its reference or workspace coordinate. One transposed coordinate can pull the whole fit away from the other points.

Use clearly identifiable features and spread them across the area where you will place jobs. Closely clustered points make errors outside that cluster harder to detect.

If the center aligns but the edges do not

Camera Mapping fits uniform scale, rotation, and translation. For perspective or lens curvature, use Align Camera and its recommended 9-point layout. Include both center and edge reference points. A low error at the supplied points still needs a physical check elsewhere on the bed.

Use a less distorted lens, increase the camera-to-bed distance, and mount the camera as square to the bed as practical.

Verify the result

• Lower RMSE is better, but do not use Quality alone as a pass/fail threshold.
• Check several known physical positions after saving.
• Run Frame before starting a job placed with the camera.

Still stuck?

• Camera Mapping dialog
• Camera Alignment dialog
• Map the camera guide
• Post in the Facebook group with the camera model, frame dimensions, and point pairs.