iRecHS2 + NI + MonkeyLogic 5-point calibration not matching exactly

Dear NIMH ML users,

I am using iRecHS2 with an NI analog input setup in MonkeyLogic.
- Raw signal looks correct: horizontal and vertical movements are fine.
- In 5-point projective calibration, the yellow pluses are centered on the cyan targets.
- But the red calibrated points are slightly smaller and shifted, so they do not match the cyan circles exactly.

I am using 5-point 2-D projective calibration with analog X/Y inputs from iRecHS2.

My question is:
What is the usual cause of this remaining scale/offset mismatch, and what settings should I check in MonkeyLogic or iRec?

Thank you.

That is the way it should be. The calibration looks good and there is nothing wrong.

-----

FYI, here is the legend for the symbols. X & Y axes of the map are in degrees.

Yellow crosses: fixation points (in degrees)
   Red circles: raw eye inputs (in voltages or unitless)
  Cyan circles: eye positions transformed from raw inputs through the calibration (in degrees)

  Open circles: results from previous calibration
Filled circles: newly calibrated points

The red circles are not aligned with the yellow crosses or cyan circles because they represent raw eye-tracker coordinates and are not expressed in degrees of visual angle.

Small distances between the yellow crosses and the cyan dots indicate good transformation performance. As the number of fixation points increases, minimizing these distances becomes more challenging. See this to compare how each 2-D transform method performs when there are many fixation points.