ARToolKitPlus

assert error in Tracker::arMultiGetTransMatHull

Bug #691600 reported by DeadVirus on 2010-12-17

This bug affects 2 people

Affects		Status	Importance	Assigned to	Milestone
	ARToolKitPlus	Fix Released	Undecided	Unassigned

Bug Description

The line ~100: "assert(iabs(idx0-idx1)>0);" in function Tracker::arMultiGetTransMatHull(...) fails. Also, idx0,idx1,idx2,idx3 should be initialized before being used (otherwise it can cause problems).

Doing "mTrackerMulti->setHullMode(ARToolKitPlus::HULL_OFF);" stopped the problem (but I don't know what this does).

This happens in the trunk version.

See original description

DeadVirus (amfcosta13) on 2010-12-17

description:

updated

Revision history for this message

Pavel Rojtberg (rojtberg) wrote on 2010-12-18:

you can look at the description of setHullMode() to see what it does. I will look at the problem as time permits.

Revision history for this message

Pavel Rojtberg (rojtberg) wrote on 2011-03-30:

just had some time to look at it: idx_ are not used unitialised; they are passed as references.
as for the assertion - I cant reproduce it here. Can you attach a test-image that triggers it?

Changed in artoolkitplus:
status:	New → Incomplete

Revision history for this message

David Cabecinhas (david-cab) wrote on 2011-04-03:

Sample image that results in segfault Edit (415.2 KiB, application/octet-stream)

I'm having the same problem.
It looks like it happens when no marker is detected.

I'm attaching a test-image so you can reproduce the problem.
I get a segfault when testing it with the samples/multi code.

Revision history for this message

David Cabecinhas (david-cab) wrote on 2011-04-03:

Making the following change in line 93 of arMultiGetTransMatHull.cpp

if( (hullTrackingMode==HULL_FOUR) && (numInPoints != 0))

gets the code running, resulting in the following output for the sample/multi code:

"0 good Markers found and used for pose estimation.
Pose-Matrix:
  0.00 0.00 0.00 0.00
  0.00 0.00 0.00 0.00
  0.00 0.00 0.00 0.00
  0.00 0.00 0.00 1.00"

Caveat: There may be better codepoints for checking for this condition, I'm not acquainted with the ARTKP code.

Pavel Rojtberg (rojtberg) on 2012-05-16