Version 3.3
Version 3.3.2, released 15 Mar 2018 (rev 16, 19 Mar 2018; 08 Nov 2018)
This release fixes a bug that essentially disabled the velocity stabilization feature in Trial Mode. While the target would "snap" to the current eye position at the start of the segment, Maestro failed to adjust the target position IAW eye movement from that point on. This bug has existed since the introduction of the velocity stabilization sliding window average in V2.7.0. Apologies!
(Revised 19 Mar 2018) When velocity stabilization is initiated with the Snap to Eye feature AND a nonzero instantaneous position change (Xo, Yo) is specified for the VStab target, Maestro now sets the target position to (Xe + Xo, Ye + Yo) -- where (Xe, Ye) is the current eye position --, regardless if the position change is ABSolute or RELative. With Snap to Eye on, ABSolute positioning doesn't really make sense.
(Revised 19 Mar 2018) A subtle change to velocity stabilization: RMVideo target trajectories must be calculated 3 display frames (roughly 30-40 ms, depending on the RMVideo frame rate) in advance. Previously, this was NOT done with the VStab compensations, which meant that the VStab effect on an RMVideo target was delayed slightly WRT when it was initiated in the trial segment table. Now, VStab adjustments are made "in advance" as well, so the VStab effect on any RMVideo target no longer lags the actual trial timeline. However, the compensation is computed using eye position samples earlier in the timeline.
(Revised 08 Nov 2018) Corrected a coding error (since V3.3.0) that made the mid-trial reward feature behave incorrectly.
NOTE: If you did not install V3.3.1, please review the Installation Note for that release (below).
Version 3.3.1, released 16 Nov 2016 (rev 21 Nov 2016, 31 Jan 2017, 22 Feb 2017)
New feature: An "image" target for RMVideo. Several researchers have asked for a target that displays a single image; some have "jury-rigged" the Movie target by creating a single-frame video file. With this release, that is no longer necessary. The new RMVideo Image target loads an image from a JPEG, PNG, BMP or PSD file and displays it at any location on the screen; the location may be updated on a frame-by-frame basis, so you can "pan" the image at any direction and speed over the course of a trial. As with video source files for the Movie target, image source files for the Image target are downloaded from Maestro to RMVideo's "media store", formerly known as the "movie store". This store lies under the "media" subdirectory in RMVideo's installation directory. [When the new RMVideo first launches, it checks for the old store directory named "movies". If present, that directory is simply renamed "media"; otherwise, a new "media" subdirectory is created.]
Note that the analysis utilities readcxdata() and JMWork were both updated to recognize the new RMVideo Image target, but the Maestro data file version remains the same at V=20. The change was very minor, and it is not necessary to get the new versions of these utilities unless you want to be able to check the Image target's source folder and file name.
(Revised 21 Nov 2016) (1) Increased the maximum allowed value for Inter-trial Delay (on the Protocol tab in Trial mode) from 2000 to 9999 milliseconds. (2) Bug fixed: Maestro would crash during exit if you had interacted with the grid control on either the Analog IO or Event Timer DIO tabs in Test & Calibration mode. This was a problematic bug, because a reboot was required in order to relaunch Maestro successfully. An RTX "blue screen" crash also occurred during testing to address the issue.
Installation Note **IMPORTANT **: This release requires that both Maestro and RMVideo (version 7) be updated. Since the auto-update mechanism for RMVideo (over the Maestro-RMVideo connector) was eliminated in Maestro 3.2.1, both programs must be manually installed on their respective machines. Furthermore, RMVideo v7 is only built for an up-to-date Linux machine running Ubuntu 14.04.x LTS. Therefore, if you want to use the new RMVideo Image target, you must commit to upgrading your RMVideo workstation accordingly.
(Revised 31 Jan 2017) Updated the maestrodoc() Matlab utility and the Import JMX Doc command in Maestro to support defining an RMVideo "Random Dot-Patch" target in the two-color contrast mode in script-generated experiment documents (JSON format). See Scripting Experiments in Matlab for more information.
(Revised 22 Feb 2017) New trial sequencer mode added: The Randomized (Repeat) sequencer mode is identical to the Randomized mode except that, if the subject fails to complete the current trial, that trial will be repeated over and over again until it is successfully completed.
Version 3.3.0, released 03 Oct 2016 (rev 11 Oct 2016)
New feature: Random Variables in a Trial. You can now define up to 10 different random variables in a trial object and assign them to a variety of parameters in the trial's segment table: segment duration; target window position, velocity and acceleration (H, V); and target pattern velocity and acceleration (H,V). Four types of RV distributions are supported: uniform, normal, exponential, and gamma. In addition, an RV can be defined as a function of zero or more distribution-type RVs. The Trial Editor has been reorganized somewhat to accommodate the trial's random variable list. All of the widgets above the segment table are now laid out in three tab panes: Main, Perturbations/PSGM, and Random Variables. For details, consult the chapter on the trial object and the Trial Editor. Note: This feature introduces some changes to the Maestro experiment document, so a document saved in this version cannot be opened in a previous Maestro release.
Bug fix: Maestro's EyeLink service thread sometimes fails during certain operations in Continuous mode -- such as when turning on fixation with one or more active targets -- with an error message "[Eyelink] Stopped recording on error: IPC buffer overflow". This can happen if the RTX-based driver takes too long to execute a command from the Maestro GUI -- so that it fails to service the EyeLink sample buffer before it overflows. Some technical changes were made in an effort to address this issue.
RMVideo bug workaround: There was a long-standing bug in RMVideo's Bar/Line target -- If the bar width > height, RMVideo displayed it with W == H. The bug is very minor and there's an easy workaround: If you want a horizontal bar with W=10 and H=1, swap W<==>H and add 90 deg to the target's "drift axis". If you're using the more modern build of RMVideo, an updated version 6 that fixes this bug is available for download. If you're using the 2009-era build, use the workaround if you need a Bar/Line target.
(Revised 10/10/2016) The random number generator that determines the trial sequence in "randomized" modes is now seeded IAW the current time to ensure that, each time Maestro starts up, the sequence of trials generated is different.