Changelog

The setup failure page used to end at a Windows error code and nothing else, which is no help to anyone. It now adds the two things actually worth trying: run the installer again, and if it keeps failing, right-click it and choose Run as administrator. This matters because of a fault in the industry-standard installer engine we build on, not in DirectRT: on some machines it occasionally fails to hand itself the administrator rights it needs, reports “The handle is invalid”, and stops before installing anything. It is intermittent — running setup a second time almost always works — and it cannot happen at all when setup is already running as administrator. Nothing is left half-installed when it happens.

On a PC where IT enforces an Application Control policy, DirectRT would open, take your input file and subject number, and then fail at the moment the study should have started — reporting that it couldn’t load one of its own files. Nothing had changed on the machine; the policy had tightened around it. The cause was ours: a program like DirectRT is a small launcher plus a set of code libraries beside it, and we were signing only the launchers. Windows let DirectRT start and then refused to let it load its own code. Everything we ship is now signed, so a policy that trusts our certificate covers all of it. If your institution locks down software, this is the build to use. One honest limit: if your IT allows programs by file location or by an exact file fingerprint rather than by publisher, they’ll still need to approve DirectRT — no amount of signing on our end changes that.

After updating, the splash screen would appear and stay there, with no window and no error, and Task Manager was the only way out. One file — the part that checks your licence — wasn’t being replaced during the update, so a new DirectRT was left running a months-old copy of it. Fresh installs were never affected, which is why it reached people before it reached us. Updates now replace every file they should. If you’re stuck on an earlier build, this update fixes it on its own — you don’t need to uninstall first, and your licence is untouched either way.

Put auto in the time column instead of counting out the clip’s length in milliseconds — the trial moves on when the clip finishes. If a row starts more than one clip, auto waits for the longest, so nothing gets cut off. Re-cut a clip and the input file doesn’t need touching.

Ctrl+→ ends the stimulus on screen and moves to the next one; Ctrl+Shift+→ abandons the rest of the trial. No more sitting through every 30-second video and every timeout on every pass. Anything you skip is recorded — responses you skipped past are logged as NULL, the session summary says how many trials you cut short and that the run wasn’t clean data, and the stimuli you cut aren’t counted against the timing figures.

It sits beside the tabs, so it’s there whether you’re on Styles or Session Options, and it tells you where you stand — it reads Save when you have unsaved changes and Saved when you don’t. Ctrl+S and File → Save still work as before.

Reports how this computer identifies itself to the licence server — which hardware trait it anchors on, and the fallbacks it uses if that trait changes. If you ever need to ask us about a licence, run this and send us what it says; the values are one-way codes and reveal nothing about your machine.

The Session Options tab is gone — display, reaction time and response feedback are all on the main page, where you can see them. Each style in the list is drawn in its own font and colours, so a style whose appearance doesn’t match its name is obvious the moment you open the file. Formatting a style now opens a dialog from the Edit button beside it, with its own live preview, and the preview at the bottom spans the full window. The colour picker gained two rows that save real time: the colours already used in this file, and the ones you picked most recently.

Two separate faults made the emergency hatch unreliable. During a long fixed-duration stimulus — a two-minute instruction screen, say — DirectRT didn’t notice the key until the stimulus had run its full course. And during a response window, Esc could be taken as the participant’s answer instead of an abort: the session moved on to the next trial and the keypress was written to your data file as a real response, with a reaction time. If you pressed Esc during a session on an earlier 2027 build, check that participant’s data for a response you did not intend — it looks like an ordinary row. Esc now ends the session immediately, wherever it is pressed.

In your style file, 200-1500! means “stop waiting for a response at 1500 ms” — and in DirectRT 2020 the trial then carried on with whatever it had left to show. In the 2027 alpha it was abandoning the trial at that point instead, which is what 200-1500!! is for. Both forms work as documented again: ! continues the trial, !! ends it. One difference worth knowing if you use !!: because the trial is being abandoned, the too-slow message is not shown.

MP3, M4A, AAC, MP2, AIFF, FLAC and WMA now play, alongside WAV. Two things were wrong and both were silent: DirectRT was adding .wav to the end of any sound filename that didn’t already have it (so !song.mp3 went looking for song.mp3.wav), and even with the right name it could only decode WAV. If a sound still can’t be played — Ogg Vorbis and Opus have no decoder on Windows — DirectRT now tells you when it reads your input file, instead of running the trial in silence.

If a clip was there but Windows had no codec for it (or the file was damaged), DirectRT showed nothing and said nothing — the stimulus just quietly didn’t happen. It’s now named in the session summary and the event log.

WebM, Ogg Theora and Flash Video files are now pointed out when the input file is read. Everything else — MP4, AVI, MOV, M4V, WMV, MPG and MKV — is unaffected.

A sound set to fire and move straight on was being graded as though it had asked to be on screen for one millisecond, so it reported running “2 frames too long” on every trial. That false alarm is gone.

If you opened a style file from an earlier version of DirectRT, changed Display in Session Options, and saved, the conversion quietly reset it to Current — and then told you it was set to Current, as though that were the plan. Your setting is now kept, and that notice only appears when Display really is Current.

DirectRT would tell you to close a file you never had open; it now says which file it couldn’t find and where it looked.

Opening an input file sometimes opened a data file as well. The console’s clickable paths responded to the button release on its own, so the second click of a double-click in the file dialog landed on whatever sat underneath — which, once you had run a session, was that session’s data path.

On the “overwrite existing data?” prompt, clicking Yes ran the session but pressing Enter cancelled it — even though Yes carried the default highlight. Anyone confirming that prompt from the keyboard could lose the run they had just asked for and reasonably assume it had started. Enter now matches the highlighted button on every themed prompt; Esc still answers No or Cancel.

Installing 2020 onto a machine that already had 2027 left 2027 impossible to remove or repair: setup answered an uninstall request by asking you to remove 2020 first. If you are in that position now, removing DirectRT 2020 makes 2027 removable again — this build prevents it happening rather than rescuing a machine already in that state.

The progress page names the destination (C:\Program Files\DirectRT), so if you’re used to 2020 living in C:\DirectRT you can see where the files are actually going.

A row (or line) is a whole trial as you see it in Excel; an element is one stimulus/location/time triplet inside that row; a line is an electrical wire. The guides had used “row” for both senses, which invited putting a TTL on its own row and expecting it to fire with the video on the next one — the ITI lands in between. Added the missing TTL-with-video example.

On a multi-seat licence, every machine after the first was refused with “fingerprint is not activated” even though seats were free — and on any licence, a hardware change (new drive, Windows reinstall, new motherboard) would have locked the owner out for good. Both are fixed, and DirectRT now recognises a machine whose hardware has shifted instead of spending another seat on it. If you were blocked on a second computer, just relaunch — it will activate itself.

The sample studies are now a numbered, work-through-in-order set — each with a plain-English READ-ME, an on-screen intro, modern styles, and full-HD layouts. Classic demos have been re-themed onto neutral topics, and the handful not yet refreshed are parked in clearly-marked “not yet modernized” folders so you always know what you’re looking at.

Leave your input file or last data file open in Excel and DirectRT tells you exactly which file to close, instead of failing partway with a cryptic error — or worse, running the whole session and then being unable to save the data.

If DirectRT 2020 is still on the computer, setup now stops and asks you to remove it rather than uninstalling it for you. The two versions share program names and the .drt file type, so they can’t coexist — but which of your programs to remove is your call, not ours.

A page that declares its own size is shown at that size, centred; a fluid page fills the screen. You can still force a size from the input file when you want one.

The open-ended answer box looks like a proper dialog again, in a readable size, and sits above the stimulus instead of covering it.

directRT, Style Editor and executeDRT are back in the DirectRT folder itself (as shortcuts), alongside a Utilities folder, so old desktop shortcuts and shared-drive habits keep working now that the program files live in bin.

The end-of-session results now lead with a link to the study folder, and the other files are listed relative to it, so the report is far easier to read.

Setup now shows the file it’s copying and a percentage, so the long step no longer looks like it has frozen.

Setup used to delete every shortcut and put an identical one back; now it leaves the ones it’s about to reinstall alone. On a test upgrade this took the install from just under two minutes to under ten seconds.

The user guide has been rewritten against the shipping software — corrected throughout, with screenshots, a three-level contents sidebar that follows you as you scroll, and coverage of a dozen features the old manual documented but the 2027 guide had dropped.

If you use those prefixes, re-check any affected studies — the first key is now registered correctly.

The Session Options tab was effectively invisible in light mode — it renders correctly now.

Installing DirectRT 2027 now automatically removes an older DirectRT 2020 if it’s on the machine, so the two can’t clash over .drt files or over which version MediaLab launches. If DirectRT 2020 can’t be removed, setup stops instead of leaving both installed.

The installer has a new look — a branded side panel and a cleaner license page — matching the Empirisoft website.

Your samples folder now opens the first time you run DirectRT, and if you already had it installed, any newly shipped sample studies are merged into Documents\DirectRT Samples automatically on the next launch — nothing you already have is overwritten.

Double-click a .drt style file to open it straight in the Style Editor — and you can now have several Style Editor windows open at once.

The end-of-session report now includes a clickable Event log folder link, alongside your data and log file paths.

The end-of-session report now shows your Data file and Log file paths, and file paths in the console log are clickable — double-click one to open the file or its containing folder.

Import a new or renewed license from the Help menu even when one is already active — it re-validates online, so there’s no need to reinstall to pick up an updated license.

On a monitor set above 100% scaling, DirectRT’s own dialogs and message boxes could come up oversized or with their buttons and text clipped. They now scale properly.

On modern Windows the legacy display engine could black-screen a running session, and if you used parallel-port TTL its kernel-level driver (NTPort) was refused by Windows 11’s security (Memory Integrity / core isolation). DirectRT 2027 is built on modern Windows graphics and drives TTL over USB, so a session runs cleanly on a stock, locked-down machine. Same study files, current operating system.

Hardware triggering no longer needs a parallel port or the NTPort kernel driver — it runs over a USB serial device, which works with Windows 11’s Memory Integrity / core isolation turned on (the setup that previously blocked the driver from loading). Pure keyboard studies never needed the driver either way.

Present a designed HTML page as a stimulus — richer layouts and styling than plain text or a single image, timed to appear on an exact screen refresh.

Use PNG, JPEG, GIF and more directly as stimuli. Existing bitmap studies still work exactly as before.

Name a trigger device once and refer to it by name in any study, and fire a timed sequence of TTL markers at exact moments — for example, at set points during a video — so recordings line up with what’s on screen. (Basic EEG/equipment triggering already worked in 2020; this adds named devices and timed marker lists.)

Activate online in seconds, with a 30-day offline grace period so a dropped connection never interrupts a session. A standard user can install the license without admin rights.

A single signed installer sets everything up, with an MSI for silent, managed deployment across a lab — and no special hardware drivers required.

Video already played in 2020; now it’s more reliable — line up multiple clips, prepared between trials so playback starts cleanly at onset. Prefer MP4/H.264.

Spoken-response timing was already supported; setup is now easier — pick your microphone and check input levels before you run participants.

Text stimuli existed before; now any script — including right-to-left languages such as Arabic and Hebrew — renders properly, with no workarounds.

Each stimulus is drawn in a single, tear-free flip locked to the monitor’s refresh, and media is prepared between trials so onset stays clean. An optional per-session report shows refresh rate, onset accuracy, and any dropped frames so you can trust your numbers.

Each response is timestamped with a high-resolution timer on its own dedicated thread, so DirectRT adds virtually no delay of its own. Your keyboard’s USB polling still sets the floor — for the tightest millisecond precision, pair DirectRT with a button box over TTL.

DirectRT now runs at your current desktop resolution by default and looks crisp on modern and high-resolution monitors. If an older study needs a specific screen size, you can switch resolution for that session in the Style Editor.

A modernized style editor and hardware tester, plus an input-file checker that catches mistakes before you run participants.

In a response list with a port listener, the numbers you write are now matched as the raw TTL values — a TTL of 2 satisfies rt:2. Previously the digits went through the keyboard number-row remap, so an author had no way to know what value to send. Keyboard responses are unchanged.