The Idea Behind Pro Tools HD Video Performance Tests

The main initial idea of the test was to compare video hardware brands, as historically many users were experiencing performance issues with Blackmagic video peripherals. My initial qualitative test results indicated that the AJA peripherals or PCIe cards were far out performing the Blackmagic cards in Pro Tools 11.

However, once Avid released Pro Tools 11.1.3, Blackmagic released driver version 10.1.1 that allegedly offered dramatic improvement to Blackmagic card users.

I widened the test to include Pro Tools 10.3.9 and its video performance so as to compare with Pro Tools 11.1.3. I decided to also include a few other interesting tests for comparison whilst I was at it (“in for a penny - in for a pound”!) so that we can end some of the video urban myths.

So Why Should Non Post People Read The Pro Tools HD Video Performance Tests?

Music guys really should pay attention to this test series, as you will soon see that video is not just a matter of “pop any old Quicktime into the timeline and Bob’s your uncle”. Sync is a complex subject – think of it as latency and that might make your ears and prick up and your expertise kick in.

How Did We Do The Pro Tools HD Video Performance Tests?

Sync test measurements were simultaneously performed with a Pharaoh Audio Syncheck 2 & 3 (thanks to UK distributor Mike Woods of Dirty Dog Audio Ltd and to Mike Wabro of ‘Audio In Sight’ for the version 2 loan!). See my review of the Syncheck 3 here.

I used the Syncheck video files, and where no file existed, I created it within my Avid Media Composer 4.05 (running on my Mac Pro 1.1).I used the fast paced “multi spaced” test files where the time between flash/pips varies between 5 and 12 frames, often with a distinctive pattern. Once a system’s AV error is known to be less than 4 video frames, the faster pace of “multi-spaced” patterns is best at revealing subtle drifting or intermittent errors (the video span).

System Spec Used for Pro Tools HD Video Performance Tests

• Pro-Tools-PC 3.5Ghz Quad Core i7 with 32GB ram (1600Mhz DDR3), OSX 10.8.5

• Avid Pro Tools HDX1

• Avid Sync HD

• Avid Omni audio interface

• Rosendahl Nanosyncs for SD syncs

• ESE DV319 for HD tri-level syncs (its HD and SD syncs are synchronous)

Video Hardware Used In The Pro Tools HD Video Performance Tests

• Avid Analogue Composite Mojo for SD video

• AJA IO XT thunderbolt for HD & SD video (with AJA 10.5.1 software)

• Blackmagic Decklink Studio for HD & SD video (with Blackmagic 10.1.2 software)

Testing Info for Pro Tools HD Video Performance Tests

Values are recorded to 1/100 of a frame and are averaged unless significant individual readings are shown.

For the MXF media no video offset was used

For the QT media for DV and H264 media, quarter frame offset values are shown, although please note if video ref is used, these are rounded to the nearest whole frame by Pro Tools (whether you like it or not!).

Let the Pro Tools HD Video Performance Tests Begin - Test 1) Pro Tools 10.3.9 with SD video

This was my baseline test. We have always been told that Avid’s preferred video format for its Mojo has always been DV PAL MXF (that’s DV SD video wrapped in Avid’s MXF container), and that Quicktime video performs badly out of a Mojo. I wanted to test this and compare it to how well a Blackmagic video peripheral would perform in PT 10. AJA peripherals do not run with Pro tools 10.

Users without a video peripheral often use the desktop video window on their computer monitor, just how accurate is this?

All SD composite video (apart from the desktop video) was output to a Sony Trinitron analogue TV, so that display delays could be all but ignored.

• MOJO - DV MXF - video ref Av = 0.10f Span = 0.5

• MOJO - DV QT - video ref Av = 0.11f Span = 0.00 (QT Offset = 12qf)

• MOJO - DV QT - no video ref Av = 0.29f Span = 0.01 (QT Offset = 14qf)

• MOJO - MP4 QT - video ref Av = 0.11f Span = 0.00 (QT Offset = 14qf)

• MOJO - h264 QT - video ref Av = 0.87f Span = 0.01 (QT Offset = 12qf)

• Desktop video - DV Pal QT – no vid ref Av = 1.51f Span = 4.70 (QT Offset = 36qf)

• Desktop video - h264 QT – no vid ref Av = 1.57f Span = 5.26 (QT Offset = 36qf)

• Blackmagic Decklink - DV QT - video ref (individual tests listed due to wide range of reults)

= 0.89f Span = 0.02 (QT Offset = 8qf)

= 0.10f Span = 0.02 (QT Offset = 8qf)

= 0.88f Span = 0.01 (QT Offset = 8qf)

= 0.10f Span = 0.01 (QT Offset = 8qf)

= 0.02f Span = 3.38 (QT Offset = 8qf)

• Blackmagic Decklink - DV QT - no video ref (individual tests listed due to wide range of reults)

= 0.00f Span = 0.02 (QT Offset = 8qf)

= 0.87f Span = 0.01 (QT Offset = 8qf)

= 0.29f Span = 0.02 (QT Offset = 8qf)

= 0.18f Span = 0.01 (QT Offset = 8qf)

Conclusion Pro Tools HD Video Performance Tests - Pro Tools 10.3.9 SD VIdeo

1. Avid Mojo is spectacularly accurate with most things, even when un-referenced (which really surprised me)

2. I was surprised at the span result on the MXF media and would have expected a zero or near zero result

3. H264 came off worst on the Mojo due to its inter fame operability (it needs to know where a perious frame’s pixel was to calculate where the next goes). No surprises there.

4. Desktop Video is appalling and almost never in sync and meanders around in play like a snaking river (see the wide span results). Avoid like the plague. Anyone who has been using DT Vid in Pro tools 10 should be asking themselves some very searchig questions. The DT Vid was tested on a Slate Raven screen, to avoid this being blamed I perform a later test to show the DT Vid on a Samsung LCD monitor.

5. The BM Card appears to be poor in its ability to resolve to video in 10.3.9. Once playing it is very stable, but the difference between audio and video is quite random between each play pass, sometimes with audio in advance, sometimes with video in advance. I was testing its SD outputs – I do a later test to see if the HD to SD down conversion is causing a problem here.

Join us in part two for: 

• further tests - introducing Pro tools 11 and the Avid Video Engine

• videos of some of the tests

Update

on 2014-07-12 09:59 by Mike Aiton

here’s a response from Richard Fairbanks of Pharaoh Audio who designed and makes the Syncheck:

“Desktop video is awful. Debates happen because some people have unfounded ideas that the companies involved have already done all the testing necessary to ensure that audio and video are in sync and nothing else need be done. Then, someone posts “my system has perfect sync and I don’t use any of that stuff” and a whole new generation of people accept it as truth. I’ve grow tired of going over the same points time and time again, only to discover they haven’t been understood or believed! And then, receiving emails that go something like “I have x and y with two 30” Cinema Displays and of course I don’t want to spend any money on video stuff, so what settings do I use to make the sync perfect?”

It’s great that you are willing to re-educate. Best wishes!”