[Joe Marler]

Thanks for the response, but that doesn’t work. You cannot “use the audio track as one of the angles” in a multi-cam source sequence.

Even though all tracks are sync’d with an IN point, you can’t select them in the project pane, right-click and pick “Create Multi-Camera Source Sequence” — the option is greyed out if there’s a separate audio-only track.

I am highly confused by this. What is the point of improving CS6 multi-cam ease of use, if it does not cover the most common situations? It is very common to have a separate audio-only track in multi-cam editing.

[Joe Marler]

OK after re-checking this I found my above statement about 99.9% drift is in error. I thought I’d measured that amount, but it’s actually much less, about 0.16 sec after 58 min, or 99.995% drift. This is within the expected timebase variance of the camcorder and audio recorder. The solution here is just use rate stretch to fix it, use more precise devices, or use common timecode. Sorry about the incorrect statement.

[Joe Marler]

[Jeff Pulera] “more precise (expensive) timing components are used in camcorders. With digital audio recorders, no one would ever notice if the speed were off by 1%”

Jeff, the audio in this case was recorded by a Korg MR-1000 professional audio recorder: https://www.bhphotovideo.com/c/product/485879-REG/Korg_MR1000_MR_1000_1_Bit_Professional.html

The 99.9% audio sync drift problem when using 2nd source audio is frequently blamed on a mis-match between 29.97 and 30 fps. This cannot be the case since video recorded at 29.97 is played back at 29.97. Whether video is recorded at 24, 29.97, 60 or 120 fps, if it’s played back at the same speed, the playback rate matches real time. Likewise whether audio is sampled at 44.1 or 48Khz if it’s played back at the same rate it also matches real time.

The video and 2nd system audio should only drift by the differences in their time bases. IOW you could hit PLAY on the camcorder and audio recorder, and they’d only drift based on their differences in time base precision, the same as two quartz watches drift apart in time.

That is not the case here, when the video and 2nd system audio are played back in Premier Pro. Audio is drifting slower at *exactly* 100.1% of the video rate — the exact proportion of 29.97 to 30.00.

It’s almost like Premiere Pro is erroneously applying some NTSC timing alteration to the audio or video tracks but not both.

Of course I can fix it with a rate strech, but I’m trying to understand the underlying cause of the problem.

[Joe Marler]

Just a follow up: I ultimately used AVCHD UpShift; it worked fine. I batch transcoded 240 1080i AVCHD files (64GB worth) to .m2t. The program is easy to use and is apparently very multi-threaded, as all four cores (8 cores with hyperthreading) were well occupied.

I’m sure NeoScene is better, but this was adequate for our purposes.

I did multiple test runs with different parameters, and ultimately used 50 mbps, variable bitrate, keyframe=1, GOP=1, and Interlace=Interleaved (IOW deinterlace during transcode).

It never crashed during all the work, which is always a plus when dealing with digital video software 🙂

[Joe Marler]

Errorlog file sent. Further testing shows problem is isolated to importing .jpg images; importing 200GB of AVI files (inc’l scene detection and thumbnail generation) works OK.

However importing a directory of about 1100 .jpg images crashes every time. I don’t think it’s corruption in the data, as other folders of different images still causes the crash.

Problem seems related to .jpg thumbnail generation. Doesn’t happen if Tools | Preferences | Previews & Thumbnails is set to “no automatic generation”.

However attempting to build the thumbnails after the import by selecting Media | Build Thumbnails still causes the crash.

On CatDV ver 8.1.10 for Windows, I don’t see an option under “Advanced Media Handling” to uncheck “Generate full size thumbnails from file system”. That may be for Mac only or a prior version.

[Joe Marler]

No, Canon AVCHD. It’s not a big deal, we can transcode to a format that’s compatible with CatDV.

[Joe Marler]

I don’t think GPU acceleration is available for your Mac config. If you use any effects, that’s a huge difference. On Macs the only officially supported card for GPU acceleration is the expensive Quadro FX 4800. On PCs you can use many different cards.

I love the Mac OS interface, but if you spend most of your time within CS5 the app appears mostly the same between Windows 7 and Mac.

At the lower end of the desktop range the price difference between Mac and PC is less. But at the upper end typical of video editing the price differential is substantial.

I’d suggest pricing a quad-core i7-based PC with a 10k rpm boot drive and RAID 0 data drive, a lower-end nVidia card supported for GPU acceleration, then price the equivalent features/hardware (inc’l GPU acceleration) on a Mac. If the difference is too much to afford the Mac, get the PC.

PCs are widely available from various vendors with factory support for overclocking. So that’s another consideration — you aren’t limited to the stock CPU frequency.

[Joe Marler]

On my machine, CS5 export of 1080i/30 .mts to MPEG-2 DVD at default settings takes about 1 min per 2 min of material — IOW about 2x real time. That’s with no effects, and without pre-rendering of timeline.

Using effects without pre-rendering would be much slower, unless you have GPU acceleration. A quick test would be turn off all effects and try to export a shorter clip, then turn them back on, reboot CS5, export and compare times.

Or you could pre-render (sequence > render entire work area). But if that’s faster on export, you just traded shorter export time for longer pre-render time — it’s still taking a long time.

If your slow export is due to effects, your options for GPU acceleration are more limited on a Mac. I think the only supported card is the Quadro FX 4800. On Windows machines there’s a hack to enable other nVidia cards, but I don’t know if that’s available on a Mac.

Quad-core i7-860 @ 3.8Ghz, 8GB DDR3, 10k rpm velociraptor boot drive, 7200 rpm 1.5 TB data drive, GTX-275 GPU, Windows 7 Home Premium 64-bit.

[Joe Marler]

My testing with Premiere Pro 5.0.1 on an nVidia GTX-275 with GPU acceleration enabled shows encoding of Blue Ray, DVD and Flash is not accelerated. Yet those are the three formats nVidia says ARE accelerated for encoding via GPU hardware.

My card is admittedly not on Adobe’s approved list, but GPU acceleration is enabled using the usual procedure involving cuda_supported_cards.txt. It greatly accelerates effects, and it obviously accelerates exporting if effects are used.

However it’s the effect rendering aspect of export that is accelerated, not the encoding itself.

There is a remote possibility an approved card like a Quadro might accelerate encoding, but I doubt it.

The previously-listed nVidia statement about GPU acceleration of encoding seems to be marketing speak, as the same video describing GPU acceleration of encoding eventually says it only happens if effects are used. IOW a back-handed way of saying encoding is NOT accelerated, only the effect rendering is.

Encoding is so time-consuming, people are obviously motivated to expedite it however they can. They could read the nVidia headline claiming GPU encoding acceleration, buy an expensive GPU, then be disappointed. It appears (in the current 5.0.1 version) that *only* effects are accelerated.

If anybody knows otherwise, please explain. Maybe someone with a Quadro could do a quick export test to Blu-ray, DVD or Flash with GPU acceleration on then off, restarting CS5 in between.