[Dave Haynie]

Not a Mac user. But if the iMac is encoding as quickly as the Xenon, and it’s really taking very long at all to encode your 25min segments, it sure sounds like your encoder is single-threaded. At least, that’s the best explanation.

I’d recommend getting a multi-threaded encoder. Also, are your sources SD or HD? It is not a trivial task for a modern PC to decode H.264 in HD, particularly if (as in most NLE and other applications, as opposed to players) the H.264 decoder isn’t using the GPU. Or multiple threads.

I just did a quickie benchmark from a project I have on Vegas on the PC. Rendering HD AVC to high quality DVD MPEG-2, it’s taking more than 2.5x as long as rendering HD Cineform (kind of the same idea as Apple’s proprietary ProRes, or Avid’s standardized DNxHD)… two tracks in the same project, everything else being equal. Chances are, you’ll get a huge speedup, no matter what else, if you start with something other than AVC, when it comes time to render. Of course, if you start with AVC (as many do… I have two camcorders and an HDSLR that shoot it), you’ll have to pay at some point for the extra overhead of AVC.

Decoding with an FPU can absolutely speed things up, if supported. Just in a player on my six core 3.2MHz AMD machine, I see about 50-60% total CPU playing an HD AVC file directly copied from one of my cameras, about 6% in a player that uses the latest Windows GPU acceleration. Pretty significant. As I mentioned, many video editors, including mine, don’t use GPU acceleration for editing or video decoding. They should. You may find the same thing on the Mac… probably, since Apple’s GPU APIs only came out last year. But it couldn’t hurt to ask around.

-Dave

[Dave Haynie]

[Carly Smith] “I want to export the highest quality settings I can that will fit onto a single DVD – the same way I would for a retail DVD.”

I don’t know anything about Premiere itself (last used it well over ten years ago). But the best quality one usually get using consumer/prosumer tools is done this way: get yourself a bitrate calculator. Enter the length of the video and the type of audio you want (most likely AC-3 for a sporting event, perhaps PCM for a concert or some other event in which audio is more critical). The calculator (you can Google and find dozens if not hundreds) will figure out the maximum bitrate you can use for encoding your video in order to fit on the selected media type (DVD5 or DVD9).

Once you have that, you want to set that birate as the average encoding rate. Set a higher bitrate as peak, somewhere between 50% and 100% higher, but don’t exceed 8Mb/s or so, there’s rarely any need. Set the minimum bitrate to 1/2-1/4 of your peak or less. Then encode. If your encoder does VBR well, your final video will fit. You can also use this technique to encode multiple segments, as long as you add up the times. If it doesn’t fit, your encoder may not be so accurate at VBR, so try again, scaling back your average based on the overrun.

[Carly Smith] “How do the pros export such high quality to a single DVD?”

A commercial DVD is encoded by a mastering engineer. They have workstations with realtime variable bitrate MPEG-2 encoders, and other tools. For example, in very fast motion scenes, you may see MPEG-2 block artifacts (one reason I like to do my sports videos in Blu-ray… I can use 60p rather than 60i, better for sports). These are the results of too many differences between each I-Frame and the subsequent P and B frames in the MPEG-2 stream. An encoding engineer can selectively apply low-pass filters (eg, well designed blur) to lower the amount of information on the whole picture, and thus, avoid the MPEG-2 artifacts. If you study the action scenes in any commercial film, you’ll find the birate goes up, but in fast motion, there’s also substantial blur added… which your eye ignores, while the emergent edges in the DCT macroblock compression overload is parsed by your brain as horizontal and vertical lines… something we’re hardwired to detect.

In short, the top tier pros do this by eye, with very powerful tools that make this a reasonable thing to do. We don’t have those tools, and have to count on algorithms to produce the best results. But very certainly, the one you can do is use the available space on the DVD to your advantage… any unused room is not contributing to a better image.

-Dave

[Dave Haynie]

You gotta love the gear specs… everyone picks their own favorite units. I’ll try to explain how I figured out what to do.. at least what works for me.

The spec for a consumer line-level output… what you might find from an RCA tape monitor output from a mixer, is -12dBV. You can get this output on some mixers, it’s maybe the best one available for a stereo tap. The typical output from a submix or other XLR or 1/4″ balanced output is probably +4dBu (international studio level). So what does that mean? Well, dBV is decibel relative to a 1V signal… the “V” means “Volt”, -12dBV = 0.25V. dBu is a decibel relative to 0.775V, the “u” means “unloaded” … +4dBu = 1.228V; -12dBV = -9.78dBu. So basically, you need a 15dB pad on a “pro” signal to deliver the same level as you get from a “consumer” signal. So that’s the first thing to factor in when you’re hooking up… at least, to a line level output.

And of course, if you have the option, you may get to hook up to a studio monitor output, which will probably have a volume control. This is probably expected to drive a studio amp or powered monitor, you’d kind of hope it’s limited to +4dBu, but I wouldn’t always put money on it, some amps can take inputs up to +24dBu. But if you have access to the volume knob, you can solve your problems there entirely… assuming the sound guy doesn’t accidently hit that button during the show.

Now, let’s look at the Zoom H4n inputs themselves. Naturally, since these are mic inputs, they’re speced in dBm (decibels relative to 1mW… the actual voltage you see depends on the impedance of an attached microphone). So anyway, balanced input is speced at -42dBm to -10dBm at 1K ohms, the 1/4″ is speced at -32dBm to +2dBm at 480Kohms, 1/8″ input is speced at -48dBm to -7dBm at 2Kohms.

Now of course, an unbalanced line level signal is usually really high impedance, so I actually run the line level (which I want) though a direct box, which delivers a 600 ohm signal to the XLR input on the Zoom… which is about what it’s looking for there — good mics are usually 150-1K ohms, more or less. At 600ohms, my handy dandy dB calculator (https://www.sengpielaudio.com/calculator-db-volt.htm) tells me I have 0.25V at -10dBm… which is just that -12dBV. There’s definitely some range there in the Zoom, so you could leave some headroom and drop another 10-20dB off the consumer signal, or maybe 30-40dB off the +4dBu output (a 40dB pad is more common, what I’ve used with the direct box, and what most people seem to recommend… if it’s not enough, you probably have a much hotter signal on that mixer).

Of course, given the fact you’re running from XLR outputs, you have at least +4dBu, but maybe more, depending on just what that output is. Most direct boxes do have built-in attenuators, and if you flip it around, you could run the XLRs in and the 1/4″ out at a known high impedance to the 1/4″ input on the Zoom. I’d stay away from the 1/8″ input, both because they break too easily, and also, that’s a 2K ohm input, not a good match to an unbalanced line level output. In my case, I have a few of the variable XLR pads, as well as the 40dB pad in the direct box, so I could actually cut 80dB from the mixer if necessary… I never needed the extra pads for this stuff, though. But I always bring a bag of audio support stuff… extra cables, adapters, etc. if I have a chance at a sound board tap when recording a show.

The other thing about the direct box is that you can flip a switch to cut ground, which can eliminate hum (you might have a ground loop between the H4n and the mixer, particularly with a stereo tap).

-Dave

[Dave Haynie]

For $500 or even $200, there’s not much to excuse the wedding shots they showed. Even with a single camera, at least he could have set up within sight of the ceremony. It looks like he arrived late, hasn’t even looked over the church beforehand, and didn’t bring a tripod.

I shot a wedding once, in a small country church, where the minister wasn’t very into the whole video thing, and basically agreed to let me (and the still photographer, once he arrived) set up in the back. I had two cameras on the ceremony (and a MiniDisc recorder hidden up front… this was a while back)… basically just providing redundancy, and it worked out fairly well, all things considered.

And, bottom line, the couple were happy. In fact, they liked one of the stills I shot better than what their photographer had (I always bring a still camera, to get coverage shots, stills for DVD/BD and cover art, that sort of thing… I hate the pre-packaged wedding clip art).

But sure, when you hire the cheapest guy you can find, you shouldn’t expect much. But you really should expect to be the only wedding filmed during your wedding… that’s the part I’d claim is downright criminal, and a big reason the video is terrible.

-Dave

[Dave Haynie]

1) Get an NDA.

2) Give him a spare flash card, take it with you when you leave. He can have the video back after it’s released, or you send him a video-only copy, if he’s concerned about being able to review his work.

3) Make it clear to him it’s the band’s decision, not yours. In the music business these days, it’s not unusual for a little paranoia surrounding unreleased music.

-Dave

[Dave Haynie]

Yeah… no details. I have shot dramatically better HMC40 video in a pretty dark nightclub, so I kind of wonder what was going on there. I would definitely be shooting at 24p and 1/24th sec. in a room as dark as that seems to be.

On the other hand, I have zero doubt the AF100 would outperform the HMC40 in low light — it’s basically a DSLR with camcorder controls. The HMC40 has a combined sensor area of about 23mm^2, the AF100’s 4/3″ sensor has an area of about 225mm^2. That’s just about 10x the area, more than 3 full f-stops worth. The difference between the f1.8 HMC40 and an f3.5 on the AF100 is only about two stops… so, big advantage. A Canon 7D or 60D has a sensor at 329m^2, close to four stops improvement, and something like the 5D, with a full-frame 35mm sensor, tops out at 860mm^2, better than a 5 stop advantage (assuming the same fill percentage and other things being similar from sensor to sensor).

And unlike exposure, noise isn’t necessarily a linear function. The basic idea is that noise is always in your sensor, even though they’ve been lowering that base level every year, particularly in CMOS sensors. In the bight light, the ever-present noise is too low-level to show up in actual video, but as you increase gain, it makes an appearance. And even an f-stop or two worth of gain can be day or night with your sensor.

And I could mention that you can buy three HMC40s for the price of the AF100… and you still to buy a lens for the AF100 after that. A 60D or Panasonic GH2 (which seems to have the same sensor as the AF100) with a pretty decent zoom will run you less than an HMC40. And the 60D, at least, records at substantially higher bitrates. No question I’d rather use the AF100 for video from a functional point of view, but that extra $4000 would buy some sweet lenses and accessories.

-Dave

[Dave Haynie]

Which Zoom recorder did you use? The H4 and some other other older ones have a reputation for a pretty dicey clock. The cheap everyday crystals used to drive microprocessors come in at about 100ppm (parts per million), which translates to a loss of sync of one frame at 24fps every 6.9 minutes. I’ve read a fair bit of anecdotal evidence that suggests that’s about what you’re going to get with the Zoom…. it’ll certainly vary from unit to unit and video device to video device… your camera may be super accurate, but it’s it’s off, there’s a 50/50 chance it errs in the same direction as the audio recorder. So you can’t know the whole story just from a few specific-case measurements.

On the other hand, the clock module I use in my digital radio designs (the day job) is 1.5ppm, which would mean one 24fps frame every 7.7 hours. The typical cheap watch crystal, usually 32.768kHz, used for the timebase in some consumer gear, is usually around 20ppm, which gets you one frame de-sync every 35 minutes or so. And again, that 20ppm is a worst-case number…. at room temperature, you’re probably sticking with a frame every hour or so.

I haven’t found any hard numbers on the Zoom H4n (I have one, and no problems to date, but I figured I’d look it up), and I’ve read claims of it being off one frame in 100 minutes, so that could well be running on a 10ppm rate clock. But again, that’s only so useful if your camcorder walks all over the place.

And of course, do check that your project sample rate and video sample rate are correct. Most of expect “24p” video to be at 23.9fps, but the iPhone isn’t really a video device.. does it maybe give you a hard 24fps or something?

-Dave

[Dave Haynie]

This is not a problem in Vegas, it’s simply the performance of your PC.

AVC is a very complex format to decode. Have you tried even playing the video files in a regular player on that PC? I have an older 2.4GHz dual-core laptop (in addition to my 3.2GHz 6-core desktop, which is just dandy at editing video from my HMC40), and honestly, AVC files can barely play without stuttering, unless I use a player that can use GPU acceleration.

As you can imagine, editing is a more involved process than simply playing things back. I would definitely make the claim that a minimum of four cores are what you want for AVCHD editing. Yeah, it will work with less, but you won’t enjoy the process.

-Dave

[Dave Haynie]

The best quality approach is to compress all of your video assets to variable bitrate MPEG-2 that you’ve worked out, using a bitrate calculator, to ensure a proper fit on that DVD. This takes some experience to get working the first time.

The one minuute solution is this. After you have all of your assets in DVDA, select the “File/Optimize Disc..” dialog. Make sure all of your video assets (or at least those you’re willing to compress/recompress) “Use default bit rate” set to “Yes” under “Recompress Settings” (on the Video tab). One this is set up, click on the “Fit to Disc” button. This will calculate a bitrate for you that will ensure everything fits.. or give you an error, if you don’t have enough assets set for recompression.

-Dave

[Dave Haynie]

[Kelly Griffin] “Dave: You just made my head explode. Now please come clean it up.
“

Yeah, that happens. I have a syllogistic mop around here somewhere…

-Dave