Keeping Up with REAPER Video

The REAPER forum is a regular stop for me. That’s where I learned about video in REAPER and where I look for the latest info. The video implementation is actually still in Alpha form, so it’s not fully supported and documented. Here’s the thread that announced the use of FFmpeg libraries.

Forum member airon has created a thread where he tracks the current state of video in REAPER, “Video Playback and Exporting” that I check often. He keeps it updated with the current FFmpeg libraries of choice, along with other news and tips.

Here’s the Cockos Wiki page describing video support. It includes details on installing the library files.

With all my enthusiasm, please remember that there are no video effects in REAPER currently. No fades, no transitions, no flipping or stretching. All we can do is move clips (and stills) around, trim them, and sync with separate audio tracks. So our style will be defined by those limitations.

Titles and Tricks

A static title is pretty easy to add to your project in REAPER. Just create a jpeg or png image in your favorite tool, heck, Windows Paint will do the job. Be sure to set the size to match your video, for instance 1280 pixels wide by 720 pixels high is a common HD video size. Start with an image or create a gradient, then add text, maybe in a couple of stages or pages. Then drag these into REAPER, drop them on the project timeline, and stretch them to match the tempo of your clip. You can see this technique in action in the tutorial video below.

You might have stills that contribute to your video, or a second or third camera shooting. Or you can extract stills from your video and use those for effects. REAPER 4 can handle multiple tracks of video along with multiple tracks of audio. The audio from all the video tracks will be part of the mix, but only the lowest (highest track number) video will be displayed. Once again, watch the video below to see these operations.

If you mute a video track it stops displaying the video as well as turning off the audio. To eliminate the audio but keep the video, just turn down the volume slider on the track, or use the Item Properties to turn down the audio.

Syncing and Slicing

The audio waveforms displayed in REAPER make syncing multiple video and audio clips a snap, especially if you create a sharp noise at some point when all the devices can capture it. The is exactly like the old “clapper board” you see in movies about making movies. You can simply clap your hands to get a spike in the audio that makes syncing easy.

Once you have tracks synced, take care to keep them that way. Use Item / Group to join items, or be careful to select all the items that belong together when you move, stretch, or trim items. In the video below observe how easy it is to select multiple items to keep them in sync. Also, I’m using ripple editing most of the time to maintain the correct relationship between items.

Cutting video items is just the same and just as easy as cutting audio. Select the item, position the edit cursor, press the S key, and your item is split at that point. As above, make sure all the items that you’re keeping in sync are selected and get split appropriately. I use ripple editing a lot when I’m trimming out parts of my videos.

Tweak the Audio

Even though there are no video effects available, the full range of audio processing definitely is present when working on video in REAPER. For instance, the clip I’m working on in the tutorial video below was shot with a Zoom Q3HD for the main video and audio. I raised the level, added some compression, applied a touch of reverb, and then put a limiter on the output to make sure I didn’t clip. The Q3HD delivers pretty clean audio but it tends to be a bit low in level and punch, so the ability to tweak the sound really adds to the usefulness of this little camcorder.

Rendering

Currently the h.264 video codec seems to deliver the goods for internet video. With the FFMPEG codec library installed in REAPER this is my preferred output. So I select the following options:

Output format: Video (FFmpeg encoder)
Format: QT/MOV/MP4
Size: 1280 x 720
Framerate: 29.970
Video codec: H.264
Bitrate: 9000
Audio codec: 16 bit PCM

Watch the Action

I used the Q3HD and a Creative Vado 3rd Gen to shoot an old Hawaiian slack key standard, My Yellow Ginger Lei in the style of Uncle Leonard Kwan. I decided to turn the video into a little tribute to my Kathy Wingert Model E guitar, so I used the guitar for the title still and shot a video of the guitar on a stand. Then I used REAPER 4 to combine those elements into a final clip for YouTube.

While I did the editing job, I shot with a couple of Sanyo Xacti HD2000s and recorded audio on a Zoom H2n. I edited those together to create a tutorial video:

and by the way, I put that tutorial together in REAPER 4 as well.

If you’re interested here’s the finished version of My Yellow Ginger Lei on my slack key channel:

I don’t think anyone is going to cut a major motion picture in REAPER, but as a way to tweak YouTube performance clips quickly and easily with an emphasis on audio, it’s a pretty decent tool.

Here’s the package:

Lens along with some magnetic circles and the covers

The lens mounts by glueing one of the magnetic circles to the lens cover of the Q3HD, then sticking the lens to the magnet. Here’s the magnetic circle in place:

The little metallic circle is a magnet with a stick-on backing

Once the magnetic circle is in place the lens snaps on and off:

Here's the lens mounted by sticking to the circle magnet.

Using the lens reduces the light and causes vignetting, a shadow effect in the corners. Here are a couple of shots with and without the stick-on wide angle lens in place:

With the lens in place note the vignetting

No wide angle, no vignetting, better picture quality

Here’s a video showing the Q3HD in action with the stick-on wide angle lens.

As I said in the video, I’m not sure how much I’ll use this trick, but it was interesting to evaluate how much we can improve the audio of our Q3HD with this fairly simple trick.

Now that I’ve done it, though, I’m pondering uses for all these gadgets. One that I’ve been playing with for a while is a “guitar cam” – a camera mounted directly on the guitar and giving an unusual view of the performance. With all these cams hanging around and Lynnie out of town I went to work.

The choice of guitar was pretty easy. The Epiphones have scattered, so the sacrificial guitar around the house these days is stage guitar #1, the Composite Acoustics OX. It’s really ideal for this project because my main tool is double stick tape, and the “raw” finish of the makes it unlikely to suffer damage.

Past Experiments

I’ve already tried out various mounting techniques for a peghead cam. I’ve used my smallest Gorillapod, but that proved too shakey. A mount that vibrates makes for a sea-sick viewing experience. One of my favorite engineering materials is packing foam of various kinds, but once again these efforts were lacking in stability, especially when I mounted the largish Zoom Q3HD on the peghead. I wound up with a stick of wood, a 1/4-20 bolt, some plastic, and lots of double stick tape, and came up with this.

Double stick tape, a 1/4-20 bolt, and a piece of wood

Success At Last

This mount was stable and gave a good view of the guitar neck. Here it is mounted.

Double stick tape attaches the device to the guitar

Adjustability could be better, but it works. Heavy duty (gray) 3M double stick tape holds a stack of plastic chunks together, guiding the 1/4-20 bolt at an angle to the stick. Then lighter duty (clear) 3M double stick tape holds the stick to the back of the guitar peghead.

But We Have More Cameras

Once I had a couple of decent test shots from the peghead mounted Q3HD, I started thinking about mounting additional cameras. First, one looking straight up from the bottom of the guitar, to get the fingers plucking, and my beautiful face!! Well, that didn’t work so well. The plucking was out of focus and at he wrong angle, and my face just isn’t that lovely, or interesting.

OK, let’s put one on the front, looking at the picking hand alone. Hey, and the Kodak Zi8 has a macro setting that I almost never use. Now all we need is one looking at the fretting hand from my view. The small and light Creative Vado 2nd Gen (or third, or however many) looks like a good candidate. Sure enough these cams were easy to mount with just a bit of clear double stick tape.

Kodak Zi8 mounted in the cutaway of the CA OX Raw

3M double stick tape holds the Vado easily

Three video cams mounted on CA OX

Lights, Cameras, Action

With the cameras mounted I scouted a location. With my sweetie out of town for a few days I had the whole house for my stage, and I found just the spot. Ordinarily I would never try to record in the close in space of a bathroom or shower, due to the strong early reflections. But for a guitar cam shoot I needed simple uncluttered monotone on both sides and this little hallway was perfect.

This little hallway provides two simple backgrounds

The back light would make it tough to get good video shooting down this hall, but the guitar cam(s) will be viewing side to side so they won’t get splashed. The Zoom H2 was nicely positioned by putting it on top of the hamper, and we were ready to shoot.

I barely had room for the guitar in this narrow space, but I was pretty happy with the shots I got. I spent a while shooting in another location, but when I looked at clips these were the winners.

Out of Rotation

When I pulled the clips into Edius Neo 2.5, my video editing software, I was dismayed to find that I didn’t have a way to rotate the clips. With the cameras mounted the easy way the resulting image has the guitar laying on its back, parallel to the floor.

The guitar is laying down on the job

Guitar still laying down on the job

I had expected to pull these into my Edius software, rotate them, crop and resize them, and build my composition with these images showing the guitar in a normal orientation. But it turns out that my Neo 2.5 version, the previous version entry level package, doesn’t have this capability. I would have to upgrade to the full commercial package to be able to rotate these clips. Time to rethink.

My first effort was a new layout for the clip, using the odd orientation for its impact. I cut a rough version but wasn’t too impressed. Next, do it the right way, change the camera orientation with a new mounting system. That took a day and wound up too awkward and shaky. Without re-engineering the whole project I needed to work with the clips I had.

Avidemux to the Rescue

I’ve done tutorials on Avidemux in the past, it’s a powerful bit of free software that continues to grow in its ability to manipulate video. The latest versions handle these H.264 based file formats easily and with a bit of tweaking and fiddling I was able to create rotated versions of my clips with minimal loss of quality.

Clip from Q3HD rotated in Avidemux

In order to create a standard 720P HD video file (1280×720) that Edius would accept I first rotated the image, giving a 720×1280 layout. I resized that to 405×720 in order to create the proper vertical dimension. Then I filled out the frame to 1280 pixels wide with black as you see in the still (this image is reduced to 680×360, half size).

And Here’s the Result

The song I chose is a sprightly version of an old Hawaiian royalty piece, Moani Ke `Ala, written by Prince William Pitt Leleiohoku in the late 1800s. Gabby Pahinui recorded this piece, as did his son Cyril, and both were a great source of inspiration.

As I’m listening to the clip just after it was uploaded, I’m appalled at the sound quality. There’s a background noise that is far greater than any on the original clip. I’ve been disappointed in the past but this one is really annoying. I wonder if I’ll see any improvement from a later, more leisurely transcode by Google.

I’m afraid that in some cases, simple physics is the culprit, helped along by our experience listening to modern recordings. We’re used to bigger than life sounds and close mic positions because that’s how most pop music is recorded. When we put our Q3HD in a location that gives us a conventional picture of a seated guitarist the camera is usually about four feet from the guitar. That four feet make the sound of the guitar weaker, and allow more of the room sound to intrude. This is called the direct to reflected ratio, and while we need some reflections to make a natural sound, the quick reflections of a small room generally don’t make for a pleasing recording. There’s another important ratio that suffers from a distant placement – the signal to noise ratio. Less guitar signal means we notice the surrounding noise more.

Several people have complained about the low level of the recorded audio and suggested that the Q should provide more preamp gain. But when you consider things for a moment you realize that more gain means more room reflections and more noise along with more guitar, because these ratios are established before the recording begins. So more preamp gain gives the same result as simply turning up the volume on playback in terms of room reflections and background noise.

The real solution is to move the microphones closer to the source. But that’s going to mean our conventional picture loses its head, and its hands. Since the conventional picture won’t work, we just have to get a little more creative, and find a new way of looking at the guitarist. Let’s start by positioning the mics in a good location for the audio, then see if we can find an interesting picture.

I usually use mic stands to position the camera – I have several boom stands, and they give me more flexibility than a typical tripod. I use the Zoom MA2 Mic Stand Adaptor or the Edirol OP-MSA1 to attach the camera to the mic stand. Interestingly, the Edirol adapter costs more and looks much more sophisticated, but the simple stick from Zoom does a great job and is handier in many ways.

I put together a video showing some different locations for the Q3HD using this concept of positioning for audio first. The first is the classic “off the 12th fret” location nearly universally recommended as a starting point for recording acoustic guitar. Actually, I used two variations of this, one below the guitar pointing up, one above and pointing down. From there I moved to the “off the bridge” lower bout position, and finally I tried the “over the shoulder” position that places the mic near the player’s ear.

In the video you can see the locations and quick samples of the audio and video that result from each position. All of these could be improved for both sound and picture by some extensive tweaking, and of course it’s a lot easier with two people than one.

In the process of creating the tutorial I realized I had enough material to put together a music video, a somewhat quirky one but perhaps an interesting project. I decided to try to edit the clips in REAPER, and with a bit of patience I was able to pull it off. I used Cockos LICEcap and the GIMP image editor to create a title, and rendered the results as a 1280×720 29.970 fps .MOV using the FFmpeg libraries. Here’s Waialua Slack Key as seen through the Zoom Q3HD at various angles:

I wouldn’t call the video an artistic masterpiece, but it serves to demonstrate the different angles and differing timbres of the audio captured from different locations. Hopefully it will serve as a bit of inspiration for other folks trying to get great audio and video from their Zoom Q3HD.

I also made a big step in my video camera arsenal with a shiny new Panasonic Lumix GH2. This little beauty is packed with a high performance sensor, lots of onboard image processing, a full range of manual controls, interchangeable lenses, and enough options and settings to keep this novice busy for a long long time.

I’m a long way from figuring out how to get the best from this wonderful camera, but I took advantage of one of the lovely “convertible weather in February” days we get here in Northern California to shoot a bit of outdoors playing. The wonderful old sentimental ballad, Silver Threads Among the Gold has become part of the slack key repertoire thanks to Leonard Kwan and Ledward Kaapana, and this song about the inevitable process of aging seemed to fit the austere scene.

By the way, I have a pretty fair collection of videos featuring Hawaiian slack key, mostly by me but with a couple of guest artists as well. You can see them at my YouTube channel.

The audio for this clip was captured using the Zoom Q3HD, it was placed about 24″ from the guitar, you can see the black blob of the windscreen in the lower right corner of the main performance shot. I also used the video from the Q3HD for the closeup of the picking hand.

I edited the clip in Edius Neo 2.5 and used a new reverb plugin from Jason Breebart’s Toneboosters line. The crows came with the house.

How About That H2

Then I remembered the excellent quality of the headphone/Line Out on my Zoom H2. While the quality is not quite at the level of the NT4 through a fine preamp, the H2 does a nice job of capturing clean, accurate audio. In the past I’ve used the H2 as a powered mic feeding the Kodak Zi8 and the Sanyo Xacti HD2000 mic inputs. This was always problematic, because the sensitivity of the mic inputs is a poor match for the output of the H2. It took a lot of experimenting to get the right level, then the camera preamp would wreck the audio anyway. Since the Q3HD audio input is called Line In, it should be a much better match for the Line Out of the H2. Another huge plus – the Q has metering!! This is a feature that is usually reserved for much pricier camcorders, and it definitely adds to the usability of the Q.

Time for a Test

With the house all to myself for the morning I set up in our bedroom to enjoy the lovely morning light, and shot video clips with the H2 as a Line In source as well as a few using the camera mic at various distances. With the construction project down the block and the heater going on an off, there’s plenty of ambient noise to stress test the various recording tactics. I shot a “making of” video using a Panasonic Lumix ZS3. I did a terrible job of setting the camera for the conditions, don’t blame the Panny.

I also didn’t bring a tape measure to the shoot (I learn a little every time, but at my age I forget a bit as well) or frame the close shots very well. I’ve thought about reshooting to get accurate measurements, but then I realized that each of us will be working with different conditions and different goals, so the exact distance will depend on the situation.

Interesingly, only the last shot with the Q3HD camera mic had audio quality similar to the H2 miked shot, and by that time the camera was at about the same distance as the H2 had been.

Field Recorder as Powered Mic

Naturally this technique of using a field recorder as a powered mic will work with any recorder that provides a monitoring signal in its headphone output. The problem might arise that the headphone output is low in quality, this was certainly true of the Fostex FR2-LE, for instance. Great recordings, but terrible monitoring sound. But if you have a recorder with a good clean headphone output, you have a tool to improve the audio in your Q3HD video clips.

I’m calling this experiment a success. I’m happy with the video I made using the H2 as a powered mic into the Line In of the Q3HD. I even have a wind sock for the H2, so I’m all set for shooting in the tropical tradewinds.

Hold the Phone

I learned a few things in the process of getting some comparison clips. First, I figured out how to use some of the mic stand parts I’ve collected as an iPhone “tripod.”

iPhone Tripod Improvised from Mic Stand Parts

The soft rubber mic clips don’t really fit any of my mics, but they’re handy for holding onto odd objects.

Then I learned that the iPhone viewfinder is not very video friendly. The right edge of the actual frame is hidden under the toolbar, so I was stuck guessing and experimenting to find the right position for my shot. And the size of the framing is off, the actual image is cropped noticeably larger than the view on the LCD, that is, if the viewfinder shows the image within some boundaries, the actual image will spill over those boundaries, and the object of interest will be smaller than expected. Once again, it was a matter of experimenting, guessing, griping, and doing it again to find the right distance to get the framing I wanted. (I’m hoping some iPhone guru will read this and call me names while explaining what I’m doing wrong!)

Here are the two clips on YouTube.

You can also download clips as delivered by the cameras: iPhone 4 Video Zoom Q3HD Video

Built-in Editing

Both cams have built-in video editing. The Q provides only bare bones functions, trim or divide. Trim lops off the beginning up to the selected point. Divide splits one clip into two at the selected point. But this is enough to pull one good performance out of a bunch of retakes, and clean up the beginning and end. It’s really all you need. The interface is a little tweaky, but I figured it out without reading the manual. The iPhone has a lot more power in the form of iMovie, an inexpensive app. It offers titling, more complex editing, inserting additional media, audio substitution. And the interface feels slicker. But I had to go online to read the FAQ to figure out how to use the thing, I didn’t need themes or titles or additional media or audio substitution, and the program appeared to transcode the video in the process of creating the output. The processing time on the Q was much shorter, making me think it left the original video compressed when trimming and splitting.

I tried out the possibility of editing iPhone video on the Q, and vice versa, but no go. I also was unable to copy a file back to the iPhone, or sync it to a location where I could pull it into iMovie. (iPhone guru, here’s another opportunity to berate and educate me. Thank you!) I was able to move files back and forth on the Q3HD by simply copying them to and from the SDHC card.

Perform for the Edit

One last thing I learned about editing on the camera – a simple “look at me play guitar” video doesn’t have much going on to mark edit points. So even though we don’t need a clap for audio syncing, it’s helpful to make a big obvious motion, like a large clapping motion, to mark the start of retakes. It’s also helpful to leave some generous and profound pauses before and after the playing so the edit point has enough leeway. These tricks will make it a lot easier to deal with the small screen and small controls when editing inside the camera or phone.

Another thing missing on both these cams and many others below the professional ($2000 and way up) level is the lack of metering and monitoring. There’s not a way to tell if the audio is being recorded well until playing back the clip, when it’s too late to make any corrections. Even when I’ve done multiple test shots I’ve always considered myself very fortunate if I manage to capture some useful audio with this kind of gear.

The Q3HD Arrives

A couple of years ago, Zoom introduced the Q3, a pocket video recorder that emphasized the audio quality. It included stereo directional mics in an X-Y configuration and didn’t inflict excess processing on the audio. Unfortunately, they released this SD format camera just when HD video took off. YouTube switched to HD format, Flip put out the MinoHD, HD video recording began to appear in phones, HD was everywhere except in the Q3. But earlier this year Zoom announced the new Q3HD, which addressed the format issue, offered improved low light performance, added special settings for difficult lighting conditions like concerts, and added a Line Input for stereo audio. In the past, new products from Zoom have arrived long after their planned date, so I was expecting to see the Q3HD sometime in 2011, but happily the first shipments have arrived in the US and are available from a number of retailers. I bought one from BSW and it arrived the day after Thanksgiving.

My initial impressions are really positive. The build quality seems a bit better than my venerable H2, and it’s noticeably more solid feeling than the Kodak Zi8. The menu system is easy to manage, there’s audio metering and monitoring during recording, it even comes with a 2 GB memory card and a set of AA batteries.

Three Way Comparison

I mentioned the other cams that I purchased because they had audio input and some degree of manual control. It seemed reasonable to use them in a comparison with the Q3HD. So I setup in Digital Duck studio (also known as the spare room upstairs) and shot some video, first using camera mics then external mics.

I chose 720p for the video format for all three cams. The Zi8 and Sanyo don’t offer control over the audio format, but on the Zoom I selected CD quality PCM (44.1 khz and 16 bit word length).

For the external mic shots I connected an Audio Technica AT822 to the Zi8, a Rode NT4 to the Xacti HD2000, and a pair of Shure KSM141s through an M-Audio DMP3 to the Q3HD. Here’s the video that resulted:

When I listen to the clips on my “junk reference” computer speakers, the difference in audio quality isn’t that great, but as soon as I put on headphones or turn on the good monitors, the audio from the older cams sounds dreadful in comparison to the Q3HD. My Zi8 sounds worse than I remember, so perhaps it’s dying, but even the HD2000 features burbling noises in the quiet moments and an odd tonal balance. I also feel that the video quality of the Q3HD was noticeably better than that of the Zi8. Looks like the Kodak is going in the drawer.

My First YouTube Clip from the Q3HD

The Q3HD has minimal built-in editing, just “Divide” and “Trim” – minimal but sufficient to create a clip for YouTube, so I figured I should put up something straight from the camera. I chose the Rode NT4 stereo mic just to simplify the setup, and fed the mic into the M-Audio DMP3 preamp. From there an adapter chain converted two 1/4″ outputs to a single stereo cable ending in a 1/8″ stereo plug, which went into the Q3HD Line In jack. I took advantage of the meters on the Q3HD to check my levels, aiming for average levels around -20 dBFS and peaks around -6 to -8 dBFS.

With that set I chose the highest resolution video, 1080p format. I chose PCM (uncompressed WAV) audio, but stayed with the basic 44.1 Khz sample rate and 16 bit word length. After a few test shots to get the framing right I recorded a few takes of an old R. Alex Anderson song, Haole Hula. With a little scrolling around in the playback, I was able to trim the clip down to the performance I chose.

Based on my experience with other camera software, I didn’t load the Handyshare package that comes with the Q3HD. I simply removed the memory card, stuck it in my card reader, and copied it to my hard disk. From there I uploaded to my YouTube account. Here’s the result:

Since I was determined to use only in-camera editing, my clip has no title, no fades, no credits, but I feel that it presents the song in fine fashion anyway.

Framing Changes with Format

One thing I stumbled on while shooting my clip – changing the video format also changes the framing of the shot. I framed the image then decided to switch from 720p to 1080p. I made the format change and shot a clip, and when I played it back my head was gone! Here are a couple of snapshots that illustrate the difference:

Framing at 1080p

Framing at 720p

As you can see, the change is pretty dramatic. So be sure you check your framing whenever you change the video format, or you might lose your head.

Not Quite Perfect, But Pretty Darned Good

On the frown side, I’m a bit annoyed that I have to buy the AC adapter separately. My brief attempts to shoot in lower light were not a big success. And while I was really pleased when I learned that Zoom added a Line In jack on this new camera, I’m already wishing it had a 1/8″ stereo mic input as well. With that extra feature the NT4 and the Q3HD would make a fantastic package for portable recording. As it is, I’m shopping for a battery powered two channel preamp with P48 phantom power that won’t bust the budget … and I’m not having a lot of luck.

On the smile side, I expect to have a lot of fun with this little camcorder. The video quality is reasonable in good light and the audio sets a new standard for a consumer cam. It’s nicely made, easy to use, an all around fun toy.

Now for the icing on the cake – the development team has linked in the FFmpeg video libraries and given REAPER the ability to do simple video editing. I’m totally happy with my video editing system since I moved to Edius Neo 2.5, but when I played around with video in REAPER I realized that this is a tool many musicians want and need.

Many of us like to post simple performance videos on YouTube – not big production numbers but simple clips showing off our latest tune or instrument or technique. And we want to do this on a budget. We use inexpensive pocket video cameras, and they have at best mediocre sound quality. Many of us also have a fairly high quality recording system of some kind. Now with REAPER we can create our videos just like they do it in the movies – recording the sound on a high quality audio system positioned for the best audio quality while capturing the video at the distance needed for framing and perspective.

Using video in REAPER is a snap. First, go to the CockosWiki Video Support page for instructions on adding the FFmpeg libraries to the REAPER program directory.

Once the FFmpeg libraries are in place, just open the video file like any audio file, or drag and drop from your file browser into the REAPER track window. I’ve tried clips from three camcorders so far, the Kodak Zi8, the Flip Ultra HD, and the Sanyo Xacti HD2000. REAPER handled all of them with ease. Only the .mts files from my Panasonic Lumix TS2 failed to open, and I was able to convert them to an AVI that REAPER liked.

If I’m shooting video around my studio, I’m all ready to record audio into REAPER, so there’s another step out of the way. If I’m shooting on location it’s a snap to bring the recordings back to the PC and drop them into REAPER. The whole process is so easy that instead of writing detailed instructions I decided to simply shoot a video of shooting a video and demonstrate the syncing process.

So that’s how easy it is. Drag and drop, drag and drop, trim, drag and drop, trim, and render. When I think of the hours I’ve spent calculating frames in Avidemux, or waiting for Premiere Elements to redraw, or being told by Windows Movie Maker that it can’t deal with my file …

Here’s the video that I created in REAPER in the process of the tutorial. The Flip would have been happier with more light, for sure, but the quality you see in this video is right there with the original from the Flip. And the audio is in a whole different, and better, league. Naturally, since I was working within REAPER I could have easily added effects, applied EQ, worked over the audio to improve the result.

By separating the mic from the camera, I should be able to position the mic for optimum sound while placing the camera for the visual effect I want. Of course, I can do this wihout an audio input on the camera by a technique called parallel recording, that is, recording on a separate audio system of some kind. The familiar clack of the slate at the start of a movie sequence is used to make it easy to synchronize the picture and sound, and this technique works very well for my homebrewed videos as well. Still it’s very tempting to think that recording directly to the camera is an easier way out, with camera audio every take and every edit lining up without any extra effort.

I’ve been interested in the range of options for attaching a mic to the Zi8, and back in December I went on a little mic shopping binge. Between my storage drawer, Ebay, and Sound Professionals I gathered a collection of mics in a range of prices. But I was struck with a glitch in my Zi8 audio, which required several round trips to attempt repair at the East Coast contract repair depot. Eventually Kodak agreed to replace my camera, but by then I was on a lovely trip to Hawai`i.

Test Design

Now that I’m back I’m ready to offer up some test clips. At first I planned to video a number of brief samples of guitar and vocal, which was very troubling to me. I strongly believe that audio sampling should be done with a single performance and carefully volume matched, in order to reduce the number of variables in the sample. As much as we think otherwise, humans just cannot judge subtle (or even not-so-subtle) differences without careful controls. My aha moment came when I realized that I was only testing audio, not video, so I could use a recorded source to get a repeatable performance. Focusing on audio only also allowed me to upload the much smaller audio clips, so the comparison will be quicker and easier.

For level matching I decided to take the easy way out and simply normalize the files. Normalizing raises the peak of a file to a given level, in this case I chose -3 dBFS. I adjusted levels like this so our ears won’t be fooled by level differences, but these samples hide the differences in the real recording you’ll get if you use one of these mics. So I’m including the original sample as well. Remember that I adjusted the sensitivity of the Zi8 for each mic, so the volume levels are affected by that even in the original, un-normalized samples.

I first created a test recording in Reaper, some slack key guitar naturally. I recorded in mono with an Audio Technica AT4050 in cardioid pattern. Then I added a bit of narration. All camera mics I know about, and many accessory mics as well, are omnidirectional. There are plenty of reasons why an omni is a good choice, but real trickery is required to get a good stereo image with omnis. So I threw in a bit of movement in the stereo field to evaluate channel separation. I arranged the narration and the stereo test ahead of the music and adjusted levels.

I placed all the mics very close to my normal listening position and played back the test recording through my normal monitoring chaing – Echo Audiofire 8, Hafler P3000, Dynaudio BM6p. A Radio Shack SPL meter showed 74 db C weighted at the mic position.

As I mentioned, I adjusted the sensitivity of the Zi8 for each mic. In each case I adjusted the sensitivity to the highest level that did not show any red. I label the default sensitivity as 0, lower sensitivity goes from -1 to -5 and higher goes from +1 to +5. I wound up using the whole range in this comparison.

Test Clips

Here’s the reference recording. It won’t get any cleaner than this.
ref.mp3
or download ref.mp3

I used the Zoom H2 as the mic in my first experiments with the Zi8, so I included it in this comparison. It’s also handy because we can record with the H2 and the Zi8 at the same time. Even though the signal in the Zoom was very low and pumped up with a lot of digital gain in the normalization process, it is quite a bit cleaner than the Zi8. I set the Zoom to M Mic Gain because tests have shown that the H setting adds noise. This still resulted in a rather low level, so raising it also raised the noise as well.
h2p.mp3
or download h2p.mp3

Original versions, no normalization:
oh2p.mp3
or download oh2p.mp3

I used the rear mics on the H2 so I could operate the controls on the front of the unit. As a result the stereo left and right are backwards in both the parallel clip and the Zi8 clip. Here’s the output of the H2 headphone jack recorded on the Zi8 with sensitivity set to -3.
h2v.mp3
or download h2v.mp3

Original versions, no normalization:
oh2v.mp3
or download oh2v.mp3

If you’re like me you’ve been fooling with recorders for a while, maybe going back to cassette tapes. Perhaps, like me, you have some kind of little mic in a drawer somewhere. My drawer mic is labeled www.reactivesounds.com but the web site is a dead end now. I bought this to use with my minidisc recorder a few years ago. The mic is omni, no cable (requires a female to male extension), and has been repaired with superglue. I used -2 sensitivity for this mic.
drawer.mp3
or download drawer.mp3

Original versions, no normalization:
odrawer.mp3
or download odrawer.mp3

I’ve used Sound Professionals http://www.soundprofessionals.com/ for these kinds of recording gear for years. I used one of their T-mics for years with my minidisc, but somehow lost it. They still sell a mic with the same model number, SPSM-1, but it doesn’t look much like the mic I remember. I set the Zi8 to -3 sensitivity for this mic.
spsm-1.mp3
or download spsm-1.mp3

Original versions, no normalization:
ospsm-1.mp3
or download ospsm-1.mp3

The SPSM-15 is the least expensive single point stereo mic sold by Sound Professionals. It appears to have a higher low frequency cutoff than the SPSM-1, and/or lower sensitivity. As a result I set the Zi8 sensitivity to -2.
spsm-15.mp3
or download spsm-15.mp3

Original versions, no normalization:
ospsm-15.mp3
or download ospsm-15.mp3

The AudioTechnica AT825 has been a popular single point stereo mic for many years, but was recently discontinued. The replacements are much higher in price, so it might be worth watching for used examples of the AT825 on Ebay or Craigslist. It seemed perfectly matched to the Zi8 and worked well at the default sensitivity setting.
at825.mp3
or download at825.mp3

Original versions, no normalization:
oat825.mp3
or download oat825.mp3

I was impressed with the performance of the AT825. I was very disappointed in the AudioTechnica Pro 24. Although it looks much more impressive than the T-mics from Sound Professionals, it was very low in output, requiring that I set the Zi8 to the maximum +5 sensitivity.
atpro24.mp3
or download atpro24.mp3

Original versions, no normalization:
oatpro24.mp3
or download oatpro24.mp3

Let’s Look at the Video

Let’s finish off with some actual video. Be careful with comparisons, but perhaps there will be some blatant differences we can take away. When I framed the shot to include the whole guitar the camera wound up about 5 feet from the instrument. I adjusted the Zi8 sensitivity for each mic using my normal procedure – the highest setting that did not show any red squares.

I had a little fun with my new Edius Neo 2.5 software putting together a compilation of the mic demonstrations.

All these mics sound better here than in the audio test with the recording, but they still show their basic characteristics. The camera mic picks up a lot more room, as it must at that distance. And since the room is a boxy little thing, the recording sounds small and boxy.

The H2 in this clip is very slightly overdriving the Zi8, to my ears. The slightly dark voice of the H2 comes through as well. The Sound Professionals SPSM-15 does a pretty decent job for very small bucks. I failed to mention in the video that the sensitivity on the Zi8 is down to -3. This means the SPSM-15 is quite sensitive, so it might be a problem if your source is loud. To my ear the AT825 is the winner. I like the clear presentation. I’m sorry I didn’t do a better job of balancing the channels.

I hope these comparisons are useful. I plan to try these mics with a different camera in a future post.

* * * Correction on 12/03/2010 * * *

The mic I keep calling an AT825 is actually and AT822, the unbalanced version. Sorry for the brain fog.