Shooting Video at The JazzSchool, Berkeley

We’ve shot some great music at the JazzSchool in Berkeley, such as The Primavera Latin Jazz Band last year.

Our most recent shoot was at a recital of jazz vocalist and educator Maye Cavallaro‘s students.

The facility is located in the basement level of a large office building.  The performance space is small – maybe 100 seats max, mostly seated around cafe tables.

The stage at the JazzSchool, Berkeley

We set up a tripod next to the sound booth in the back and had one roving camera on a monopod.  When the audience fills up the place it’s very hard to move around, but you can usually squeeze into an aisle on the right or near the main desk on the left.

The audio engineer that night (Lou Judson) was kind enough to give us a good mic-level XLR feed to the camera.  Sound was good but dry, so I also captured audio through a Zoom H4n to get audience reaction and mix in a little room sound.

Unfortunately the lighting is mixed, with stage lighting mounted on the somewhat low ceiling, but also recessed compact-fluorescents overhead on the stage, and some daylight coming in from windows near the ceiling on the right.  The stage lighting was uneven that night, with a bright spot on the left front but falling off by about a full stop on the right side of the stage.  We didn’t have the time or authorization to mess with the lights so we shot it as-is with a tungsten white-balance.  It was evening so the outside light didn’t intrude much.

I was using the small Canon Vixia HF-G10 on a Manfrotto monopod because it’s less conspicuous and easier to maneuver than the 5DmkII with a big lens on a tripod.  It worked well and intercut nicely with the AVCHD footage from the Sony HDR-AX2000 in the back.  The only thing I missed with the G10 is a decent zoom control.  I was using both the zoom lever on the top and the remote control, but neither gives a very controllable zoom.    I might look into a simple external LANC zoom control.

The JazzCafe is adjacent to the performance space and serves beer, wine and light meals.  And of course downtown Berkeley has dozens of restaurants within walking distance.  There is ample garage parking nearby and street parking if you’re lucky.  Bring a good equipment cart.  There’s an elevator to the left of the main entrance.

 

Compressor: Looking for Speed — Update: Matrox CompressHD

In the continuing quest for speed using Compressor, after testing to find the fastest workflow for me I started looking for hardware acceleration.

It looks like the main player in this is Matrox with their Max Technology as an option on their hardware In/Out boxes and cards.  The cheapest implementation is the Matrox CompressHD card ($499 retail) which only gives you H.264 acceleration, no I/O.  After finding several bad reviews of the video capabilities of the next cheapest product, the Matrox Mini02, I decided to just try the CompressHD card.

I found one on eBay for about $100 less than retail.  Being used, I was taking a chance, but the guy said it tested perfectly.  The card arrived today in a plain brown box with lots of bubble wrap but nothing else.

I installed it in an empty PCI-e slot in the Mac Pro and fired it up.  Then I went to the Matrox site to download the software.

(Note to self:  Write down the serial number of a used card before installing it!  Matrox required the serial number to download the software & I had to open the Mac back up and pull the card to find it.)

There were about a dozen choices of software to download, based on the Mac OS you’re using and whether you’re using Compressor or Adobe Media Encoder.  None of them matched OS 10.6.8 that I’m using, but I picked the closest one before it.  Software installation was straightforward but caused some anxiety when it automatically updated firmware on the card, since I didn’t even know if the card was working yet.  Luckily things went smoothly and after a restart I was ready to go.

The software installs the Matrox Max H.264 settings into the Custom folder of Compressor’s settings.  I loaded up the 90 second ProRes LT test file from my previous tests, grabbed the High Quality 720p/29.97 Matrox setting and started the compression.

Previous fastest encode was 03:07.  Encoding with the Matrox Max CompressHD card:

01:22 — More than twice as fast with the card, and 8 seconds faster than real time!

I also tried Fast Encode, which finished in 50 seconds but the quality suffered a bit.  I could see some stairstepping in the titles and the image appeared softer.  So I’ll probably stick to High Quality unless it’s just for a quick approval clip.  Good to know it’s there if I need it.

Matrox Max High Quality Encode - 720p 100% Crop

 

Matrox Max Fast Encode - 720p 100% Crop

I’ll try a longer file soon, but this is very promising.

Tascam DR-40 Sync Drift / FCPX Audio Fix Missing

I was editing an hour+ long performance video in FCPX using audio from both an audience mic into a camera, and audio from the front-of-house board recorded on a Tascam DR-40.  I synced everything with PluralEyes, but I noticed an echo toward the end.   Zooming in to the waveforms I could see that the two audio tracks were in sync at the beginning, but slightly out of sync after an hour.

I couldn’t find a way to change the speed of the Tascam track in FCPX.  In FCP7 there’s a Change Speed feature where you can input the percentage or duration change down to 3 decimal points.  I searched the forums and blogs and it looks like FCPX can’t do it.

I went back to FCP7 and re-synced everything there.  Zooming in tight to the waveforms and using Modify: Change Speed (command + J) I adjusted the speed of the track with trial & error until it matched up.  Speed change was 100.008 percent.

Out of curiosity I ran a test with a Sony HDR-AX2000 camera, a Zoom H4n and the Tascam DR-40.  I found that the Tascam track was 9 frames slower than the audio from my camera after one hour.

Sync Drift After One Hour: Tascam +9 Frames, Zoom +2 Frames

Interestingly, the Zoom H4n was 2 frames off too, so eventually it would also need to be adjusted.  But at one hour the sync difference was not noticeable.

I went to Change Speed and entered 100.008 %, re-synced at the beginning and got the Tascam track to within 1/2 frame.

Sync After Speed Change

Interestingly, I couldn’t change the speed of the Zoom track less than 100.005%, so it couldn’t be improved.

So for sequences of less than 15 minutes or so the Tascam should be ok, but for long form shoots (which I do a lot of) it’s kind of a pain.  As much as I like the DR-40 (for the line/mic switch, the locking XLRs, and light weight) I think I’ll have to get rid of it and stick with the Zoom H4n.

Tascam DR-40/A-Data SDHC FAIL

I bought a Tascam DR-40 audio recorder to try out after reading some good reviews.  It is the cheapest recorder ($165 on eBay) I could find with XLR inputs that would record to 48kHz/16bit to match my cameras.

There’s a lot I like about the DR-40 like the mic/line switch for external inputs, running off of USB bus power to save batteries, ability to record at two levels simultaneously to avoid distortion, etc.

I used it on a long performance shoot last week to take audio from the board.  I was recording to an A-Data 16gb SDHC card which had been reliable up until now.  During the shoot I had it set to automatically start a new file after recording 1gb.

After about the first hour I looked at the Tascam and it had stopped recording.  There was an error message saying “Write Error – Press Enter.”  I pressed enter and kept getting the error.  Finally I powered down and restarted the recorder but THE FILE WAS NOT SAVED!  When I got back to the studio and checked the card there was a file name but 0 kb in the file.  I tried a couple of file recovery programs but there was nothing there.

Luckily I had backup audio from one of the cameras, but lost a lot of confidence in the Tascam and A-Data.  I’m not sure which was the problem, and haven’t been able to duplicate the issue.  I think I’ll stick to the Zoom H4n after this.

Next:  Tascam DR-40 Sync Drift

 

Compressor: Looking for Speed

Apple Compressor has always been a mystery to me.  Sometimes it takes forever to transcode, but when I’m in the middle of a job I don’t have time to figure out why.  I just plan my workday to transcode overnight and hope nothing crashes at 4am.

So I took some time today and did some tests with a 90 second clip from FCP7.  Send to Compressor 3 and transcode to H.264/720p took 09:14 with a lot of other open programs on my 8 Core Mac Pro — 6.15 times the length of the clip.

After restarting and having only FCP7 and Compressor 3 on, same file took 08:54

Saving the Compressor 3 settings and reopening with Compressor 4 with Qmaster on (7 instances) yielded 09:01 — Seven seconds worse than Compressor 3 without Qmaster (I’ve never been able to get Qmaster to work with Compressor 3 — another story).

Next I went back to FCP7 and Exported a QuickTime file with current settings, saved as a self-contained movie.  It took 00:33 to export, or about 1/3 of the length of the clip.

Transcoded the file with Compressor 4 with Qmaster on.  03:15

Transcoded with Compressor 3 without Qmaster: 03:07

So the best workflow for me right now is:

1.  FCP 7: Export QuickTime file with current settings.  Turn off FCP and any other open programs

2.  Transcode file with Compressor 3

This yields a total transcode time of about 2.5X length of the clip, so a one hour video should take 2.5 hours to transcode.

Update:  A one hour clip took 02:40:30 to create the QT file and transcode.   A lot better than before.  But I’m going to look into hardware acceleration.

Compressor: Reset Background Processing

I’ve been working on a large video project and have been frustrated by extremely long Compressor sessions.  Taking a one-hour video from FCP 7 edited in ProRes LT to H.264 720p was taking 15 hours or more in Compressor 3.53.  I tried a trick (from CreativeCow Forums) to take the file to Compressor 4 by saving the Comp 3 file and reopening it in 4, but it was worse — the time remaining to finish kept growing until I finally canceled it when it grew to 30 hours.

Tried another suggestion from CreativeCow to Reset Background Processing and that helped a lot.  Transcode times dropped to 8 hours for a one hour video.

 

Shooting Video at the DNA Lounge, San Francisco

We were hired to shoot a 3 hour event with multiple performers at this South of Market club.  The DNA Lounge is a cavernous former warehouse that has been built out with a large stage, impressive lighting and sound systems, and a mezzanine that runs around 3 sides.  It’s painted completely black inside.  They do a wide variety of shows there, from hip-hop dance to indie rock to burlesque reviews.  This night was a relatively tame benefit for a local charity and a going-away party for a well-known cabaret singer.

I made a site visit a couple of weeks before and shot some test footage.  I also talked to the audio tech so I thought I would be able to get a good feed from the board.

I brought 4 cameras — 2 Sony HDR-AX2000s, a Canon HF-G10 and a Canon 5DmkII.  I had planned to lock down one of the Sonys and the G10, have my second cameraperson run the other Sony on the mezzanine and I would shoot with the 5D on the dance floor.

The first problem was that the management wouldn’t let me shoot with a tripod on the dance floor.  I had to either stand on a 4×4 plywood box for the whole night or shoot from the mezzanine.  I decided to put the G10 on a tripod on the box, unattended because it has a long run time with its internal 32gb and two SD slots that will run consecutively.  That became my wide covering shot.  (Click on photos to see a larger image.)

Coverage shot, Canon HF G10

I mounted one of the Sony’s behind the stage pointing past the performers toward the audience.

DNA Lounge, Rear camera

The other Sony was on the mezzanine stage left and I was stage right with the 5D.  My 70-200 couldn’t get very close from that distance so I had the other camera shooting tight closeups while I covered two- and three-shots.

The audio was supposed to come from the board into my Tascam DR-40, but I had a write error and lost about an hour of audio.  My backup was that the DNA always captures the audio on an mp3 file that they post on their website for a week, but that turned out to be distorted.  So I had to make do with audio from the G10 which was the closest camera to the action.  It’s not great, but it’s usable.  Luckily I remembered to turn on the mic attenuator when I set it up.

The DNA is great for shooting from the mezzanine, especially that night when they had it blocked off so we had most of it to ourselves.  Unfortunately there’s no elevator so you have to lug your gear upstairs.

The lighting is mixed, with an LED color DMX system and a single quartz follow-spot operated by a guy on the mezzanine.  When the spot is on and lighting the main performer you get some nice backlight and hair-lighting from the LEDs that are constantly changing.

DNA Lounge, Front Spot Light On

I shot 1080p30 at 1/30 sec at around f4 (5DmkII at ISO 1250, Sony’s at +6 gain) with the white balance set for tungsten.  Sometimes the operator turned the spotlight off, and then you get some horrible blue, magenta or yellow color on people’s faces that’s impossible to fix in post.  I think those LEDs must be missing some part of the spectrum because it is really ugly by itself, plus they are probably “daylight” balanced so that doesn’t help.  If I shoot there again I’ll ask the light operator to keep that spot on all the time.

DNA Lounge, Spot Light Off

When we got there at 4:30 on a Tuesday there was plenty of street parking, but when I did my site visit on a Friday at 10pm parking was impossible, since it’s on the same block as Slim’s nightclub plus a couple of restaurants.

The DNA runs a pizza cafe that’s open 24/7 next door to the club, and there’s a full bar and snack bar in the club itself.

 

Freelancers Union – Contract Creator

The Freelancers Union is a New York based group working to improve the lot of those of us working outside the corporate world.  Some of the issues they address are:  affordable health insurance for freelancers, fair taxation, unemployment protection, unpaid wages and electing political allies.  They sponsor member events and offer member discounts, mostly in the NY area.

One of the free tools they’ve developed is Contract Creator, a fill-in-the-blanks way to generate a straightforward contract between you and your client.  I’ve used it several times and people seem to appreciate knowing the specific product they will get, when to expect it and how any problems will get resolved.  It also specifies what they are responsible for & when.

Once you fill in the basic data, you can save it as a Word file.  I usually customize that file with my own letterhead and soften up the formality a little.

The Freelancers Union is free to join, but consider making a donation so they can keep up the good work.

Why I can’t use the Canon T3i

I picked up a Canon T3i (D600) last year in a bundle deal that dropped the cost of the body to under $500 after I sold off the rest of the bundle.  I was hoping to use it as a backup to my 5DmkII and was excited about the added capabilities like the 1.6 crop and digital crop — giving me a lot more range with my lenses. The T3i has been well-reviewed by many and is a very capable and affordable DSLR for general use.

Unfortunately I shoot mostly in low-light situations and the T3i doesn’t hold up.  Shooting video at ISO 1600 (1/30 sec, f2.8, L lenses) the blacks are noisy and ugly and it just doesn’t intercut with video from my other equipment.  Too bad.  I’ve seen some very good low-light footage shot with this camera but for my purposes it just doesn’t cut it.

I sold it and picked up a Canon HF-G10 prosumer camcorder instead.  It’s pretty great for a sub-$1200 video camera.  More on that later.

Welcome

I’m starting this public journal to collect and share experience gained from shooting and editing location video and photography here in the San Francisco Bay Area.  I’ll post things I learned, gear that works and doesn’t, shortcuts, location issues, and anything else that I want to keep track of or that someone else might find useful.

My training is in still photography and I worked as a corporate/industrial photographer for many years.  I started shooting video on the job about 10 years ago, and built up skills as needed and upgraded equipment as technology improved.  I started editing with Final Cut Pro soon after I started shooting, and I’m continually looking for better, faster and easier ways to edit.

These days I shoot mostly video of live performances, with an emphasis on jazz singers and musicians.  You can see a sample reel and full clips of some of my recent work here:

http://www.behance.net/paulkagawa

See the “About” page for a list of gear I use.

Please feel free to leave comments or questions.