The moon in transit, moving in real time across the frame. By “real time” I mean that the tripod was locked down and the camera was stationary, pointing at one point in the sky. All the movement here is caused by the motion of the earth and the moon.

Except for some of that shaking. I’m pretty sure that was caused by walking within 15 feet of the camera. Sorry about that.

Shot with a Canon T3i at 24fps, ISO 200 and a FD ƒ/8.0 500mm reflex lens in Brookly, NY.

We got our first major snowfall here in Salem and, just like last year, I wanted to try and capture it. If you’ve ever scrolled through a long grid of photos in iPhoto, Aperture or Lightroom you might have seen something that looks a little like the video above. That’s where I got the idea anyway. Unfortunately, it didn’t really turn out like I had imagined. I was hoping the difference in color would be really apparent as the snow blew in and night fell. But, at this point, I’ve already spent two days on this and I’m sick of looking at it.

I used the iPhone’s camera and let it shoot every five seconds for two hours. The phone was supported by my Glif on top of a camera extender on top of my Manfrotto 501 baseplate.

I used a replicator in Motion, each offset by one frame in a six by six grid. That was enough for my computer to beg for death, so I baked it out and finished the rest in Final Cut. The music is Dirty Glass by WHY? [iTunes].

I made another title animation in Motion, this time 20 seconds instead of 10. That turned into around 16 hours of work and another 9 or so for export.

The texture I used for the floor and wall came from cgtextures.com, who appear to have a very reasonable terms of service regarding commercial and non-commercial use. The reason it took so long to export was because of the text. Since I wanted the camera to fly in so close to the title I needed the text to render at a very high quality. That takes time. The settings I used also did some hardcore anti-aliasing for object intersections which helped sell the point where the wall and floor met.

To compress for the web I ran it through an iOS droplet I made from Compressor with a data rate of 1Mbps. Any higher and my web host seems to have a problem delivering the video fast enough to iPhones or iPads. The Ogg Theora (Firefox) version was encoded in Terminal with FFMpeg2Theora at 3Mbps. The WebM version was done with Make Web Video, a Firefox extension (also at 3Mbps).

If you have any questions, feel free to write.

I’ve never spent a ton of time working in Motion. I’ve always assumed After Effects was a superior application for compositing, so trying to become as skilled in Motion as I am (or at least as I think I am) in Final Cut seemed like a waste of time. I recently watched this tutorial by Andrew Kramer and wondered if it was even possible to create those kind of elements within Motion. He, of course, was working in AE.

Turns out you can, more or less. I spent a few hours yesterday poking around in Motion and made a proof-of-concept title animation. This is what it looks like.

Here’s something I worked on recently. Shot with the DVX100a equipped with polarizing and UV filters. Edited in Final Cut Studio and graded using Magic Bullet Looks.

Our friends, Justin & Elise, asked us to watch their friends while they went camping for a few days. Everything here was shot on iPhone 4 and finished in Final Cut Studio.

A quick video shot and edited on iPhone 4 in Beverly, MA. If you have trouble with the video above, watch it on Vimeo.

Our eyes don’t see all of the subtle variations between every color. We do, however, see the difference in shades of a color. In video, this is called luminance. The color is referred to as chroma, or chrominance. Since the dawn of video, cameras have been using a technique to reduce the amount of color information captured to reduce the amount of bandwidth needed to deliver it. Since our eyes won’t see these colors anyway, we’re not missing much.

How is it done?

The most common model of color subsampling uses an area that’s 4 by 1. Four pixels wide and one pixel high. For each of those pixels measurements are taken, and values assigned. The first is for how luminous each is, measuring once for each pixel. When we’re done we’ll have four unique measurements for each of our four pixels.

Next, color information is measured. We already know that our eyes aren’t going to see all of the color that we’ trying to record, so instead of getting a color value for each pixel we’ll be averaging the colors together. This dramatically reduces the amount of data needed to transport the image. Colors accuracy is also significantly reduced. When we’re done we’ll have one measurement that represents four pixels.

This example assumes we’re using 4:1:1 sampling. It’s commonly found in consumer DV cameras and for most applications it’s just fine. The real problem starts when you get into color correction and — even more so — when working with green screen. It’s then that the more color information you have to work with the better off you’re going to be.

Below are brief descriptions for some of the most popular color sampling methods.

4:1:1

DV video, represents a highly subsampled image. Sample area is four pixels wide and one pixel high. Found in cameras like the DVX-100.

4:2:0

Some HD cameras use this including the Sony EX-1 and EX-3. Technically, its the same as 4:1:1. The camera is still averaging the color of four pixels down to one. The difference is that the sample size is two pixels high by two pixels wide. Commonly used in HDV cameras.

4:2:2

Vastly superior to 4:2:0 and 4:1:1. Each color sample size is further broken down from four pixels to two, yielding much higher color accuracy. Used by DVCPRO HD and Digital Betacam.

4:4:4

Every color is represented as accurately as possible. No subsampling. Used by RED, Viper and Genesis camera systems.

4:4:4:4

Same as 4:4:4, but includes an alpha channel, typically used for chroma keying.

Why care?

Maybe you’ve noticed artifacts in your image; blockiness where colors bump up against each other. This is because the same algorithms are trying to average the colors the best it can. Sometimes, especially in these transitional places, it may guess wrong.

This is important to understand when doing color correction, and absolutely essential when doing any kind of chroma keying. If you’re working with a camera that has low color sampling resolution you’ll need to know what kind of limitations you’re up against. You may not actually be able to get the seamless VFX shot you originally had planned. But that’s filmmaking, right?

Now what?

I encourage you to figure out what color sampling method your camera uses. The more you know about your camera, the more you’re going to be able to get out of it. You’re also going to make things easier on yourself once you get to editing.

Additional resources

Color Subsampling on DVXuser

Subsampling notation PDF

Halloween in Salem, MA, 2009. Occurring on a Saturday, it was said to have been the largest event in town for nearly a decade. An estimated 100,000 people showed up to celebrate.

Used our Flip Ultra HD to capture our cat’s holiday excitement.

Same basic setup as last time. But instead of scaling up for 1080p I scaled down for 720p. I thought maybe this would make the picture a little sharper, and that seems to be the case. It was still light outside which makes the image cleaner anyway.

I wasn’t sure if I was going to capture anything interesting out the window with this one, and it looks like I was right. The most note-worthy observable change here is how the snow on the street melts away — not terribly enthralling.

I used the iTimeLapse Pro [iTunes, $2.99] iPhone app to capture the beginning of the first major snow fall of the season here in Salem. 760 frames total, playing at 24 frames per second. The phone was standing vertically in its dock looking out the window.

Each individual frame started 1536 wide by 2048 high. Which is almost wide enough for a standard 1080p video (at 1920 by 1080). After assembling the frames in a custom FCP timeline, I exported the video and dropped it in Motion where I added the camera movement. The video was still very red from all of the street lights, so I ran it quickly through Color to make it more blue. The graininess is from shooting in such low light.

The final version was exported using the Apple TV preset in Compressor.

An interview with Charles Woodbury Wallace, B-17 pilot in WWII. Filmed at the Wings of Freedom event held September 19th, 2009 in Beverly, MA.

See it on Vimeo.

I wasn’t super happy about how I color graded the Boat Festival video. So I thought I’d take another stab at it. Afterwards I was curious to see just how much I had deviated from the first grade — quite a bit it, it turned out. You can see a side-by-side comparison below. Some shots will jump out at you more than others.

A video snapshot from an antique boat show here in Salem, MA. Shot with DVX100a at 24pa, with a 1/48 shutter, using polarizing and UV filters. Audio captured with a super cardioid shotgun microphone. Edited and graded in Final Cut Studio 2. See the rest of the series here.