New video for Seattle Cancer Care Alliance

Here’s an informative little cancer video project my crew and I just delivered to Seattle Cancer Care Alliance, a client that’s been keeping us very busy this summer. We’ll be doing a lot more videos of this kind for SCCA over the next few months, uncovering powerful human stories about these innovative healthcare providers and the patients they serve.

How it was lit

I typically start to light a scene by deciding which light to take away. In this case,  the lab was grossly over illuminated (for filmmaking purposes) with ugly overhead fluorescents. I wanted to separate the doctor from the background, through both luminosity and color. So I turned off half the overheads, the half behind the subject, and left the half in front of him on. Then, by setting my Sony FS5’s color temp to 3200K, I was able to create a nice blue wash in the background from light leaking through the closed windows. That created nice color separation. But his face was now too dark. I fixed that by filling  strongly with my LiteGear LiteMat2 placed camera left, through 1/2 grid cloth. Next, I gave a little blue kick to his shoulder with my lovely new Dedolight DLED7 through Hampshire frost. Then I just set the aperture on my Zeiss 50mm with Speedbooster to f/2.0, and boom! Done.

B-Roll

B-roll was all shot very quickly, handheld with Zeiss primes and a 100mm Canon macro for the tight stuff. When I got to cutting the piece, I was reminded why my Canon 17-55mm f/2.8 IS lens is usually my go-go b-roll lens: because it is image stabilized! My FS5 is so ergonomically awesome, that I find myself shooting handheld more than ever. But when doing so, a lens with IS really is nice to smooth things out. Nevertheless, I sure do love that full-frame look of those Speedboostered Zeiss primes.

Sidebar: I am taking another look at Canon’s 24-70 f/4 IS lens, not because of its optics, but because of its image stabilization. In theory, with a Speedbooster it should look virtually identical to and be just as fast as my Canon EF 17-55 f/2.8, only with much better handholding because of it’s 4-stop hybrid IS. Will be experimenting with that combo on my Sony FS5 on an upcoming shoot and look forward to sharing the results.

Thoughts on 4K

I’m finally coming around to shooting my interviews in 4K. It really is nice to have the option to punch in for a tight shot pulled from a medium shot. But the drawback is focus. Focus is really, REALLY important to nail when shooting 4K. And on both the FS5’s LCD and my SmallHD 502 monitor, it’s harder to gauge focus in 4K, because the image looks a little smudged compared to HD. So it really does require careful monitoring during a shoot, especially when you do what I do, which is shoot damn near wide open all the time to visually separate the subject from their background.

I’ve shot with both the internally recorded 4K and with external recorders, and I gotta say, for the type of work I’m using it for (well lit interviews), the drawbacks of external recording outweigh the benefits of internal. Smaller file size counts for a LOT.

Many thanks to my crew – producer Sara Finkelstein, camera assistant Kollin O’Dannel (who took one for the team when he nearly fainted as the doctor described a surgery), and intern Alexandra Watkins who rolled sound.

Dedolight DLED7: the ultimate LED shadowcaster?

Dedolight DLED7: precise and versatile

The new DLED7 Turbo is an exceptionally precise and versatile light with all the Dedolight features plus optional battery operation

Lighting involves two big things: putting light where you want it, and NOT putting it where you don’t want it. To say a scene is lit means more precisely that an important part of it is lit, and an equally important part remains hidden. Placing shadows is just as important as placing highlights. And no light allows you to place shadows with less fuss than a Dedolight.

Original Dedolight

Original Dedolight

Dedolights have earned a reputation for precision in the film industry. The venerable 150-watt DLHM4-300 tungsten hot light can throw a 3200K beam exactly where you want it, with virtually no light spill, thanks to the company’s patented aspheric optics. What’s more projector attachments are available to further focus the light. With one of those, you can literally hit a nail on the head from across a room.

I’ve lusted after one of these beasts for a long time, but decided to wait until an LED version arrived, because more often than not, I’m mixing daylight on location rather than tungsten. The wait is now over, and it’s called the DLED7. (Note, Dedolight’s DLED4, which shares the same form factor,  has been available for more an a year but wasn’t powerful enough for my needs).

Dedolight DLED7 with DP2 projector

DLED7 rigged with optional DP2 projector

On paper, the DLED7 gives you everything its tungsten predecessor gave you in terms of precision while adding a number of new benefits:

  • Low power draw (90 watts) means it will run for a little over an hour at full power on a high-output v-mount battery such as the Switronix Hypercore 98WH.
  • Dimming is greatly improved, because the LEDs don’t suffer the red shift that tungsten lights do when dimmed.
  • The DLED7 is also runs much cooler, although it does require a very quiet fan. The light gets warm to the touch when operating, but never too hot to hold.
  • Color temperature is adjustable from 2700K to 7400K without gels.
  • The even beam of light produced by this instrument is gives you consistent light meter readings from nearly edge to edge. So you can use it to light a background with a single source. And, the light can be spotted from 1:20 (a typical fresnel is only 1:5) and cut very precisely with barn doors. Exciting stuff!

How does it work?

It’s a remarkable design achievement, really, that Dedo can make a bi-color light that remains fully focusable for both tungsten AND daylight. They achieve this by checkerboarding the different color LEDs at the focus point of the light behind the two aspheric lenses. I removed the lenses to have a peek at the guts:

DLED7 aspheric lens

It’s a simple matter to remove the two aspheric lenses and get a look at the LEDs themselves.

The LEDs are arranged in a checkerboard shape.

The DLED7 LEDs are arranged in a checkerboard shape. Half of the LEDs are tungsten and half are daylight (90 watts total for each)

DLED7 battery ballast

DLED7 battery ballast is rather large and heavy. It offers dimming and color temp controls and on/off switch.

One thing to note about the dimmer: it is not stepless,  changing in small increments like aperture control on a Canon. The steps are small enough to not be an issue, but it’s impossible to do a completely smooth fade up or fade down the light without visible stepping, just as smooth iris pulls are impossible with Canon still lenses.

DLED7 ballast controls

DLED7 ballast controls

The first time I switched on the light, I was surprised it wasn’t brighter. But then I realized just how bright it gets when focused. Just how bright is it?

DLED7 vs Lowel Pro Light

By way of comparison, I put the DLED7 head to head with my Lowel Pro Light, a small 200-watt light that throws a very nice beam a long way. I found that the Pro Light is about 3 times brighter when fully spotted, to the DLED7 in tungsten mode spotted equivalently (which is about half spotted). However, it’s possible to focus the DLED7 MUCH more precisely. But daylight is the great equalizer. With a dichroic filter attached to the Pro Light to bring it to 5600K, the two lights are equally bright, and with equal throw.

Wide color range

DLED7 color temperature indicator on the ballast

DLED7 color temp indicator

It’s worth noting that my DLED7 is slightly less bright at 3200K than it is at 5600K, by about 1/3 stop. The DLED7 goes even further up and down the Kelvin scale, though, advertised to go from between 2700K and 6800K. As it turns out, it actually goes quite a bit further up the Kelvin scale than that, as I discovered next.

Color accuracy scores

I tested the DLED7 with a Sekonik Spectromaster C-700R color meter, which reveals some interesting details about the quality of the light.

First test: How accurate is the ballast color temperature indicator vs. the actual measured color temp?

What I observed is that the DLED7 tends to run cooler than the ballast readout, and that the more you turn up the dimmer, the cooler the color temp gets. The only way I could hit 2700K, in fact, was to dim the light all the way down to 2 percent (the light is off at 1 percent), where it read 2733K.

Color temp set at 2700K:

Dim 2 percent: 2733K
Dim 10 percent: 2775K
Dim 25 percent: 2776K
Dim 50 percent: 2827K
Dim 75 percent: 2845K
Dim 99 percent (brightest setting): 2860K

Going to the far blue end of the scale, with my color temp set to 6800K, we find the same thing, only it’s even more blue-shifted. Here’s my readings:

Dim 2 percent: 6738K
Dim 10 percent: 6905K
Dim 25 percent: 7089
Dim 50 percent: 7395K
Dim 75 percent: 7459K
Dim 99 percent: 7422K

So where the DLED7 is furthest from its advertised color temp is when it’s at it’s bluest and brightest, at which point it’s about 600K over the mark. Definitely something to be aware of.

In practice, it’s pretty easy to adjust for this to hit 3200K. Just set the temp readout to 3100K, which put my readings within 100K of the target at all dimmer settings. For 5600K, though, things get trickier. At 25 percent dim, set to 5900K. At 50 percent, it wants to be at 5500K. At 99 percent, setting to 5300K got me real close. In practice, as long as you don’t mind being a few hundred K off your mark, you’ll be fine. But for precise color work? Carry a color meter and be ready to use it a lot.

How does the DLED7 stack up against other lights in terms of CRI rating?

I tested the CRI and color spectrum of the DLED7 at both 3200K and 5600K (as measured on meter, not by the color temp readout on the ballast, because those numbers are slightly different as outlined above). For this test, I used a Sekonic Spectromaster C-700R.

At 3200, the DLED7 scores exceptionally well. CRI: 97.3

The DLED7 measured at 3200K

The DLED7 measured at 3200K

At 5600, things slip just a little a bit. CRI: 95.3

DLED7 at 5600K color rating

DLED7 measured at 5600K

Most notable is the R9 readout in the far right column, above. The R9 shows how well the light does in reproducing saturated reds. These are important colors for skin tones. So we would ideally like to see as high a rating as possible for R9.

It turns out, though, that scoring 75.9 in R9 is quite good for an LED light. By way of comparison, my lovely bi-color LiteGear LiteMat2 has a measured CRI of 93 and a R9 score of 73.2 for daylight and only 64.4 for tungsten. And yet, it still produces fabulous results with skin tones.

To illustrate why, consider this: professional gaffers wouldn’t hesitate to place a full CTB filter over a tungsten light to match daylight. But guess what that does to the R9 score of a tungsten light? See for yourself (measured with my Lowell Pro Light and full CTB):

A tungsten light with full CTB filter shows even lower R9 and lower CRI

A tungsten light corrected to daylight with full CTB filter reveals an even lower R9 score of 68.4 with a CRI of 89.4.

Suffice to say, if the Dedolight DLED7 is posting scores that easily beat a tungsten light corrected with CTB, it’s more than good enough for film production work.

For information about how this light stacks up against other LED film lights, check out the very comprehensive light test results published by Indie Cinema Academy. As of this writing, their tests do not include ratings for the DLED7.

A few tips for getting the most from the DLED7:

Glue black felt on the  barn doors (the two larger ones, not the small adjustable ones). Because they no longer get hot, adding light-absorbing material to the surface of the doors makes them less reflective and improves their light shaping ability tremendously.

The barn doors feather the light when it is spotted, and cut it when it is flooded. For example, if you rake the light along a wall at full spot, closing the barn doors on the wall side, you can feather the light  to spread it evenly across the wall, achieving the same effect as a graduated scrim.

DLED7 as rim light

DLED7 as motivated rim light, with barn doors used to feather light as it falls off subject’s camera-right shoulder.

DLED7 as back light.

DLED7 as back light, visible as the splash of blue light on subject’s camera-right shoulder.

The power connectors on the ballast are well designed, and have plenty of length to allow you to run the light up a stand without fuss.

DLED7 battery ballast

DLED7 battery ballast

DLED7 road kit

DLED7 road kit with DP2.1 projector, DPEYESET with filters, battery, battery ballast, barn doors, and scrim bag for hanging battery on light stand.

So, here’s what it all boils down to:

CONS

  • Pricey and doesn’t include required accessories.
  • LED version isn’t as bright as its tungsten predecessor.
  • Currently it’s a special order item in the US.
  • Measured color temperature runs cooler than ballast indicators display.
  • Dimming, especially at cool end, shifts the color temperature.

PROS

  • Wide color temperature range from 2700k – 7400K (measured)
  • Very portable
  • Dimming at warm end without significant color shift
  • Battery operable for long periods
  • Controllable, even light spread
  • Exceptionally focusable

The DLED7 is very special location light that can do what no other LED light can do in terms of focused throw with virtually zero light spill. It’s very easy to place a light or a shadow exactly where you want it with this light. It excels as a background light, and as a back light, rim light or hair light. For some kinds of commercial work, such as photographing a beer label illuminated just perfectly, it’s in a class all by itself. It’s portability, battery operability, and huge color temperature range make it the swiss army knife of location lights.

But at $1700, it’s also quite expensive and doesn’t ship with required accessories (you have to add a ballast, and the one for battery use costs over $500 and the one for AC use costs more than $700). The DP 2.1 projector with 85mm lens adds another $477. DP EYESET diffusers adds another $116. Barn doors: $50. Also, the ballast color temperature readout isn’t fully accurate, and especially at the cooler end of its spectrum, the light shifts color temperature as it is dimmed.

Nevertheless, this is a light that gives you real lighting superpowers: the ability to decide exactly where to place the light and shadow in your frame, at whatever color temp you want, without a power cord. I recommend it without reservation. It’s exceedingly well made and precise, exactly what you’d expect from a Dedolight. And the premium price is something we in the film industry expect to pay for top gear. I look forward to recouping that expense over the many years of service this light is likely to give me.

Want to try it out for yourself? Rent my Dedolight road kit, which includes all the above items, via Cameralends.com. (Seattle pickup only please.)

DLED7 road kit

DLED7 kit ready to roll in Nanuk 935 case

 

Shooting raw on Sony FS5 with Atomos Shogun Flame

Atomos Shogun Flame with sun shade on Sony FS5

Atomos Shogun Flame with sun shade on Sony FS5

I have been eagerly awaiting the raw upgrade to Sony FS5 that was released last week. What does 12-bit 4K raw look like on this camera? I purchased the $600 CBKZ-FS5RIF upgrade as soon as I could, installed it, and over the weekend, had a chance to shoot a little on a music video project. Here’s my initial observations, based on recording with a Atomos Shogun Flame.

First insight: the Flame isn’t actually a raw recorder just yet. You are limited to recording 4K 10-bit ProRes or DNxHR from the Slog 2 or 3 12-bit raw signal that the camera sends. This means you don’t have the option, if you want it, to “develop” the digital negative to your liking from CDNG using raw tools like Resolve or Adobe Photoshop. Atomos says CDNG raw will be supported in a future firmware upgrade.

If you want 12-bit CDNG raw now, an Odyssey 7Q+ with the latest firmware and an FS5 will do the job. I look forward to testing that combination soon. However, from experience I know the Odyssey screen isn’t as bright as the Atomos Flame, and I love me a bright monitor. However, the Odyssey is OLED, which the Flame is not.

Atomos Shogun Flame sun shade

Atomos Shogun Flame sun shade allows touch screen to be accessed through an opening in the bottom.

So, just how bright is the Flame? The sun was shining in Seattle yesterday, so I had a chance to really work it. And I found that it’s noticeably brighter than a SmallHD 702 (1500 nits vs 1000 nits).  In direct sunlight, though, you still need the hood. That’s because the screen is highly reflective, and in glaring sunlight, you need to cut down on sources of reflection. The Atomos hood creates a box with a little window you can peer through, which effectively reduces the area of reflection to a small rectangle in the middle. I found that wearing a ball cap while operating further shades this opening, making it possible to view the screen pretty effectively even in direct sunlight. But the best option? Shade the camera and operator with a flag or umbrella.

The HDR mode is…interesting. It does look cool, visibly brighter than in rec 702 mode. But it feels a little 1.0 to me. The pixels aren’t OLED, so brightness feels like it’s coming from over cranking the LCD pixels. I’ve seen a real OLED Sony HDR monitor, and wow, it blew me away with its brightness. This monitor doesn’t even come close. It’s more like a preview of coming attractions than the real deal. But it’s a start, and I applaud Atomos for shipping it as soon as possible, because HDR is going to be amazing when it gets here and is widely supported on consumer monitors.

There’s a display mode on the Flame that tries to convert Sony SLOG2 or SLOG3 to Rec709 using Sgamma 3 or Sgamma 3.cine color modes. The problem is, it doesn’t really work for me, because I (properly) overexpose my SLOG. To visualize that, you need a LUT, and the Flame happily does support LUTs, allowing you to load up to 8 custom ones.

I found that the LUT I normally use with SLOG3 looked slightly different when applied to the raw signal of the FS5. So that will need some tweaking. But the good news is that your LUTs will work and can be applied, not only to the screen view, but you can burn them in to the recorded footage if you choose.

Slow Motion is a non-starter on this Shogun. The Flame will display a 60p signal, but the record button is disabled, meaning it does not support any slow motion frame rates above 30p. The Flame doesn’t display anything but noise when the S&Q option is enabled on the camera. You’ll need a Shogun Inferno to record at 4K 60p, in burst modes, and in the 2K slow mo mode.

There’s also no way to down-convert the raw 4K -> 2K on this recorder, which means you HAVE to record 4K, fwiw. Not a big deal since you can record 10-bit on the camera’s sd card. But it would be nice sometimes for workflow reasons to go straight to ProRes in 2K instead of having to transcode the footage first, something the Odyssey supports.

Neither SDI nor HDMI output is available when shooting raw, which means you can’t send a feed to a director monitor (at least not to the SmallHD 702 I used for testing).

Battery: The Flame chews through batteries. Bring lots of them if you plan on shooting for very long.

Sound: The Flame has a fan, but it runs nearly silently after noisy boot up.  In a hot production environment, this might be an issue.

So, how about that 4K image? Jury’s still out. I definitely noticed the same kind of noise in the shadows that you see with SLOG recorded in-camera. Apparently 10 bits isn’t enough to solve for that. I’ve been told that 12 bits isn’t enough either – that you need 16 bits before those shadows really come clean right out of the camera. So just be aware that you will still need to overexpose the same way you do with in-camera 8-bit or 10-bit SLOG. Beyond that, the benefits of shooting 4K in 10-bit are all there: you can push the file around in post more without degrading the image, you get better color information and no macro blocking and artifacting that you can see when pixel peeping your XAVC footage.

Conclusion: I wouldn’t buy this recorder, but I would happy rent it again – until the Inferno is available.

Chasing perfection with a probe lens for beer promo

A few months ago I had an opportunity to shoot some ingredients for a Sierra Nevada beer video using a probe lens. The results of the work are now public, and I’d like to share the process we used to make a few of the images with this unconventional optic.

Director Mark Bashore had most of the road trip footage for this spot already in the can, but he wanted some super sexy shots of the ingredients that go into the beer, and a beauty shot of the beer itself. So he convinced his wife that it would be a good idea to set up a studio in their living room for a few days and go crazy with lights, track, and cameras. Fun stuff!

Mark wanted shots that literally felt like flying through the ingredients, and he had the idea to use a probe lens. I’d never heard of a probe lens before. Turns out it’s a specialty lens, designed for special effects and tabletop photography. It’s basically a long telescope-like snout that optically allows the lens to sit on the tip, apart from the camera. This allows for otherwise physically impossible shots, such as driving the lens through openings that are too narrow for a camera body. Incidentally, this type of lens was used to shoot many of the special effects in the original Star Wars.

probe2pluspic3I went searching for a rental house that carried such a beast, and after researching the options, I picked Innovision Optics, an LA based company that helped pioneer the development of probe lenses in the 80s. They sent us their Probe II + package, which covers Super 35mm sensor cameras and comes in PL mount. I hooked it up with my Sony FS5 camera with PL adapter, and everything tested out great.

The kit comes with a full set of miniature lenses, that vary in focal length from 9mm up to 55mm. In testing I found you really want to be shooting wide when you’re doing tabletop shooting, because it’s the wide focal lengths that give you that immersive experience with the subject. I initially wasn’t overly confident that the tiny lenses could produce great results, but I was very impressed by what I was seeing on the monitor.

One of the nice features of separating the lens from the camera is that it allows you to get the lens dirty. The snout is designed to be waterproof (according to Innovision), so splashes of ingredients on the lens and lens housing could easily be wiped off between takes, while the camera stayed dry. However, I don’t think you’d want to fully immerse it in water.

I rigged the lens with a DJI Focus wireless follow focus, to allow focusing the lens during pushes without touching the rig. The prevented unwanted camera shake, and worked flawlessly. But one note of warning: if you’re renting the DJI Focus, be sure to allow plenty of time to familiarize yourself with the unit beforehand. It’s not immediately intuitive how to calibrate it and get it responding to pulls just the way you want.

So let’s take a look at a few of the shots:

Grapefruit being ripped apart while lens pushes through the middle

Grapefruit being ripped apart while probe lens pushes through the middle

Mmm. Isn’t that a mouthwatering shot? We had to do a lot of takes to get just the right energy, and we went through a lot of grapefruit in the process. The lens was seriously sticky by the time Mark moved on to the next shot. I placed a Arri 650 as the backlight (which you can almost see providing the rim light) and filled with a LiteGear LiteMat2 at 3200K. Notice the lovely flare the probe lens produces, and the tiny details. The grapefruit, from this perspective, becomes a monumental presence.

Grapefruit push in shot

Grapefruit push in shot in 4K

For the pink grapefruit on pink background, I rigged the camera on an arm and rolled on it in 4K with the FS5. It’s lit from one side by an Arri 650 bounced into a 4×4 foam core, with bounce on the opposite side. The push was done in post. That 8-bit 4K works great as long as you expose correctly!

composited grapefruit with neon sign

Composited grapefruit with neon sign

For the grapefruit in this composited shot, we used my Rhino Motion slider as a dolly for the grapefruit. It was easier to control the motion by moving the ingredients for shots like this than it was to move the camera. To create the platform, I mounted a Matthews laptop table on the Rhino slider’s 3/8″ spud using a ball head.

The neon sign is real, and was made specifically for the shoot. We shot it against black foam core. The compositing was a little painful for this, requiring each frame to be individually selected in Photoshop to clean up edge artifacts.

Beer shot in an aquarium

Beer shot in an aquarium

Beer poured into an aquarium

Beer poured into an aquarium

For the beer shots, we poured a lot of beer into an aquarium, and tried lighting it different ways. Most of these shots are played unnaturally – i.e., backwards, upside down, or sideways, which supports the larger than life vibe of the piece.

We shot most of these at 240 frames per second with the Sony FS5. I used CineGamma 3 for most of these shots, because we were pushing the limits with the light. I wanted as much depth of field as possible, so I was stopping down here to like f/11 and it was everything my pair of Arri 650s could do to push enough light. We were using the lights without any diffusion, just blasting at the tank from above, below and beside at different times.

slow push into glass of beer

The Money Shot: a slow push into carefully lit glass of beer

We spent a LOT of time trying to get the lighting right for this shot. I had the camera on a Dana Dolly, with I think a 35mm Zeiss Contax prime lens on IV Metabones adapter. I was shooting at about f/11 or f/16 to get as much depth on the sign as possible. To get the lovely rim light down the side of each glass, I used a pair of Kino Flo single-lamp 4′ fixtures, which were very carefully flagged to be just out of frame on the right and left. It has to be almost visible to get that rim light. To get the beer to glow in the middle was the hardest part of all. It required backlighting, of course, but every angle we tried wasn’t working. The light was always hitting the black plastic counter and blowing it up. After about an hour of trying things, it hit me: cut a custom flag for the backlight, basically a notch in a piece of black foam core. Then the light would fall only on the glass and not on the counter. It worked.

Tabletop photography is an exercise in patience and repetition. It’s tedious as hell, and requires more light stands than I think I’ve ever used for any other project! Every imperfection becomes visible, and has to be taken care of. So tons of flagging, and very specific lighting throughout the frame. But the results can be very, well, refreshing! Luckily we had lots of Otra Vez beer on hand during the shoot, and no, we weren’t just shooting it!

Have you ever shot with a probe lens? How did you get the best results?

One way to modify Contax Zeiss lenses to work with Metabones Speedbooster

dremel-lens-1

One of the things I’ve noticed now that I’m doing the bulk of my shooting on super 35 sensor cameras instead of a full frame DSLR is that my set of Contax Zeiss primes just aren’t getting used that much any more. Why? The lenses, most of which are f/2.8, just don’t look as good when adapted with the crop factor of f/1.5 using the Metabones IV adapter. So, why not just just use them with a Speedbooster?

Answer: because of protruding rear elements (see below) that prevent them from fitting on the Speedbooster.

dremel-lens-0

It’s really quite extraordinary the difference in look that a full frame vs. a super 35 sensor makes with this glass. Using the IV adapter, my 35mm f/2.8 Zeiss becomes about a 52mm lens in terms of field of view. Sometimes the change in field of view is fine – when I want a longer focal length. But another thing happens that’s rarely desirable: what was f/2.8 on full frame now has the depth of field appearance of f/4. And the difference between apparent f/4 and f/2.8 on full frame? Huge.

Over the weekend, it occurred to me that there might be a way around this impasse. Why not get a Dremel tool and grind those nasty rear elements out of the way? A close examination revealed that the elements are useless for video work anyway. I will never be using these lenses, which are permanently cine modified already, for shooting stills on a Contax camera. So I ran across the street to my hardware store and got busy.

dremel-lens-2

First thing was learning about griding wheels – turns out the one you want for this job is the aluminum oxide grinders that are designed for metal work. Pictured above is the B132 grinding stone, which worked great.

To prevent fine metal shavings from falling into the lens, I carefully used painters tape to mask off the rear elements of the lens and block the openings into the lens. Then I made sure my glasses were on, and got busy.

dremel-lens-7

The metal pretty much just melts away with carefully applied pressure.

dremel-lens-4

Bits of metal become flattened and hang over the sides of the real element, which I was able to scrape away with a small screwdriver. I also used a vacuum cleaner to suck out any stray metal dust.

dremel-lens-5

Pulling away the tape reveals the flattened elements.

dremel-lens-6

ground-down-side-view

Now the moment of truth: It fits! I ground down all 5 of my lenses in my set (25mm f/2.8, 35mm f/2.8, 50mm f/1.4, 85mm f/1.4, 135mm f/2.8).

zeiss-iris

zeiss-only-wide

I tried all 5 of the lenses on, and all fit, but I noticed the 35mm in particular was very tight. This and the 135mm were the most difficult to work on, as they were very tight when attached to the adapter. But they seemed to fit OK after repeated grinding sessions. Woo hoo! I’ve got a set of dreamy Zeiss glass with the full frame look on my FS5!

contacts-screwed-up

But before you get as excited as I did, let me tell you the rest of the story. When I put one of my Canon L lenses on the Speedbooster, everything seemed fine … except the image stabilization didn’t work. Crap. What did I mess up…

A close look revealed the despite my best efforts at grinding, something had remained on at least one of the lenses that tangled with the last contact in the row. For the Zeiss glass, it doesn’t matter – it’s all manual anyway. But for the Canon lenses, you need every one of those contacts in working order. Sigh.

dremel-lens-10

I’m still not sure what messed that up – but it obviously did. So my solution has to make this Speedbooster my dedicated  “Zeiss only” adapter.

zeiss-only

It now lives in the same Pelican case as the Zeiss Contax set. I had to spring for another one to use with my Canon L glass. If were going to do this over again, I’d start by purchasing one of the older Speedboosters (you can find them on Ebay for a few hundred bucks cheaper than the new Ultra Speedbooster that I used), and plan from the beginning to dedicate a Speedbooster to the Zeiss glass, instead of thinking I could continue to use the same one with L glass.

dremel-lens-11

All in all, though, I’m thrilled to have the full frame look back with my Zeiss glass on Super 35 and am looking forward to shooting my next project with this glass. Here’s why:

25mm f/2.8 Zeiss adapted to Sony FS5 with Metabones IV adapter (1.5 crop factor makes approximate field of view of 37mm):

Screen Shot 2016-05-24 at 11.29.14 AM

35mm f/2.8 Zeiss (after grinding) adapted to Sony FS5 with Metabones Speedbooster (.071 crop factor makes approximate field of view 37mm):

Screen Shot 2016-05-24 at 11.29.27 AM

Notice how much more selective the focus is on the orange umbrella in the second frame. If you’re not noticing the difference, click on each image to view at full resolution.

PS. Since this post went live, I’ve learned that Metabones makes a Contax CY -> Sony E mount Ultra Speedbooster. Duh! I would have been much better off just buying that, saving the contacts on my EF Speedbooster, and leaving well enough alone. But then I wouldn’t have had anything to blog about, would I?

 

 

Sony FS5 firmware 2.0 is out. And the Auto-ND is killer.

Sony has been pretty sneaky in their release of firmware version 2.0 for the FS5. There’s been no big public announcement that I’m aware of, aside from this brief post in Alister Chapman’s blog. Maybe that’s because it isn’t fully ready to roll out – for example, there’s currently no way to purchase the upgrade to the raw-output option contained in the 2.0 firmware. So what’s the big deal?

Well, for me, the auto-nd feature is pretty epic. It’s the first time I’ve gotten really excited about an auto feature on a camera since Canon released dual-pixel autofocus. Auto-nd on the Sony FS5 is like having a pair of sunglasses that automatically get darker or lighter, almost instantly, in response to brightness. You can now do things like go from indoors to outdoors seamlessly, without affecting ISO, shutter or iris. This is huge: a creative tool DPs have never had before.

Sony Catalyst Browse displays GPS info

Sony Catalyst Browse displays GPS info contained in video clip metadata.

Also cool: The GPS feature, which is now enabled by default. What’s it useful for? I’m not sure yet. None of the GPS info is showing up in FCPX. But it does appear in Catalyst Browse, the free app from Sony for logging and transcoding footage. I look forward to finding useful ways to use this info. Chapman has suggested that one possible use will be the universal time that each clip is tagged with. However, it does not appear to be frame accurate – the frame field is empty on all clips.

Note: The GPS feature seems to be quite unreliable, with the signal coming and going and most of the time I’m seeing “No GPS” on the screen. I know that there is GPS in my area, because my cell phone is working with it fine. So the receiver on the camera seems pretty weak.

This 2.0 firmware has me really excited about shooting raw. I can see one really great use for it: shooting interviews in 4K, to enable getting two shots (medium and tight) frame same camera.  I can’t wait to get my hands on the raw upgrade as soon as it’s available.

Why Inclusion Matters

When the Seattle School District cut funding for a beloved special education program, teachers and families were shocked. Then, they rolled up their sleeves and started a campaign to save it. This is the story of why inclusion matters at the University of Washington’s Experimental Education Unit.

I produce a video every year to support the EEU’s annual fundraiser. This video is screened at the beginning of the event, to help set the stage for fundraising. In other words, to prepare participants to open their wallets a little deeper. So the point of these pieces is to tell a human story. This year, we had a potentially awful story to tell, but one that has a very happy ending, thanks to the efforts of teachers and parents.

Technical notes: Traveling shots made with DJI Osmo; interviews and b-roll Sony FS5 in SLOG3. Interview mic: Audix SCX1-HC.

Is DJI Osmo more than a toy?

My first step on the journey from still photographer to cinematographer was to simply grab a camera and start shooting. I quickly learned to put it on a tripod instead. Or a shoulder rig. Or a slider. More recently, a Movi. But the reality of gimbal-stabilized shooting is that it can end up looking like this:

bigrig

Hardly freeing! So when DJI announced the Osmo late last year, with the promise of a stable camera that fits in your palm, I was fully intrigued.

Osmo is priced aggressively enough that I didn’t have to think about it too long before deciding to buy. Over the Christmas break I had a chance to do a little shooting with it, to answer the question: is Osmo more than a toy? Could the footage it produces work on a real documentary production?

DJI-OSMO-01-650x489

The first thing I noticed out of the box is that Osmo is cute. Sexy, even.  The sleek black device invites holding and feels great in your hand. If it were white, I would swear it was an Apple product. But then I turned it on, and the next thing I noticed was that the horizon wasn’t level. And it kept crashing when I touched the screen in the wrong place. Definitely not an Apple product!

Based on my early observations, it’s clear that DJI released Osmo before the firmware was fully ready. DJI embodies a trend in which camera manufacturers use early adopters as beta testers for firmware that is still in development. But within a few weeks, and just in time for Christmas, DJI released new firmware, and most of the problems I observed are much improved, if not fully corrected.

The Osmo’s lens is fixed-focus, like a GoPro, which keeps everything more than 1 meter away in focus. Everything close than that falls out of focus. A meter is almost 40 inches, and to get the camera that far away, you’ll need to use the optional DJI selfie stick (currently in short supply) to get sharp images of yourself. I’ve ordered one and look forward to the option of getting the camera up high (drone like) for close aerial perspectives with the Osmo.

selfiesoft

Selfie mode at arms length

Shooting in selfie mode produces a slightly back-focused subject in frame. Click the image above to view at full resolution, which reveals the full extent of the problem (the downsampled image as rendered above disguises this).

Another challenge: the Osmo with X3 doesn’t do closeups. It doesn’t even do medium closeups. Remember, the Osmo field of view is the equivalent of a 20mm lens on a full frame camera. That’s wide!

tooclosecamera

Distortion and soft focus ensues when you get the Osmo closer than 1 meter

Moving Osmo too close both distorts your subject and throws it slightly out of focus. The slight defocusing is made all the more apparent because whatever is behind it will be perfectly sharp.   So, with Osmo, I have to fight my instinct to get the camera close, especially when shooting people.

The closest thing to a workaround (until the X5 camera for Osmo is released) is to shoot in 4k, and leave lots of room around the subject, then crop into the shot in post. But I find it’s best to just keep people a meter or more away. That’s just how this camera was designed to be used.

Slow motion

In 1080p, you have some nice framerate choices for conforming to 24p slow motion: 30p, 48p and 60p. You can also engage the slow motion mode which shoots at 120p in windowed mode.

The best quality slow motion results happen from shooting no more than 60fps. Anything more than that, and the image suffers from significant noise. To my eye it looks like 720p that’s been scaled up: a little soft, and very noisy. Still, with some denoising, you can get some usable results. The field of view when shooting in 120fps is much narrower due to the windowing of the sensor.

Battery life

You’ve got two batteries to stay on top of with Osmo: one in the handle that powers the camera, and one in your smart phone powering your screen. I find the Osmo batteries last about 30-40 minutes in constant use. So you’ll need a handful of them to get much shooting done.

iPhone battery life is greatly reduced when shooting with Osmo. During an afternoon of shooting, I was able to keep going by plugging my iPhone into a portable battery quick-charger. It’s easy to operate the Osmo with a charging cable running to your pocket. Your smart phone battery life may vary, but if you plan to shoot for more than 45 minutes or so I recommend something like this.

4K

I was unable to get the camera to record in cinema 4K at the DCI standard of 4096 x 2160. I thought at first that my card wasn’t fast enough, during the time I was writing this post, DJI released firmware version 1.4.1.8, and this resolved the issue (and many others). All 4K modes work perfectly now!

Selecting 4096 x 2160 changes the aspect ratio into a slighter wider field of view than 16×9 (approximately 1.9:1) and sets shutter to 24p. The more common UHD-1 (16×9 aspect ratio) 4k standard is supported in this camera in frame rates of 30p and 24p.

Screen Freezing issue

For some reason, the DJI Go app freezes occasionally when I tough the screen anywhere that isn’t a menu. This is definitely a bug, because the screen is designed to allow setting a spot-meter by touching the screen when in auto exposure modes. The v. 1.2.1.8 firmware update does not fix this problem. UPDATE: This issue has been fixed with the latest firmware update.

Manual controls

One of my favorite things about the Osmo is that it gives you full manual control over exposure. In practice, this means you will be able to get cinematic results from this camera that GoPro users can only dream about. For example, shooting in dark forested areas and emerging into more brightly lit areas, I was able to set the exposure to be correct on the brightest areas and allow the rest to go dark. Great for reveal shots. Trees and foliage tends to overexpose when set to auto, and going manual resolves this. The controls are easily accessible by touch, and unlike many other cameras, you can punch the screen while rolling without disturbing the shot because the brushless gimbals are doing their thing. So it works very well.

Another nice touch is the option to set shutter priority – which allows locking in a cinematic framerate of 50p, while letting the Osmo set aperture and ISO.

ND filtration

Although it doesn’t come with any ND, Neutral density filters can easily be screwed onto the Osmo’s threaded filter (it ships with a clear filter). The Osmo’s X3 camera uses the same filters as the Inspire 1. Luckily, there are quite a few options for ND filtration available for Inspire 1. I purchased a set of 3 Renaat ND filters for US $69. These filters weigh just 3 grams apiece, and won’t unbalance the gimbal. To purchase a set of your own, send an inquiry email to ceulenaere@gmail.com.

For overcast conditions, I found the 1/8 ND works great. For more brightly lit conditions and direct sun, the 1/16 and 1/32 filters will give you access to those cinematic shutter speeds. At 1/324, IR pollution becomes a bit of an issue, but I prefer to deal with a little color correct than to live with staccato shutter effect that results from high shutter speeds.

D-log

Osmo gives you a whole bunch of picture profiles to choose from. The most interesting one is d-log, which gives you a flatter image, holding onto highlight detail in post. I find it’s  more like Canon’s C-log than it is like Sony’s S-log: less dynamic range, but easier to grade. It feels to me like d-log gives you a ballpark of around 10 stops of dynamic range. Exceptionally good for this little camera!

In my original testing, d-log had a noticeable red shift, which had to be corrected in post. But this seems to have been corrected in firmware updates and I’m no longer seeing this issue.

What it’s really good at

Tracking shots! The Osmo excels when you’re in motion filming something that’s also in motion. For big establishing shots where you want everything in frame to be in focus, it’s killer. Think of it as a drone that you can hold in your hand, and shoot accordingly. So, passing between things or through things, big sweeping reveals, that sort of thing.

I’ve used the Osmo for one commercial project, and found it to be an incredible tool for getting shots I wouldn’t have been able to get any other way. In the video below, all of the traveling shots were shot with Osmo, and intercut with Sony FS5 footage. The cardboard walls in the space were 8′ tall, and I wanted a way to see over the top of the walls while moving, to reveal the participants interactions. I put the Osmo on a painter’s pole, which allowed me to extend it about 6 feet. This allowed me to get some dramatic crane-up reveal shots, like the shot that opens with the team huddled over an iPhone screen, booming up to reveal the view over the walls. Pretty sweet, and a shot I couldn’t have dreamed of getting with a larger camera rig in the cramped space.

What it’s not so good at

It doesn’t do closer than 1 meter, or selective focus.

Conclusion

The Osmo X3 is an exceptionally useful tool right now. For wide sweeping shots with movement, it’s killer. But the need to keep the camera at least a meter away from your subject limits it’s storytelling capabilities. The small sensor means everything in the frame is in focus.

When the X5 for Osmo is released, things will get a lot more interesting. Footage should then intercut seamlessly with higher quality cameras, and focus will be controllable, opening up a world of possibility.

But I’m not waiting for that. I’m carrying Osmo today on many of my commercial projects, looking to bag those special wide shots that I otherwise wouldn’t have time or budget for. The Osmo is the gimbal you can always have with you. And that makes it a winner with a very bright future.

C300mkii cfast file corruption issue – and one tedious solution

meNo shit. There I was. At the end of a long day of shooting for a commercial client with a rented C300mkii, and I get this error message: “Buffer overflow.” I tried to dismiss the error but the camera’s OS had frozen. I had to pull the battery to cycle power and restart. I hit record and same error. Additionally, another error popped up with something like “some files need to be recovered.” WTF?

My rented 256gig Lexar 3400 Cfast card was 6 minutes away from being full in slot A. I had a fresh card (same brand and capacity) in slot B, and my first thought was to switch cards. I did that, and everything was happy. I finished the shoot without further incident.

But when I inserted card A into the reader that evening, much to my surprise, my Mac froze and had to be hard-rebooted. Damn! Now I was getting nervous. This was a big project for a client who had flown to Portland from Austin, at considerable expense. I started to imagine the extremely uncomfortable conversation I was going to have to have with him.

“Um, you know that second day of shooting we did? The one where you kept telling me the shots looked so much better than our first day of shooting? Yeah, that extra day of shooting that you didn’t plan to pay for but did because I convinced you it would be worth it?”

I did some Googling, and discovered that the only Cfast cards officially supported by Canon for the C300mkii are all SanDisk cards, in capacities only up to 128GB.

SanDiskCampatability

 

I was extremely disappointed to find discover this fact, because my rental house, Lensrentals.com, had listed the Lexar drive as “works well with” the c300mkii. But who knows whether the Cfast card error thrown by the camera was related to the media anyway? Maybe it was something else. But I didn’t want to point fingers: I wanted to find a solution.

I put the bad card back into the camera, switched into media review mode, and held my breath as I tried to open the files. And guess what? I could open them! Big sigh of relief. If the data was readable on the camera, that meant the files would be recoverable. But how?

lexar3400After several attempts to do disk first aid and data recovery on the card failed with the same OS freeze as before, I got another idea. Would it be possible to copy, one at a time, the files from the Cfast card onto the camera’s SD card, which normally records proxy media? Nope. No can do. There isn’t even an option for that. But as I was exploring the C300mkii’s file options, I discovered that it IS possible to copy files using the camera menu from Cfast card A to B. I took another deep breath and gave it a try and…bingo! The file copied and the copied card was readable on my Mac.

It took me about half a day to painstakingly go through all 128 files and copy them over, but it was the best half day I’ve ever spent.

I alerted the rental house to the potential card compatibility issue, and they said they’d investigate. I still have no clear optic on what caused this data drama. But I do know a couple of things. The first one is that a problem like this has never happened to me while shooting with my Sony FS5. The second is that the next time I rent a C300mkii, you can be damn sure I’ll be sticking with the cards that Canon has officially supported.

UPDATE: After investigating, the fine folks at Lensrentals.com told me “it looks like the card was defective” and issued me a refund for the card rental.  “We haven’t had any higher issues with these cards/cameras than any other. I wouldn’t hesitate to use them again – just perhaps with a backup or two.”

New king of the ring: Cool-Lux gears

cool-lux-gears

I’m always surprised when I see documentary shooters working without a follow focus unit. To me, the precision and added stability are essential. But plenty of documentary shooters roll with Canon zooms, and gearing those lenses has until recently required making an awkward compromise (i.e., Red Rock gears) or an expensive permanent choice (Duclos mod). I’ve written about one ring to rule them all, but those hard-plastic gears are really for primes, and won’t fit on my favorite documentary zoom, the Canon EF-S 17-55mm f/2.8 (above right).

Enter Cool-Lux Lux gears. These gears are unique in my experience in that they are made of a firm yet flexible silicone, which has just enough stretch to slip on over protruding zoom lens elements. Yet when finally worked into place, they stay put. Even with constant ins and outs from a camera bag, they don’t budge and have the look and feel of a permanent solution. Plus, they can still be removed when desired.

peel-on

The 76-77mm gear slides over the protruding stabilization switch on the 17-55mm f/2.8 Canon EFS zoom

slide-on

After I ordered one of the gears, and raved about it on Facebook, Cool Lux’s product manager Patrick Fee dropped me a thank you. Very kind of him. So I seized the opportunity to ask him a few questions about the company and how the gears are made.

Q: What kinds of gear do you make, who do you make it for, and how long have you been doing this? 

Cool-Lux started off as a lighting company way back in 1977 specializing in on-camera lights for video professionals, the type of thing you would see on the shoe mount of an ENG camera to add some fill light or get an exposure in the dark.  Advances in LED technology has since led to Cool-Lux’s current line of Pro LED Panel Lights.

In fall of 2013 I was hired on to focus on making innovative new camera rig and accessory products for cinema video professionals like yourself.  In that time, I designed a completely new shoulder mount camera rig system for Cool-Lux based on the idea of going from a tripod to the shooter’s shoulder with the press of a button.  It’s actually a really cool system with a flip out action and chest support aimed at rental houses and production companies that use a lot of different cameras because it works on everything. I also designed some very simple but effective lens gears called Lux Gears that are just now starting to take off.

Q: The market for gears seems pretty crowded. What inspired you to jump in? 

You’re right in saying the market for lens gears is pretty crowded but everything out there was just a copy of the same old terrible design.  I always hated the one size fits all type.  Yeah they are easy for a distributor to stock but the damn buckle is not only hideous to look at, it also prevents the product from doing what its intended to do, especially on lenses that don’t have built in stops.  The only other option was to permanently fix a gear to a lens, which is fine if you’re only going to use it in a cinema type setting, or have something custom made to your specific lens that could, in theory, be removable and still give you a secure 360° rotation.  The Lux Gears filled that gap by providing a simple range of 16 follow focus lens gears that look professional and perform like a very secure permanent gear but can also be easily removed in seconds without having to get out the tools.  There is nothing else like them.

Q: The striking thing about these gears is how they stretch over the lens to fit perfectly and stay put. How are the gears made?

Lux Gears are a precision molded product.  We use a high durometer silicone material to get the elasticity needed to stretch over the lens barrel but also have enough rigidity to hold up to the amount of torque needed to turn the lens with a manual follow focus or motor driven gear.  Most people might think that a flexible gear would lead to a lot of slop or backlash when it comes to dialing a follow focus back and forth.  I actually found the opposite to be true.  Consider that with two rigid gears the engagement has to be perfect.  Either there is space between the teeth which leads directly to “slop” between the gears or the teeth are too tightly engaged which leads to a grinding of the gears.  With the high durometer silicone the engagement can be slightly too tight and the gears will not grind because of their slight flexibility.  They provide more margin of error when engaging the driving gear which is a benefit I didn’t see coming.

Q: Where are these gears made? 

We partnered with a local Chicago company that specializes in elastomers to help with the product development and make the production molds and gears for us.

So there you have it. Cool-Lux gears are the clear winner in the quest to make the perfect follow focus gear. They run about $28 each. Learn more at http://www.cool-lux.com.