Category: FocusPulling Original Video

16 February 2021

Canon EOS R5 8k Camera: Review / Field Tests / Menu Guide

Written by Paul Moon

Somehow it’s started feeling like we live in an 8k world now. I thought 6k was pushing it when I showed off the newest Blackmagic Pocket in September 2019, but things began to settle into 8k about a year after that. Sharp never delivered on their Micro Four Thirds prototype, but Android phones began shooting 8k video, and then this Canon EOS R5 dropped. Dropped is a good word for it, because the release got dogged by scarcity, silent recalls, and suspicious outrage by that usual tribe of sponsored bloggers who never quite seem to have any public body of creative work to show for themselves. Years prior, some had argued hard for Sony’s overheating 4k cameras. Others pontificated that no “professional” would ever want to run a movie camera for half an hour anyway — so when the R5 couldn’t make it past ten minutes or so, Canon got canceled at the gate. Folks also freaked at the R5’s price tag at just under four grand. But fast forward to a couple weeks ago, when Sony had played their usual long game, getting their bought-off army behind a new “flagship” 8k Alpha One, at almost double the cost of the R5.

So am I late to the party? Well, maybe we weren’t ready for the R5 when it arrived last year — and then it wasn’t ready for us either. But just like the “old” 4k revolution, it took that second and third camera to shake us into realizing: damn, I’m gonna start needing more hard drive space. The R5 is finally shipping (in limited quantities but faster than ever), and it doesn’t overheat any significantly worse than Sony’s flagship at 8k. By the end of last year, Canon changed how it regulates overheating, and the operating times dramatically improved. So in most ways actually, it’s the better camera, for just over half the cost. And here (tracking the video embedded here), I won’t only go down its list of specs, like some kind of advertisement. Yet it can’t hurt to quickly brush up on the features with some commentary:

One of the first things you see that sets the R5 apart, is its top-panel display — a Canon tradition, and something that Sony and other cameras skimp on. It helps to have more at-a-glance options. I also really like the placement of the record button, and the center-press MODE button on the dial.
The ON/OFF button has a nice, solid, firm click to it that can’t accidentally go one way or the other.
The back panel is what you typically get on Canons. The only thing I hated was the way you have to simultaneously press two distant buttons, just to switch into video mode.

Something that the Sony 8k camera lacks, is a fully articulating screen. This one flips out, and can pivot up and down, off to the side, avoiding glare. And while you can leave it pointing backward, you can also face it forward which might come in handy. People yammer that it’s only for Instagrammers who wouldn’t buy an 8k camera anyway, but that’s a dumb argument because having the additional option doesn’t hurt at all.
The ports are just as you expect: simple headphone and microphone ports, USB-C and micro-HDMI (sadly not full-sized), along with flash photography stuff that I’ll never use because…well, it’s the 21st Century?
You get a CFexpress slot alongside an SD card slot, the latter and much cheaper of which can handle most of the shooting modes, which I’ll get to later.
The battery is a clever upgrade to the classic LP-E6, adding an “NH” suffix for more capacity but keeping the same form factor so that you can use older batteries in the R5 too. Just one of these new battery types lasted me this entire day’s shoot.

One last thing to show you is inside the RF lens mount. True to Canon tradition, the sensor stays shuttered for protection while the camera’s off. This is another great feature that Sony and others could have easily enabled on their cameras prior to the new 8k Alpha One, but hadn’t for mysterious reasons. Maybe it’s part of that whole game they play for enticing you to buy fake upgrades…

Now that we’ve skimmed the camera, we could ask, “why bother with 8k?” But the density of Manhattan’s skyline answers right back! I shot the video embedded here from the R5 in DCI-aspect 8k at 29.97 frames per second in 10-bit 4:2:2 compressed, and you can watch and pause at this resolution, if you choose it from the YouTube gear icon. You’ll still be watching on a 4k monitor or probably less, but one of the main reasons to shoot 8k is for latitude in post. This hand-held excursion I took, walking from my new downtown studio, across the Manhattan Bridge, then back across the Brooklyn Bridge, is a great example why. Without a tripod, I can use the extra resolution to apply stabilization without loss in post, like Adobe Warp. I can also adjust my horizon by punching in a little and rotating, since it’s tough to judge in the field with so many lines of convergence between bridges and shores and skyscrapers. And of course, I can punch in to simulate a closer focal length to begin with, with less perceptible loss.

Speaking of focal lengths, this entire shoot used the stock 24-105mm continuous f/4 RF-mount zoom lens, which performs surprisingly tack-sharp, and adds in-lens image stabilization. That’s important, because it’s always better than sensor stabilization, something Sony refuses to admit as it keeps skimping out on lenses. And when you combine this stock lens’ stabilization with the R5’s best-in-class sensor stabilization, it beats the Sony Alpha cameras by a long shot (or at least a few stops).

But there are some things that stabilization can’t fix, and one of the big ones is rolling shutter. Since this is a full-frame sensor, that can be a problem. Above you can see a freeze frame from the train that shares space on the Manhattan Bridge. I find this result to be better than full-frame Sony Alpha cameras up to the a7S III, and on par with the Blackmagic Pocket’s smaller sensor readouts. It’s not something you’d notice in typical usage.

Another reason Manhattan is a great test subject, is all the moire patterns from fences and lines and grids that push sensors and compression to the limits. The R5 does have a low-pass filter and it works; likewise the compressed video format holds together well. Speaking of which, I’m using a UHS-II SD card at a V90 speed rating, which let me shoot internally on the cheap, at the 680 megabits per second bitrate you’re seeing here. Between frames, it’s compressed at IPB, rather than ALL-I, but I found the difference imperceptible.

I didn’t have a CFexpress card handy, so I didn’t test RAW format, but I’m saving my RAW shoots for my Blackmagic cameras, since BRAW has game-changing efficiency. Even so, the internal RAW recording capability on the R5 gives it a major leg up over the new Sony Alpha One, which is limited with severe H.265 compression. And you don’t want to add an external recorder that magnifies the Sony’s weight and cost. Meanwhile, internal recording on Sony’s 8k competitor is 10-bit 4:2:0, compared to the R5’s 10-bit 4:2:2.

All of these specs relate to dynamic range, so an outdoor shooting excursion is also an ideal way to test the sensor’s limits. There’s never a choice but to only shoot in LOG, if we do ignore grumpy old-skool videographers (and Sony’s ballyhooed S-Cinetone is a distraction from that principle). But Canon has left out its latest-generation color profiles, so we’re stuck with using what could be called “C-LOG 1.” (Oddly, Canon’s official downloads for the R5 include C-LOG2 and C-LOG3 but not the original LUT; you can get the proper conversion at this Canon Japan link for now.) It’s still better than the S-Logs and V-Logs, and Canon’s color rendering is simply better than Sony’s, and we’ve been living with that fact for a very long time. (UPDATE on March 29, 2021: Canon has now added C-LOG2 and C-LOG3 in a firmware update.)

But when using C-LOG 1 to match the sensor’s dynamic range (limited because of the resolution density), it’s decent but not exceptional. If looking into the sun, the shadows suffer (as seen in the above 8k still frame). Or I can expose for the shadows, but I lose details in the highlights. At dusk, the R5 performs reasonably well in lower light — it’s a full-frame sensor, after all. Once the sun’s fully set, I see very low noise at f/4 while still obeying always the 180-degree shutter rule at 1/60 for 30 frames per second.

In the remainder of the associated video, I try out another location for testing auto-focus and audio, then wrap up by going through each of the camera’s menu pages, with commentary. As the R5 continues living on, there is a user group dedicated to it, with frequent news updates, and promotion of your work if you end up using the R5. Please follow and share at: facebook.com/r5user & twitter.com/r5user & reddit.com/r/CanonR5, while properly tagging your videos and adding them to the Vimeo Group via vimeo.com/groups/r5user.

12 October 2019

Comica WM300A Dual Wireless Microphone Kit: Review / Audio Test / Feature Guide

Written by Paul Moon

This is a quick review, and associated video, of a dual-transmitter wireless microphone kit by Comica, the WM300A. Besides its affordability, what this kit brings to the table is the ability to feed two transmitter packs, into one receiver. My point of comparison is the RodeLink system I’ve been using for about a year, which has worked out well for me because it offers the option of hooking straight into a shotgun microphone with phantom power. But it’s worth mentioning that you can get a similar module from Comica for this system, too, that can simultaneously transmit with one of the lavalier packs. You’ll notice looking at the Rode kit that there are no external antennas, partly because of its high 2.4 GHz frequency band, which limits its range a little, and it’s vulnerable to interference from Wi-Fi devices that share the same unlicensed spectrum, while preferring walls to bounce off. In this post, we’ll explore all the features of the Comica system, which includes that second transmitter sending at the same time, which isn’t possible using the RodeLink.

Starting with the receiver seen at right, besides twin antennas, you’ll also notice an IR or infrared port that’s one way to synchronize the transmitters with the receiver. The 3.5mm output jack is stereo, which is especially important here because you can make it allocate one transmitter to the left channel, and the other to the right.

The transmitter also has a micro-USB port, for charging its internal lithium battery. Unfortunately there’s no backup alternative to get power from alkaline batteries, but you can always try tethering a USB power bank.

In the picture below, you can see both of the transmitters, with one included lavalier microphone plugged in and secured with a screw-in connector. The other included transmitter is identical, with its own lav mic, and just like the receiver, they both have internal batteries charged through a micro-USB port on the bottom.

I like how there’s not only the usual belt clip on the back of each transmitter, but also a 1/4″-20 mounting screw socket, which is something you normally only see on the receiver side. The listed frequency range of 520-580 MHz tells you that this system uses older UHF frequencies, which have pros and cons compared to the RodeLink’s 2.4 GHz spectrum. The signal travels farther, but the frequency hopping to avoid interference, is less sophisticated.

We see this in the transmitter menus, with the first option to select between Group A at the lower end of 500 MHz, and Group B at the upper end. But also, there are channels within those ranges, and you can either manually select them, or automatically sync them using that IR port. Really the goal is to avoid interference if you literally hear it, whereas that sophisticated RodeLink system listens thousands of times per second and hops around to avoid interference before you hear it. So basically, this is another a case of pros and cons.

In the transmitter menus, you can specify the impedance at the audio input, between microphone and line level. You can also decide whether to add a low cut filter that can reduce hum or wind noise, at low intensity or high intensity. And you can also boost each radio’s power output to a high setting if necessary, using more battery. Finally, there’s a muting option in the menu, but it’s easiest to just press the power button with one tap, and you’ll see the top Audio light turn red.

One really important menu option on the receiver is to select between Stereo output and Mono output. If you’re only using one transmitter, then Mono is just fine, and you’ll get the same audio in the left and right channels at the physical stereo mini-plug output. But if you’re using two transmitters at once, then it’s wisest to choose Stereo so that one transmitter is on the left side, and the other on the right — this way, you can control the levels separately later on. But you’ll need to be really careful in post-production, to reallocate those channels onto separate tracks so they aren’t sounding hard-left and hard-right in the final export.

Another nice feature is the ability to power off a transmitter remotely from the receiver. And you can also set the volume at this pre-amp phase, especially to avoid clipping in really loud environments. There’s a bluish backlight on the LCD display, but to save energy, it automatically turns off after a time you can set. And of course, there are the usual language settings, and menu-based ability to do a factory reset, besides a reset pinhole at the bottom of the unit.

In the associated video with this post, you can hear a series of sound quality tests using live audio from both the Comica and RodeLink systems, for comparison, fed into a Zoom F6 32-bit float audio recorder.

(When you click on this picture, you get taken to the relevant timecode in the video.)

To wrap things up, I do hear a small quality difference that’s better on the RodeLink probably because of its Rode lavalier microphone, so it’s worth getting better lavs and trying that out on the Comica transmitters. But the versatility of being able to get two separate transmitters feeding into one receiver, is the best feature of this Comica kit. Other folks like Sony are just starting to launch products with the feature, but at a much higher price point. The all-metal chassis on each of these Comica units feel durable, and performance seems adequate, so you might want to save a few bucks and give these a try.

October 12, 2019 FocusPulling Original, FocusPulling Original Video Leave a Comment

04 September 2019

Blackmagic Pocket Cinema Camera 6K: Footage & Review

Written by Paul Moon

By now, the dust has settled after Blackmagic Design dropped their latest bomb, the 6K version of their Blackmagic Pocket Cinema Camera that uses Canon EF-mount lenses. The product was ready to ship upon its announcement (compared to waiting months after that 4K launch), and just as quickly, this space got flooded with anxious, speedy reviews that tended, as usual, to be binary: buy it, or not?

Well, let’s slow down and most importantly watch a formal test shoot, closer to real-world usage. The featured video here builds upon practices learned (often painfully) using the BMPCC4K, because they’re almost exactly the same camera. Also importantly, it’s shot in the way I’d shoot something I care about (as I’ll be using much of this in a new addition to “Whitman on Film”): following formal cinematography principles in framing, exposure and color. As I note in the optional commentary (or, you can just mute it!), the footage is essentially raw from raw: I ingested the Blackmagic RAW (BRAW) 6K footage using Autokroma’s BRAW Studio plugin (about which I created a long-form tutorial), to stay within my preferred NLE Adobe Premiere Pro, but I did not apply any plug-in, or even Lumetri Color. Given the nature of raw video in a log profile, it’s a mandatory step to coddle the footage with source-level adjustments to luminance and chrominance clip-by-clip, so I essentially color-graded — and applied the official Blackmagic LUT — at the BRAW Studio level by only adjusting the raw metadata. It’s really a faithful representation of the camera’s base capability without interference. Normally I’m tempted to stack something like FilmConvert for stock emulation (even though that causes most of my crashes), but this was a camera test, not art.

So, the images speak for themselves (and I speak over them, about some key issues). But the following written sections dive into a few trailing subjects that you may be interested in.

CANON EF MOUNT & SUPER 35MM-ISH SENSOR

The permanent adage holds, that you will/should spend way more effort + money on building your lens arsenal, than any given camera body. So the choice of camera is radically affected by your current lens kit, right? Of course. That’s the number one hugest factor why you’d go for the BMPCC6K, apart from the bump in resolution and sensor size, if you’re already a Canon user. In general, investing in EF-mount glass (especially for cine lenses) is the better proposition, because it adapts down to just about anything (E-mount, Micro Four Thirds, etc.) due to its outdated gigantic flange distance that was originally designed to accommodate deep inset single-lens-reflex prismatic shutters (i.e., something irrelevant to video). Also, lots of EF-mount glass is full-frame (compared to EF-S crop), which gets you ready for that inevitable time when we’re all finally shooting full frame, like I do on my Sony a7 III whenever I want better low-light sensitivity and depth-of-field latitude, if I can live with Sony’s inferior 8-bit 4:2:0 H.264-based codec.

So yeah, longer flange distance, and bigger sensor coverage, are in the plus column for EF-mount and the BMPCC6K. But in practical application, I see little difference between the Pocket 4K and 6K mounts, and sensors. I already have a toxic stew of EF, MFT, and E-mount glass, and since these Blackmagic cameras lack in-body image stabilization at the sensor, in-lens optical image stabilization (O.I.S.) is especially critical given the Pockets’ ornery form factors (e.g., no third point of contact without an eyepiece). MFT lenses have a distinguished history of nailing O.I.S., generally better than EF lenses, and if we really do care about the Pocket being (nearly) pocketable, MFT lenses are also more portable. (Many blogger types also emphasize here, that focal reducer mount adapters, including but not limited to Metabones, further close the gap between these models. Yet I’m personally, permanently averse to sandwiching more layers of glass between sensor and lens, and risking incompatibility where modern lenses rely upon electronic control of aperture and focus. Basically, native-lenses-or-bust is a smart practical decision; life is short.)

As to sensor size, it was never so simple. I was getting bad vignetting at the 12mm end of a MFT zoom lens on the BMPCC4K, because that camera’s sensor is simply bigger than a typical MFT sensor, especially if you’re shooting in DCI aspect. Meantime, Super35mm was never an exact measurement, including variances like APS-C, and so the difference in sensor size further narrows — both still far away from full frame. And after increasing resolution on the BMPCC6K sensor, the photosites are more dense and less sensitive. All told, low-light capability is basically a wash. Same for control over depth of field. Whichever aperture value you choose relative to lens quality, is practically everything (especially given the bonus feature of dual native ISO).

TOO DARN HOT VS. HVAC

One of the biggest surprises, arriving into the BMPCC6K, was its substantially bigger fan/ventilation, compared to the BMPCC4K which actually didn’t ever produce reports of overheating. Maybe the marginal resolution bump demands enough extra number crunching to push the chassis into furnace territory, but an obvious concern now is weatherproofing. I don’t shoot for NatGeo or anything, so maybe it doesn’t matter much, but what I did run into is the fairly universal need to mount the camera onto a tripod, using a common quick-release plate such as the Manfrotto seen at right. All that’s left is to poke it out the back, to avoid blocking circulation, which really throws things off balance especially if you’re mounting a heavy lens. But there’s a little good news, at least: the fan is whisper-quiet, even when it’s hot outside and when your shot runs long.

COMPATIBLE STORAGE MEDIA

Let’s start with the complaint that CFast 2.0 cards are just too expensive and too scarce, if we have other options. And we do. The obvious choice is simply to plug a portable SSD into the BMPCC6K’s USB-C port. I’ve done that on my BMPCC4K, but it required some awkward rigging, with both a cage, and a proprietary SSD mounting sled — going even farther away from a “pocket.” But for this shoot, I really needed to run-and-gun, powering off those short-lived LP-E6 batteries by simply stashing one-half dozen best-in-class knock-offs into my pockets. I actually burned through only half of them; when shooting short bursts (e.g., gathering coverage, etc.), freed from the liability of missing one second out of unrepeatable minutes, those Canon batteries work out just fine (and the new battery sled designed for the 4K, will work on the 6K). But, what about portable storage media?

Same strategy: buy a bucketful. Strangely, there’s just one sensible option on the market at this time, pictured here. Reason is, if you avoid CFast 2.0, and go with UHS-II as the internal alternate, it’s critical to buy the speed class labeled “V90” which guarantees a minimum sequential write speed of 90 megabytes per second; theoretically, this Transcend reaches 180 megabytes per second for headroom, but you’ll need as much speed as you can get for both ProRes and BRAW at 6K resolution. (And it doesn’t hurt that the read speeds, when properly connected via USB 3.2 Gen.2 UHS-II card readers, go through the roof as pictured at right.) Weirdly, at this spec Transcend maxes out at the 64 GB capacity, but it’s stunningly cheap at around fifty bucks each, while competing higher-capacity cards offer bad cost-per-gigabyte value — and if you can manage it, there’s wisdom to breaking apart your shooting day into multiple smaller cards, reducing risk of loss anyway. Given the extraordinary efficiency of BRAW at reasonable compromise ratios like 12:1, you’d be surprised how much can fit onto a 64 GB card (I got about a one-third-hour) — everything you see in the video here, fit onto about three of them. But…

WHICH CODEC? THIS TIME, IT DEPENDS

I’m on a mission to kill ProRes, which is an unspeakable offense among the Church of Jobs, but let’s get real: it’s a monstrously inefficient codec, originally designed as a technology compromise for slow Apple computers to keep up in Final Cut as an intermediate codec. It was never meant to be an acquisition codec (i.e., shooting video), but for a while, it was the best we had, starting with the Atomos Shogun. Yet now we have BRAW, which fundamentally moots ProRes — especially now that Autokroma’s BRAW Studio gets the stuff seamlessly into Adobe Premiere. Case closed!

Well, no. When you really think about it, raw means raw: if you want to store what you shoot into a format that doesn’t match the native sensor specs, given any manageably simple conversion formula . . . then it’s raw no more. So with this BMPCC6K, the odd result is that you can shoot full 6K footage in BRAW, but if you want to shoot standard 4K-DCI or 4K-UHD footage, you’re stuck with ProRes. You can see this in the available menu options above, between resolution and codec combinations. It’s really frustrating! When you consider how 4K-UHD is now the standard benchmark resolution, having this camera that can’t shoot natively into its benchmark BRAW codec, is a real limitation. And it offsets the argument that many are making (and have acted upon), to upgrade from a BMPCC4K to a BMPCC6K.

Yet ultimately, are we any worse off shooting 6K?

6K: WHY NOT (AND WHY)?

The first, gut reaction to any 6K camera arriving in 2019 — Blackmagic is soon joined by the Z CAM E2-S6 and the Panasonic S1H — is that no one watches content in 6K. Of course, but once you’re past that, it comes down to over-capturing resolution into the actual video footage, to reap the benefits when you finally render out to UHD or even HD (besides the other benefit of being able to crop in post). Storing more resolution than you’ll eventually watch, is better than: in-camera reading out more resolution than you’ll watch, and down-converting it to a lower resolution for storage. That’s literally the history of bad video cameras, dating back to the high-megapixel sensor of the Canon 5D Mark II designed for stills, bayering badly down to two-megapixel HD video footage with harsh aliasing, etc.

One of my favorite reveals to film school students, is that the classic ARRI Alexa that still shoots the majority of all movies we watch today, doesn’t even record in 4K. Does that mean my Samsung Note10+ shoots better video at UHD+? The weird conclusion is, resolution matters and yet it doesn’t: the reason the Alexa looks so amazing, is that its sensor and processor deliver such immaculate definition and color science for every precious photosite — fewer, but better. But! When you don’t have/can’t afford an Alexa, there’s no getting around the fact that a wide shot with lots of tiny little tree leaves fluttering in the background, look better from the BMPCC6K, than from the BMPCC4K, when their respective maximum resolution footage gets scaled down to HD. This compensates for its lack of elite ARRI juice.

True, 6K footage takes a lot more grinding to churn through. Last month I built a new workstation around the epic Intel-killing AMD Ryzen Zen2 3900x (ignore Puget Systems), paired with Nvidia’s new RTX 1070, and this 6K footage flies fast off my PCIe SSDs. But that’s not a typical workstation, and you might end up needing to use proxies. Even with my Death Star machine, Adobe Premiere crashed — of course — reporting that it was a GPU error, which is just Adobe saying as usual, “not our fault!” But I had applied no GPU-accelerated effects, besides Adobe’s own Warp for which they hold sole responsibility. Point being, computer editing technology keeps crawling neck-and-neck, barely keeping up with this Moore’s Law-like rise in video resolution where 6K is becoming the norm — 8K is next. We keep needing to carefully balance our demands and values, against the rage of tearing our hair out.

GETTING EXPENSIVE: PARTY OVER?

Speaking of which, in life, when things seem too good to be true, they freaking are. The BMPCC4K is $1,295, and the BMPCC6K, double that. What’s different, performance-wise? A little more resolution — but not even doubling down across the whole feature set. Another fact: notice how the 6K is being shot out like T-shirts at a baseball game, but the 4K still has a line going outside the stadium since the past year? Business is business: there’s more profit in a camera that rakes in twice as much, but costs hardly any more to make (with most of the R&D already sunk into the first one).

It’s totally unfair and unreliable to prophesy that Blackmagic is the next Apple, becoming a beloved boutique brand that simply cannot afford not to become unaffordable (!). But I’ll never forget seeing that fancy new keyboard at NAB, thinking to myself “Blackmagic being Blackmagic again!” expecting a $100 price tag . . . then, seeing it’s the most expensive QWERTY keyboard known to humankind, at $1k. If RED folds under the weight of their extraordinary arrogance soon, Blackmagic will further dominate the market (and most of us cannot afford ARRI anyway). Is China next, after these years down under? I’ve got Z CAM’s new E2C and will be posting a review here soon.

This is a render from an unprocessed, straight-to-storage still frame capture on the BMPCC6K. By clicking this image, you can download the original raw DNG file, and process it yourself.

September 4, 2019 Blackmagic Cinema Cameras, Camera User Groups, FocusPulling Original, FocusPulling Original Video Leave a Comment