Tyler Purcell

I agree, price vs performance is such an essential thing. The Alexa is the best, but it's also the most expensive on that list. I too made the pocket camera investment because I know how important glass is. I understand that NOBODY is going to see anything I shoot in 4k, nor would I ever edit anything in 4k. So the whole "resolution" game, is irrelevant to me. Imager quality, lenses and codec are FAR more important in my eyes then resolution. The lenses are part of what makes the pocket so killer. It's easy to source high quality Super 16 coverage glass because very few people want it. Where full frame glass is MUCH more money for equal quality. It's nice to have a big professional camera, especially for bigger shows. Having just been involved with a feature shooting on the Canon C300MKII, I miss the more professional camera, with a decent viewfinder, that's easy to shoulder, that balances nicely and has more robust codec's. So there is absolutely a positive to shooting with better camera bodies, it's just... they don't need to be SOO expensive. The URSA Mini 4.6k is near the top of a very short list of cameras that do these things right, for not much money.

I remember seeing the first trailer for this movie quite a while ago. It kinda disappeared off my radar as the release date kept getting pushed back. I finally got an opportunity to see it last night at the Arclight cinema in Sherman Oaks California. It was put into one of their smaller houses for the 8:05 screening, which was depressing for a relatively new release. It was funny when the guy at the ticket counter said, this is where the art films go to die. He was joking of course, but maybe he wasn't? So THE LOBSTER... It's deadpan humor was pretty spot on. The story is of course, takes place in our world, todays date, the only thing that's changed is that the government controls relationships. If you aren't in a relationship, you are sent out of the city to this boot camp for single people and if you can't find a new perfect match in under a certain amount of time, you are killed. They say "transformed" into an animal, but in reality, they kill you. The cinematography is very plain, with the use of very few focal length lenses and very little added light. Most scenes seem naturally lit and there are some scenes of pretty decent sized grain particles during day exteriors, making me think it was shot at 800 ISO all the way through. The grain did make it seem more filmic, though I could tell right away it was shot with the Alexa. I was surprised with some of the clearly underexposed material which was pushed in post. Never the less, it did look good and very filmic for digital. Since there was very little camera movement in the movie, it was all about cutting and acting, rather then whiz bang camera moves, which was nice to see. Over-all I enjoyed the movie, even though it could have lost 20 minutes and been the same product (translation: it was too long).

1) BluRay uses a 8 bit, 4:2:0 MPEG 4 Transport Stream codec, with around 35 - 45Mbps. Web based streaming services are almost all flash based, which allows for lower bandwidth playback, with the same quality content. 2) Web-based streaming is generally Flash based, but non-web based streaming is similar 8 bit 4:2:0 MPEG 2 or MPEG 4 program stream, but bandwidth is no more than 10Mbps. Almost all web-based content providers ALSO transcode to a lower-resolution .h264 file for streaming on mobile devices (ipad/phone, etc). If you only have a mobile device, you may be seeing better quality video then your desktop computer, which is again, flash playback in most cases. 3) Refer to answer 2 4) Refer to answer 1 and 2 5) Well, you can do the math yourself. If you're streaming at 10Mbps... times 60 seconds in a minute, times 60 minutes in an hour... 6) Refer to answer 2 7) What holds people back from streaming UHD content is the cost to them. Remember, they have to pay for bandwidth. So if they send you a 50Mbps file, they are paying 5 times the amount to do that, then a 10Mbps file. If the 10Mbps file looks fine, why would they send you a 50Mbps file? Also, our infrastructure can't support it. There is no way we will see UHD streaming to the quality of UHD BluRay anytime in our future, it just won't happen. Even Gigabit internet, which is right around the corner, can't deliver anything near the bandwidth of UHD BluRay. 8) Disks are dead already. The video stores went out of business. The retailers have decreased their media sections to a few hundred titles and big name movies are being pushed out the door for $4.99 on BluRay, just to clear inventory. The disk market is a zombie walking around with no direction today. The vast majority of people want instant access to media, quality doesn't really matter to them. They want to sit down, press a button and watch content. The only way for streaming services to reach the bandwidth of BluRay is if they charge consumers for that bandwidth. This would mean $50/month netflix for everyone. I don't see that happening anytime soon.

Exactly, tests only say so much, in reality things can be slightly different. The comments you made about filtration as well are so important. A lot of people don't realize what happens to an image with over filtration.

Well yes David. You are right that the over-all images dynamic range as a total number, doesn't "decrease" per say. However, the "effect" of going over or under the base ISO, IS a detractor to the over-all image. Lower the ISO and you have less highlight dynamic range. Increase the ISO and you have less black dynamic range. So where the total number (lets say 14 stops) doesn't change. There is a significant difference in the imagers ability to capture the same dynamics as it's base ISO. http://www.provideocoalition.com/alexa_iso_settings_the_least_you_need_to_know/ Still, that's the Alexa... the most advanced cinema camera on the market. In reality, most cameras and imagers don't function anywhere near the Alexa.

Think of it a different way. Have you ever had a dimmer on lights before? You know, a knob or switch that you can vary which increases or decreases the brightness of light? An imager works in a very similar way. The dimming light switch is called a potentiometer and on an imager it's called a photosensor. So imagine your dimming light switch is in the middle, you have range to go up to full bright and down to full darkness. The middle would be considered your base ISO. It has a lot of headroom at the top and lots of room at the bottom. Now, imagine if you turned that dimming light switch almost to the top, what happens? Well, you don't have anymore room to go right? You have hit the peak of the switch and it's the same thing with a photosensor. If you over-saturate it with light, you are loosing your dynamic range. You've just lost that rage from the middle of the switch to the top and bottom. You've raised the black levels considerably AND over-driven the imager to the point near distortion. This is why on some cameras, you actually get color shift with lower ISO's and bright sunlight cinematography. Is that a good enough analogy to explain how it works? With digital, there is only one ISO per camera. To get different ISO's, the electronics change the way they process the data coming from the imager. Again, use the dimming light switch analogy from above. If you shoot 6000 ISO with an imager which is designed to work at 800, you have the dimming light switch almost all the way at the bottom, make sense? So if you wanted to see in a room with the lights all the way down, it would be really difficult wouldn't it? The way they compensate for that is by taking the little bit of data it's receiving and forcing it to create an image. Since it's an analog process in many ways, noise is introduced into the image, which is why high ISO has noise. ISO has nothing to do with a lens, it's all the imager and processor. All the lens and filtration do, is decrease the amount of light going into the camera. You don't really NEED an iris in a lens if you use filters and electronic compensation in the camera to make up for the light coming in. In fact, the whole premise of an Iris didn't come until much later in the age of photography. Zooming in digitally in post production.

Yea, I hear ya. So they're producing custom ASIC's like Canon's DIGIC which perform specific tasks. I always thought manufacturers used a more basic processor like Go Pro does with the A7. It has a multi-core GPU on board, which is part of the magic.

Well as Miguel said, you can't close the aperture that much. Most digital cinema cameras are so sensitive even at their lowest ISO, it's hard not to use filtration. The other issue is depth of field. The more you close the lens, the flatter the image. Cinematographers in general, kinda like shallow depth of field. This is what separates our subjects from the backgrounds and foregrounds. This is why a lot of cinematographers will tend to shoot at a certain stop for the whole show and compensate for light differences with ISO and filtration. As Miguel pointed out as well, most glass doesn't like to be all the way open or all the way closed. Glass in general is best used in the meat of the stop range. This in conjunction with keeping the aperture more on the open side for depth of field reasons, tends to push people towards T4 through T8 in a lot of cases. The reason why any of this is an issue is because unlike film cameras where you change the stock to get different native ISO's, digital cameras have a built-in native ISO that you can't change. The more light you let into the lens OVER it's native ISO, the less dynamic range you'll have. That extra light saturation, over-drives the imager and as a consequence, the highlight detail is lost. This isn't such a problem on cameras with a native 800 ISO, working at lets say 400. But it's a huge problem with cameras at a native 1600 or more, like some of the lower-end Sony cameras. The imager really looses it's ability to deal with highlights when you lower the ISO to a normal working outdoor ISO without several stops of filtration, which is not only expensive, but kind of a waste. In terms of your resolution question... I mean, the more resolution you have, the more you can manipulate in post. For instance, if you shoot something and you want to push in on the subject, you CAN do that with a 6k camera delivering in 4k, you've got lots of room. Plus, the rate technology is going, some people think we'll have 8k televisions and theaters soon, in that case we should be shooting everything today in 8k, to protect our assets for the future. Unfortunately, that's not what's happening at all. Most everything is shot in 4k and finished in 2k (1080p) for delivery. One funny side note, my friend works at one of the top content delivery/archival houses and he said 90% of his work today is still 1080p, including features. Which is really sad to think of, we went from shooting 35mm film for most features, which is arguably 4.5k when the original negative is scanned, to basically finishing everything in 2k or worse. The only reason they do it is because 2k is cheaper all the way though the process. Less storage space needed, visual effects take less time to render, you don't need super powerful computers to compile/composite your work either. This trend will eventually change, but for the time being, it's what most people do. The interesting part is, so far digital distribution has not succeeded the best quality of film and I doubt it will in my lifetime.

I have never done active camera logs on a documentary. I simply download the material every day, organize via subject/interview. Then when I have the time to watch material, I will log what I see in an editing program. As a cinematographer for hire, I generally make sure my camera has decent metadata.

I agree with Adam. I always shoot at the proper ISO for the given situation and then add filtration if needed. At the same time however, I've found it important to under expose slightly, just enough to protect your highlights. The noise at 800 ISO is unnecessary if you're shooting bright daylight scenes.

Pricing sounds right to me, blank VHS tapes aren't very expensive at all.

You need to read wikipedia: https://en.wikipedia.org/wiki/Super_35 https://en.wikipedia.org/wiki/Anamorphic_format https://en.wikipedia.org/wiki/35_mm_film Just read those three articles, they will answer ALL of your questions.

In the past, the high grade tapes had better back coating on them and denser particles. With HiFi audio, you needed those denser particles to reduce FM modulation noise. Today however, I don't think there is any difference. I bet both tape stocks are identical, the only difference is how much tape is on the reel. Back in the day T160 tapes were actually thinner, but today nobody uses tape so they probably don't make multiple thicknesses anymore. I don't know of any multi-machine VHS duplication facilities anymore. I pulled out several facilities over the years. :(

Long GOP can be encoded by the GPU portion of a processor, leaving the CPU portion to crunch the imager data. Plus since these cameras mostly record 8 bit 4:2:0, there is far less actual processing involved. The conversion happens on the fly in real time. Pro Res by contrast, can't be encoded with a GPU, it's CPU only. So if your CPU is busy dealing with the data coming off the imager, then it also has to deal with encoding Pro Res, that's a huge problem. I mean just do a test with 10 bit Pro Res encoding on your desktop vs .h264. On mine, Pro Res takes ALOT longer then using a GPU friendly .h264 program. I can spit out compressed 8 bit, 4:2:0 Long GOP files VERY quickly. Raw is MUCH simpler to deal with if was simply streamed onto the card. However, now you're dealing with bandwidth issues related to media. I'm not sure if Cinema DNG (which is just a string of tiff files), can be compressed with the GPU efficiently. I think the reason Blackmagic gets away with a low-cost pro res license camera is because they make so few cameras compared to the other brands. That's not an excuse in my mind, Sony COULD produce a low-yield camera line, specifically with those codec's for a lower price. Make a "version" of the A7SMKII with Pro Res, why not? I just don't think they CAN. TO make it work on the FS7, they needed more hardware. To make it work with the F5, they needed a software patch. Thus showing you how much better/faster the processor and media system is in the F5 compared to the FS7. Also, Canon has ZERO excuse. The C series cameras all use media that's plenty fast enough. They absolutely don't make or sell enough C series cameras for the Pro Res licensing issue a problem with pricing either. If the C500 was Pro Res XQ and Cinema DNG Native, but was $12,000, people would still buy it. They would actually probably dominate the market because as I said before, the C500 is just a magnificent looking camera. None of the issues of the Sony imager (in that price point), which I'm not a fan of what so ever.

The problem is, unlike PC's which don't have any sub-system for media, you have to add it with plugins, apple has a built-in media plugin system called Quicktime. Almost all editing programs use quicktime as the base for decoding media. Since DNx is NOT native to quicktime, you have to install a special plugin to get DNx to work at all in your editing program. So the DNx is translated into something quicktime understands and then you can playback. That translation is slow, it's not really "native". Similar to Pro Res on windows. Some programs like Avid and DaVinci, don't use the quicktime engine, so they don't care. They can take DNx files without any 3rd party plugins. Of course, if you have 4500 media clips and you want to watch them before importing, it's nice to have file level viewing capabilities, so that's where being quicktime native is nice. I do a lot of work on Windows systems, reading pro res works fine. Exporting is a problem because it's only 32 bit. DaVinci supposedly has a 64 bit Pro Res export tool, but I don't know anyone who would run DaVinci on windows besides yourself. Everyone I know uses Linux or Mac OS to run a coloring solution. Partially because they're far more stable operating systems that can run for 24/7 without needing a reboot. In terms of graphics cards, your card is pretty powerful. Most computers have Intel onboard graphics, with shared RAM as video memory. The rest of them, have super low-end mobile graphics units, either on PCI boards or soldered onto the motherboards. So the vast majority of computer owners, don't understand why their DNx files don't playback fine. So they simply don't play them back. You are 100% accurate that DNx will perform better on your windows 10 machine. The problem is, the rest of the computers in the world with $19 graphics cards, will need to understand that they need upgrades. Easy for a computer person to understand, but not easy for a consumer, someone who can barely keep a portable phone working, let alone a complex editing environment.

Well, name another camera that shoots Cinema DNG Raw and/or Pro Res HQ that records 60fps, which isn't gobs of money. There aren't any... and there aren't any for a reason. Pro Res is very difficult to encode for smaller/lower end processors. Blackmagic couldn't afford to source an A7 for instance, it's too expensive. They put all their money into the packaging, imager and software. Blackmagic struggled to get 30p out of the pocket camera, I use to drop frames constantly until they came out with a software patch. The URSA has a much faster processor, it finally has decent slow mo capabilities as a consequence, but in a much bigger package. The reason why the MPEG cameras can record slow-mo is because the processing is such low bitrate, the lighter weight processors don't have any issues. It wasn't a licensing problem that prevented them from having Pro Res, it's simply the processor. This is why the FS7 requires a different piece of hardware to record pro res, rather then just a simple software patch like the F5.

I second David's theory. A lot of my friends who use anamorphic on digital, are using it for the lens flare and other optical artifacts. One of my friends uses anamorphic's exclusively and his stuff is really pretty. If I had my druthers, I would use the Hawk 1.3x anamorphic's on a digital feature without question. The idea of matting down the top and the bottom to get 2.35:1, doesn't really interest me as much. Even though for that aspect ratio, the digital projector will be cropping the top and bottom.

Well it all comes down to the fact Pro Res is native to mac's AND quicktime on every single operating system dating back more than 10 years. So no special 3rd party drivers/plugin's necessary to work with Pro Res on a mac. DNx by contrast doesn't work at all on a mac without special drivers. In fact, DNx codec wrapped in MXF is one of the most incompatible professional formats for a mac. It requires different driver combo's to work and SOME editing software is very slow when using because the drivers translate the codec for quicktime. This is so every application on the mac (including finder level) can use the DNx codec, which is nice, but slow. On windows, you still need those drivers, but you can't just hit the spacebar and preview the DNx files, like you can on mac. You need to open them up in a program to watch them. The ability to quick preview anything that works with the quicktime engine, is so nice. I'm not sure if Windows has fixed that issue with 10 service packs, but on the few windows 10 Avid bay's I've installed, the DNx files were not playable by the operating system without an application to view them. So if all you do is use specialized programs all day long, DNx works fine on both mac and windows. The problem is, a lot of an editors time is spent organizing and watching files at the file level. Without the computers ability to natively playback files at that level, it makes organizing very challenging and a real pain in the ass. This is why handing DNx files to random clients to work with, isn't such a smart idea. In most cases, non-Avid clients will never have seen an MXF wrapped DNx file in their entire life. They aren't going to visit Avid's website to download a driver because most industry clients who use business computers, don't have the ability to install new software. This is very common place in the film/broadcast industry. At the same time, Pro Res has been around for quite a while, so most clients already have the driver on their machine and most IT departments, install quicktime as a default. I send pro res files to random people on mac's and windows machines on a regular basis and rarely have a problem. Sure, you get the occasional non-industry client who can't play it back, but if you install iTunes, you have quicktime and the ability to playback pro res. The other problem is laptops. The vast majority of people in the film/broadcast industry, use laptops. Not for just for editing on the go, but also for viewing clips their editors send them. In this case, they can't upgrade the graphics card so it's powerful enough to handle the very GPU intensive DNx codec. Where Pro Res works great, even on 10 year old mac laptops, because it's designed to work properly on multithreaded 64 bit systems. At the same time, the vast majority of computer owners, don't even know what a GPU is. Educating all of them is impossible, most of them just buy a new computer when it gets slow, instead of upgrading at the component level. So you can say; buy a better GPU all you want. The reality is, very few of them will. On the Mac side of things, nobody does. The vast majority of computers I work with, have stock graphics cards. This is because the clients buy 10 at a time and they don't spurge for the higher end computers. DNx is great for people who understand computers, and have put in the time and effort to make those computers work well. For everyone else, Pro Res is a far better format, not just because it's native to mac's (most of the post production industry is mac) but also because it's CPU based, which means you don't need to go out and find a special graphics card to make it work. This is especially a problem with laptops which have junky Intel built-in graphics.

This is how I'd rate them. Alexa SXT Epic Dragon Red One MX F55 F5 C500 Ursa Mini 4.6k FS7 A7SMKII GH4 Blackmagic 2.5k The Alexa is by far the best over-all camera on that list. Some may argue the newer Red Weapon-X is on par, but the Epic is not. The Red One MX is very similar to the Epic in quality, even though it's bigger. The F5 and F55 are great cameras, but not QUITE up to par in quality to the Alexa and Red. Those cameras aren't really going after that market anyway, sony makes other cameras for the purely "cinema" market. If you put the F5/F55 footage next to the Alexa and/or RED, even with a good colorist, the difference is clear. Not as much with the F65, which competes directly with the Alexa. I'd put the C500, Ursa Mini 4.6k and FS7 very close to each other. They all have strengths and weaknesses, but final output quality is very similar. The FS7 is more ENG (electronic news gathering) style, the C500 is a hybrid with the functions of an ENG camera, but in a much smaller form factor. The Ursa Mini is more of a cinema camera with a sprinkling of ENG features. I really like the C500's imager, it's a gorgeous kit, but the Canon electronics let it down I feel. They haven't quite nailed getting the imager quality onto the screen like the other manufacturers have. The C500 is by far the best attempt, but it's not there yet. I'd put the GH4 and A7SMKII as an identical finish, they are the competition to one another and they are the most similar looking cameras after coloring. The A7SMKII has a few features like the mechanical stabilization and ultra low-light capabilities which make it stand out. However, the A7SMKII also has some issues, one of which is the imager being so sensitive, it's hard to get decent quality high dynamic range images out of it, when you're in direct sunlight. It's as if the designers never took into account natural light when designing, which is very strange. I've worked with both cameras and colored many shots with them, I've found them both to have similar highlight clipping issues on bright patches and look more like "video" then cinematic. Of course, both suffer from the very restrictive, low quality internal recording format, which is a real shame. The expense, weight and size of external recorders to get better quality, makes the fundamental purpose of buying a small camera, negated. The Blackmagic 2.5k is a great looking camera, but it stands alone for many reasons. Partially due to the physical issues; internal battery, big LCD display in the back which is worthless, odd size which makes it hard to hand hold, big/expensive SSD as the only recording device, 1/4" audio which is annoying to adapt. Nobody has any idea what Blackmagic was thinking when they designed this camera, it's almost a joke of an idea, but they sold like hotcakes and still do today believe it or not. The imager looks outstanding for the price bracket and it shoots real 2k 12 bit RAW and Pro Res HQ, which is awesome. Missing from this list is the Blackmagic Pocket camera, which I would honestly put above the 2.5k because it has so many positives in comparison. From the ability to run lower-priced, but higher quality S16 lenses, to the form factor and of course, the fact most of the mistakes made in the 2.5k design are missing in the Pocket. Having spent quite a bit of time with the 2.5k, 4k and pocket camera, I can attest to how much I really love the pocket in comparison, it's a world of difference in my eyes. Not something you'd use to make a feature film with, but absolutely something I'd use on the go.

DNx in itself is a great codec actually, I've been told it's JPEG2000 based which means it requires specialized hardware to playback. This is the reason why Avid has always been a hardware based editor. Today however, with super fast GPU's available, it's less of a problem. The only real issue is with people who don't have fast GPU's. The GUI tends to gobble up all the system resources and the first thing to stop working is the DNx playback engine. On systems with fast GPU's, this isn't an issue and DNxHR is really a great codec for those people. One small side note, it does require 3rd party plugins to work in the OS properly. Very much unlike DNx, Pro Res is a CPU based codec. The whole reason it works well is because it's multithreaded. This splits the tasks against multiple threads and cores within each of the processors. This requires a 64 bit operating system and multiple cores, which is actually easier to deal with then fast graphics cards, which eat up power. Apple wanted Pro Res to be workable on all their computers, including laptops, without the necessity of having a super fast GPU. Bit rate vs bit rate, Pro Res and DNxHR are identical quality wise. The only difference is how they used by the system. From my experience, DNx works a lot better on windows computers then Pro Res. It's visa versa on Mac's, where most people prefer Pro Res because it's 100% native without 3rd party plugin's like DNx has. Another kinda important side note is that, I don't believe you can render effects to DNx with any non-avid editor. That's one of the limitations that Avid put on licensing DNx. This means, when you render effects on your sequence, you are most likely going to working with a different codec. This isn't an issue in Avid, where DNx native, but it's something to think about. Obviously DaVinci only renders on the fly, so there is no reason for an intermediary codec like other editors. I'm glad Pro Res doesn't work good on Windows, it separates the platforms even better. Keep the professionals using the Mac OS.

Right, so when you shoot 2x anamorphic and project it, you are getting 2.39:1. I guess Stewart's point is that with a digital imager the aspect ratio would be 2.66:1.

If you shoot academy and make a print, you aren't cropping to get 2.39:1.?

Well, the agency I worked for owned the rights, since it was all commercial work. So I couldn't just go in there and get a copy of my stuff, it really sucked. The one time I did, they handed me a 3/4" tape with burned in timecode. I'm sure somewhere in a box, in someone's attic, that tape still exists, but I doubt it works.

It's really down to the imager's ability to translate photons of light into data and how that data is then translated into a visual image. Most of the still cameras like the Lumix, focus on creating still images, rather then video. If you were to take a still frame from the video and compare it to a still frame shot with the camera, the difference is night and day. The video side of the DSLR's has always been very weak, the GH4 and A7SMKII being slight exceptions. In my opinion, 4k isn't as important as the imagers ability.

I don't have anything really good online, I'm mostly a for-hire filmmaker. So I can't post anything and a lot of times, I never get a copy of my work, which sucks. Almost all of my decent film stuff was lost due to the company who financed it, filing bankruptcy and somehow mysteriously loosing all the masters. It's an unfortunate side effect of the industry we live in. But if you're bored, you can check out my company page: http://tpproductionfilms.com