By Jendra Jarnagin, June 8, 2013
{Editor's Note: Jendra Jarnagin is a talented cinematographer who has a growing list of independent feature films and other projects to her credit. This article is from her blog and is published here unedited and with her permission.}
I usually use this space for announcements about projects I am shooting or screenings of films I have shot. But I have had so much interest in my camera tests for the new Sony F55, that I wanted to share my results publicly.
I am shooting a hilarious comedy feature called “A Bet’s A Bet” in Rhode Island, where I grew up, and 2/3 of the film is being shot within 2 miles of my dad’s house, and the rest within 10 miles. It is my first opportunity to use the new Sony F55, so of course I shot some tests to learn the ins and outs of what settings to use and how I want to expose it, etc.
Disclaimers:
I tested what formats were available at the time, June 2013. New firmware with new options is imminent, but that is being released while we are in production, so I am sticking with what the camera can do today. Because I am swamped with pre-production for this film, I don’t have time to get my videos uploaded anywhere to share. Besides, I believe watching someone else’s tests, with online video compression, doesn’t tell you anything anyway. So instead, here are my own written assessments.
I have spoken below about editing proxies, etc. I want to note that FCP X can edit 4K XAVC natively, and Premiere will soon, but for what we are using right now, the proxy solution is our preferred workflow, but you have other options.
Since I go into production the day after tomorrow, forgive my lack of responsiveness for comments and questions.
What I Tested:
I shot over and under exposure tests, gain tests (including in-camera noise suppression settings) and codec and resolution comparison tests. I also shot some filter tests to decide which ones I liked best for this camera and this project. I shot the test at Adorama (thanks to Henry Bornstein) and projected the test at Goldcrest’s DI Theater with colorist John Dowdell using a Quantel Pablo IQ and a 2K projector. (Thanks to Mark Doyle.)
I shot everything in Slog2/S-Gamut.
Things I didn’t test and why:
Some things that are exciting about this camera, I didn’t test because I am already familiar. I know what a global shutter looks like. And I am a huge fan of the extended color gamut of the F65 and F55. For the movie I shot on F65, I have never seen colors like that come out of an electronic camera before, namely the tertiary colors: purples, red-oranges, and yellow-greens. And the 16 bit color depth of skin tones blew me away. In the side-by-side comparisons I saw projected that the ASC & PGA did (that I participated in as Behind the Scenes stills photographer) it was the first time that I saw skin tones that looked as good as film. It was a subtle but a powerful thing when you saw all the subtle distinctions in the color texture of a human face.
Codec Test:
My first order of business was to decide what codec to shoot. I was biased when I heard that the new XAVC was a long GOP h.264 type codec. I have had such bad experience trying to grade h.264 DSLR footage, and remember nasty temporal artifacts when we first started seeing Long GOP codecs in cameras. I need to adjust my thinking about this, because this codec turned out to be quite robust. I was also concerned that the 4K XAVC has such a low data rate for 4K that I was worried about compromises with compression artifacts to get it there. I wondered if the less compressed HD XAVC might be a better choice. (Whereas HD SR codec would be even better once it is available in a future firmware release.) 4K RAW isn’t an option in terms of amount of data for this low budget indie to handle, especially since we are shooting 2 cameras much of the time. We just can’t afford that many hard drives… So I did some tests designed to stress the codec and I couldn’t break it. I was pleasantly surprised and ended up choosing to shoot 4K XAVC for this film.
I shot an assortment of different shaped flowers and greenery, (which was probably the ugliest bouquet I ever bought) with a fan blowing on them, in the background of a wide shot, and also in close up. I also shot the front page of the NY Times, that has a variety of font sizes on the same page, starting far away, and walking closer to camera, and then rotating the newspaper slow and then fast. (Thank you to Mitch Gross from Abel for that idea.) I compared the 4K RAW, 4K XAVC, and HD XAVC. We were looking for any weirdness around the edges of the areas of fine detail. Surprisingly, we couldn’t see any differences at all, even when zoomed in on the screen.
Resolution:
I had heard that the HD coming out of the camera was so sharp, that it looked as good as the 4K. “Yeah, yeah sure,” I thought, “to the untrained eye, viewing on a laptop in a non apples-to-apples situation, “yadda yadda yadda… But the thing that surprised me the most out of this entire test experience was that I actually agree that the HD looks just as good as the 4K! Myself, John Dowdell, and Tim Spitzer, (managing director at Goldcrest) surprisingly couldn’t tell the difference when we cycled back and forth between the HD & 4K XAVC . We stuck our noses up to the screen, we zoomed in 200% (and more) on the image, we went back and forth with the slate framed out and lost track of which was which. Again, we were looking at a 2K projector, but this project (and MOST projects) will finish in 2K or HD. I imagine you could see a difference on a 4K projector. But we couldn’t. So my conclusion is that the downsampling is so good in-camera, it looks as good as the 4K. My decision to shoot this movie on 4K was very carefully considered and for the following reasons:
A) For re-framing in post (which I used to be offended by someone else re-composing my shots, but have seen editors work magic and save some scenes this way, so I am open to it, especially for a comedy.)
B) To future proof the film. I have high hopes this is going to be a big indie hit. What if there was only a standard def copy of Animal House and they couldn’t re-master the re-release?
HOWEVER:
I only made this choice after confirming the staff editor at the production company could conform it internally (from the HD edited proxies back to the 4K original) before sending it off for finishing. I didn’t want to cost the production company the extra money of a conform at the color correction facility. If the answer had been no, that there was no one to do that without substantial extra cost, I would have comfortably chosen to shoot in HD, which would forego the need to edit proxy material. (See my note in the disclaimer section above, that working in proxy is OUR workflow and doesn’t have to be your workflow.)
Dynamic Range:
I agree with Sony’s assessment of 14 stops Dynamic Range. Set at native ISO 1250, I read recoverable detail at 5.5 stops over exposure and 7.5 stops under. Of course, dynamic range is subjective, as to how much noise are you willing to tolerate in your shadows before you don’t consider it usable image information and you want to crush it into solid black when it gets ugly, but I would say, as long as you leave those last 2 stops of shadow detail pretty dark, you can still make out some detail there.
Post Processing Problem:
I did have an unfortunate error in my test. There was something wrong with the way the RAW files got processed. Goldcrest is talking to Sony’s engineers about getting to the bottom of what went wrong. At 6 stops underexposed, the RAW showed no more detail, and the XAVC did. In fact at 7 and 8 stops under there was detail in the XAVC and not the RAW. I immediately called in their tech people and called Dhanendra Patel at Sony, because that doesn’t make sense. RAW would have as much, if not more readable detail in the shadows. The XAVC material was processed through Resolve and we saw the extra shadow details. The RAW was processed at the default settings of Sony’s RAW viewer software. We don’t want people to think the camera has less range than it does, so this investigation and follow up testing is still underway. I will update here when I find out what’s up.
Because of this issue, I couldn’t tell if the RAW has more usable range than the XAVC. I compared the RAW and XAVC 4K shots that were 5 stops under exposed and graded up and they looked the same to me. The characteristic of the noise (which we pushed hard for the purpose of judging it) looked the same.
Noise and Gain:
I wouldn’t hesitate to intercut material shot 1 stop pushed (2500 ISO) with normal 1250 ISO footage. And I totally wouldn’t mind shooting a whole scene at 5000 ISO. (Seriously, 5000 ISO!!!!) It has a look to it, in terms of seeing some noise, so I wouldn’t mix it within the same scene as material shot with ISO 1250, but I don’t find the noise objectionable at all. When I went one more stop to 10,000 ISO, it was CONSIDERABLY more noisy than at 5000. A very big difference, compared to the other one stop changes. It was definitively quite noisy and I wouldn’t want to shoot it for a narrative movie.
Noise Supression Settings:
I tested the camera’s internal Noise Supression settings at 10,000 ISO. When set to low & mid, I didn’t see much happening. Which I am told is an improvement in the last firmware update, because people previously found it to be too aggressive and softened details in the image. The High setting definitely improved the noise somewhat, but not to the degree that it would have me choose to shoot 10,000 ISO. I would rather leave it off, and have a finer degree of control with post Noise Reduction tools. But if I was shooting something with fast turnaround and no color correction, like a documentary, I would go ahead and use it at High.
Filters:
This is totally subjective. I prefer using diffusion filters in camera. I like the optical alchemy of it vs. the electronic image manipulation. It feels more organic to me.
I consider it should be irrelevant to anyone else what my personal taste is in this situation, but in case anyone cares, I liked the Black Diffusion FX. At least in this lighting set up, this lens, this actress, this frame size, for this movie, etc. I still liked my favorite Schneider Classic Softs (which along with Tiffen Soft FX I call the “invisible filter,” the idea being you explicitly can’t tell you are using filtration.) I’ll carry a variety of diffusion filters for a variety of people & situations. But the BDFX I just liked on this camera. It’s usually not my favorite one. It’s my least favorite filter when shooting film for example. And on Alexa, because of its inherent softness (which I LIKE, don’t get me wrong!) I find all the diffusion filters look remarkably similar to one another with very little to distinguish the flavor of each.
Ease of use:
There are lots of good cameras out there that make great images, and that I am happy to use. I am always a believer in “the right tool for the job” and despite some close vendor relationships I have had over the years, I remain philosophically format agnostic. But it always drives me crazy when you have to put a million attachments and rails and brackets and doo-hickeys to make the camera into something to shoot with. That’s a HUGE turn off for me. I want a camera that’s shaped like a camera. And Sony has finally given us that with the F55 and F5. And so far, I really like the OLED viewfinder. Another plus is how versatile the camera is with the choices of resolutions and codecs. I decided to shoot 4K XAVC for this film, and part of that decision was the convenience of the simultaneous HD recording. It can record HD XDCAM 50Mbps at the same time ON THE SAME CARD. No need to rent an external recorder, no need to transcode dailies! How smart is that?!
Conclusion:
Prior to this camera, I felt the Sony F65 made the best images, and that the Arri Alexa was the best all around camera to use, when all things were considered. (In other words, in many production situations, the tradeoffs of using the F65 were not worth the increase in image quality when so many other cameras look so good.) I so far feel like this camera is the best of both worlds, with serious plusses and so far no minuses that I can find or think of. I expect seriously fast adoption of this camera as the new professional camera of choice.
And just so you don’t think I’m strictly a Sony convert, I’m still excited about Panavision and Arri’s upcoming cameras and Red’s new Dragon sensor!
For more information about me, please visit my website.