![Lost - The Cinematography Team [DVD]](http://ecx.images-amazon.com/images/I/51Wg8DqbHML._SL75_.jpg)
User Review: SONY f65 Cinealta Camera
Friday, July 20, 2012 at 11:15AMDoP David Proctor used the new Sony f65 camera on Roots which needed a very raw and naturalistic look
Handheld f65
ROOTS (extract) from Geoff Bellhouse on Vimeo.
I tested the F65 on its release in January and again in April with the support of Filmscape Media, Panalux, CVP, Mission Digital & Dirty Looks, both in the studio and on location. Since this, I have been keen to give it a proper run for its money on an appropriate film.
Our early testing used the full RAW S-LOG recording at 24p, offlined in DNX and onlined on Dirty Looks’ Baselight with colourist Tom Balkwill. We were instantly impressed with the camera’s colour rendition, natural skin-tones, native sensitivity and excellent dynamic range. The mechanical rotary shutter also completely eliminates rolling shutter, the plague of all CMOS sensors.
Whilst rated at ISO800 the sensor produces very clean images. However, the F65 seems, in my opinion, to have a slightly more textural quality to it - a subtle, grain like noise giving it a distinctive character, somehow less clinical than other cameras on the market. When director Geoffrey Bellhouse approached me with the script for Roots he was after a very raw, naturalistic feel, taking heavy inspiration from the works of the Dardenne Brothers and their cinematographer Alain Marcoen. Aware of the F65’s naturalistic image rendition, and with the rotary shutter well suited to the demands of handheld work, I felt it perfectly suited to the project.
Workflow-wise I had tested the S-LOG RAW 4K with Mission Digital & Dirty Looks but in the days leading up to the Roots shoot Sony released the SLOG2 update. This opened up an array of new options for recording compressions. After some further testing with DIT Mark Kozlowski at Filmscape Studios, we opted for SLOG2 RAWLITE which seemed the best compromise of quality and file-size, in the region of 10GB per minute, as opposed to around 17.5GB per minute of the full RAW. Safe in the knowledge that we had no heavy post-work I felt comfortable that the smaller file sizes, giving us more valuable stock, so to speak, was a worthwhile trade-off.
Again based on previous testing, I opted for a set of Cooke Panchros as our primary lens set with a set of Arri/Zeiss Ultraprimes for a couple of night exterior shots.
Before the REDs, ALEXAs and countless other 35mm sensor cameras hit the market, for years I used the Sony HDW-790p and F900R as regular workhorses, coupled with the P&S Technik PRO35 adapter. Whether in desert sandstorms or humid jungles I have come to expect true reliability from Sony’s cameras and the F65 is no exception.
Across the shoot we didn’t experience dropped frames, overheating, shutdowns or any of the other problems seemingly common in early release cameras. Also of note, is the simple and intuitive menu design that is both fast to navigate and clear.
Handheld
The two day shoot was 100% handheld and I found the camera ergonomic and well balanced. Aside the images, a real stand-out feature is the viewfinder, which is large, crisp and comfortable with exceptional peaking, useful on scenes where the operator is left to self-focus - either within confined spaces or for more intimate scenes, working without an LCS. The internal ND filters also helped speed and avoided stacked filtration.
Inspired by Malick’s recent works, Geoff had specifically requested that lighting be kept minimal. Whilst necessary for the schedule, around 25 scenes across the two days, this was also an aesthetically lead decision. A lighting package was generously supplied by Panalux, and I worked closely with regular Hungarian gaffer, Aaron Szogi. Many scenes were photographed simply manipulating the available light, and others lit with a range of fluorescent, HMI and LED fixtures.
SLOG2 seems to work slightly differently to other LOG formats. It is neither overly flat, nor completely de-saturated - a very strong look, straight out of camera. This made on-set monitoring extremely close to our desired aesthetic and left me to light with my meter and regularly check the waveform.
As space was tight on location and the budget was limited, DIT Mark Kozlowski devised a workflow using the Sony Controller attached via Ethernet cable to a MacBookPro 2.5ghz i7, 16gb RAM and an SSD with two G-Tech drives attached via Esata. Davinci Resolve was used to export the SLOG2 RAWLITE 4K files to PRORES 422HQ for the offline. Maintaining a tight shooting ratio, this laptop setup was able to handle the workflow and all back-ups and transcodes were completed by the time the camera gear was packed.
Overall, I am incredibly impressed with the F65’s sensor and thoroughly considered design. It is interesting to see Sony release a camera that is such a strong competitor to the Alexa on every level. The fact that it is possible to get the full quality from the camera without the need of an external recorder is a real selling point. For now I see its place predominantly in drama and features but once its high-speed capabilities have been unlocked it could well make waves in commercials and promos. I certainly look forward to using it on upcoming commercial and dramatic projects alike.
Title: Roots
Format: Sony F65 – Cooke Panchros / Arri/Zeiss Ultraprimes
Ratio: 16:9
Form: Short Fiction
Duration: 13 mins
Country: UK
Producer: Emily O’Connor
Director: Geoffrey Bellhouse
Production Company: The Beat Picture
Camera: Filmscape Media
Lighting: Panalux
Synopsis: Jaco has to make space so his family around him can breathe.
Reader Comments (8)
Interesting test for the F65, be some parts seems unclear for me.
What’s about the new SLOG2 RAW from Sony. Why its called SLOG, when there are no colors or fixed gamma in a raw-file? The color and gama is only on HD- SDI-Out, but not recorded. I know the SLOG from the F35 but it’s only for RGB or YUV recording, the same for the Alexa workflow.
I think Sony is using this term for better promotion, but it’s again the technical basics of RAW.
I also known the Sony Ethernet Controller, but for real-time backup of 4k/10GB/min. it’s to slow.
Sometimes we need to backup two Alexa cameras, recording ProRes4444. This is only 4GB/min. and almost impossible to backup and doing dailies without two new MacBook Pro at the same time.
When you are shooting 1 hour a day, that’s 600GB and over 1 Gig Ethernet you will need
more them two hours for copy one backup and one more hour for transcoding.
Sorry it’s hard to believe, that you did this in the time you packed up the camera gear.
The raw format is a term classified as a digital negative do therefore it is an unprocessed image ready for manipulation to a more positive image. S-log2 has 1300% latitude which equates to around 13 stops of dynamic range or latitude coupled with 16bit colour so it has a lot of information an colour depth. I think your confusing RED raw with raw files in general.
And yes all raw and 4444 compressions are RGB, YUV is film (generally).
With regards to backing up I think you need better connections on your system, eSata connection should be able to handle 3GB/s on a MacBook Pro. So you should be downloading full 32GB Alexa cards in 5 mins of you are doing it properly. I regularly run 2x Alexas and transcode and am done 30 mins after wrap. (btw 1 hr of Prores 4444 is 140GB not 600GB).
I think you are also misunderstanding gigabits per second and gigabytes per second. There is a factor of 8 difference.
With regards to the F65 I was transcoding at 12fps on my laptop which you means 30 mins of footage was an hours transcode (roughly). So efficient reloads would mean your last mag wouldn't be that big. Which gave us plenty of time to finish, download, transcode and wrap the kit.
I think Mark DIT is equally confused as Valentino.
1- "S-log2 has 1300% latitude which equates to around 13 stops of dynamic range or latitude coupled with 16bit colour so it has a lot of information an colour depth."
a) What is a 1300% latitude? You cannot have percentage latitude. What is 100% latitude then? Besides, where did that 13x figure come from? 13 stops is 2 to the power of 13 which is 8192.
b) If the format is 16 bit there is no need to use a log curve. Log curve is used when you do not have enough space to fully represent the signal, like in 8 or 10-bit signal. It is not logical to have a 16-bit log signal.
2- You do not have "a lot of information an colour depth" you have a lot of colour depth.
3- "all raw and 4444 compressions are RGB, YUV is film (generally)."
a) 4444 means RGB + Alpha and it has no place in a camera.
b) YUV is 'video' never film. Film has always been RGB and video has always been YUV.
Amendment: YUV is now considered to be YBrCr as it figuratively replaced it. YUV was used originally on the old film cameras for broadcast transferring analogue information for light and colour (black and white, Y and the addition of colour later UV). This has now broadly changed so please excuse the sweeping definitives in my last post.
My objection to the article was not for the historical difference between the terms YUV or YCbCr, iIt was the use of them for film cameras. You are still talking about "old film cameras", which is totally wrong. You should have said 'old broadcast cameras'.
Film has never, ever had colour difference representation. Film has always been represented with RGB, which is what 444 is. Whereas YUV or YCrCb is 422. It is only on video cameras that the colour difference term YUV (which is now referred as YCbCr) can be used. See the related Wikipedia articles for YUV and YCbCr for detailed explanation.
I also note that you have ignored the other errors I raised on the article. Does that mean you agree with me?
Actually no, I'd firstly say being aggressive and then contradicting yourself isn't really smart.
1. These figures are taken from sonys documentation directly, have a read it's quite interesting.
B. Chips record linear but cameras make adjustments to the curves (looks, luts) as its mimicking film which isn't really a linear curve as our perception of light differs from logarithmic light. (color correction handbook). Also the bit depth is just the amount of colours used in the gamut so it's an interpretation of the image with the palette nothing else.
2. This is just grabbing at phrasing, both those terms are correct and say the same thing. This is unnecessary commentary.
3. Film is not RGB, RGB is digital representation of light. Film is chemical and based on light exposure,so the YUV traditional explanation works better. film gets scanned into RGB.
The codecs in question can handle alpha channels so sometimes cameras quote 4444 although they can physically handle an alpha channel, so agreed on that.
These forums are to educate or humiliate so let's play nice.
First of all I was not being aggressive. English is not my mother tongue and if my words comes out as such I apologise. I am simply arguing with you about your article. I had no intention to humiliate you. Pointing to someone's wrong is not humiliating that person. Otherwise how can you debate?
1- Maybe you can share that Sony document with us?
b- The sentence "the bit depth is just the amount of colours used in the gamut so it's an interpretation of the image with the palette nothing else" is also wrong. Bit depth does not represent the gamut. It represents what is the maximum number of colours possible in a gamut. Bit depth is the definition of the signal. Due to the physical limitation of the sensor or display gamut is always less than what the signal can carry. You may be falling into the trap of the consumer definition of 10bit TVs having billions of colour. No they do not. They have the possibility to have billions of colours if their panels would have allowed, which none has.
3- I have to insist that you are very wrong about film. Film is RGB. It has been since the demise of Technicolor. You can read about it on any book but Wikipedia has a decent explanation:
In color negative films there are three main color layers: the blue record, green record, and red record; each made up of two separate layers containing silver halide crystals and dye-couplers.
How modern color film works
An old article but I though I would try to clear a few things up.
S-log and S-log2 are gamma curves that transform the output signal levels from the sensor to a smaller signal range more suited to conventional video recording and monitoring.
Raw recording is the ability to take the sensors raw, un adjusted data before it's been decoded from bayer/Q67 etc and record that. So that means recording data that has no gamma, gain or white balance applied.
There is no such thing a S-Log raw, the two are mutually exclusive.
I suspect what is being referred to is recording raw but using S-Log2 as the MLUT (look up table) for monitoring purposes. So the recording is raw, but to make raw viewable on set (the camera has to convert the sensor data to a viewable video signal via some form of gamma transformation) you must use a look-up table and the "normal" look-up table would be S-Log2 as this can handle the cameras full 1300% dynamic range. 100% range is referenced to REC BT.709 the standard gamma for HD video.