This is how Canon’s Dual Gain Output works on the new EOS C70 4K camera
Sep 27, 2020
John Aldred
John Aldred is a photographer with over 25 years of experience in the portrait and commercial worlds. He is based in Scotland and has been an early adopter – and occasional beta tester – of almost every digital imaging technology in that time. As well as his creative visual work, John uses 3D printing, electronics and programming to create his own photography and filmmaking tools and consults for a number of brands across the industry.
Share:
There’s been a bit of misunderstanding over exactly what Canon’s Dual Gain Output (DGO) is on the newly announced Canon EOS C70 4K Super 35mm cine camera. Some have speculated that it might be something akin to dual ISO (given that gain and ISO are often interchangeable in the video world) and, well, sort of, but not quite.
Canon has released a new white paper detailing what Dual Gain Output for their new 4K Super 35mm sensor is and exactly how it works to provide over 16-stops of dynamic range.
Dual ISO, as we hear the term in other cameras like the Blackmagic Pocket Cinema Camera 4K, essentially means that a camera two base ISOs. For one ISO, the information coming from the sensor goes through one circuit, and for the other ISO it goes through a different circuit. This way, each is optimised for that different base ISO level to provide the maximum quality when you raise it up.
Canon’s Dual Gain Output actually has two separate photodiodes at each photosite with each sending its signal through its own separate circuitry and processing to protect both the shadows and the highlights. It nd then combines the two signals to increase the overall dynamic range of the entire frame – and it does this however many times per second it needs to in order to meet the required frame rate.
The sensor combines two methods of processing to create the final result, both of which are detailed in the white paper. The first is Saturation Prioritizing Gain mode where the goal is to retain the maximum dynamic range of the sensor as a whole, paying attention to protect the highlight details. The other is Noise Prioritizing Gain mode which aims to minimise the amount of noise present in the final footage.
The way DGO works might explain why there’s no raw video output on the EOS C70. Because each pixel on the sensor is essentially pulling information from two separate photodiodes, it doesn’t really fit the traditional raw model, where you have one photodiode per pixel. Even if it could output a raw file, those files, in theory, simply wouldn’t have the dynamic range of the files the C70 puts out.
Or, they’d be twice as large as they would otherwise be in order to store twice the amount of data from pairs of photodiodes instead of individual photodiodes. That would not only increase storage requirements but make actually editing the footage more difficult, too. Not to mention new software would need to be developed in order to decode the files on your desktop.
It’s a fascinating technique, although Canon isn’t the first to implement it. ARRI’s been using it for a decade and it’s not too dissimilar to the way some Quad-Bayer smartphone cameras shoot HDR video – a camera like this is obviously going to be much better quality than your phone, though. Hopefully, we’ll start to see this method being applied in more cameras to help increase dynamic range and improve overall noise levels, especially in low light.
You can read the complete white paper, here.
[via No Film School]
John Aldred
John Aldred is a photographer with over 25 years of experience in the portrait and commercial worlds. He is based in Scotland and has been an early adopter – and occasional beta tester – of almost every digital imaging technology in that time. As well as his creative visual work, John uses 3D printing, electronics and programming to create his own photography and filmmaking tools and consults for a number of brands across the industry.
Related Posts
Canon EOS 5D mk IV is the best of Canons (but still behind Sony and Nikon), DxOMark says
Canon issues service notice over lockups with Canon XF605, EOS C70 and EOS R5c cameras
The Canon EOS C70 is Canon’s first RF mount cinema camera – 4K 120fps (without overheating?)
Battle of the $2,499 Canons: Canon 5D Mark 4 vs. Canon R6
Join the Discussion
DIYP Comment Policy
Be nice, be on-topic, no personal information or flames.
11 responses to “This is how Canon’s Dual Gain Output works on the new EOS C70 4K camera”
Hmmm… a+b = raw? I think they can save two raw frames, or only one raw as sum of its pair. Something like this in mp3 stereo, there is only one channel (mono) with differeneces between left and right stereo channels. But in image we have only one plane (for displaying), then it is not necessary have two data: sum + difference. It possible to save only wide 16-bits data, DNG for example
But the issue would be one of bitrate to write out two files simultaneously. Just a single 4K 120fps raw video stream would likely already be pushing the limits of CFexpress. To output two simultaneously would be impossible to a single card slot.
And then you’d still have the issue that neither raw would be as good as the combined baked-in dual-gain footage it presently puts out. And if you were hoping to be able to recreate that look in post, a whole new set of software would need to be created (Canon would need to provide one themselves, obviously, but then there’s also Resolve, Premiere Pro and FCP support) to achieve it on the desktop.
So, I don’t think it’s really as simple as just splitting it up into two files. Not if you actually want the benefit of dual gain output. And if you don’t… Well, there are better suited cameras out there for this kind of money. :)
I guess some file formats can keep any dataset with any bit depth.
Bit RATE not bit depth. The amount of data being written out per second.
Bit depth = how much data for one color channel of pixel. They say the Canon C70 allegedly use two different channel for wide dynamic range. It needs more bit depth, but they say it impossible to save its data in classic RAW, DNG or something else. I guess they lie.
I know exactly what bit depth is. I’m saying that the issue is bit RATE. It wouldn’t be able to write the data out fast enough.
C70 has 16 stops DR, it’s about 16 bits, not 12 + 12. Of course camera needs good alghorithm for packs this data.
Not really RAW? Apple and BM show HalfRAW, QuarterRAW, etcRAW, but… they say it’s RAW.
Ok. Let’s check 24 bits per pixel. FullHD 24p with FULL RGB color is about 427 MB/sec. But Canon’s sensor doesn’t produce FULL RGB, because Bayer. 1920×1080 * (12 + 12) bits – it’s true color data of these pixels, only 142 MB/sec for 24 fps. 569 MB/sec for 4K 24p. It is true 12+12 bits of Canon’s DGO. But, I guess there is not true useful signal of all 12 bits in both pixels . If they say about 16 stops DR, I guess it’s only useful 8 bit of each data set (I read it is saturation and lightness, I can’t understand how they split it).
You’re completely missing the point. But never mind. Best of luck to you. :)
The canon c300 mk 3 has dgo and raw.
and with the firmware update, now the C70 does as well.
Just FWIW, these has been analog amplification prior to A/D conversion in the previous products (look up the C300 whitepapers). Yes, it is true that this analog amplification can’t be changed in RAW files (which are post-A/D digital information. Also, the two photosites in a pixel have traditionally been added together (which cancels some noise and was the original idea behind researching multiple photosites).
The difference here is that BOTH photosites provide a high and low gain signal which are converted to digital, added together (separately) and then adjusted and combined into a single value per pixel. That value would be what goes into a RAW file (on the 300 for example – one value per pixel) or subsequently converted to a 10 or 8 bit value and compresses by one of the other codecs.