Canon R5 Announced with 8k

What I wanted to say above was that the sensor readout is a first stage of the image pipeline that happens before and independently of further processing.

ahalpert said:
But I know, from reading Sony sensor datasheets, that Sony sensors can always achieve progressively higher frame rates at progressively lower bit-depths, which I assume indicates a faster readout time and therefore faster rolling shutter?
Click to expand...

I'd say that a camera's ability to read the sensor will cap the max frame rate of the system. But I think we agree that this isn't exactly a revolutionary insight.

But what you say: "a camera can increase frame rate if it lowers bit-depths", won't let us draw any conclusions about readout speed. It might also be a data issue, for instance. Perhaps buffer? The mk3 has a substantial buffer.

We know the 1Dx mk3 reads the sensor at 12bit in raw up to and including 60fps. In 60 fps it reads the sensor at 16ms. This is confirmed, so there should be any technical limitation to why it wouldn't work with 24-30.

As to the current implementation at lower fps, I don't know. But I'm also curious about it. I've tried reaching out to Rudy Winston at Canon (technical advisor and presenter) to see if we can get further, but Canon is historically difficult with these things.

And as mentioned in this or if it was some other thread: sometimes these forum discussion go down the rabbit hole and outward facing brand representatives just don't have a chance answering the questions. It won't help, of course, when most of them are just 'people off the streets' hired to do "social media".

I tried interacting with Canon Europe on Twitter a while back (it was actually their initiative). What a farce.
 
What you're saying makes sense, but looking at the Sony sensor datasheet, it seems that there are multiple A/D converters at various bit-depths built into the chip (thus there is no down convert step?)

This data is camera-independent, so I don't think it depends on camera buffer size.

I have no idea if Canon does things the same way, I just happen to have read multiple sony sensor datasheets and they all look similar with regard to max frame rate vs bit depth

But so my thinking was that perhaps Canon read the 24fps at 14 bit depth for max quality and then down coverted to 12 due to codec limitations, especially since the Sony figures show double the framerate when dropping from 14-bit to 12-bit readout, which parallels a halving of rolling shutter on the Canon when switching from 24fps to 60fps

https://www.sony-semicon.co.jp/products/common/pdf/IMX410CQK_Flyer.pdf
 
You could say that I've done similar reading, but on the Canon side, since that is the system I'm in.

I've read most of their white papers since the original C300 regarding Cinema EOS including papers on cameras, sensors and transfer curves. Oh, some papers on lens design too.

I don't say this in an effort to 'drop the mic' and leave, nor am I trying to convince you of something. I can't say for sure how much correlation there is between Sony's and Canon's methods. We can assume there is a great deal of overlap.

Since I like to be right =) mostly for my own piece of mind, I try not to speculate too much (or at least make sure it's clear that I'm only speculating). I don't mind not knowing things.

I don't feel we have enough information here to go much further. The critique is valid and I just wish we could get Canon to comment. Not sure how...
 
Well if you do get through to Canon - and this theory is correct - I would imagine that they could drop the readout bit-depth of the 24/25/30 mode to 12-bit to vastly improve the RS with a modest IQ penalty. (Or hell give you both readout options in menu)

I'm generally souring on these new cameras. Drawbacks to all. Maybe I will stick with what i have. Either way I'm sure I'll be watching/reading a lot of reviews to come.

Keep us informed of what you find out!
 
ahalpert said:
I would imagine that they could drop the readout bit-depth of the 24/25/30 mode to 12-bit to vastly improve the RS with a modest IQ penalty.
Click to expand...

Perhaps to round off this little tangent... =)

I'm not onboard with the bitdepth connection to readout. In my mind, the photo diode gets charged by incoming photons which generate an analog voltage that gets read out. This will then be converted from analog to digital, but effectively after the, in this case rolling shutter, has gone over all the pixels via one or many 'taps' out of the sensor pixel array. So, once the digital image construction begins, the sensor has long been read out. That is my mental model.

Anyway, if lightning strikes and I hear something, I'll keep you posted.
 
AndreeOnline said:
Perhaps to round off this little tangent... =)

I'm not onboard with the bitdepth connection to readout. In my mind, the photo diode gets charged by incoming photons which generate an analog voltage that gets read out. This will then be converted from analog to digital, but effectively after the, in this case rolling shutter, has gone over all the pixels via one or many 'taps' out of the sensor pixel array. So, once the digital image construction begins, the sensor has long been read out. That is my mental model.

Anyway, if lightning strikes and I hear something, I'll keep you posted.
Click to expand...

There is one additional variable before engaging the camera's buses and processors, which is the bus the sensor uses to deliver data TO the camera's buses. If that bus is limited in its data rate, then it might be dropping the bit depth for higher frame rates in order to fit the data stream within the constraints of its bus.

In other words, though bit depth my have no affect on readout time, the sensor might not have enough bandwidth to support a higher frame rate without lowering the bit depth to compensate.

The sensor doesn't read the well potential out and THEN convert the result to a digital value. Converting it to a digital value is part of the readout process. Since everything behind the sensor is digital, there's really no other option. Even in CCDs that was the case, what differed in a CCD is that the ADCs were all attached to one edge of the sensor and the photosites passed their data down the line to the ADC like a bucket brigade, which among other things lead to the CCD smear artifact.

CMOS sensors don't have that constraint, they can read out data from each photo site directly... but the controller still needs to pull that data, package it up, and stuff it on a bus to send to the camera's processors. It's already digital because of how CMOS sensors work. When a photon strikes the photosite that photo site outputs a value describing that photon's energy into an accumulator. When the image capture is done the data is already digitized, but stored in a small buffer on each photo site. The controller goes through each row and grabs its data like in DRAM and packages it up to send out to the camera's processors over its bus. That part is what determines the readout time, and also explains why lowering the image height on most cinema cameras allows for higher frame rates.
 
This may be relevant and, if I'm reading it correctly, show a direct link between A/D converter bit-depth and read ou ht speed:

"The speed is governed by a clock - often an incoming communication clock for external SPI or I2C controlled ADC chips - so the speed of them is usually limited to the speed of the communication bus. It usually takes as many clock cycles to capture a value as the sampling resolution, so a 10-bit ADC would take 10 clock cycles to sample. Typical rates are in the hundreds of Ksps up to 1.1Msps in some MCUs."
 
The technical aspects are certainly interesting, but we can't escape this simple fact:

> 1Dx mk3 reads the full sensor in 16ms and stores this information in 12bit Cinema Raw light files at 60 fps.

Whatever the underlying conditions are, most people would be happy if the camera did the same for 24, 25, 30 fps.

So, whatever it does at 60 fps—do the same now that you have twice the time, or more, to do it.
 
Can't argue with that! Hopefully they listen to the many users I'm sure will be requesting it

What's your feeling on the raw bitrate? How does it compare to p6k? Not too much data to deal with?
 
AndreeOnline said:
The technical aspects are certainly interesting, but we can't escape this simple fact:

> 1Dx mk3 reads the full sensor in 16ms and stores this information in 12bit Cinema Raw light files at 60 fps.

Whatever the underlying conditions are, most people would be happy if the camera did the same for 24, 25, 30 fps.

So, whatever it does at 60 fps—do the same now that you have twice the time, or more, to do it.
Click to expand...

Feeling a lot like the 1DC days... ='(

I say this almost every time I bring up the 1DC, if they had implemented 2K DCI in raw..... I digress. Magic Lanter, we were so close. so close.
 
ahalpert said:
What's your feeling on the raw bitrate? How does it compare to p6k? Not too much data to deal with?
Click to expand...

Braw is much superior. Not in a 'quality' sense, because that part the Canon gets right. I haven't looked at many files from the 1Dx mk3, but the file shared in the other thread is all I need to know—it's good.

But my P6K is 6K and still much lighter on the system. I've got an older, but top spec late 2015 iMac 5K that I look at files on, when I'm away from my Mac Pro. It deals OK with the P6K, but the 5.5K crm is a slide show.

And Braw is more configurable AND can be trimmed to a smaller braw in Resolve, which is great for slimming down projects.
 
Meanwhile, Sony is going to swoop in and announce 2-3 new cameras including the A7SII successor 1-2 weeks before Canon makes their announcement
 
ahalpert said:
Meanwhile, Sony is going to swoop in and announce 2-3 new cameras including the A7SII successor 1-2 weeks before Canon makes their announcement
Click to expand...

At the same time, does it matter?

I know nothing about Sony’s cameras now and I will know nothing about their cameras after the launch.

I’m not sure there is much crosstalk between systems?
 
AndreeOnline said:
At the same time, does it matter?

I know nothing about Sony’s cameras now and I will know nothing about their cameras after the launch.

I’m not sure there is much crosstalk between systems?
Click to expand...

Does it matter to you/me or does it matter in general?

Certainly it matters to someone looking to invest in their 1st camera or even their first mirrorless camera.

What do you mean "crosstalk between systems"? Do you mean the capacity to adapt Canon lenses to Sony with good AF? Yes, it is limited (though not entirely useless). I own all EF mount lenses but most are manual focus, so for those lenses it's a moot point. (Tho i have 4 EF AF lenses, which isn't nothing.) I reckon that if I buy one Sony mount midrange zoom (say, the Tamron 28-75 f/2.8), I could cover 75%+ of the situations I would want to use autofocus and with admirable performance.

Right now I'm in the uncomfortable position of owning the FS7 as my A camera. I've never been a Sony fan but it's very popular in my sector (along with the C300mk2 which, sadly doesn't shoot 4k60p). I also have Panny S1, GH5, GH4, GH2, and Canon C300, 5Dmk3 and 60D.

If I buy another camera, it will be either to get higher rates and/or improve image quality.

If I got an R5 or a 1DXmk3, they would both seem like a B camera to an eventual C300iii or C500ii. That's one possibility but an expensive route and I'd have to see clearly that the money was there vs keeping the FS7. (Much as I'd love to shoot on c500ii). It's not that the R5 and 1DXmk3 won't have incredible image quality, it's just that the type of clients who know/care what I shoot on (which, granted, isn't everyone) will presumably still see those cameras as DSLRs and therefore an inferior hire, especially when their client sees what we're shooting with in person.

If I got Komodo, it's a bit of a lateral move and I'm not sure how marketable it will be to the type of clients who like the FS7. Same for blackmagic cameras, but they have a lesser reputation than RED.

So, the A7Siii is a natural B camera to the FS7. I could sell my S1 and buy it to gain autofocus + a better-known camera line + having a matching brand with the FS7 (even if the footage doesn't ACTUALLY match easily, which is probably the case given what I've read about difficulties matching FS7 and FX9, but nevertheless I think it would be marketable to have matching brand A/B cameras). I think the FS7 will remain marketable as the codecs and form factor are still current.

So, my heart is with Komodo. Or even BMPCC6K. But I also need to consider what's the best business move. But even so, if the Canons mop the IQ floor with Sony, I will migrate back to Canon bodies.

And there's an outside chance that the A7Siii will use quad Bayer architecture to do a DGO HDR implementation by having alternating high and low gain pixels. That could be cool as long as the whole thing doesn't deliver dismal rolling shutter.

So yeah, if I were more heavily invested in other, current Canon bodies or lenses, i suppose the Sony would be off my radar. But I'm straddling multiple systems and considering which direction to head in long-term.

Sorry for the rambling. I'm an external processor.
 
Last edited:
I'm with Andree on this one. A person could go their whole life never using specific brand/s and only sticking to their two or three favourites and never ever suffer for it. All of the brands stay relatively competitive, and even when they make a good system, sometimes they still disappear or get bought up/out. But someone could stick to say Panasonic their whole lives and always have the "best" gear. They may even have a better quality of life: no fomo. no learning new brand's ways of doing things. No panic purchases.

I still think all digital video kind of sucks. As far as replicating and capturing the world in a semi realistic way. Yes, even Arri sucks. And yet, all digital video cameras are also kind of amazing. I still enjoy the impressionistic paint brush that is the original Varicam. With care and craft, one can make realistic images with it, but it is definitely not a camera that one can carelessly expose with and get a realistic image. The DVX100? Even more so. If the Varicam is Manet's brush, than the DVX100 is Monet's. Absolutely no detail captured to tape, constantly asking for the ECU. (Although the sensor does actually have near 720p resolution, it is never captured to tape). But the idea is, that if you don't care what other people use, you can always get amazing results with the camera you have.

Glad the upgrade cycles are slower and bigger than before, from a business side of things, but I wish tech had moved faster, because now cameras are juuuust starting to get good, and we are all being forced out of the job. So, robots, AI, and "talent" get the best cameras. But considering how incremental these companies dole out product improvements, it really doesn't matter which one releases the new feature first, because they'll all be close enough to each other. One might as well save themselves the time and energy and just not worry about it all. Just go shoot, because soon, we may get paid less and less for doing it.

Carpe diem
 
Last edited:
Unclear re: paid less and less. Maybe some positions disappear and others persist.

But certainly I am wary of the value proposition of every piece of gear, how long it would take to pay for itself, and how much risk is involved in a longer payoff cycle. But could overly conservative thinking be detrimental?

In any case, my issue is that I'm trying to second-guess the minds of clients and which configuration of cameras I own will maximize my pay. I'm not particularly good at that, and I'd rather just pick my favorite cam to work with.

I was happy with my Scarlet but it was a lousy investment.

I would have stayed in Canon over Sony for the color science, but the FS7 has been marginally more hireable.

I was interested in the EVA1, but I'm not sure anyone else in NYC was. I hear that lots of folks have C200s but I've never seen one in the wild. I hate Sony colors, ergonomics, menus, but they're super popular.

Sure, every brand may be qualitatively competitive. But if your particular client pool isn't hiring it...therein lies the rub. And I'm trying to figure out which camera will make me the most money and perhaps realizing that it's not any of the cameras I really want.

I've never owned a Fuji or a Nikon or a BMD, but I'm aware of their offerings. One doesn't have to pay attention to every brand, but I think it's wise to be aware and flexible.
 
ahalpert said:
Unclear re: paid less and less. Maybe some positions disappear and others persist.
Click to expand...

I've talked to a couple ASC members, and they talk of the drop in prices as the talent pool expanded. Now we are seeing a huge talent pool and a lot of talk about jobs going away. So, not sure what is unclear about being paid less and less. Ok, sure, we might get the same day rate that has been standard for the last 10years, but those jobs are fewer than before, which is just another version of being paid less. And some may go away completely with current trends. Sure we could find a new way. But I don't see the need to act like it isn't happening and that it is all just because of a negative attitude.
 
ahalpert said:
Unclear re: paid less and less. Maybe some positions disappear and others persist.

But certainly I am wary of the value proposition of every piece of gear, how long it would take to pay for itself, and how much risk is involved in a longer payoff cycle. But could overly conservative thinking be detrimental?

In any case, my issue is that I'm trying to second-guess the minds of clients and which configuration of cameras I own will maximize my pay. I'm not particularly good at that, and I'd rather just pick my favorite cam to work with.

I was happy with my Scarlet but it was a lousy investment.

I would have stayed in Canon over Sony for the color science, but the FS7 has been marginally more hireable.

I was interested in the EVA1, but I'm not sure anyone else in NYC was. I hear that lots of folks have C200s but I've never seen one in the wild. I hate Sony colors, ergonomics, menus, but they're super popular.

Sure, every brand may be qualitatively competitive. But if your particular client pool isn't hiring it...therein lies the rub. And I'm trying to figure out which camera will make me the most money and perhaps realizing that it's not any of the cameras I really want.

I've never owned a Fuji or a Nikon or a BMD, but I'm aware of their offerings. One doesn't have to pay attention to every brand, but I think it's wise to be aware and flexible.
Click to expand...

Sincerely, I wish you could open your eyes about this camera thing because you're struggling.

I have worked in Manhattan my entire life for many people; we are in the same market and from first hand experience I know that so many people, companies do not care about which camera you own. Are you really that rare of an example that it matters for everything you shoot? Maybe.

But I truly believe you need to invest in a better business plan and find niches that will maximize your pay, not cameras.
 
You must log in or register to reply here.
Back
Top