Canon has a global shutter camera in the pipeline

The a9 III also has dynamic range issues at high ISO where it is most important.
I am not sure why so many people fail to mention this.
The A9III is fantastic for video. Perhaps Sony’s best sensor. The exposure latitude matches the Red V-Raptor-W. The only thing that lets it down is the less robust internal XAVC-S codec. That said, it’s all most need. The external raw removes the limitations. FYI: I own both the R1 and A9III.
 
  • Wow
Reactions: 1 user
Upvote 0
in the world of sub 3ms readout speeds, I am trying to wrap my head around where GS makes sense.

The answer I am coming too:
High readout/High MP/fast action.

Thinking though the math the R1 is reading its sensor at 2.7 ms: 24MB/0.0027 sec =8,889 MB/sec. The R5 II: 45MB/0.0063 sec=7,142 MB/sec. Likely higher as I am using 1:1 MP to MB here.

Say Canon has a 60 MP sensor where they want sub 3 ms readout speeds? That would be 60 MP/0.003 sec = 20,000 MB/sec. Perhaps beyond the next generation of Digic.

Or, as these decisions usually boil down to cost, getting a digic/other readout hardware to do ~20,000 MB/sec is more costly than a global shutter.

All speculation, but this fits for a high resolution R3 or R7. With R3 it makes some sense to me as the R1 would be ultimate IQ and R3 would take an IQ hit but give higher MP. I’ll happily take ultimate IQ.
As I own both I can tell you, the A9III seems to work in almost all lighting conditions without flicker. It works exceptionally well with LED walls, with no artifacting. The R1 works in most conditions, though cheap LED lights will cause flicker and LED walls must be shot in Raw to avoid artifacts. The global shutter makes these otherwise challenging conditions a non-issue.
 
Upvote 0
You shouldn\'t see banding like that at anything at or below the flash sync speed of whatever shutter mode you\'re using.
Global shutter certainly is one way to eliminate that banding, the other is to just keep making the electronic shutter readout speed faster, until it\'s \'fast enough\' for most of us. R1 has an ES flash sync speed of 1/320, so you would have been almost there.
 
  • Like
Reactions: 1 user
Upvote 0
I’ve posted this elsewhere, but global and leaf shutter cameras are the only cameras that display no artifacting when shooting scenes with PWM-dimmable theatrical LED lights when using high enough shutter speeds to freeze motion.

For my use case, I don’t particularly care about the high fps that GS can achieve, but shooting with worrying about these kinds of banding is nice.

View attachment 221250

View attachment 221249
Great pics for comparison :) Thx for sharing!

I do actually like the first pic better because it has warm light and the colors. I don´t know if this is due to the cameras or the lighting on stage.
 
Upvote 0
Great pics for comparison :) Thx for sharing!

I do actually like the first pic better because it has warm light and the colors. I don´t know if this is due to the cameras or the lighting on stage.
The photos are unedited RAWs screenshotted straight out of LR, but the R3 did choose a warmer white balance for this scene than the A9III. That said, I would caution against making any generalized IQ comparisons based on that alone.
 
  • Like
Reactions: 1 user
Upvote 0
It was on ES in this case, but I’ve had the same issue in specific lighting situations even with DSLRs like the 1DX2 (and that certainly has a mechanical shutter).

The only fix I found is really to slow the shutter speed all the way down but then you can’t freeze motion especially in dance, so it really is a case of pick your poison.

I also like ES for theater and dance for silence reasons (it can be distracting for some of the student actors I work with) so a GS camera would “solve” all my problems.
You may have shot the 1DX2 in electronic shutter mode as well ...
 
Upvote 0
Thinking though the math the R1 is reading its sensor at 2.7 ms: 24MB/0.0027 sec =8,889 MB/sec. The R5 II: 45MB/0.0063 sec=7,142 MB/sec. Likely higher as I am using 1:1 MP to MB here.
Each pixel is at least 14 bits of resolution (to provide the DR) so the result is 15.6GB/sec (24E6 pixel * 14bit/pixel / 2.7msec). However, that's after the A/D converter. Most likely they have multiple ADCs converting in parallel and a wide databus (say 16ADC and a 16*14bit =224bit databus, but now they have 1/40sec - 2.7msec = 22.3msec to store the entire image in memory, so a ~67MHz conversion rate and databus speed would suffice).
 
  • Like
Reactions: 1 user
Upvote 0
I don't think the 1DX2 even had an electronic shutter or even EFCS mode.
It didn’t. Canon called it ‘silent shutter’ even for live view, but with the advent of electronic shutters (the 1D X III was the first 1-series with it), Canon said on DSLRs like the 1D X II it would be better referred to as ‘soft shutter’.
 
  • Like
Reactions: 1 user
Upvote 0
Going for APS-C sized sensors first may also be a yield issue in the fab ie smaller sensors are more likely to have a higher yield during the manufacturing process vs full frame. A R7ii (or R7gs) at a higher pricepoint than the R7 could be possible.

Canon have history bringing out new features in lower level cameras first but most of that is gone with regular firmware updates assuming the hardware doesn't need to change.
 
Upvote 0
Here is an example with the 1DX2 + 70-200, I am not sure what you would classify this as:

View attachment 221253
.
I have similar issues in some indoor sports locations with ES on my R5. No way to fix it in post. The lighting shouldn't be dimmable though.
Going to anti-flicker slowed down the shooting speed way too much so I stick to mechanical now.
 
Upvote 0
Huh, this would jive with the "R7 is going upmarket" wouldn't it?

I feel like an APS-C global shutter would be a perfect compromise to get started with the tech, since no one is choosing APS-C for "image quality" in comparison to full frame models. That way, the R1 still is the "highest-speed, highest-quality" option, even if the R7II outpaces it with a global shutter.

The 70d also introduced DPAF back in the day, so I feel like introducing a global shutter in that market is on par with the advent of DPAF, which really changed the game in terms of AF.
 
  • Like
Reactions: 1 users
Upvote 0
Each pixel is at least 14 bits of resolution (to provide the DR) so the result is 15.6GB/sec (24E6 pixel * 14bit/pixel / 2.7msec). However, that's after the A/D converter. Most likely they have multiple ADCs converting in parallel and a wide databus (say 16ADC and a 16*14bit =224bit databus, but now they have 1/40sec - 2.7msec = 22.3msec to store the entire image in memory, so a ~67MHz conversion rate and databus speed would suffice).
Fair enough. I'll stand corrected on that one. 8 bits to a byte. 24MP * 14 bit/pixel /8 bits/byte / 2.7 msec.... 15.6 GB/sec coming off the R1 sensor. But, the proportions I think hold comparing larger MP if you want fast readout speeds.

As for the 22.3 msec to store the entire image...do you think the process might be a bit faster including AF between images? And, how many times does it AF between images. This is the stuff I wonder about.

I'll add LCD flickering to my definition of "fast action." :) Another valid point.

I am mulling the benefits of Global shutters vs rolling shutter with 2.7 ms readout speeds. Ok, LCD flickering and flash sync speed. But even global shutters are going to hit a bottleneck at some point, Demosaicing? A/D converters? Perhaps a GS is not that expensive compared to a rolling shutter so it is just a matter of time for Canon to work out the IQ?
 
Last edited:
Upvote 0