Here We Go Again, More EOS R3 Mark II Chatter

[Canon Rumors]

Over the last few days there have been all sorts of new claims about an upcoming Canon EOS R3 Mark II. All of this started with images that Olympic photographer Jeff Cable posted from inside the Canon CPS room. People do like seeing hundreds of thousands of dollars in lenses and cameras stacked together. There […]

See full article...

 

[AUGS]

Edit: Current R1 firmware is 1.2.0, so I think simply FV120 = Firmware version 1.2.0.
Labelling a specific FW version (especially new) would make sense in this environment.

 

[Canon Rumors]

Edit: Current R1 firmware is 1.2.0, so I think simply FV120 = Firmware version 1.2.0.
Labelling a specific FW version (especially new) would make sense in this environment.

I hadn't thought of that! Nice. Edit incoming.

 

[EthenJ]

...The new machine supports switching between the two native resolutions of 54 million pixels and 24 million pixels, respectively, to achieve 40 shots per second And high-speed continuous shooting of 90 shots per second. It uses an enhanced Bayer array to pass through adjacent images in 24 million pixel mode! The combination of elements increases the sensitivity by about 80% compared to EOS R3...

Whoa, what's Canon doing here? What exactly is this "enhanced Bayer array"?

It doesn't seem to be quad-Bayer like a smartphone. 54MP -> 24MP is a conspicuous 2.25X reduction, not 4X.

They could be doing it in software... there's actually a lot of unpicked "low hanging fruit" in software downscaling, like the dpid algorithm that I absolutely adore for keeping sharpness: https://github.com/mergian/dpid

But they're wording it like it's done on the sensor level? Especially with the sensitivity claims?

I can't wait for some more analysis on this.

 

[rukes]

I hadn't thought of that! Nice. Edit incoming.

It does actually say FW, not FV. The left part of the W is squished against the middle, just like the “IV” in “III” so the same person who wrote that did that also. Probably put the tape on the cabinet then awkwardly wrote on them at an angle.

 

[neuroanatomist]

Whoa, what's Canon doing here? What exactly is this "enhanced Bayer array"?
It doesn't seem to be quad-Bayer like a smartphone. 54MP -> 24MP is a conspicuous 2.25X reduction, not 4X.
I can't wait for some more analysis on this.

The analysis is in. The person that posted that rumor that was passed along to CR was smokin' the ganja hard.

 

[roby17269]

But those Leicas have a 60mp (ish) sensor... that's their native resolution. The other modes are either binning or cropping, not "native" as I would call it.
AFAIK, IMHO, etc.

 

[DhlcadR6]

I don't think R3ii is available in CPS booth, but dual native resolution that isn't cropping is a welcoming move.

 

[Flyingskiguy]

the pixel math at least makes sense. Downscaling the 3x3 pixel array to a 2x2 array is a 2.25x reduction, which happens to multiply by 24 to make 54.

as for 6k120 - that is technically a very similar data rate to 8k60, so should be possible on the current CFexpress implementation.

 

[Traveler]

24 MP is the exact APS-C crop resolution of a 54 MP full-frame sensor. However, Canon uses a 1.6x crop factor. The 54 -> 24 MP reduction corresponds to a 1.5x crop factor.
So if it was not Canon, I'd be pretty sure that someone saw a cropped version of the image.

I don't think dual resolution makes much sense. It would have to be pixel-binned (something like 80/20 MP, four neighboring pixels would share the same color in the Bayer mask). In that case, I don't think it would bring much of an increase in detail in the 80 MP image.
If the binned pixels don't share the same microlens, then you'd still have to store data from all the pixels to recreate the color information.

 

[EthenJ]

24 MP is the exact APS-C crop resolution of a 54 MP full-frame sensor. However, Canon uses a 1.6x crop factor. The 54 -> 24 MP reduction corresponds to a 1.5x crop factor.
So if it was not Canon, I'd be pretty sure that someone saw a cropped version of the image.

I don't think dual resolution makes much sense. It would have to be pixel-binned (something like 80/20 MP, four neighboring pixels would share the same color in the Bayer mask). In that case, I don't think it would bring much of an increase in detail in the 80 MP image.
If the binned pixels don't share the same microlens, then you'd still have to store data from all the pixels to recreate the color information.

If this isn't totally made up, they could be using something more complex than the quad-bayer array like smartphones normally use for pixel binning.

The simplest scheme that fits the rumor would be some kind of 3x3 pixel array that could be sampled as either 9 pixels or 4 pixels, lightly software corrected for the full 9-pixel resolution. But I'm not qualified to even guess what they would look like.

If that's the case, all our existing editing software will need updates to process RAWs. We might be able to get hints by peeking at changelogs.

 

[EthenJ]

Okay, so I've been thinking about what this array might look like. One thought is:



Reading that as an array of 3x3 pixels would be normal: RGGB.

Reading it as a 2x2, would only be slightly uneven: 2R 5G 2B, 3R 4G 2B, 3B 2R 4G, and 2R 5G 2B. Still more green in every pixel, as it should be. This would be similar:



Those big blocks of 2 green, in particular, could be binned as one more sensitive pixel each.

Alternatively, it's possible that Canon is doing something funky like sampling adjacent pixels or using the autofocus individually diodes in "low resolution" mode, to hit that massive ISO gain figure.

...I don't really know about this stuff, I'm just spitballing. But it's an interesting idea.

 

[John Wilde]

"I haven't talked to anyone that thinks there will ever be a Mark II of the R3."

In a nutshell.

 

[neuroanatomist]

"I haven't talked to anyone that thinks there will ever be a Mark II of the R3."

Except for the people here on this forum who think there will be one. They can't agree on what it will be, but they are sure we'll see it.

Actually, in one way they can agree on what it will be...the camera they personally want. It's just that they all want something different.

 

[ShowMeTheEagles]

My take, Canon is trolling us, knowing someone would take a pic, but just using an R1 casing for testing out features for future cameras in general, not the R3 II specifically. No way the R3 is going to have all those features and cannibalize the bodies above and below it. I'm really dubious of any R3 II though.

 

[Quack]

No way the R3 is going to have all those features and cannibalize the bodies above and below it. I'm really dubious of any R3 II though.

Assuming the R3 II ever comes into being, it will be a high-end camera at least as expansive as the R5 II, so Canon will not care one bit if it "cannibalizes" R5 II sales. It can't in actuality, if one means that Canon will lose money. Nor do I think Canon worries if R3 II sales take away from R1 sales. It's all camera sales and money in Canon's pockets, so as long as it''s a Canon being sold, there is no cannibalization. Why people keep thinking this is a "thing' is beyond me. What it means in reality is that at least some people will buy an extra camera that they might not have before.

 

[Slappy_G]

Okay, so I've been thinking about what this array might look like. One thought is:

View attachment 228022

Reading that as an array of 3x3 pixels would be normal: RGGB.

Alternatively, it's possible that Canon is doing something funky like sampling adjacent pixels or using the autofocus individually diodes in "low resolution" mode, to hit that massive ISO gain figure.

...I don't really know about this stuff, I'm just spitballing. But it's an interesting idea.

It's an interesting idea, but heterogenous color layouts like this will almost certainly lead to color aliasing and moire effects. This would be a pretty bad solution, while admittedly academically interesting. I think whoever is claiming multiple "native" resolutions just doesn't understand what they are saying. We have had downscaled Raws and crops for decades now. That's likely all this was.

 

[zardoz]

The firmware versions make sense - I've had some weird camera reset issues where my camera will freeze and reboot ever since I've updated to 1.20. I wonder if they've got cameras on older firmware for that very reason....

As for the R3 II - my #1 gripe about the R1 is that the sensor has less latitude at high ISOs than my 1DX Mark II from ten years ago, so I'm still pulling out the EF body in low light situations where I know I'll be at 12800 and need to make ANY adjustments to exposure or shadows (including vignetting correction). I preordered the R1, assuming I'd get the same results as the R3 I evaluated from Canon, and was bitterly disappointed when I got my shots home and couldn't get the results I was able to get with the R3.

So I'll absolutely jump on the R3 II if I can get an improvement in low light performance that's equal to that of the R3 with the AF performance of the R1. Don't need to change anything else; just shut up and take my money. If it had a high resolution mode, then I get to sell my R5 which makes me even happier.

Cautiously optimistic that I may have a solution moving forward.

 

[Spikeman512]

Re: R3 Mark II.

I was curious about the average # of days from a Mark I announcement to a Mark II announcement for a variety of Canon cameras. This is based on the announcement dates of each camera...culled from putting the squeeze on cgpt.

EOS 5D — 1,122 days
EOS 6D — 1,746 days
EOS 7D — 1,840 days
EOS R5 — 1,469 days

EOS R3 Mark I (days since announcement) — 1,621 days to 2/23/26.

There seems to be modest time left, yet, at this point, the strongest evidence for an R3 Mark II is a blank cell in my complex spreadsheet and my optimism — and those are each questionable.

 

[Surab]

Since we are having fun speculating about how such a dual resolution idea could come to fruition: My first instinct was something akin to Foveon, in particular Quattro Foveon for which some of the color layers had less resolution.

Anyway, just a fun idea and that's it.