Quad Pixel Autofocus is ready for production [CR1]

[Del Paso]

I suppose it is as usual:
In a lot of cases it is sufficient, and sometimes you wish you have a 100MP sensor for this one pic.
I think you can recognize the sensor performance with this and this dragonfly post of mine.

The main reasons for me not to buy an R6 II are:
R5's EVF is noticably better. (for me: very important)
Lack of upper LCD control panel. (I hate that!)
Hoping for the R5 II to have eye-controlled EVF.
The higher MP count? Since I'm not into BIF, I don't really care. 24 MP sensors in recent cameras are more than I need.

 

[Exploreshootshare]

The R6II you’re coveting has the exact same sensor as the R8, which is why I mentioned it, not to sway you towards the R8

Yes, I got that
I just mentioned it, because if the R8 had a joystick, I'd probably just get it. But after nearly five years of cramping my fingers sometimes, I really do want it badly

 

[Exploreshootshare]

The main reasons for me not tobuy an R6 II are:
R5's EVF is noticably better. (for me: very important)
Lack of upper LCD control panel. (I hate that!)
Hoping for the R5 II to have eye-controlled EVF.
The higher MP count? Since I'm not into BIF, I don't really care. 24 MP sensors in recent cameras are more than I need.

I´m with you on the upper LCD control panel. I love this thing on the R and I use very often. But is not a feature I couldn't live without or pay an extra premium for.

 

[Bonich]

Really? What was the point in going mirrorless then if not for the fantastic AF system?

The point going mirrorless is the superior chance using manual lenses:
- the user experience with TS-lenses is on another level compared to DSLR
- the chance to adapt M-mount lenses, FD-mount lenses, .... opens a universe of options using glass with very different rendering style
- the minimal size opens great options, may it be i.e. RF28 pancake, Leica 28mm Summaron, Voigtländer 28mm Ultron, if you want the 28mm Otus.

 

[Bonich]

You too?

Yes, me too

 

[frankchn]

That…and AF point-linked spot metering.

I am still slightly annoyed that this was not available in the R3

 

[simonella_virus]

Thx for the input. I looked at the R8, but after having used the R7 over the summer (which I now sold to my soon to be father-in-law) I absolutely fell in love with having a joystick! One of the two things seriously missing from the R. Therefore, R8 is not for me, but still a very impressive package.

I have both of these bodies (r8 and r7). I use the r8 for real estate and general use and the r7 for wildlife.

Let me tell you: the AF is do good on the r8 that i don’t use the (non-existent) joystick . Most of the time I’m in continuous af and spot and it just locks-on. Even with the r7 i find that i don’t even use the joystick.

 

[VegasCameraGuy]

Now the only decision will be to buy an R1 or a R5II. Assuming the R1 comes in above 45mp then I'll bite for the R1 as a low mp pro camera like the R3 doesn't move me in the least.

 

[Quarkcharmed]

Does anyone actually care about DR?

I do. QPAF may have some impact on read noise (and therefore DR), as well as readout speed.
In fact the modern sensors are already approaching the theoretical limits for the DR, but there's still room for improvement.

Will we see 15-bit or 16-bit ADCs with minimal read noise?

 

[snapshot]

Now the only decision will be to buy an R1 or a R5II. Assuming the R1 comes in above 45mp then I'll bite for the R1 as a low mp pro camera like the R3 doesn't move me in the least.

I suppose my decisions will be based upon resolution and sensor read out speed. If i can start taking tennis pictures with electronic shutter then the new camera can replace the R5 i am using with electronic front curtain. R3 didnt make it into my bag because I wasnt sure how reduced resolution would affect the distance/focal length required for eye detect auto focus to work.

 

[Loswr]

I do. QPAF may have some impact on read noise (and therefore DR), as well as readout speed.
In fact the modern sensors are already approaching the theoretical limits for the DR, but there's still room for improvement.

Will we see 15-bit or 16-bit ADCs with minimal read noise?

Sorry, my post was a facetious reference to the DRone wars of a decade ago, when Canon’s ‘poor low ISO DR’ (1-2 stops less than the Sony/Nikon cameras at the time) made them essentially unusable ‘for critical work’…at least, according to the aforementioned DRones.

That group notably included DPReview. I haven’t read their coverage of the a9 III, but I’d be surprised if they don’t give Sony a pass on the lower DR and higher base ISO. They’ll probably cite speed/action as the use case that excuses it. And no one will remind them that was a totally unsatisfactory excuse for the 1D X.

 

[snapshot]

Does anyone actually care about DR?

I am not sure I know what care means, but DR wasn't the reason I moved from 5D4 to R5. Nor was it why I moved from 5d2 to 5D4.

 

[Dragon]

So a 45MP QPAF sensor is actually a 180MP Quad Bayer sensor (think phone sensors). The left/right/up/down looking pixels might have a small impact, but I see no reason the sensor output could not be processed to produce the equivalent of a 180 MP 4:2:0 image quite easily and with some clever AI, you should be able to extract a full 180 MP image. I don't do smartphones, so haven't looked to see what Adobe is doing to support Quad Bayer phone images, but if the support is not there yet, it will be soon. Canon already has the option of storing full data from DPAF RAW files, so QPAF would be a small step. I, for one would find a 45/180 MP sensor quite handy.

 

[David - Sydney]

So a 45MP QPAF sensor is actually a 180MP Quad Bayer sensor (think phone sensors). The left/right/up/down looking pixels might have a small impact, but I see no reason the sensor output could not be processed to produce the equivalent of a 180 MP 4:2:0 image quite easily and with some clever AI, you should be able to extract a full 180 MP image. I don't do smartphones, so haven't looked to see what Adobe is doing to support Quad Bayer phone images, but if the support is not there yet, it will be soon. Canon already has the option of storing full data from DPAF RAW files, so QPAF would be a small step. I, for one would find a 45/180 MP sensor quite handy.

There was always a promise that DP raw images would be useful but I am yet to see anything. I wouldn't hold your breath for a 180mp file.
The 48mp promise for the Mini 3 Pro just takes up space and slows down write speed over the 12mp files.

 

[Quarkcharmed]

Sorry, my post was a facetious reference to the DRone wars of a decade ago, when Canon’s ‘poor low ISO DR’ (1-2 stops less than the Sony/Nikon cameras at the time) made them essentially unusable ‘for critical work’…at least, according to the aforementioned DRones.

That group notably included DPReview. I haven’t read their coverage of the a9 III, but I’d be surprised if they don’t give Sony a pass on the lower DR and higher base ISO. They’ll probably cite speed/action as the use case that excuses it. And no one will remind them that was a totally unsatisfactory excuse for the 1D X.

Yeah I actually understood that given the previous messages you were responding to, but I'm still concerned about the DR in the prospective R5II. Could it be an ultimate landscape camera?

Actually I'm also concerned about the dark currents/hot pixels which are an issue in my R5. Long exposures (1 second or more) are increasingly impacted by the hot pixels.

A lower dynamic range in the R5II would be disappointing, persistent hot pixel problem would be disappointing too. Maybe if it's both QPAF and BSI, it'll make things better.

It's all highly speculative at the moment.

 

[Jethro]

It's all highly speculative at the moment.

Absolutely - although if Canon are (per this rumour) actively discussing / presenting this with 'select dealers', then it suggests some sort of more general announcement must be coming sooner rather than later. That should mean that the camera specs are locked in, and it's therefore a matter of timing. The R1 was always going to be in the wild in some form pre-Olympics, and probably well pre-Olympics.

 

[Dragon]

There was always a promise that DP raw images would be useful but I am yet to see anything. I wouldn't hold your breath for a 180mp file.
The 48mp promise for the Mini 3 Pro just takes up space and slows down write speed over the 12mp files.

I suspect DJI just doesn't have the software perfected. The iPhone 15 looks pretty decent at "48" MP https://www.dpreview.com/sample-galleries/9434362346/iphone-15-pro-sample-gallery/5421721019 . The 48 MP files are huge because they are stored as .dng files, but that is not necessary with raw compression where Canon actually leads the market. The sensors for the Mini 3 Pro and the iPhone 15 are likely both made by Sony, likely from the same family, and could even be the same sensor (haven't researched the sizes). The point is that Quad Bayer can be made to work with the right software, so 180MP may not be a pipe dream.

 

[sanj]

Canon's in camera manual focus focusing aids would work better than they do with DPAF, if you use them.

How better? I mean focusing aids are very straight forward...

 

[Dragon]

How better? I mean focusing aids are very straight forward...

Just like with AF, the DPAF structure providing feedback for manual focus doesn't work well on mini-blinds unless you turn the camera sideways. a QPAF structure would give you response to horizontal structures as well as vertical and if you are looking a random items like pebbles, you will get twice the level of response in the EVF, so, yes the MF aids should work better.

 

[sanj]

Just like with AF, the DPAF structure providing feedback for manual focus doesn't work well on mini-blinds unless you turn the camera sideways. a QPAF structure would give you response to horizontal structures as well as vertical and if you are looking a random items like pebbles, you will get twice the level of response in the EVF, so, yes the MF aids should work better.

Thank you.