Canon lays out their corporate strategy

[roby17269]

If there's little to no improvement over the existing EF 35mm f/1.4 L II, why would those who already own the EF lens be interested in an RF version? One would expect that the vast majority of potential interest in an RF 35/1.4 or 1.2 would from those who already own an EF 35 L?

The EF 35 L works just as well on RF bodies as it does on EF bodies.

That may be why we haven't seen it yet. The EF 35mm f/1.4 L II is only 8 years old.

We will see... the RF 50 and 85 1.2 primes are, IMHO, materially and significantly better than their EF predecessors (I had the EF lenses, I have the RF ones).
So the future RF 35mm L may be worth it. We do not know.

But for me, I do not have a EF 35 L. I had the mkI but did not like it on the R5, so I sold it - I never bought the EF 35 L II while I was shooting with the 1D X since in that timeframe I was "investing" in my HC system. And I am not buying the EF 35 L II at this point since I am not willing to put new money in an obsolete mount... Nor I am buying the RF 35mm 1.8 since I would eventually buy the RF 35 L and the non-L would need to be resold or kept gathering dust.

We will see. I hope that the 35mm, being a "mainstream" focal length, will be prioritized over other lenses, but that just me hoping

 

[koenkooi]

[...] All academic to me unfortunately, as I'm never likely to be able to afford an R1...

For me it's not only the price, I also find the 1 series style body impractical to pack and use outside of my garden. I very much like the ergonomics of my 1D, but it's just too much camera for me.

There are a few exceptions, I feel that the 180L can use a bit more body support, so I attach a battery grip from time to time. Like @neuroanatomist I don't like the extra bulge on the bottom.
The R3 looks like it might be small and light enough to be practical for me. If in a few years an RF180L gets released with a similar weight and center of mass, I would strongly consider switching from small body + R5 to small body + cheapest R with built-in grip (R3II/R4/whatever).

Given the RF pricing, I expect the RF180L to eat up a few years worth of toy budget, so it will very likely live on the good old R5

 

[entoman]

For me it's not only the price, I also find the 1 series style body impractical to pack and use outside of my garden. I very much like the ergonomics of my 1D, but it's just too much camera for me.

There are a few exceptions, I feel that the 180L can use a bit more body support, so I attach a battery grip from time to time. Like @neuroanatomist I don't like the extra bulge on the bottom.
The R3 looks like it might be small and light enough to be practical for me. If in a few years an RF180L gets released with a similar weight and center of mass, I would strongly consider switching from small body + R5 to small body + cheapest R with built-in grip (R3II/R4/whatever).

Given the RF pricing, I expect the RF180L to eat up a few years worth of toy budget, so it will very likely live on the good old R5

When I first pick up a 1Dx, it feels absolutely great - I love the huge grip and chunky, widely spaced controls. Unfortunately it's far too heavy to use for more than a few minutes at a time, and takes up a lot of my luggage allowance when flying.

I don't think the R1 will necessarily be a huge heavy camera. It could be the same size and weight as the R3, which is perfectly manageable and IMO has the best ergonomics of any Canon body. I'd love to have a R3-based camera with a 45/60MP sensor, as it would make an ideal birding/safari camera, and would have more cropping ability (effectively more "reach") than the R3 with the same lens. A big advantage for me would be that I could use shorter focal length lenses (cheaper, lighter, more manoeuvrable) than I'd need on a R3.

But a gripped body would be unsuitable for a lot of my work - for insects, landscape, botany and general travel photography the R5 makes a lot more sense for me, so the cheapest option would be to get an accessory grip for the limited occasions when it would be valuable.

I'd definitely be tempted by a RF 180L macro, but I'm finding now that the RF 100 macro and the RF 100-400 (which goes down to 1:2 scale) together cover all my needs. The RF 100-400 is extremely sharp at 1:2, and has very effective stabilisation.

I actually have a lot of duplication of focal lengths in my set-up, but it works well for me: for landscapes in the UK, when I can take a lot of time setting up a shot, I like my T/S-E 24mm. My walkabout/travel lens is the RF 24-105 F4L, which provides a huge amount of versatility. For smallish insects such as hemipterans, grasshoppers, beetles and caterpillars, I favour my RF 100mm macro. For birds and mammals from safari vehicles, the RF 100-500 is my go to lens. I also have the RF 800/11 which is usually the best choice for handheld bird photography (except BIF) if the light is good enough. For butterflies, by far my favourite lens is the RF 100-400mm. Here is a (cropped) shot I took yesterday at 400mm/F8/ISO 800:

 

[neuroanatomist]

What would be really nice, is if the R1 provided the following user-selectable options:

90MP RAW
45MP pixel-binned "RAW"
22.5MP pixel-binned "RAW"

Wish lists are great, but this is one where the technicalities matter.

The middle one could not pixel-binned RAW, even in quotes. It could be “RAW” like mRAW and sRAW. Binning two pixels would give distorted output, either 3:1 or 3:4 aspect ratio depending on the binning direction.

Even binning four pixels is complicated, because if you bin adjacent pixels you must demosaic at that time (i.e., it’s not RAW after binning), or use a quad-Bayer CFA. The latter is the approach taken by OM-1 and by Apple in the iPhone 14 Pro main camera (sensors made by Sony)



That’s a great approach when the main application is the binned output. The OM-1 is sold as a 20 MP camera, though it has 80 million photodiodes. Likewise the default iPhone 14 Pro main camera output is 12 MP (matching the 12 MP sensors in the other two cameras), it only outputs 48 MP in RAW.

The ‘problems’ with the quad-Bayer array are spatial and color resolution when outputting the full MP count. The four pixels of each color block (in Sony’s implementation) are under a single microlens, so the non-binned output is closer to the binned output in terms of real resolution – e.g., using the OM-1 in high resolution mode (80 MP output) gives a spatial resolution above 20 MP, but closer to that than to 80 MP. Color resolution of the full MP output is also lower than the standard CFA because each color block is further apart, meaning a greater magnitude of color interpolation.

OTOH, the OM-1 leverages those four pixels under one microlens in a similar manner to the two sub-pixels under one microlens used by Canon for DPAF, giving the OM-1 quad-pixel AF i.e. every pixel in the 20 MP array functions as a diagonal cross-type AF point.

I’m not sure how Canon will choose to implement binned output, if they do so at all…

Will they use a Sony-type solution with a quad-Bayer CFA and QPAF (they patented horizontal-vertical that differs from Sony’s diagonal) but functions best for binned output?

Or a true high MP sensor with one microlens per pixel, with either four subpixels or alternating orientations of dual subpixels for cross-type AF, and the binned output would be demosaiced and thus not RAW?

Or a middle-ground approach such as a quad-Bayer array with each pixel having its own microlens and alternating DPAF, enabling full MP spatial resolution while sacrificing some color resolution at full MP for a true RAW 4-pixel binned output?

Time will tell, but it’s important to keep in mind that, “90 MP RAW with 22.5 MP binned ’RAW’,” is more involved than simply combining blocks of four pixels.

 

[neuroanatomist]

When I first pick up a 1Dx, it feels absolutely great - I love the huge grip and chunky, widely spaced controls. Unfortunately it's far too heavy to use for more than a few minutes at a time, and takes up a lot of my luggage allowance when flying.

When I first picked up the R3, it felt like a toy compared to my 1D X mostly because it weighs 1/3 less (~1 kg vs. ~1.5 kg).

Put another way, the 1D X weighs about the same as two R5’s, but the R3 weighs only 1/3 more than the R5.

 

[entoman]

Wish lists are great, but this is one where the technicalities matter.

The middle one could not pixel-binned RAW, even in quotes. It could be “RAW” like mRAW and sRAW. Binning two pixels would give distorted output, either 3:1 or 3:4 aspect ratio depending on the binning direction.

Even binning four pixels is complicated, because if you bin adjacent pixels you must demosaic at that time (i.e., it’s not RAW after binning), or use a quad-Bayer CFA. The latter is the approach taken by OM-1 and by Apple in the iPhone 14 Pro main camera (sensors made by Sony)

View attachment 207980

That’s a great approach when the main application is the binned output. The OM-1 is sold as a 20 MP camera, though it has 80 million photodiodes. Likewise the default iPhone 14 Pro main camera output is 12 MP (matching the 12 MP sensors in the other two cameras), it only outputs 48 MP in RAW.

The ‘problems’ with the quad-Bayer array are spatial and color resolution when outputting the full MP count. The four pixels of each color block (in Sony’s implementation) are under a single microlens, so the non-binned output is closer to the binned output in terms of real resolution – e.g., using the OM-1 in high resolution mode (80 MP output) gives a spatial resolution above 20 MP, but closer to that than to 80 MP. Color resolution of the full MP output is also lower than the standard CFA because each color block is further apart, meaning a greater magnitude of color interpolation.

OTOH, the OM-1 leverages those four pixels under one microlens in a similar manner to the two sub-pixels under one microlens used by Canon for DPAF, giving the OM-1 quad-pixel AF i.e. every pixel in the 20 MP array functions as a diagonal cross-type AF point.

I’m not sure how Canon will choose to implement binned output, if they do so at all…

Will they use a Sony-type solution with a quad-Bayer CFA and QPAF (they patented horizontal-vertical that differs from Sony’s diagonal) but functions best for binned output?

Or a true high MP sensor with one microlens per pixel, with either four subpixels or alternating orientations of dual subpixels for cross-type AF, and the binned output would be demosaiced and thus not RAW?

Or a middle-ground approach such as a quad-Bayer array with each pixel having its own microlens and alternating DPAF, enabling full MP spatial resolution while sacrificing some color resolution for a true RAW 4-pixel binned output?

Time will tell, but it’s important to keep in mind that, “90 MP RAW with 22.5 pixel binned ’RAW’,” is more involved than simply combining blocks of four pixels.

Thanks for the explanation

I think the solution that would suit me best, would be the "true high MP sensor with one microlens per pixel, with either four subpixels or alternating orientations of dual subpixels for cross-type AF, and the binned output would be demosaiced and thus not RAW", as I wouldn't want to sacrifice colour resolution.

Whatever method Canon chooses to employ, if they launch a 90MP (ish) camera, I think they'd be highly advised to incorporate a low-res non-JPEG option, and an ability to instantly switch between (e.g. 90MP to 22.5MP).

 

[neuroanatomist]

Thanks for the explanation

I think the solution that would suit me best, would be the "true high MP sensor with one microlens per pixel, with either four subpixels or alternating orientations of dual subpixels for cross-type AF, and the binned output would be demosaiced and thus not RAW", as I wouldn't want to sacrifice colour resolution.

Whatever method Canon chooses to employ, if they launch a 90MP (ish) camera, I think they'd be highly advised to incorporate a low-res non-JPEG option, and an ability to instantly switch between (e.g. 90MP to 22.5MP).

Maybe “bRAW” for binned ‘RAW’? The problem IMO is that demosaiced “RAW” formats like mRAW and sRAW are not supported by 3rd party RAW converters (at least some e.g., mine – DxO) although the non-demosaiced cRAW is supported by DxO, for example.

 

[koenkooi]

Maybe “bRAW” for binned ‘RAW’? The problem IMO is that demosaiced “RAW” formats like mRAW and sRAW are not supported by 3rd party RAW converters (at least some e.g., mine – DxO) although the non-demosaiced cRAW is supported by DxO, for example.

Linear DNGs would be the obvious solution, those can even include extra metadata about colour profiles. This is what DxO will output when picking ‘optical corrections/denoise only’.

But that would make things ‘Just Work’ outside of DPP, which Canon (and most other brands) seem to dislike.

 

[entoman]

Maybe “bRAW” for binned ‘RAW’? The problem IMO is that demosaiced “RAW” formats like mRAW and sRAW are not supported by 3rd party RAW converters (at least some e.g., mine – DxO) although the non-demosaiced cRAW is supported by DxO, for example.

I don't call what they call it, as long as the colour and tonal gradation are not compromised.

Very interested to read about Panasonic's organic sensor tech, which promises much better colour accuracy and elimination of cross-talk.

Panasonic blogs about organic sensor's improved cross-talk, but doesn't talk photos

A blog post from Panasonic touts the ways its organic film CMOS sensor can control color spill between pixels, giving more accurate color in challenging lighting, but doesn't propose photography as something the sensor is suited for.

www.dpreview.com

 

[neuroanatomist]

Linear DNGs would be the obvious solution, those can even include extra metadata about colour profiles. This is what DxO will output when picking ‘optical corrections/denoise only’.

But that would make things ‘Just Work’ outside of DPP, which Canon (and most other brands) seem to dislike.

Indeed. Canon took a step forward with HEIF, so a step backward with a new, proprietary RAW format is probably next.

 

[koenkooi]

Indeed. Canon took a step forward with HEIF, so a step backward with a new, proprietary RAW format is probably next.

I also wish that outputting HEIFs was better supported in photo software. I haven't found a workflow yet that takes advantage of HEIC. The closest I can get is exporting PNGs from lightroom and using heic-convert to turn those into HEIFs. The issue with that is LR not supporting extended colour spaces and gamma curves, so the HEIF will still look like a JPEG on your phone. With better gradients, but not with better colours.

And of course Apple uses a low-res image embedded in the HEIF metadata for HDR mapping, which is specific to them. So getting a picture of a sunset to look as vivid as the one your phone took isn't possible yet, without writing all the software glue yourself. All the bits are there and open source, it just needs someone to write a wrapper for libheif that takes DNGs and outputs HEIFs.

Rant over

 

[Kit.]

The swipe bar of the Canon was criticized by many. Upon closer examination, almost all of those that complained about it, did not understand how it could be customized because they never read the manual. I customized mine to work just like a button.

Can you configure it to work just like a joystick?

 

[neuroanatomist]

Can you configure it to work just like a joystick?

A joystick constrained to one dimension, yes. So more of a moderately happy stick.

 

[David - Sydney]

Just a wild guess, but maybe it's Sony's 2-year model replacement cycle vs. Canon's 4-5 year model replacement cycle? A products expected lifespan might be defined by when the replacement model is typically expected?

No, it is based on the "reasonable" lifespan rather than product lifecycle. You could bring out a new fridge every year but a reasonable lifespan would be 5+ years and mine has been going for at least 10 years so far.

 

[mxwphoto]

If there's little to no improvement over the existing EF 35mm f/1.4 L II, why would those who already own the EF lens be interested in an RF version? One would expect that the vast majority of potential interest in an RF 35/1.4 or 1.2 would from those who already own an EF 35 L?

The EF 35 L works just as well on RF bodies as it does on EF bodies.

That may be why we haven't seen it yet. The EF 35mm f/1.4 L II is only 8 years old.

The ones clamoring for the RF 35L from what I see do not have the EF 35L II. The performance difference from EF ver I to II is night and day.

A RF 35mm 1.2L would be a Halo lens in the category, but if Sigma's anything to go by at 1.1kg, it will weigh like a brick.

 

[Del Paso]

We will see... the RF 50 and 85 1.2 primes are, IMHO, materially and significantly better than their EF predecessors (I had the EF lenses, I have the RF ones).
So the future RF 35mm L may be worth it. We do not know.

But for me, I do not have a EF 35 L. I had the mkI but did not like it on the R5, so I sold it - I never bought the EF 35 L II while I was shooting with the 1D X since in that timeframe I was "investing" in my HC system. And I am not buying the EF 35 L II at this point since I am not willing to put new money in an obsolete mount... Nor I am buying the RF 35mm 1.8 since I would eventually buy the RF 35 L and the non-L would need to be resold or kept gathering dust.

We will see. I hope that the 35mm, being a "mainstream" focal length, will be prioritized over other lenses, but that just me hoping

I just cannot imagine the 1,2/35 to be anything else than superb.
As you wrote, the 50 and 85mm f1,2 are immensely better than their predecessors. The 1,4/35 L II being a great lens already, it could be hard for Canon to improve it like they did with the EF 50 and 85. But I have not a single doubt the 1,2/35 will be an extraordinary lens, but $$$...
Fortunately, I don't need it, can't afford a divorce .

 

[dolina]

@entoman

We succeeded! DPReview is shutting down April 10!

 

[blackcoffee17]

Sounds a lot like the EF-S and EF-M selections, doesn't it?

EF-S at least had lenses like 17-55 2.8, 15-85, 10-22, 60mm macro, 24mm pancake. Canon couldn't even be bothered to make a 24mm equivalent kit zoom like every other manufacturer.

 

[neuroanatomist]

EF-S at least had lenses like 17-55 2.8, 15-85, 10-22, 60mm macro, 24mm pancake. Canon couldn't even be bothered to make a 24mm equivalent kit zoom like every other manufacturer.

You’re considering the EF-S and EF-M lineups from the perspective of a mount that’s been around for many years. The RF-S mount (using the term loosely) is less than a year old.

When EF-S was a year old, there was an 18-55, a 10-22 and a 17-85. When EF-M was a year old, there was an 18-55, a 22/2 and an 11-22.

RF-S now has an 18-45, 18-150 and a 55-210.

Patience, Grasshopper.

 

[roby17269]

I just cannot imagine the 1,2/35 to be anything else than superb.
As you wrote, the 50 and 85mm f1,2 are immensely better than their predecessors. The 1,4/35 L II being a great lens already, it could be hard for Canon to improve it like they did with the EF 50 and 85. But I have not a single doubt the 1,2/35 will be an extraordinary lens, but $$$...
Fortunately, I don't need it, can't afford a divorce .

Me too, that's why I have perfected the art of smuggling toys! Luckily for me most RF L prime lenses look pretty similar one to another apart from minor size differences... Otherwise the SWMBO wouldn't be best impressed.
I can tell you I'll order on the morning of the announcement day, that's for sure