Are 400mp stills coming to the Canon EOS R5?

[entoman]

Art photography would be a market.

My sister is a painter. Every so often, she has a pro photographer shoot her paintings. Static flat object, tripod, controlled lighting, and the time it takes the camera to take the extra photos is negligible. If the firmware would allow moving the sensor to align photo-sites of different colors to get better color reproduction, that makes the R5 that much more attractive compared to Sigma Foveon and medium format cameras.

It would make the same 45MP photos, no need for better lenses. It would produce four times the amount of photos, but would save time on color interpolation and handling moire, so it wouldn't necessarily take 4 times as long to process the images.

And, finally, target audience is probably small, but the price of firmware development is probably low enough for the extra sales to make a profit.

Agreed - one possible application is artwork digitalisation. Another is product photography e.g. automotive. And yes, we've discussed here recently the option to merge pixel-shifted shots to produce a 45MP image with better colour. But these are very much niche usages, and as such would have virtually no impact on sales. The reality is that pixel-shift, (just like 8K) for 99% of users (OK I'm guessing), is just a marketing tool, something that they'll probably never use, but which makes them feel better just because it's there.

 

[entoman]

Canon R5
Sigma 150-600C with adapter.. that one is at 293mm
1/2000s F5.6 IS800 EV0 (note should use more like F8 as the more you zoom in the narrower the depth of field.. its fine to shoot at higher ISOs)
this was handheld, panning.

For menu settings some random ones:
RAW only
Mechanical shutter (though fine with electronic, but it can warp when panning)
Image review: off
Auto ISO max 12800 limit
Eye Detection: Enabled
Eco mode: off

AF Operation: SERVO AF
AF Method: AF [[]]
Subject: Animals
Eye Detection: Enable
Continuous AF: disable
Focus Guide: On
AF assist beam: off
SERVO AF mode: CASE 2 with tracking sensitivity set to lowest value (-2) and Accel/decel set to max (+2)
Switching Tracked Subjects: 0 (vs default of 1, locks on better)
Lens Drive when AF unavail: off
Initial Servo AF pt for [[]]: Initial AF pt set for [[]] (small square centred)

Those are almost exactly the settings I use myself, saved as a custom mode for animals and birds. I've also got my M-fn button mapped to switch between micro-spot AF, full area animal-eye AF or 9-spot surround AF. I generally use a RF100-500mm, and usually at maximum aperture. Shutter speed typically around 1/3000 or 1/4000 for BIF.

 

[usern4cr]

What would this feature be used for? Is it for printing large billboards of static subjects? I wonder how many Mega Bytes the RAW files will be. Since Canon are adding new features the R5 II must be quite some time away yet.

Who said it will be a raw file? I'd bet it's a jpg file.
I doubt I'll be able to just drop the huge file into PhotoLab for raw processing. But we'll see.

 

[Jethro]

Any views on why there would need to be 9 images (equating to 400MP)? Is that somehow necessary to achieve the 'best' result from this sort of process, rather than a smaller number (2, 3, 4) which would result in a more manageable sized final image, and presumably would allow more 'composite' images to be taken allowing for FPS and whatever buffering effects result in-camera? If it was (say) 3 images, then that would strike me as increasing the possible use scenarios to hand-held shots of moving objects.

There has been talk for a number of years of higher MP sensors being introduced (eg c. 100MP), so query whether this is seen as an alternative to such a niche body.

 

[neuroanatomist]

Any views on why there would need to be 9 images (equating to 400MP)? Is that somehow necessary to achieve the 'best' result from this sort of process, rather than a smaller number (2, 3, 4) which would result in a more manageable sized final image, and presumably would allow more 'composite' images to be taken allowing for FPS and whatever buffering effects result in-camera? If it was (say) 3 images, then that would strike me as increasing the possible use scenarios to hand-held shots of moving objects.

Four 1/2-pixel moves would be the minimum to keep the same 3:2 aspect ratio. Canon has chosen nine 1/3-pixel moves for a higher MP output.

I’ve previously posted that I suspect even the 2x2 array, though it yields four times the MP, does not deliver four times the resolution. The reason is that gapless microlenses already collect light from the full spatial area projected into each pixel. Many years ago, my Zeiss cameras would capture 2x2 or 3x3 sub-pixel arrays with their 1.3 MP CCD sensors that did not have gapless microlenses, and where only about 1/3 of the pixel surface was photosensitive. That means the resulting 5 and 12 MP images were real resolution increases.

IMO, a 2x2 sub-pixel sampling array with a modern sensor will have a significant proportion of empty resolution, and Canon’s 3x3 sub-pixel sampling array will be mostly empty resolution.

I look forward to seeing some detailed testing.

There has been talk for a number of years of higher MP sensors being introduced (eg c. 100MP), so query whether this is seen as an alternative to such a niche body.

Given the requirement for a static subject, I’d say no.

 

[Czardoom]

Not saying the image from Canon EU is fake but if the 9 images = 400 MP image is correct, than Canon is doing things differently than anyone else. "Pixel" shifts that result in higher MPs use a shift that moves the sensor the distance of 1/2 of a photosite. Some use shifts that move the sensor i photosite width which results in no higher MPs. For those cameras using 1/2 photosite shifts:

Fuji cameras take 16 images, but the files are only 4 times larger than the sensor resolution.
Olympus cameras take 8 images, files are 4 times larger.
Panasonic cameras take 8 images, files are 4 times larger.
Sony cameras take 16 images, files are 4 times larger.

In all cases the number of final MPs is never the original sensor MP multiplied by the number of images. So 45 MP x 9 = 405 MP (or 400 MP) does not compute. Unless they are doing things totally differently, I wouldn't expect more than 180 MP - which , in my opinion, is a much more realistic and USABLE number.

Of course, this could be totally off base, but throwing it out there anyway!

 

[neuroanatomist]

Not saying the image from Canon EU is fake but if the 9 images = 400 MP image is correct, than Canon is doing things differently than anyone else. "Pixel" shifts that result in higher MPs use a shift that moves the sensor the distance of 1/2 of a photosite. Some use shifts that move the sensor i photosite width which results in no higher MPs. For those cameras using 1/2 photosite shifts:

Fuji cameras take 16 images, but the files are only 4 times larger than the sensor resolution.
Olympus cameras take 8 images, files are 4 times larger.
Panasonic cameras take 8 images, files are 4 times larger.
Sony cameras take 16 images, files are 4 times larger.

In all cases the number of final MPs is never the original sensor MP multiplied by the number of images. So 45 MP x 9 = 405 MP (or 400 MP) does not compute. Unless they are doing things totally differently, I wouldn't expect more than 180 MP - which , in my opinion, is a much more realistic and USABLE number.

Of course, this could be totally off base, but throwing it out there anyway!

The math is fine, it’s all about the sampling method chosen.

The others are using a combination of 1/2 sub-pixel shifts (to increase apparent spatial resolution, although likely not as much real resolution as described above) and full pixel shifts to increase color resolution by eliminating the need for interpolation of the Bayer color filter array. Full pixel shifts are simply merged and do not increase the MP count.

Combining four 1/2-pixel shifts with four full-pixel shifts is 16 captures and four times the native MP.

Combining four 1/2-pixel shifts with two full-pixel shifts is 8 captures and four times the native MP (with some color interpolation still needed, but less than with a single capture because each pixel space is captured in two of the three colors).

Canon is apparently using 1/3 sub-pixel shifts without any full pixel shifts to yield a 9-fold larger MP count.

FWIW, the Zeiss cameras I referred to above could also combine sub- and full-pixel shifts for increased resolution and elimination of color interpolation. The 3x3 array of 1/3-pixel shifts combined with full pixel shifts meant 27 captures for a single image (only three full-pixel shifts are actually needed with a Bayer array, one sample each for blue, red and one of the two green lenses).

 

[unfocused]

scratching my head on a couple of posts about autofocus improvements.. the autofocus on the R5 is miles ahead of my previous SLRs.

Tests by many reviewers have it on par top tier Sony and Nikon..

maybe there is something else at play like the settings you use or the lens you use or the conditions you use it in? Maybe try higher shutter speed? I was initially frustrated with some of the behaviours on my Sigma 150-600 on the R5 .. but you have to remember the more you zoom and the more you move the more things get blurry esp when you pixel peep a high megapixel camera.. you need to up your shutter speed quite a bit. Autofocus is also slower on 3rd party EF lenses adapted (though still work for me). High megapixel cameras are less forgiving when you make errors, hence all the tools to improve your odds (IBIS, IS, AI autofocus etc)

my real world experience has not had any issues with autofocus with fast moving animals, people, jets, cars etc.

<edit> adding what is normally a very difficult thing to keep in a focus.. zoomed in (highly cropped) bird in flight in dense branches.. yet check out the eyes, in focus.

That's a nice shot. However, in my experience the problem I have had with both the R5 and R3 is getting the initial focus when the bird is sitting amongst branches or in otherwise busy surroundings, such as grasslands. This is especially problematic if there is a lot of foreground grass and if it is a small bird. For example, a sparrow or goldfinch half-hidden (as they often do) in prairie grass. For large birds, especially if the eye is in the frame, it's usually not a problem, but with smaller birds it can be an issue. Also, interestingly, movement can help the camera lock on, such as in the case you are showing where the Eagle is in flight.

It's not impossible by any means and we all have a good selection of shots in focus. But, there are times when the cameras just cannot seem to "find" the bird and end up "hunting" until the bird has moved on. There is also the frequently frustrating instance where the camera locks on to the bird's eye, but then loses the focus a second later (for example if you attempt to reframe the image for a better composition.) This is a known issue with mirrorless cameras and has been discussed frequently on this forum. If memory serves me correctly, some on this forum have written that it is due to the nature of how dual-pixel autofocus finds a subject (Contrast Detection vs. Phase Detection) and it will require quad-pixel autofocus to really address the problem, but I am no expert and am only repeating what others have written on this forum.

You are correct that in most instances, the mirrorless cameras have a much higher success rate than DSLRs, but there are exceptions.

 

[unfocused]

A few semi-random thoughts.

• I doubt if I will have much use for this, but I appreciate Canon adding a major feature through firmware on a three-year-old model (If they do so.)
• I expect that they will add it to the R3 as well.
• I don't see any reason why they would add it to the Cinema R5. Cinema is about motion. The blur that results from shooting at 24 or 30 frames per second at 1/60 second or similar shutter speeds helps blend the frame together so that the eye seems a smoother image. I'd be interested if someone who shoots video has a reason why they would find this feature desirable.
• I see this as a sign that the R5II is at least a year away.
• As for applications, I recall Canon giving Martin Parr a 5Ds a few years back to take life-size portraits of people and, of course, there is always the example of Andreas Gursky and his use of digitized high resolution images shot in large formats. We also have the longstanding example of GigaPixel images that stitch separate images together into a single super high resolution photo. So, I imagine creative people will find new ways to use this. It may be like Canon's dual fisheye lens. I think sometimes Canon brings products and features to the market to challenge creators and see how they will be used.

 

[neuroanatomist]

I expect that they will add it to the R3 as well.

I hope so.

I don't see any reason why they would add it to the Cinema R5. Cinema is about motion.

Since the feature uses the IBIS system to shift the sensor, and the R5C lacks IBIS, I don’t see how it would be possible to add it to that camera regardless of the rationale.

 

[snappy604]

That's a nice shot. However, in my experience the problem I have had with both the R5 and R3 is getting the initial focus when the bird is sitting amongst branches or in otherwise busy surroundings, such as grasslands. This is especially problematic if there is a lot of foreground grass and if it is a small bird. For example, a sparrow or goldfinch half-hidden (as they often do) in prairie grass. For large birds, especially if the eye is in the frame, it's usually not a problem, but with smaller birds it can be an issue. Also, interestingly, movement can help the camera lock on, such as in the case you are showing where the Eagle is in flight.

It's not impossible by any means and we all have a good selection of shots in focus. But, there are times when the cameras just cannot seem to "find" the bird and end up "hunting" until the bird has moved on. There is also the frequently frustrating instance where the camera locks on to the bird's eye, but then loses the focus a second later (for example if you attempt to reframe the image for a better composition.) This is a known issue with mirrorless cameras and has been discussed frequently on this forum. If memory serves me correctly, some on this forum have written that it is due to the nature of how dual-pixel autofocus finds a subject (Contrast Detection vs. Phase Detection) and it will require quad-pixel autofocus to really address the problem, but I am no expert and am only repeating what others have written on this forum.

You are correct that in most instances, the mirrorless cameras have a much higher success rate than DSLRs, but there are exceptions.

agree and in some oddball cases the older tech is better.

I imagine they will continue to improve this. In meantime for busy stuff, it's better to use spot focus to get the lock, but don't think it will track (haven't tried). It would be cool if the eye tracking focus could start with the spot focus as it is a smaller area and more accurate usually.

mostly was puzzled why several folks piped up that it was more important than this feature as if there was a major problem. I think this is a bonus and the continued improvements in focus in past releases has been a bonus.. I don't have many cases where the focus is an issue, but rveryone's experience is bound to be a bit different.

Personally I was stoked when someone highlighted other features I was pining about from Magic Lantern and wasn't aware the R5 had... focus bracketing! Was using it today with some nice results. Managed to do many 14 picture focus bracket shots hand held today with good results! DPP even has a built in focus stacking tool that works quite well! so.. I like to dabble in so many areas so I find these bonuses great!

 

[sanj]

Most likely JPEG

 

[koenkooi]

Most likely JPEG

The Canon website stated is was jpeg, so I don’t know why there is still discussion on it.

 

[another_mikey]

I for one will welcome this feature. I also appreciate Canon putting these kind of updates into a camera already on the market for this period of time. And this super resolution mode is just one of the updates from everything I have heard, so it should in no way preclude the possibility of many of the other firmware feature additions mentioned above, many of which have been reported to also be upcoming for the R5.

 

[Quarkcharmed]

The Canon website stated is was jpeg, so I don’t know why there is still discussion on it.

If it's jpeg only, it's almost useless (at least for me)
However it can be similar to HDR, where you get out of camera jpeg but also the source raw files that can be merged in DPP.

 

[Birdshooter]

I'm the first to admit that I have difficulties in this area.

I've experimented a great deal with AF settings on my R5 - I've tried a myriad of combinations of "cases", AF zones and other AF-related settings. The problem I have is nothing to do with shutter speeds, it's simply that the camera often fails to lock on to birds in flight, even when they are quite large in the frame, and often jumps to trees etc in the background.

With slow-moving birds it's easy to get the eye-AF to lock on and track, even when the bird is quite distant, if the background is uncluttered. But when the background is fussy and close behind the subject, I'm not having much sucess.

So how about sharing all the AF settings that you used on the shot you posted?

The exact reason for my buying an R3, as the R5 fails miserably at times to lock on fast birds in flight. (NOT big slow eagles) lol

 

[koenkooi]

If it's jpeg only, it's almost useless (at least for me)
However it can be similar to HDR, where you get out of camera jpeg but also the source raw files that can be merged in DPP.

That's what I'm hoping for as well: an in-camera jpeg, but with the 9 source images available for processing on a computer. I fear it might be a CRM 'roll', like the RAW burst feature.

 

[Eriknoteric]

R5 has enough buffer to handle a few normal raws, no one is saying the camera is creating the huge file btw, could be entirely done in Canons DPP software.

It must be doing this in camera otherwise this is nothing new, and can already be done simply by taking a burst of images and stacking them in post. Astro photographers do it all the time. I do agree that the buffer is large enough though 400mp is only 9 images

 

[LDS]

Really? You believe that? How much is "a lot"? Do you have data to back that up? I'd guess maybe 1%...

Architectural, landscape, and still life - including advertising, art reproduction, and the like - most of the times is done with the camera on a tripod, because precise and repeatable framing is needed. Or the classic long exposure ND filter is used to remove moving subjects. Today, the camera may be even remote controlled and techniques like focus-stacking automated. And that's not guessing. If I can attempt a guess, I'd say that in the field of commercial photography there are much more shots taken on a tripod than among amateurs. By the way, using a tripod sharpens the image on its own, reducing camera shake and delivers images that stands better enlargements. Oh well, even a lot of wildlife is made on a tripod, since large lenses are hard to use without.

people often complain that even 50MP is "too much" and that there is "no need for more than 20MP"

Which people? I agree that often you don't need more than 20mpx. If a photo has to be shown on a web page you don't need more. There was a reason why Canon 1D line never tried to deliver much more. Photojournalists and the like don't need more, they need speed. On the other end of the spectrum there are photographers who may need to produce an image 3mx2m (9.84 x 6.25 ft) or even more, to be shown in a gallery, who will welcome the added resolution. There was a reason photographers worked with medium and large format cameras - and still use cameras like the Phase One or Hasselblad that can deliver higher resolution. When I worked on the digitalization of the collection of ancient textiles of a museum, all photos were taken with a digital scanning back to achieve the desired resolution. With a shrinking ILC market, it does make sense for Canon to attack markets it could ignore before. IMHO it's good to see a feature that doesn't appeal Vloggers and birders only.

I’m pretty sure this feature will not offer flash sync just like focus stacking in the R5.

This is important - it is true that LED lights are often replacing flashes, but it's also true a lot of studio photography is still done with flashes. It is true the camera would need to know when lamps are ready to fire before each shot, or let the photographer trigger it manually, or let setting an interval between shots.

 

[Skyscraperfan]

I am very sceptical about taking 9x9 photos, although the Bayer filter only has a 4x4 grid. That means that the pixels of the final image have to be smaller than the actual pixels of the sensor. So sharpening algorithms will have to compensate for that. That might work though, as those sharpening algorithms can use much more information than usual, as nine times as many pixels are recorded. There are still limits to this though. Otherwise you could make the resolution as high as you want by making 100x100 shots for example, if your IBIS is capable of doing small enough steps. The result would be a 450 gigapixel image.