Is Pixel-Shift coming to the Canon EOS R5?

entoman

wildlife photography
May 8, 2015
1,998
2,438
UK
Some advances (especially quality of life ones) are pure software. One such example is false colour or waveform. AF nowadays is similar as the image is getting read out super fast and the analysis is also done split second. Simply the algorithm (or the NN weights) might change. As recognition NN aren't that big actually (e. g. compared to generative ones), the R5 should be quite capable of handling the newest ones.
True, I was considering mainly the enhancements/features that would be valuable to me personally, such as in-camera merging with RAW output, more MP, better DR, higher res "natural" zero-lag EVF.
 
Upvote 0
Jul 21, 2010
31,217
13,079
Since in-camera compositing still is jpeg-only, I’m only interested in the camera doing the pixel shifting and driving the AF. I’ll let helicon focus deal with composing the images.
My point was that I’m skeptical that such processing is feasible. Since other cameras have in-camera compositing, if Canon adds pixel shift then I suspect it will be pixel shift OR focus stacking, i.e., focus stacking will not be available with pixel shift enabled.
 
Upvote 0

koenkooi

CR Pro
Feb 25, 2015
3,652
4,234
The Netherlands
My point was that I’m skeptical that such processing is feasible. Since other cameras have in-camera compositing, if Canon adds pixel shift then I suspect it will be pixel shift OR focus stacking, i.e., focus stacking will not be available with pixel shift enabled.
I agree, we still don’t have an option to to exposure bracketing during focus stacking either. If such a straightforward thing isn’t there, a much more complicated feature won’t be coming either.
But I can hope!
 
  • Like
Reactions: 1 user
Upvote 0

entoman

wildlife photography
May 8, 2015
1,998
2,438
UK
I agree, we still don’t have an option to to exposure bracketing during focus stacking either. If such a straightforward thing isn’t there, a much more complicated feature won’t be coming either.
But I can hope!
We don't even have an option for exposure bracketing with electronic shutter, at least, not with the R5... it would be handy to be able to bracket exposures with noise-sensitive small creatures, by using silent shutter.
 
Last edited:
Upvote 0
Jul 21, 2010
31,217
13,079
We don't even have an option for exposure bracketing with electronic shutter, at least, not with the R5... it would be handy to be able to bracket exposures with noise-sensitive small creatures, by using silent shutter.
I suspect that’s an intentional choice by Canon, and I can see the logic. The ‘typical’ use case for exposure bracketing is when scene DR exceeds camera DR, so why would one want to bracket exposures with electronic shutter when that drops the bit depth from 14 to 12? The R3 records the full 14 bits with electronic shutter, and allows AEB with full electronic shutter at 30 fps.
 
  • Like
Reactions: 2 users
Upvote 0

entoman

wildlife photography
May 8, 2015
1,998
2,438
UK
I suspect that’s an intentional choice by Canon, and I can see the logic. The ‘typical’ use case for exposure bracketing is when scene DR exceeds camera DR, so why would one want to bracket exposures with electronic shutter when that drops the bit depth from 14 to 12? The R3 records the full 14 bits with electronic shutter, and allows AEB with full electronic shutter at 30 fps.
Yes, I'm sure that was Canon's logic, but folk who spend $4000+ on a camera often don't restrict themselves to "typical" use cases. Much of the reason to purchase a high end camera is to get maximum versatility for non-standard usages. I'd chance a bet by saying that there's probably at least as much practical need for exposure bracketing and variable fps in ES, as there is for pixel-shift or 8k video, but the latter get a lot more attention as "features", because they involve the "numbers game", i.e. megapixels.
 
  • Like
Reactions: 2 users
Upvote 0
entoman said:
It's academic to me, as I only shoot RAW, but while Highlight Tone Priority might be preferred in some situations, Auto Lighting Optimiser (which places more importance on shadow detail) might be preferred in others. They can't be selected simultaneously. I don't think it would be a great idea to have either of them engage *automatically* based on ISO. It's up to the user to decide whether shadows or highlights are more important.

Most of the other items on your list seem reasonable and useful. Naming custom modes would be particularly useful if the name was displayed as an overlay in a corner of the EVF as well as on the flippy screen, although with only 3 custom modes available, I've never had issues remembering which did what...
I appreciate your thoughts, entoman. I shoot RAW exclusively too. What many users don't realize is that while Auto Lighting Optimizer doesn't affect RAW files, Highlight Tone Priority does. It is easy to demonstrate this to yourself. Take two RAW photos of your computer screen, one with Highlight Tone Priority off, the other with it on. Use the exact same exposure settings making sure to completely blow the images out. Import them into Lightroom. The two photos will appear identical—until you drag down the Exposure slider. You will find significantly more recovery latitude in the file shot with HTP engaged. I've done exactly this in the screenshot below. (Note: Recover using the Exposure slider, not the Highlights or Whites sliders.)

How does the feature work? The actual sensitivity of your sensor is fixed at its native ISO. When you increase the ISO setting, the camera takes the data coming off the sensor, amplifies it, and writes the resulting data to the RAW or JPG file. With Highlight Tone Priority engaged, the camera essentially tone maps the data, amplifying the darker tones but not the lighter tones when it records the image file. This is why the feature doesn't work at ISO 100 where there is no amplification happening. Some people argue that this is no different than shooting at ISO 100 and boosting the shadows. The principle is similar. The difference is your shutter speed. If you're shooting at ISO 800, it is typically because your shutter speed at ISO 100 is too slow.

Is there a downside to always using HTP? Not that I'm aware of, aside from being locked out of lower ISOs. The manual indicates that you may notice noise in the shadows. I've always assumed that this is the same high ISO shadow noise you'd notice with the feature turned off (shooting at the same ISO) but it's never been a problem so I've never bothered to do a comparison.

So do I always shoot with HTP engaged? Sadly, no. I typically shoot in Manual mode with Auto ISO engaged. If I'm shooting landscapes, for example, I have the time to turn the HTP feature on and off based on my ISO. But if I'm shooting street, things are happening quickly. I don't have the time to monitor ISO and change this setting. Hence, my feature request...

The screenshot below demonstrates what I have been describing. Both CR3s were shot with the exact same settings. They looked identical before I applied a -3.75 adjustment to the Exposure slider to both images. The image on the left shows the typical dull whites you get when there is no data to recover. The image on the right, shot with Highlight Tone Priority engaged, shows how much more data is available to recover. No other adjustments were made to either image.

Screenshot 2023-03-11 at 12.16.20 PM.jpg
 
  • Like
  • Wow
Reactions: 4 users
Upvote 0

entoman

wildlife photography
May 8, 2015
1,998
2,438
UK
I appreciate your thoughts, entoman. I shoot RAW exclusively too. What many users don't realize is that while Auto Lighting Optimizer doesn't affect RAW files, Highlight Tone Priority does. It is easy to demonstrate this to yourself. Take two RAW photos of your computer screen, one with Highlight Tone Priority off, the other with it on. Use the exact same exposure settings making sure to completely blow the images out. Import them into Lightroom. The two photos will appear identical—until you drag down the Exposure slider. You will find significantly more recovery latitude in the file shot with HTP engaged. I've done exactly this in the screenshot below. (Note: Recover using the Exposure slider, not the Highlights or Whites sliders.)

How does the feature work? The actual sensitivity of your sensor is fixed at its native ISO. When you increase the ISO setting, the camera takes the data coming off the sensor, amplifies it, and writes the resulting data to the RAW or JPG file. With Highlight Tone Priority engaged, the camera essentially tone maps the data, amplifying the darker tones but not the lighter tones when it records the image file. This is why the feature doesn't work at ISO 100 where there is no amplification happening. Some people argue that this is no different than shooting at ISO 100 and boosting the shadows. The principle is similar. The difference is your shutter speed. If you're shooting at ISO 800, it is typically because your shutter speed at ISO 100 is too slow.

Is there a downside to always using HTP? Not that I'm aware of, aside from being locked out of lower ISOs. The manual indicates that you may notice noise in the shadows. I've always assumed that this is the same high ISO shadow noise you'd notice with the feature turned off (shooting at the same ISO) but it's never been a problem so I've never bothered to do a comparison.

So do I always shoot with HTP engaged? Sadly, no. I typically shoot in Manual mode with Auto ISO engaged. If I'm shooting landscapes, for example, I have the time to turn the HTP feature on and off based on my ISO. But if I'm shooting street, things are happening quickly. I don't have the time to monitor ISO and change this setting. Hence, my feature request...

The screenshot below demonstrates what I have been describing. Both CR3s were shot with the exact same settings. They looked identical before I applied a -3.75 adjustment to the Exposure slider to both images. The image on the left shows the typical dull whites you get when there is no data to recover. The image on the right, shot with Highlight Tone Priority engaged, shows how much more data is available to recover. No other adjustments were made to either image.

View attachment 207889
Great reply, I had always believed that HTP only worked with JPEGs. I'll make more use of it in future, when shooting RAW, as I often shoot in very contrasty lighting conditions (tropical rainforests). Usually in these situations, highlight detail is vital - glossy leaves in rainforests reflect like mirrors, and metallic insects or white-winged butterflies are even more of a problem. In dense forests, dark backgrounds are unavoidable, but I try hard to avoid the black background look, as I want the images to look as natural as possible. Reflectors and fill-flash are not available options, neither is HDR (which requires shooting a burst and merging in-camera), as I'm shooting nervous twitchy wildlife.
 
  • Like
Reactions: 1 user
Upvote 0
Great reply, I had always believed that HTP only worked with JPEGs. I'll make more use of it in future, when shooting RAW, as I often shoot in very contrasty lighting conditions (tropical rainforests). Usually in these situations, highlight detail is vital - glossy leaves in rainforests reflect like mirrors, and metallic insects or white-winged butterflies are even more of a problem. In dense forests, dark backgrounds are unavoidable, but I try hard to avoid the black background look, as I want the images to look as natural as possible. Reflectors and fill-flash are not available options, neither is HDR (which requires shooting a burst and merging in-camera), as I'm shooting nervous twitchy wildlife.
Your shooting scenario sounds fun—and quite challenging!
 
Upvote 0

TAF

CR Pro
Feb 26, 2012
491
158
Either better color resolution or more megapixels. Shifting a full pixel and merging 4 images samples every part of the image with all three bayer filter colors, meaning no color interpolation is needed. Shifting half a pixel and merging 16 images does that and gives you 4x the MP, e.g. if you start with a 50 MP sensor and merge 16 pixel shifted images, you get a 200 MP output image (with no color interpolation).

What an interesting approach. What sort of file(s) would you expect to end up with? All the images saved and you assemble them in post? Or perhaps the camera assembles them and you have the option to save all the pieces or just the final one? Sort of like the HDR approach? I am intrigued. I think it would be very nice for landscapes and other stationary things when using a tripod. Not so sure about moving objects (cat photos, how will it enhance cat photos??!)

I hope they provide it.

(t.linn's list would also be really nice, as would my plea, a 640x480 high speed video capture at 1200x or so)
 
Upvote 0
While I agree with you in this specific instance, let's not pretend that the Cripple Hammer doesn't exist. Yes, Canon does want to maximize the sales of every product in their line up but only within the context of maximizing their overall profits. This is the whole point of the Cripple Hammer. Segment the market carefully and then cripple key features to force customers to buy a more expensive product or buy more than one product. This strategy works when you are the dominant player in a market but it also leaves you vulnerable to a company like Sony who is willing to aggressively compete for business. To be clear, I'm not arguing that Sony is a better company. It is just in a different position. As Sony gets closer to overtaking Canon, it is reasonable for Sony to become less aggressive even as Canon becomes more aggressive. I would argue that the R5 is Exhibit A in this shift.
"let's not pretend that the Cripple Hammer doesn't exist."

I guess the converse is to believe that "the cripple hammer" DOES exist. The reality is that companies differentiate products based on features, the features have value. You can call it whatever you want but it amounts to wanting something for nothing.

FWIW: it is hard to believe that the heat issue in the R5 was purely imaginary when the put a fan in the R5C to improve the cooling.

Conspiracy theorist is right.
 
  • Like
Reactions: 1 user
Upvote 0
Let’s just be done with the crippling of these cameras once and for all.
When the word ‘cripple’ is used it indicates more about the poster’s knowledge than any serious criticism.

The fact that canon has made so many firmware updates with new features (particularly video but also af etc) are just bonuses to our initial purchase negates your comment
 
  • Like
Reactions: 3 users
Upvote 0
Well, I can on one hand understand that, yet the overall price for the entire package will go up then. There has practically never been an instance where a company lets you rent/upgrade something where it wasn't significantly more expensive in the end.
The 5div initially had a paid clog option but needed hardware to be installed. It didn’t take too long before the price went up and clog was standard
 
Upvote 0

Michael Clark

Now we see through a glass, darkly...
Apr 5, 2016
4,722
2,655
Either better color resolution or more megapixels. Shifting a full pixel and merging 4 images samples every part of the image with all three bayer filter colors, meaning no color interpolation is needed. Shifting half a pixel and merging 16 images does that and gives you 4x the MP, e.g. if you start with a 50 MP sensor and merge 16 pixel shifted images, you get a 200 MP output image (with no color interpolation).

Color interpolation is still needed. All three RGB values for each pixel in a digital image created using a CFA are interpolated from the raw monochromatic luminance values collected by the sensor.

This is because the colors of the filters in the Bayer array are not the same as the colors emitted by our screens. They're closer to the spectral sensitivity of the cones in our retinas (our L-cones are not remotely as sensitive to "red" at 640nm as they are to lime-green at 565nm).

Blue and green are kind of close, but the "red" filters in Bayer masks are somewhere between yellow and yellow-orange, usually centered around 590nm, rather than the "Red" output of our monitors, which is around 640nm.

But even if the colors of the filters matched our emissive screens perfectly, there would still need to be interpolation done to account for the color temperature, white balance, and CRI of various illuminating light sources, just like everything must be interpolated when translating from one RGB color space to another, or even from an emissive system like RGB to a subtractive system like CYM.
 
  • Like
Reactions: 1 user
Upvote 0

Michael Clark

Now we see through a glass, darkly...
Apr 5, 2016
4,722
2,655
I don't doubt that there is a use case for pixel shift. That said, I would be willing to bet that most people that think they want it have never used it and don't realize how perfect conditions have to be for it to be effective. If we do get this feature, don't be surprised if it is limited by the same 12-bit electronic shutter mode that limits the focus bracketing option on Canon bodies. I believe most users would be better off if Canon focused its efforts on more practical improvements. Here are eleven examples of other refinements that I believe would benefit more users.
  • Fix Fv Mode so that it respects the minimum Auto ISO shutter speed setting. Right now it doesn't.
  • Give users the option to assign DOF Preview to the shutter button half-press
  • Give users the option to use the mechanical shutter to capture 14-bit images when focus bracketing
  • Display a pixel pitch-based DOF indicator on the focus distance scale. This is useful for zone focusing and manual focus bracketing.
  • Make the often unused Lock button programmable for other functions
  • Give users the ability to independently customize Subject to Detect and Initial AF Point when configuring buttons for dual back button focus or Register/Recall Shooting Function
  • Give users the option to manually focus with the lens stopped down. This would serve as a work around to any focus shift issues.
  • Give users the option to display focus peaking when focus mode set to AF. This is useful when manually focus bracketing.
  • Allow Highlight Tone Priority to engage automatically when Auto ISO chooses a value that allows its use
  • Enable discrete on/off control over every icon on the EVF and rear LCD. Each one of us can probably identify more than one icon that is never relevant to our use cases that is grouped with another one we use all the time.
  • Allow users to name each custom mode

The second most of them realize they need not only a tripod, but a very stable tripod to take advantage of pixel shift, 98% of those who think it might be cool would be out.
 
  • Like
Reactions: 1 users
Upvote 0

Michael Clark

Now we see through a glass, darkly...
Apr 5, 2016
4,722
2,655
Yea I don't see that: "Hey could you also do a video of me doing this cool jump? - "Yes just give me five minutes, I need to get my credit card, login online into my Canon account, Pay for that feature, create a WiFi Hotspot to tell my camera about it, a damn Camera Connect can't find it again, let me reset my stored WiFi devices....hey where are you going.."

That would still be better than having to buy a brand new camera every year to get the incremental updates like Sony owners do.
 
  • Like
Reactions: 1 user
Upvote 0
Jul 21, 2010
31,217
13,079
Color interpolation is still needed. All three RGB values for each pixel in a digital image created using a CFA are interpolated from the raw monochromatic luminance values collected by the sensor.

This is because the colors of the filters in the Bayer array are not the same as the colors emitted by our screens. They're closer to the spectral sensitivity of the cones in our retinas (our L-cones are not remotely as sensitive to "red" at 640nm as they are to lime-green at 565nm).

Blue and green are kind of close, but the "red" filters in Bayer masks are somewhere between yellow and yellow-orange, usually centered around 590nm, rather than the "Red" output of our monitors, which is around 640nm.

But even if the colors of the filters matched our emissive screens perfectly, there would still need to be interpolation done to account for the color temperature, white balance, and CRI of various illuminating light sources, just like everything must be interpolated when translating from one RGB color space to another, or even from an emissive system like RGB to a subtractive system like CYM.
I was referring to spatial interpolation of color.
 
  • Like
Reactions: 1 user
Upvote 0

shadowsports

R5 C - RF Trinity
CR Pro
Jan 15, 2023
173
147
Bay Area, CA
The 5div initially had a paid clog option but needed hardware to be installed. It didn’t take too long before the price went up and clog was standard
David,
The addition of cLog on the 5D4 was software, not hardware.

And, [Respectfully[ Canon like other manufacturers does "cripple" [call it what you like] or omit features that could be included on body's in order to "protect" sales of higher end models. This doesn't have to do with someone's knowledge. Firmware updates that add functionality does not negate what was initially [purposely] omitted. It does enhance the user experience and value, but only after the fact.
 
Last edited:
  • Like
Reactions: 1 user
Upvote 0