Is Pixel-Shift coming to the Canon EOS R5?

[neuroanatomist]

Many test were done back when the R5 was first released.

And I'm sure you believe they were done well. Just like you believed that YouTuber who posted that the 30 fps spec of the R3 was 'marketing hype'. Except that his testing turned out to be flawed because he couldn't be bothered to RTFM.

 

[entoman]

That is possible via software update? Isn't it quite complicated to shift a sensor exactly one pixel? I thought that would require special hardware.

The distance and direction in which IBIS shifts a sensor is determined by data feedback from the sensor. There is no reason I can see why a camera couldn't use alternative data (in the form of new firmware) to shift the sensor by a specific amount. A shift of one pixel (or less) is perfectly possible, if the IBIS until has enough precision. If pixel-shift becomes available via firmware, it would indicate that this is indeed the case.

 

[entoman]

It’s nice to see the R5 getting all these features added. Three years later Canon finally admits that the whole overheating and recording limits where all simply firmware cripples.

Now Canon, can we add opengate mode to the $6K Canon R3? And while you are at it, how about enabling the extra two audio channel when using the Tascam audio adapter?

Let’s just be done with the crippling of these cameras once and for all.

Here we go again, sigh...

One of the reasons why Canon sells more cameras than anyone else, is that they are masters of market segmentation. They know exactly how much to include, and how much not to include, to make a product sell well. Thus they leave certain features out of cheaper or current models, and save them for higher-end or future models. That's just sound business practice. Would you prefer that they sold less cameras, became less profitable, and ended up in Pentaxland?

Another possibility that you may not have considered, is that all cameras are under constant development (even after release) and that manufacturers have to launch cameras earlier than they would prefer, in order to keep up with, or stay ahead of, competing brands. Nikon, Sony, Fujifilm etc all add functionality and fix bugs via firmware updates, but rather strangely, when Canon do it, you and a few others get moaning and groaning.

It's quite possible that Canon believed that R5 shooters wouldn't want/need to shoot 8k for half an hour, and that they decided it was better to limit it, rather than have the camera overheat (which damages sensors and increases noise). At a later date, and under pressure from a tiny but vociferous minority of folk, they may then have decided to invest more time and money in developing firmware that could control the overheating - I doubt very much if it was just a case of altering a timer or a thermostat.

 

[entoman]

Pixel-shift would be a great feature that will be useful to certain users, but personally I'd rather see:

• exposure bracketing with electronic shutter
• variable fps with electronic shutter
• major reduction in EVF lag from standby
• AF initiated within any AF zone and tracked across entire frame
• animal-eye AF that worked with a much wider variety of animals
• much stickier AF tracking with less tendency to hop onto backgrounds

Now, if we got ALL of these, I'd be a very very very happy bunny!

 

[AccipiterQ]

I still don't get what pixel shift does, practically. In other words, how is a 'pixel-shifted' picture of a bird sitting on a branch that I might take going to differ from that same image if I took it non-pixel-shifted? Is it going to be more saturated? Sharper? Have increased dimensions? I've never understood why people salivate over this feature.

 

[neuroanatomist]

I still don't get what pixel shift does, practically. In other words, how is a 'pixel-shifted' picture of a bird sitting on a branch that I might take going to differ from that same image if I took it non-pixel-shifted? Is it going to be more saturated? Sharper? Have increased dimensions? I've never understood why people salivate over this feature.

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).

 

[Czardoom]

I don't shoot movies and haven't followed the issue closely, but one thing I'd mention is that if Canon gives an ability officially, it has to work right and not damage the camera via overheating or what have you. It's possible Canon didn't have the data, or did have the data and it failed the internal criteria of the time. So if it's been allowed since then, it could be an engineering change such as new algorithms that use less CPU and thus make less heat. Or perhaps a new business rule as to reliability requirements, or a business decision that at this point in the product cycle they'd rather make more sales even it means more warranty repairs. Finally I recall part of the problem was heat from the memory cards? I can't recall, but if so, perhaps new memories over the last few years include some which make less heat?

I totally agree you could be right that it's purposeful artificial limitations. I'm just saying it's not the only possible reason.

Other than the conspiracy theorists, there were those who showed with thermal photographs and other means that the cameras reached a very high temperature. So the overheating was (and probably still is) real. Initially, it seems pretty clear to me, that the time limits were implemented to prevent the camera from reaching a certain high temperature. In this respect, the limit is arbitrary. But there were probably temperatures where Canon did not want the camera to reach because a) users burning their hands could mean lawsuits, and b) high temps probably bad for may of the camera internals resulting in damage. I think we all know that Canon is a conservative company, so the temperature they chose was conservative. After lots of backlash, they decided to raise that temperature limit - and possibly found other ways to manage the heat.

One thing the conspiracy theorists miss (as they often miss the most logical information) is that Canon - or any other company - wants to maximize the sales of a new product. The last thing they want is a weakness or failure in a product that will limit sales. So anyone thinking that the highly publicized, first really revolutionary mirrorless camera in the Canon lineup was intentionally crippled is just a sterotypical Canon basher. If there is one thing we have learned about Canon, it is that they are not stupid. Conservative, yes. Product differentiaters, yes. But not stupid.

 

[davision]

Many test were done back when the R5 was first released. Running the camera off AC power in full 8K it could run for hours without issue. Battery door switch defeats. Just tons of things that people did to show that the overheat was indeed caused by a firmware cripple… Canon maintained that the limitations were real and not artificially imposed via firmware. Well, if real physical hardware limitations, then the release of firmware 1.6 could not have magically fixed the R5 overheating issue… Yet it did.

With the release of firmware 1.6 Canon admitted through actions that the assumptions that they had imposed artificial recording limitations on the R5 were in fact true. There is simply no other conclusion that could reasonably be reached.

The artificial limitations were removed with the very first firmware update and since then it was temperature limited only.

 

[fasterquieter]

As an R6 user, sniff sniff.

 

[t.linn]

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

 

[t.linn]

One thing the conspiracy theorists miss (as they often miss the most logical information) is that Canon - or any other company - wants to maximize the sales of a new product. The last thing they want is a weakness or failure in a product that will limit sales. So anyone thinking that the highly publicized, first really revolutionary mirrorless camera in the Canon lineup was intentionally crippled is just a sterotypical Canon basher. If there is one thing we have learned about Canon, it is that they are not stupid. Conservative, yes. Product differentiaters, yes. But not stupid.

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.

 

[keithcooper]

Pixel-shift would be a great feature that will be useful to certain users, but personally I'd rather see:

• exposure bracketing with electronic shutter
• variable fps with electronic shutter
• major reduction in EVF lag from standby
• AF initiated within any AF zone and tracked across entire frame
• animal-eye AF that worked with a much wider variety of animals
• much stickier AF tracking with less tendency to hop onto backgrounds

Now, if we got ALL of these, I'd be a very very very happy bunny!

View attachment 207871

No reason not to get both, then we'd both be happy, since none of those matter one jot to me

 

[entoman]

No reason not to get both, then we'd both be happy, since none of those matter one jot to me

Ah yes Keith, but you shoot mainly architecture and static macro, whereas I shoot living moving animals, birds and insects

 

[kaihp]

It's not just the presence of it which would interest me, but how well it is implemented from a usability POV

As a primarily sports shooter (motorcycles) pixel shifting is (A) unlikely to be useful for me (too much movement) and (B) I generally have enough pixels-per-motorcycle to find value in more megapickels.

But if it helps other people, fine. I'd rather have the things on @entoman's list.

 

[entoman]

As a primarily sports shooter (motorcycles) pixel shifting is (A) unlikely to be useful for me (too much movement) and (B) I generally have enough pixels-per-motorcycle to find value in more megapickels.

But if it helps other people, fine. I'd rather have the things on @entoman's list.

I forgot to include pre-capture, which would be invaluable to both of us!

That might not be doable on the R5 though, as it might require a bigger buffer and more powerful processor.

 

[entoman]

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

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...

 

[AlanF]

I forgot to include pre-capture, which would be invaluable to both of us!

That might not be doable on the R5 though, as it might require a bigger buffer and more powerful processor.

The R7 with a much smaller buffer has pre-capture at 30 fps compared with 20fps for the R5 and so has the same rate of mpx/s with the same processor so it could be on.

 

[entoman]

The R7 with a much smaller buffer has pre-capture at 30 fps compared with 20fps for the R5 and so has the same rate of mpx/s with the same processor so it could be on.

Fingers crossed then...

 

[Blorp]

• AF initiated within any AF zone and tracked across entire frame

Are you aware you can use spot AF on the shutter button, start with that and then press AF button on the back set to "full auto AF" and it will take over tracking from the spot AF position across the entire frame? Should work with zone too.

 

[Del Paso]

If pixel shift, why not for handheld pictures as well?
Olympus' OM1 has proven it's possible.