Is Pixel-Shift coming to the Canon EOS R5?

We reported earlier this week that Canon would release a major firmware update for the Canon EOS R5 this year, likely sometime in Q2. While firmware feature leaks are rare, we’re receiving a lot of chatter about what is coming. One feature that may be coming to the Canon EOS R5 for stills shooters is

See full article...
Canon, can we have an option to turn off the 'self timer lamp' please? Please?
 
Upvote 0
Jul 21, 2010
31,098
12,860
  • Like
  • Haha
Reactions: 3 users
Upvote 0

Michael Clark

Now we see through a glass, darkly...
Apr 5, 2016
4,722
2,655
I was referring to spatial interpolation of color.

Every color value is still interpolated for every image pixel as long as there is a Bayer mask, or any other CFA, in front of the sensor. That's because of the wide overlap between what each of the three CFA filter colors allows to pass, which is based on the way our trichromatic vision uses cones with a wide overlap in sensitivity to different wavelengths.

The way to increase spatial color resolution with a sensor having the same pixel pitch is to take multiple exposures with the entire sensor filtered by a single color filter covering the entire sensor, then swapping to another color filter for the next exposure, and so on.
 
Upvote 0

Michael Clark

Now we see through a glass, darkly...
Apr 5, 2016
4,722
2,655
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

Here in the U.S. it's still $100 more to get a new 5D Mark IV with C-Log than without it. They're two separate SKUs.
 
  • Like
Reactions: 1 user
Upvote 0

Michael Clark

Now we see through a glass, darkly...
Apr 5, 2016
4,722
2,655
Canon had a major firmware upgrade for the 7D in 2012, three years after the camera had been released. See for example: https://www.dpreview.com/articles/0868043083/canon-eos-7d-firmware-v2-major-update The 7DII wasn't released until over two years later.

It was closer to three years than two, from mid March 2012 to late November 2014.

Edit: I went back and double checked. The firmware announcement was in late June 2012, but the release was in July. The file creation date for version 2.0 is July 11, 2012. Ver 2.03 was created Sept 5, 2012, and Ver 2.05 was created May 22, 2013. I have no copies of Ver 2.01, 2.02, or 2.04 if such versions were ever released.

I bought my 7D in late February 2012 and thought I remembered the firmware announcement about a month after I got it. Maybe that was the last Ver 1.xxx update?
 
Last edited:
Upvote 0

Michael Clark

Now we see through a glass, darkly...
Apr 5, 2016
4,722
2,655
Thanks for this reference. I'm still inclined to believe the article's update.

Update: Canon has clarified its original statement. It now says that no hardware change is made but that, 'due to the software architecture in the camera we need to connect and use our service tool to upgrade.' The company also says that non-C-Log versions of the 5D IV will continue to be sold.

But back on topic. Hopefully R5 owners will get the pixel shift option should they want to use it.

There's probably a security level one step deeper than firmware updates can reach that required a change, possibly something as simple as field that identifies the camera's internal model number. Canon understandably would not want to release that software tool in a firmware update.
 
Upvote 0

Michael Clark

Now we see through a glass, darkly...
Apr 5, 2016
4,722
2,655
The Canon Log and Voice Tag upgrades are mutually exclusive, meaning your camera cannot operate with both upgrades at the same time. If you have previously purchased a Canon Log upgrade and then chose to purchase a Voice Tag upgrade you will lose Canon Log functionality (and vice versa).
https://www.canon-europe.com/pro/services/upgrades/eos-5d-iv-voice-tag-upgrade/

That points to requiring different hardware ID fields in the chip that confirms to the firmware what hardware model the camera is.
 
Upvote 0

Michael Clark

Now we see through a glass, darkly...
Apr 5, 2016
4,722
2,655
I'd agree with most of that, it *does* seem more of a gimmick than something of real value to most of us (you could say the same about 8K).

But there are circumstances where pixel-shift does work, e.g. architecture, product and artwork photography. It's quite feasible that resolutions far in excess of 45MP would be needed for such applications, depending on the size and viewing distance of the eventual print.

It certainly does no *harm* to include the feature, and if it helps Canon to sell more cameras, it's to all our benefit, as it makes Canon a stronger company, with more money to invest in development.

It only helps sell more cameras if the increased cost doesn't cause just as many who would otherwise buy it to not buy it.

In the end, Canon's ultimate goal is not to increase the number of cameras they sell. Canon's goal is to maximize net profit over the long term.

If the increased functionality of something as complex as pixel shift causes Canon's cost to produce the camera increase to the point that the market price can not absorb that cost increase without losing more buyers than they gain, then Canon won't consider it a viable business option.
 
Upvote 0

entoman

wildlife photography
May 8, 2015
1,998
2,438
UK
It only helps sell more cameras if the increased cost doesn't cause just as many who would otherwise buy it to not buy it.
There is NO increased cost.
It's a firmware update.
No extra cost to existing users.
No increase in cost of R5 - in fact the cost has *dropped*, as it's been on the market since July 2020.
So the extra feature will make it more appealing to many potential purchasers.

In the end, Canon's ultimate goal is not to increase the number of cameras they sell. Canon's goal is to maximize net profit over the long term.
Obviously. Adding extra features at no cost to the consumer will enable them to sell more units at the same price, and thereby maximise net profit over the lifetime of the product.

If the increased functionality of something as complex as pixel shift causes Canon's cost to produce the camera increase to the point that the market price can not absorb that cost increase without losing more buyers than they gain, then Canon won't consider it a viable business option.
Pixels-shift isn't "complex".
It uses IBIS which is already a feature of the camera.
A bit of fairly simple code in the new firmware instructs the IBIS unit to shift by one pixel in each direction, and then merge the shots.
And as previously pointed out, we have no reason to believe the firmware won't be *free* to users.
The cost of the R5 is extremely unlikely to increase - it's actually dropping, as it will ultimately be replaced by the R5ii/R5s.
 
Last edited:
  • Like
  • Haha
Reactions: 2 users
Upvote 0

entoman

wildlife photography
May 8, 2015
1,998
2,438
UK
Half a pixel.
And merge = interleave pixels from the two shots.
But yes, what you said.
No exposure or focus changes.
That would depend on what mode of pixel-shift was employed (the manufacturer could allow users to select either).

If e.g. you had a 50MP sensor:

Shifting a full pixel and merging 4 images results in a 50MP image, but as every part of the image is sampled with all 3 bayer filter colours, it has better colour than an unshifted image.

Shifting half a pixel and interleaving 16 images also results in better colour, but results in a 200MP image, i.e. double the linear resolution.
 
Upvote 0
Jul 21, 2010
31,098
12,860
Shifting half a pixel and interleaving 16 images also results in better colour, but results in a 200MP image, i.e. double the linear resolution.
I wonder if it does actually double the linear resolution, or if some or even most of the increased MP count is really empty resolution (more MP without an actual increase in real spatial resolution).

Pixel shift in the days before gapless microlenses certainly increased resolution, since any given pixel was really only sampling about 1/3-1/2 of the corresponding area in object space. But whole point of a gapless microlens is to collect light from the whole pixel area.
 
Upvote 0

entoman

wildlife photography
May 8, 2015
1,998
2,438
UK
I wonder if it does actually double the linear resolution, or if some or even most of the increased MP count is really empty resolution (more MP without an actual increase in real spatial resolution).

Pixel shift in the days before gapless microlenses certainly increased resolution, since any given pixel was really only sampling about 1/3-1/2 of the corresponding area in object space. But whole point of a gapless microlens is to collect light from the whole pixel area.
Personally, I'd only be interested in full pixel shift, because my only aim would be better colour - the 45MP of the R5 is more than enough for my needs.

In the case of lower resolution cameras (e.g. R6ii), if pixel-shift became available, I think that for most users, the increase in MP provided by half-pixel 16 image interleaving would be the major attraction.

Regretfully, I'm not qualified to comment about whether there is a gain in spatial resolution - I think that's probably a question that only the designers and engineers can answer.
 
Upvote 0
Jul 21, 2010
31,098
12,860
Regretfully, I'm not qualified to comment about whether there is a gain in spatial resolution - I think that's probably a question that only the designers and engineers can answer.
If implemented, it could be empirically tested. Possibly someone has done so for current cameras with the feature.
 
Upvote 0

koenkooi

CR Pro
Feb 25, 2015
3,574
4,110
The Netherlands
If implemented, it could be empirically tested. Possibly someone has done so for current cameras with the feature.
I think Gordon from Cameralabs showed examples in his Fuji XH2 review , you’ll have to check the video to see 100% crops.
The improvements shown in the video were better than I had expected, but not mind blowing.
 
  • Like
Reactions: 1 user
Upvote 0

Michael Clark

Now we see through a glass, darkly...
Apr 5, 2016
4,722
2,655
There is NO increased cost.
It's a firmware update.
No extra cost to existing users.
No increase in cost of R5 - in fact the cost has *dropped*, as it's been on the market since July 2020.
So the extra feature will make it more appealing to many potential purchasers.


Obviously. Adding extra features at no cost to the consumer will enable them to sell more units at the same price, and thereby maximise net profit over the lifetime of the product.


Pixels-shift isn't "complex".
It uses IBIS which is already a feature of the camera.
A bit of fairly simple code in the new firmware instructs the IBIS unit to shift by one pixel in each direction, and then merge the shots.
And as previously pointed out, we have no reason to believe the firmware won't be *free* to users.
The cost of the R5 is extremely unlikely to increase - it's actually dropping, as it will ultimately be replaced by the R5ii/R5s.

Firmware developers and testers work for free? Who knew?

Those who develop the process that combines the images also do so gratis?

I thought it was only us photographers who are often expected to work for nothing because "all we have to do is push a button"?

Not to mention that the level of precision needed to pull it off may be greater than the level of precision needed for more generic use of IBIS to compensate for camera movement during a single exposure.
 
Last edited:
Upvote 0

entoman

wildlife photography
May 8, 2015
1,998
2,438
UK
Firmware developers and testers work for free? Who knew?
Those who develop the process that combines the images also do so gratis?
Those costs are built-in from the outset.
You pay NOTHING extra for firmware updates, after having bought the camera!
When was the last time you were charged an additional fee to download/install a firmware update?
 
Last edited:
Upvote 0

entoman

wildlife photography
May 8, 2015
1,998
2,438
UK
Not to mention that the level of precision needed to pull it off may be greater than the level of precision needed for more generic use of IBIS to compensate for camera movement during a single exposure.
We are discussing *pixel-shift as it would apply to a firmware update on an R5*, on the assumption that the R5 IBIS has the necessary precision.

Obviously, if the R5 IBIS was not precise enough (unlikely), then pixel-shift would require a new camera body, and could not be achieved by a firmware update.

The title of the thread is "Is Pixel-Shift coming to the Canon EOS R5?", NOT "will pixel-shift appear in the successor model R5 Mkii"
 
Last edited:
Upvote 0

HMC11

Travel
CR Pro
Sep 5, 2020
159
195
Firmware developers and testers work for free? Who knew?

Those who develop the process that combines the images also do so gratis?

I thought it was only us photographers who are often expected to work for nothing because "all we have to do is push a button"?

Not to mention that the level of precision needed to pull it off may be greater than the level of precision needed for more generic use of IBIS to compensate for camera movement during a single exposure.
I doubt very much that Canon would introduce pixel-shift only for the R5. My guess is that pixel-shift capability is very much in their software development roadmap. As such, introducing it in R5 would be likely an incremental cost rather than an all-hands-on-deck, high-cost effort to make it work only for the R5. Besides, adding it to the R5 would allow Canon to do an extensive field test of the software before it becomes a mainstay for their higher end cameras (or hopefully, for all bodies with IBIS).
 
  • Like
Reactions: 1 user
Upvote 0
Those costs are built-in from the outset.
You pay NOTHING extra for firmware updates, after having bought the camera!
When was the last time you were charged an additional fee to download/install a firmware update?
The firmware / testing engineers total costs are built into the company's overall finances and head count. A new product will have projected R&D costs but the people will be allocated to projects as needed.

That said, there are detailed tacit knowledge in each product that only some engineers have so that they tend to be tethered to certain products or replacement product R&D until end-of-support dates. There would be engineers for the R5/R5ii family for this reason for fixing "phenomenon".

Each firmware build besides bug fixes would have costs for internal design/code writing/testing plus a bit of marketing. That incremental cost is not offset by new revenue. There could be a marketing budget where these costs are borne in lieu of new product implementation but that can be a bit advanced for older companies. The costs are probably absorbed in the R&D overall budget with management just approving the budget/cost difference on a strategic basis.

It is possible that the new fiware features were already being written for the R5ii and were "easily" ported to the current model although the cost would be included in the R5ii introduction.

All this assumes a reasonable length product lifecycle eg the ~4 years that Canon has done in the past. Shortening lifecycles changes SW engineering costs towards an "Agile" basis. The Agile SW development manifesto came out >20 years ago and it build on Rapid Application Development and Scrum methodologies from years before that. It sounds like nirvana but getting it right is really hard.

The Manifesto for Agile Software Development is based on twelve principles:
Customer satisfaction by early and continuous delivery of valuable software.
Welcome changing requirements, even in late development.
Deliver working software frequently (weeks rather than months).
Close, daily cooperation between business people and developers.
Projects are built around motivated individuals, who should be trusted.
Face-to-face conversation is the best form of communication (co-location).
Working software is the primary measure of progress.
Sustainable development, able to maintain a constant pace.
Continuous attention to technical excellence and good design.
Simplicity—the art of maximizing the amount of work not done—is essential.
Best architectures, requirements, and designs emerge from self-organizing teams.
Regularly, the team reflects on how to become more effective, and adjusts accordingly.
 
  • Like
Reactions: 1 user
Upvote 0