Canon to release major firmware update for the Canon EOS R5

[Loswr]

Always fun to find out. As usual, a mixture of a few facts and a lot of guesswork!

I expect we'll see a development announcement for the R1 in early 2024, but they won't divulge the MP count until the full announcement.

 

[entoman]

Another CP+ interview by French site Phototrends, which seems to indicate that Canon are willing to open RF mount, subject to negotiation:

Question: "Many photographers are eagerly awaiting the opening of the RF mount. At present, Canon is the only brand to maintain the exclusivity of its frame. Will we see RF optics from other manufacturers in the near future?"

Tetsuji Kiyomi : "It is true that we are approached by many third-party manufacturers. We hear their demands. However, we have no direction or policy as such on this subject. Naturally, I cannot share more details with you, but what I can say is that if this aligns with our strategy, we will take the necessary action on a case-by-case basis. I can’t tell you more."

Go Tokura (General Manager of Imaging Division at Canon): "An example of this is the launch of RF optics by Cosina [the Voigtlander 50mm f/1 is the first “third-party” lens to be able to communicate with the body via the RF mount]: we are in the process discussion with other lens manufacturers. This is the situation at present."

 

[entoman]

I expect we'll see a development announcement for the R1 in early 2024, but they won't divulge the MP count until the full announcement.

Yes, MP count is always the last thing to be announced.

 

[roby17269]

What type of clients are demanding more than 45MP?

Are their demands logical, or have they been hyped by the media into believing ultra high res is vital?

If you are a professional, you can try to educate paying clients and risk losing them to another photographer with higher-mp cameras, or you can buy a tool of your trade that will delight your paying clients.

I am no professional, but I love the 80mp output of my IQ1-80 back and prefer it over my R5's 45mp, when I am in conditions of being able to control the light being the only caveat. If I'll ever win the lottery, I will buy a IQ4-150 since for me 150mp is better than 80mp (and the IQ4 has other advantages over the IQ1 to booth). I will buy a R5 II or maybe a R1 or a Rwhatever only if they will have a FF sensor with more than 45mp, otherwise I will stick with my R5. Plenty of lenses to buy anyhoo

If more mps is a negative thing for you, that's totally fine, but do not assume that consequently more mps are worse for everyone.

 

[PinholeR5]

Yes, when the R5ii is eventually released (late 2024 is my guess) existing R5 firmware caveats will be removed, but I would also expect significant changes e.g. a BSI stacked sensor of around 60MP, a new generation processor (possibly twin processors, if economically viable), focus-breathing correction, better IBIS, and a much faster burst rate (circa 40fps with electronic shutter).

Such a combination would enable hand-held stacking and merging with more accurate registration than is currently possible - due to minimal movement between frames. It could feasibly be used with images showing a modest amount of subject movement. These benefits could also be applied, with greater effectiveness to a hypothetical R7ii, as the latter would be capable of even faster burst rates that would virtually eliminate camera movement between frames.

Stacking and merging, in conjunction with AI, can enable higher DR, less noise, sharper images, more pleasing bokeh and numerous other benefits.

Yep. When I said features with big caveats, I meant that the HW isn't really adequate for them, but a reduced functionality version can be implemented with the current HW. Then the new HW would allow for them to be fully realized. I'm not crazy about in-camera processing because SW evolution of computational photography tends to be faster than FW updates and it is always subject to the in-camera HW limitations.

However, in-camera processing to guide the overall multi-shot capture generating a HEIF image, but also a set of RAW images would be awesome.

Late 2024 seems like a reasonable time frame for me. Though it might be earlier in the year.

 

[Scenes]

The removal of the 30 minute limit is wishful thinking. It's not a technical issue. It's a TAX issue.

Canon would have to reclassify from stills camera to video camera and back tax all R5 units sold ever, to be able to do this.

It's not going to happen.

This was a European law that ended in 2019. I suspect the development cycle of various Canon products has been so long it’s only now we are seeing it in released cameras. There’s no reason legal tax reason they couldn’t add it if they wanted too.

 

[DrD]

Wouldn't there be additional licensing costs for Canon to add the other codecs?
There is one thing to add features that "only" need internal coding/Canon resources but it would be another to decrease their margins especially as it looks like there are discounts available on the current model.

Panasonic GH6 uses ProRes and the Nikon Z9 shoots ProRes RAW, so if these two companies can license Apple ProRes why cannot Canon? Especially now with new laptops with dedicated ProRes acceleration.

 

[cohenfive]

If there is no R52 this year that is a disappointment. The R5, as great as it was, is now getting to be old tech. It needs to be faster to focus and have better performance in electronic shutter mode, among other rumored updates. Software won't do much to address the performance shortcomings. Nikon will have their Z8 out soon, and from what I have heard it will be approaching Z9 performance and an R5 price point. That's an issue for canon until they update the body with newer and better performing hardware imo.

 

[Famateur]

Yes, for focus assist to work, there would have to be electronic talk between body and lens.

This doesn't make sense to me. The camera evaluates what's on the sensor to determine whether the image is in focus. No information from the lens is necessary to confirm that the place in the image surrounded by the focus assist box is sharp and so produce a visual confirmation (e.g. box turns green).

What am I missing?

Anecdotally, I was excited to inherit some of my granddad's vintage manual-focus Nikon lenses. I ordered an adapter and then was shocked to find that the focus assist arrows weren't accurate. If I snap the image when it confirms focus, it's always back-focused.

Thankfully, focus peaking is at least accurate.

With both features on at the same time, you can see stronger peaking behind what's in the focus box when the arrows converge and turn green. If I go with the arrows, focus is missed. If I adjust until the peaking is brightest, the arrows are diverged, but the image captured is properly focused.

It makes me wonder: Why would it confirm focus when it's not in focus, especially when it's evaluating what's actually on the sensor? It works for peaking but not the arrows and box. Makes no sense at all to me...

 

[entoman]

If more mps is a negative thing for you, that's totally fine, but do not assume that consequently more mps are worse for everyone.

That's a VERY strange knee-jerk reaction to my post!

What leads you to believe that more MPs is a "negative thing" to me? Or that consequently "more MPs" would be "worse for everyone"?

I simply asked whether the demand for more MP came from the *clients* or from a genuine need for high resolution by the *photographer*!

In the post immediately below the one you are referring to, I stated "Personally, as a stills-only shooter, I don't need/want more than 45MP". Note the word "personally", which surely makes it very obvious that I was NOT trying to impose my own preferences on anyone else.

Also, in several posts, I've stated that I'd prefer if future cameras offered a *choice* of resolutions. The example I gave on page 6 of this very thread, suggests a native resolution of 90MP, with an option via pixel-binning to produce 22.5MP RAWs as and when required...

 

[entoman]

This doesn't make sense to me. The camera evaluates what's on the sensor to determine whether the image is in focus. No information from the lens is necessary to confirm that the place in the image surrounded by the focus assist box is sharp and so produce a visual confirmation (e.g. box turns green).

What am I missing?

Anecdotally, I was excited to inherit some of my granddad's vintage manual-focus Nikon lenses. I ordered an adapter and then was shocked to find that the focus assist arrows weren't accurate. If I snap the image when it confirms focus, it's always back-focused.

Thankfully, focus peaking is at least accurate.

With both features on at the same time, you can see stronger peaking behind what's in the focus box when the arrows converge and turn green. If I go with the arrows, focus is missed. If I adjust until the peaking is brightest, the arrows are diverged, but the image captured is properly focused.

It makes me wonder: Why would it confirm focus when it's not in focus, especially when it's evaluating what's actually on the sensor? It works for peaking but not the arrows and box. Makes no sense at all to me...

You could, using only the sensor data, provide a confirmation LED to indicate when the subject under the AF point was in sharpest focus.

But if you want an "AF assist box", by which I assume you are referring to the Canon "arrow indicator" found in R series cameras, using the sensor data alone would not tell you which *direction* to rotate the focus ring. For that, you need feedback from the lens.

FWIW, I sometimes use focus-peaking when using a tilt/shift lens for landscapes, but for images in which it's important to have a particular part of the subject in focus, I find the "arrow indicator" and AF confirmation box faster and more accurate. For macro, where focus precision is vital, I use magnified view (unless the subject is moving, as is often the case with tiny insects).

It's not unknown for MILC AF systems to be inaccurate, although in theory it should not be possible.

For example, when using my EF 100-400mm on the R5, at close distances and maximum focal length, the subject under the AF spot was back-focussed and I needed to switch to manual focus to correct it. I eliminated unintentional forward/backward camera movement by having the camera on a tripod, and focused on a page of printed text that was parallel to the sensor. The issue existed whether I use small spot, standard spot or zone AF. I had similar problems with my 100mm macro lens on the R5, but the error was smaller, and depth of field at F8 was enough to cover it. When using my native RF lenses (RF100-500mm and RF100mm macro) on the same body, the issue doesn't exist.

 

[koenkooi]

This doesn't make sense to me. The camera evaluates what's on the sensor to determine whether the image is in focus. No information from the lens is necessary to confirm that the place in the image surrounded by the focus assist box is sharp and so produce a visual confirmation (e.g. box turns green).

What am I missing?

Anecdotally, I was excited to inherit some of my granddad's vintage manual-focus Nikon lenses. I ordered an adapter and then was shocked to find that the focus assist arrows weren't accurate. If I snap the image when it confirms focus, it's always back-focused.

Thankfully, focus peaking is at least accurate.

With both features on at the same time, you can see stronger peaking behind what's in the focus box when the arrows converge and turn green. If I go with the arrows, focus is missed. If I adjust until the peaking is brightest, the arrows are diverged, but the image captured is properly focused.

It makes me wonder: Why would it confirm focus when it's not in focus, especially when it's evaluating what's actually on the sensor? It works for peaking but not the arrows and box. Makes no sense at all to me...

While the AF is on-sensor, it is not image based. It uses the phase difference between 2 adjacent subpixels, which can be thrown off by things like UV filters.

I would welcome a mode where the camera does use image data to fine tune the last steps of acquiring focus, it would fix a number of Canon AF annoyances.

 

[entoman]

First to loose 2 steps of dynamic range by using electronic shutter to get them back via exposure bracketing?

I do think the camera needs different modes for fast speed (i.e. reduced DR) and "lazy shooting".
The combo of shutter mode with a lot of R5's functionality does not make sense.

My understanding is that the loss of DR with ES typically amounts to about 1 stop. The loss of DR is due to a reduction in bit depth (12 bit mechanical, 10 bit electronic in most cases), and that is done purely to increase readout speed. AFAIK, there is no reason why electronic shutter can't run on 12 bit and thereby retain the full DR. Presumably it's dictated by sensor design and processor power. Anyone with greater knowledge please contribute.

 

[entoman]

While the AF is on-sensor, it is not image based. It uses the phase difference between 2 adjacent subpixels, which can be thrown off by things like UV filters.

I would welcome a mode where the camera does use image data to fine tune the last steps of acquiring focus, it would fix a number of Canon AF annoyances.

Maybe I'm wrong, in which case please correct me, but I thought that Canon used a combination of phase detect and contrast detect, i.e. phase detect to achieve approximate focus, and contrast detect to do the final fine-tuning.

Interested to learn that UV filters can cause inaccuracy with PDAF - if this is true, I wonder if there are suitable "protection filters" that won't cause such issues? I'm very reluctant to expose the front element to wind-blown sand and dust on safaris, so always use a filter. Also it's very easy to scuff lens coatings when shooting macro and poking the lens in amongst thorny or prickly vegetation!

 

[koenkooi]

Maybe I'm wrong, in which case please correct me, but I thought that Canon used a combination of phase detect and contrast detect, i.e. phase detect to achieve approximate focus, and contrast detect to do the final fine-tuning.

For DPAF cameras, it's phase detect only. And worse, the algorithm seems to be open loop: DPAF will estimate the subject distance and tell the lens focus to that distance. When the lens confirms it is at that position, the shot will get taken. No DPAF or contrast confirmation at all
So if your lens is bad at moving to the correct distance, you'll get consistently OOF images. Like the RF70-200 f/2.8 had before a lens firmware update fixed it.
That a lens firmware update can fix consistent back focus issues seems to match my open-loop theory, but I haven't found anything in Canon white papers or marketing that explicitly confirms it.

Interested to learn that UV filters can cause inaccuracy with PDAF - if this is true, I wonder if there are suitable "protection filters" that won't cause such issues? I'm very reluctant to expose the front element to wind-blown sand and dust on safaris, so always use a filter.

I'm not aware of a list of 'good' filters, I only see the few instances online were someone shows focus issues, gets told to remove the filter and *presto* the issues are gone. This makes be believe that the 'good' filters outnumber the 'bad' by a very large margin.
It is straightforward to test at home, since it's the same procedure as checking for front/back focus or if you had a fancy DSLR, calibrating AFMA. Setup a target and something at a diagonal, focus on it, shoot, add filter, focus, shoot, compare pictures.

 

[Loswr]

My understanding is that the loss of DR with ES typically amounts to about 1 stop. The loss of DR is due to a reduction in bit depth (12 bit mechanical, 10 bit electronic in most cases), and that is done purely to increase readout speed. AFAIK, there is no reason why electronic shutter can't run on 12 bit and thereby retain the full DR. Presumably it's dictated by sensor design and processor power. Anyone with greater knowledge please contribute.

For the R5, per the Canon specs, “14-bit with Mechanical shutter and Electronic 1st Curtain, 13-bit A/D conversion with H+ mode, 12-bit A/D conversion with Electronic shutter.” According to Bill Claff, the difference between mechanical (6 fps or less) and electronic shutter is ~1.5 stops of DR.

 

[AlanF]

Maybe I'm wrong, in which case please correct me, but I thought that Canon used a combination of phase detect and contrast detect, i.e. phase detect to achieve approximate focus, and contrast detect to do the final fine-tuning.

As @koenkooi says, you are wrong.

For DPAF cameras, it's phase detect only. And worse, the algorithm seems to be open loop: DPAF will estimate the subject distance and tell the lens focus to that distance. When the lens confirms it is at that position, the shot will get taken. No DPAF or contrast confirmation at all
So if your lens is bad at moving to the correct distance, you'll get consistently OOF images. Like the RF70-200 f/2.8 had before a lens firmware update fixed it.
That a lens firmware update can fix consistent back focus issues seems to match my open-loop theory, but I haven't found anything in Canon white papers or marketing that explicitly confirms it.

I remembered that 11 years ago lensrentals had on their blog an article that discussed these points and concluded that Canon had changed over to a closed loop systems and better AF. https://www.lensrentals.com/blog/2012/08/autofocus-reality-part-3b-canon-cameras/

 

[entoman]

For DPAF cameras, it's phase detect only. And worse, the algorithm seems to be open loop: DPAF will estimate the subject distance and tell the lens focus to that distance. When the lens confirms it is at that position, the shot will get taken. No DPAF or contrast confirmation at all
So if your lens is bad at moving to the correct distance, you'll get consistently OOF images. Like the RF70-200 f/2.8 had before a lens firmware update fixed it.
That a lens firmware update can fix consistent back focus issues seems to match my open-loop theory, but I haven't found anything in Canon white papers or marketing that explicitly confirms it.


I'm not aware of a list of 'good' filters, I only see the few instances online were someone shows focus issues, gets told to remove the filter and *presto* the issues are gone. This makes be believe that the 'good' filters outnumber the 'bad' by a very large margin.
It is straightforward to test at home, since it's the same procedure as checking for front/back focus or if you had a fancy DSLR, calibrating AFMA. Setup a target and something at a diagonal, focus on it, shoot, add filter, focus, shoot, compare pictures.

Maybe there's something significant transmitted via the extra RF pins, because I don't have any focus accuracy issues with any of my RF lenses, only with the older EF lenses (that I've since sold).

 

[roby17269]

That's a VERY strange knee-jerk reaction to my post!

What leads you to believe that more MPs is a "negative thing" to me? Or that consequently "more MPs" would be "worse for everyone"?

I simply asked whether the demand for more MP came from the *clients* or from a genuine need for high resolution by the *photographer*!

In the post immediately below the one you are referring to, I stated "Personally, as a stills-only shooter, I don't need/want more than 45MP". Note the word "personally", which surely makes it very obvious that I was NOT trying to impose my own preferences on anyone else.

Also, in several posts, I've stated that I'd prefer if future cameras offered a *choice* of resolutions. The example I gave on page 6 of this very thread, suggests a native resolution of 90MP, with an option via pixel-binning to produce 22.5MP RAWs as and when required...

Well...

Your post I replied to read a bit different though. You wrote "hyped by the media", implying that the need is not real, just a fad.

So sorry if I misunderstood you, but the post I replied to had a very different vibe from this one.

In any case, as I wrote, it's entirely your prerogative to decide how many mps are enough or too many for you.

For me, I can totally be fine with 45, but I'd be happier with 200+

 

[bbasiaga]

If there is no R52 this year that is a disappointment. The R5, as great as it was, is now getting to be old tech. It needs to be faster to focus and have better performance in electronic shutter mode, among other rumored updates. Software won't do much to address the performance shortcomings. Nikon will have their Z8 out soon, and from what I have heard it will be approaching Z9 performance and an R5 price point. That's an issue for canon until they update the body with newer and better performing hardware imo.

Only slightly surpassed by the R3 and R6II and R7, but yes older as relative to those. Still light years ahead of the previous generation, and more than competitive with the other brand's offerings.

Z8 will be interesting. Its been rumored so many times, will it ever come? What specs will it have? the last published rumor, IIRC, was debunked as a wish list. If you believe what you hear around the forums, there are still people waiting for a Z9. So even if announced, can they actually produce them? The Z9 AF system is not quite up to par with the R5/R6, and further behind the R3/R6II. Will the Z8 be the first to join to modern AF performance club?

Sony's latest offerings are slower than the R5, which is quite important to some.

R5 is still extremely competitive. That's probably why the replacement isn't coming quite yet.

Brian