Is a Canon EOS R7 Mark II coming this year? [CR2]

[koenkooi]

You are right, as usual. Thanks. The Nikon Z7 and Z9, also 45mpx, have a lower DR performance as measured by 'Photons to Photos' compared to the R5. In fact, the Nikon D850, which is also BSI but released in 2017 and performs almost identically to the Z7, performs better than the Z9 at ISO settings up to about ISO500, and thereafter comparable to the Z9. I also tried to compare the Z5, FSI but only 24mpx, has comparable or better DR performance as the Z7 . As it is not 45mpx, I am not sure if that is due to the effect of pixel size (improve DR compared with 45mpx) counter-balanced by the Z5 using a 'cheaper' sensor. Overall, it does seems that the R5 has one of best DR despite being FSI. It would be interesting to see how Canon's BSI perform in terms of DR when they appear. Based on Nikon, however, the BSI does not seem to confer a DR advantage, but again, the comparisons might be flawed as I don't quite know how the 'quality' of the sensors compared. Overall, it is perhaps better to just stick to actual measurements rather than making assumptions about DR performance based only on written specs.

The DxO article on the R6II/R8 sensor expresses a similar "Wait, is this FSI sensor on-par or better than the BSI sensors used by the competition!??!" sentiment: https://www.dxomark.com/canon-eos-r8-sensor-test/

Like with most specs: nothing beats actual testing for your intended use case.

 

[neuroanatomist]

You are right, as usual. Thanks. The Nikon Z7 and Z9, also 45mpx, have a lower DR performance as measured by 'Photons to Photos' compared to the R5. In fact, the Nikon D850, which is also BSI but released in 2017 and performs almost identically to the Z7, performs better than the Z9 at ISO settings up to about ISO500, and thereafter comparable to the Z9. I also tried to compare the Z5, FSI but only 24mpx, has comparable or better DR performance as the Z7 . As it is not 45mpx, I am not sure if that is due to the effect of pixel size (improve DR compared with 45mpx) counter-balanced by the Z5 using a 'cheaper' sensor. Overall, it does seems that the R5 has one of best DR despite being FSI. It would be interesting to see how Canon's BSI perform in terms of DR when they appear. Based on Nikon, however, the BSI does not seem to confer a DR advantage, but again, the comparisons might be flawed as I don't quite know how the 'quality' of the sensors compared. Overall, it is perhaps better to just stick to actual measurements rather than making assumptions about DR performance based only on written specs.

The main point of BSI is to allow a higher density of photosites, and any DR advantage it confers is only apparent at those higher densities, basically with pixel sizes under 2 microns. At this point, that’s smartphone cameras only. IMO, the use of BSI on APS-C and FF sensors amounts to a marketing tactic.

 

[AlanF]

The main point of BSI is to allow a higher density of photosites, and any DR advantage it confers is only apparent at those higher densities, basically with pixel sizes under 2 microns. At this point, that’s smartphone cameras only. IMO, the use of BSI on APS-C and FF sensors amounts to a marketing tactic.

You mean the use of BSI is use of BS indeed?

 

[3kramd5]

The main point of BSI is to allow a higher density of photosites, and any DR advantage it confers is only apparent at those higher densities, basically with pixel sizes under 2 microns. At this point, that’s smartphone cameras only. IMO, the use of BSI on APS-C and FF sensors amounts to a marketing tactic.

These days I suspect it is done to improve electronic shutter operation. BSI and stacked (which are BSI) read faster than FSI, typically (though not exclusively).

 

[EOS 4 Life]

BSI and stacked (which are BSI) read faster than FSI

Not necessarily but it is a pretty safe assumption.
Stacking a sensor does not automatically make an image sensor any faster.
However, modern stacked sensors are designed to be faster by using techniques that are not possible with FSI.

 

[3kramd5]

Not necessarily but it is a pretty safe assumption.
Stacking a sensor does not automatically make an image sensor any faster.
However, modern stacked sensors are designed to be faster by using techniques that are not possible with FSI.

Right. Thats essentially what I’m trying to convey. I don’t think BSI is about Dynamic Range, and I don’t think fabs would change their process (expensive) for marketing reasons. *Telling consumers* that it’s BSI or Stacked is certainly a marketing tactic, but the *use* is for overall functionality, in my estimation speed and elimination of mechanical shutters being the goal.

 

[neuroanatomist]

These days I suspect it is done to improve electronic shutter operation. BSI and stacked (which are BSI) read faster than FSI, typically (though not exclusively).

Stacking, yes. BSI, no. Again, the R5 along with R6II/R8 are FSI with a faster readout speed than FF BSI sensors. So…more BS.

 

[3kramd5]

Stacking, yes. BSI, no. Again, the R5 along with R6II/R8 are FSI with a faster readout speed than FF BSI sensors. So…more BS.

Can you stack an FSI sensor?

 

[neuroanatomist]

Can you stack an FSI sensor?

I’m not sure, but it’s not germane to the issue at hand – namely, the rationale for using BSI (not stacked) sensors in ILCs. If they offer no better DR and no faster readout than FSI, what do you think is the reason?

You mention ‘speed and elimination of mechanical shutters’, that certainly makes sense for stacked sensors (which are BSI, as you state), but that does not apply for BSI alone, as the readout speed of the R5/R6II/R8 show.

BSI and stacked (which are BSI) read faster than FSI, typically (though not exclusively).

To reiterate, the above statement is incorrect. Stacked sensors typically read out faster than FSI sensors. BSI confers no advantage in terms of readout speed.

 

[3kramd5]

If they offer no better DR and no faster readout than FSI, what do you think is the reason?

My thinking is this, and obviously I could be wrong: Ultimately, they want to obsolete mechanical shutters. I suspect that BSI architecture is required to facilitate stacking since it puts the wiring elements in the correct location. Therefore they set up their fab process for BSI. Whether they stack or not then becomes a question of what they're looking to achieve from a given product, but they don't need to maintain two fabs, one for FSI and one for stacked.

Perhaps my verbiage isn't accurate as it pertains to BSI and speed, but I don't think there's any world in which they would eat the cost of setting up a BSI fab as a marketing tactic.

 

[P-visie]

My thinking is this, and obviously I could be wrong: Ultimately, they want to obsolete mechanical shutters. I suspect that BSI architecture is required to facilitate stacking since it puts the wiring elements in the correct location. Therefore they set up their fab process for BSI. Whether they stack or not then becomes a question of what they're looking to achieve from a given product, but they don't need to maintain two fabs, one for FSI and one for stacked.

Perhaps my verbiage isn't accurate as it pertains to BSI and speed, but I don't think there's any world in which they would eat the cost of setting up a BSI fab as a marketing tactic.

Here (link) is a 2021 DPR article which explains the benefits of a stacked sensor (it’s about the upcoming Z9, sorry ).A stacked sensor has the fast readout speeds needed for shutterless camera’s.
Stacked sensors and BSI sensors are more expensive to produce than FSI sensors. Canon might maintain FSI sensor production for “lower end“ sensors.

As for BSI marketing, Sony has beaten that drum to their advantage.

 

[neuroanatomist]

Here (link) is a 2021 DPR article which explains the benefits of a stacked sensor (it’s about the upcoming Z9, sorry ).A stacked sensor has the fast readout speeds needed for shutterless camera’s.
Stacked sensors and BSI sensors are more expensive to produce than FSI sensors. Canon might maintain FSI sensor production for “lower end“ sensors.

As for BSI marketing, Sony has beaten that drum to their advantage.

Not questioning the advantages of stacked sensors. Still not seeing any technical advantage to BSI sensors in the APS-C to FF range.

 

[Dragon]

I've just booked flights to Hong Kong for a conference and will have a day or two at either end for a spot of bird photography. Do I travel with just a single piece of cabin luggage with the R7/RF 100-400mm in it or take the R5 or R7 with the RF 100-500 in a small allowable second bag? It's not a bad choice either way - we are so lucky now for light travel possibilities. (The big boy 200-800mm is not for this trip.)

In Hong Kong, you need something wider. The canyons of the city are very deep. Take the R7 and at least the 18-150. If you have the 10-18, slip that in somewhere as well. there are places where you can use the 100-400, but not much in the city proper.

 

[AlanF]

In Hong Kong, you need something wider. The canyons of the city are very deep. Take the R7 and at least the 18-150. If you have the 10-18, slip that in somewhere as well. there are places where you can use the 100-400, but not much in the city proper.

Thanks for the advice. I've been a regular visitor to Hong and visit places like Mai Po nature reserve, the Peak and various parks where I use telephotos. it's not very good for birding but I can usually get some bulbuls etc I don't see back home.

 

[Denevans]

I would love to see an R7II. I use the R7 frequently for birding and backpack hiking where its low weight especially with ”kit” lenses and greater reach are greatly appreciated. What I would really like is an APS -C ultra zoom. The reduction in weight over the RF100-400 f5.6 -8 could be wonderful. An RF-S 100-400 could be as much as 0.244 times the weight - 1 quarter! Let me explain. For the same f-stop the 1.6 crop factor says the same f-stop lens has 1/1.6x the diameter (and focal length). The volume scales as the cube of the diameter so: 0.244. Woops! You say surely refraction is different. No. The refractive angles are all controlled by Snells Law which says the ratio of the sines of the angles of refraction is controlled directly and only, by the ratio of the refactive indices of air and glass, nsub_air/nsub_glass - whatever the lens is made out of (sometimes optical plastic), if these materials are the same in the FF and the APSC lens the refractive angles are the same in both lenses. In both lenses all the distances scale by 1.6 but all angles remain the same. (Similar triangles etc). The main thing I know of that doesn’t scale between the two lenses is the wavelength of light so diffraction will generally be worse in APSC. However if we operate both lenses in the regime where diffraction is too small to be a concern diffraction is irrelevant. Other things thatr don’t scale are screw sizes, lens housing material thickness and the actual lens mount which is RF for both FF and APSC. If someone knows about something I am forgetting about, I would love to know. Apart from these few items everything else should scale approximately as 0.244! I would love to see an RF-S 100-400 f5.6-8 or the RF-S version of the recent RF 200-800 Which would in fact be 320-1280!

 

[Dragon]

I would love to see an R7II. I use the R7 frequently for birding and backpack hiking where its low weight especially with ”kit” lenses and greater reach are greatly appreciated. What I would really like is an APS -C ultra zoom. The reduction in weight over the RF100-400 f5.6 -8 could be wonderful. An RF-S 100-400 could be as much as 0.244 times the weight - 1 quarter! Let me explain. For the same f-stop the 1.6 crop factor says the same f-stop lens has 1/1.6x the diameter (and focal length). The volume scales as the cube of the diameter so: 0.244. Woops! You say surely refraction is different. No. The refractive angles are all controlled by Snells Law which says the ratio of the sines of the angles of refraction is controlled directly and only, by the ratio of the refactive indices of air and glass, nsub_air/nsub_glass - whatever the lens is made out of (sometimes optical plastic), if these materials are the same in the FF and the APSC lens the refractive angles are the same in both lenses. In both lenses all the distances scale by 1.6 but all angles remain the same. (Similar triangles etc). The main thing I know of that doesn’t scale between the two lenses is the wavelength of light so diffraction will generally be worse in APSC. However if we operate both lenses in the regime where diffraction is too small to be a concern diffraction is irrelevant. Other things thatr don’t scale are screw sizes, lens housing material thickness and the actual lens mount which is RF for both FF and APSC. If someone knows about something I am forgetting about, I would love to know. Apart from these few items everything else should scale approximately as 0.244! I would love to see an RF-S 100-400 f5.6-8 or the RF-S version of the recent RF 200-800 Which would in fact be 320-1280!

You are missing one critical point. The diameter of the objective in a telephoto cannot be smaller than FL/f, and FL is FL, so the lens will not get shorter. Hence, there is essentially no saving in size for an APS-c telephoto lens with the same actual focal length and brightness. If what you say were true, then the Tamron 18-400 (APS-c lens) should be much smaller than the RF 100-400 (FF lens), but it is actually a bit bigger and substantially heavier. Another example is the new OM systems 150-600. It is a rebadged and mount-adjusted FF Sigma lens and that is for M4/3, which is even smaller than APS-c. All you have to do is look at the OM systems, Panasonic, and Fuji lens lineups to see that what you are suggesting is not real. This is also the reason Canon, Sony, and Nikon do not make APS-c specific telephoto lenses. There is so little to gain that there is no point. It is true that some of the rearward elements could be made somewhat smaller, but clearly not by enough to make such lenses worth the effort.

In a bit of irony, one way to make a telephoto shorter is to add magnification (as in internal teleconverter). This is the strategy for the RF 800 and 1200L lenses. The consequence is that with only slightly larger elements in the TC section,. such a lens will illuminate a sensor twice the size of FF, so a smaller lens works on a bigger sensor.

 

[neuroanatomist]

Let me explain.

You can argue with physics, but physics will win.

Every. Single. Time.

 

[Dragon]

My thinking is this, and obviously I could be wrong: Ultimately, they want to obsolete mechanical shutters. I suspect that BSI architecture is required to facilitate stacking since it puts the wiring elements in the correct location. Therefore they set up their fab process for BSI. Whether they stack or not then becomes a question of what they're looking to achieve from a given product, but they don't need to maintain two fabs, one for FSI and one for stacked.

Perhaps my verbiage isn't accurate as it pertains to BSI and speed, but I don't think there's any world in which they would eat the cost of setting up a BSI fab as a marketing tactic.

The discussion and development of FSI vs BSI has been going on for over 15 years and both processes have improved dramatically. The challenges for BSI lie in the difficulty of uniformly back-lapping wafers to extreme thinness (and the challenge of handling a 20 or 30 cm wafer that is only 10 microns think) and also in controlling pixel crosstalk. FSI has the challenge of keeping wiring out of the way of the pixels and/or concentrating the light with micro lenses to avoid the wiring. This article from over a dozen years ago does a very good job of explaining the tradeoffs up to and including stacked and multi-stacked sensors. Sony was not the first out the gate here, but they did get on the bandwagon fairly early. https://www.vision-systems.com/home...mizing-manufacturing-for-a-sensitivity-payoff The bottom line is that BSI sensors have some advantages, but there are a number of tradeoffs that make the choice much more difficult. Clearly an FSI sensor is going to be inherently more mechanically reliable than either a BSI or stacked sensor. One approach that is not mentioned in the article is a BSI sensor that is completely grown on an insulating substrate (i.e. glass or sapphire). This would resolve the thickness uniformity issue and provide excellent mechanical stability. The light would pass through the transparent substrate layer and strike the "backside" grown photosensitive layer. I have not seen any literature along these lines, but I have no doubt that someone is pursuing that path as silicon on sapphire is a very mature technology.

 

[Denevans]

You are missing one critical point. The diameter of the objective in a telephoto cannot be smaller than FL/f, and FL is FL, so the lens will not get shorter. Hence, there is essentially no saving in size for an APS-c telephoto lens with the same actual focal length and brightness. If what you say were true, then the Tamron 18-400 (APS-c lens) should be much smaller than the RF 100-400 (FF lens), but it is actually a bit bigger and substantially heavier. Another example is the new OM systems 150-600. It is a rebadged and mount-adjusted FF Sigma lens and that is for M4/3, which is even smaller than APS-c. All you have to do is look at the OM systems, Panasonic, and Fuji lens lineups to see that what you are suggesting is not real. This is also the reason Canon, Sony, and Nikon do not make APS-c specific telephoto lenses. There is so little to gain that there is no point. It is true that some of the rearward elements could be made somewhat smaller, but clearly not by enough to make such lenses worth the effort.

In a bit of irony, one way to make a telephoto shorter is to add magnification (as in internal teleconverter). This is the strategy for the RF 800 and 1200L lenses. The consequence is that with only slightly larger elements in the TC section,. such a lens will illuminate a sensor twice the size of FF, so a smaller lens works on a bigger sensor.

 

[Denevans]

You miss the point. f = dia/fl. In apsc the when the sensor size reduces by 1/1.6 the dia goes down by 1/1.6 and so to keep f fixed the actual fl must reduce by 1/1.6 too. This is why telephoto lenses in camera phones r so small.
The terminology is also at fault. If I put a ff 100-400 on apsc is delivers 160-640. If it had its actual focal length reduced by 1/1.6 it would be shorter (which is what I am talking about above) but still deliver 400 mm equivalent (not. 640). In what I said in my post above I am talking about equivalent of 100- 400 delivered not 160- 640 which would need the same actual physical fl in mm as a ff 100-400. To deliver the equivalent of the same f number and the same fl on Apsc, everything scales down by 1/1.6 - dia and fl.
I think we r in agreement. Its just a confusion in language.