Is a brand new 32.5mp APS-C sensor from Canon on the way? [CR1]

AlanF

Desperately seeking birds
CR Pro
Aug 16, 2012
12,346
22,521
Diffraction? Don´t give me that BS. Look at what sensors are mounted in mobile phones. Obviously, these are diffraction limited, but that means that 32Mpx APS-C sensor is nowhere near being diffraction limited. I compared my EOS M and EOS M6, and the difference was significant (small resolution difference considered). Right now, I sold my M, M6, and use my mums M100 I gave her. Any other camera needs to have more resolution in order to motivate me to buy it...

Mobile phones are usually not diffraction limited as they have have very wide aperture lenses, like f/1.8 for the new iPhones, which is less than their DLA of f/2.25.
 
  • Like
Reactions: 1 users
Upvote 0
Why all the negativity about 33% more resolution? Canon has almost always been slightly less DR than Sony sensors while being slightly higher pixels. Somewhere around 5 years ago they missed knowing how big 4k video was going to be, and how popular mirrorless was going to get. But a new sensor design from the ground up takes time. The lack of truly new sensors in the past few years shows that most of their team are working on something else. This and the associated full frame version of this sensor will most likely be better than the current Sony sensors and probably as fast or faster too. However I'd expect the next generation of Sony to be faster yet.
 
Upvote 0
Jul 28, 2015
3,368
570
32.5 MP on APS-C? I am not sure, if this is too much.

When you realize that 24 MP already delivers a DLA (Diffraction Limited Aperture) of f/6.0, according to TDP., as mentioned in the EOS 80D test.
What's the use in more resolution when you lose it again at an apeerture somewhere between f/5.0 or f/6.0?

It is always a battle between optical resolution from increasing the pixels vs diffraction. IMO ore pixels always wins (30D vs 7D vs 7D2) - unless you can show me the same image shot with high and low MP presented at the same viewing size that shows me any different.
 
  • Like
Reactions: 1 user
Upvote 0
I notice a number of comments about the sharpness, or lack thereof, of the 7DMll. I’ve been using one for several years along with a 5DMll and Fuji X-T10. Quite frankly, I find the sharpness of the 7D more than sufficient. I don’t find softness an issue at all. This is a matter of personal perception, I guess. I would prefer a camera with an AA filter. On the other hand, I’m ambivalent about the files coming from the Fuji. They sometimes have an unnatural “graphic” quality to them that I find unpleasant. Each to his/her own.

Agreed. I've always found crappy lenses to be the limiting factor in the images I get from my 7D II. If the lens is sharp, the images will be sharp (the Tamron G2 70-200, Canon 100-400 MII, the Sigma 60-600, and the Canon 500 F4 II I just bought come to mind as being sharp lenses).
 
  • Like
Reactions: 1 users
Upvote 0
I'm anticipating an EF-M shake-up in 2019. I think Canon is possibly making an M7 with this sensor, which will launch alongside the M5 MkII that comes with a new 24-26MP sensor. Both these cameras will use the E-17 battery used by the current EOS RP, M5, M6, 250D.

Canon M7:
This M7 will be the EF-M flagship, effectively replace the 7D MkII, and add some weather-sealing as well as a top display. Such a camera will also make room for some new M-mount lenses down the line that wildlife shooters will appreciate since they will be a better fit for this larger M body. This will also be the first M camera with a battery grip option. Regardless, I expect the now patented EF-M 15-130mm f/3.5-6.3 to launch with this M7 as a kit option too. Metal mount of course, unlike the current 18-150mm. The M7 will be pricey, costing more than the EOS RP.

Canon M5 MkII
As for the next M5, it will come with a vari-angle screen, new 24-26MP sensor, and all the dials found on the existing M5. It won't be enough for existing M5 users to upgrade but it will seriously target the XT-3 and A6400 at $1300 with new 16mm-50mm kit lens. The elephants in the room will still be the lack of IBIS and cropped 4K but it will a nice upgrade for those looking to step up from an M50.

Photokina 2020:
  • 90D with the new 32mp sensor, still a DSLR with headphone jack
  • M6 MkII
  • M200
  • Wide angle EF-M prime
  • 50mm EF-M prime without IS but hopefully with f/1.4

There is imo no chance something like 7D III is going to happen in an EF-M mount. First - ergonomics / size aspects - you want your 7D III being larger than M5, right? Second - with EF-M, there is no upgrade path to the RF mount lens. I think that if something like the 7D III is going to be released one day, it is going to be the RF version. And once that happens, it is going to be the last nail in the coffin of an EF-M mount, not that it will die off, but anchoring it definitely in the hobby segment.
 
  • Like
Reactions: 1 users
Upvote 0
If you are reach limited with telephotos, every small increase helps. A 600mm lens with a 1.16x linear resolution is equivalent to a 700mm lens. For me, that is worthwhile and I would welcome it.

To put it in perspective, a 1.4xTC gives only a 20% increase over a 1.16x increase in pixel resolution, and that comes with a degradation of IQ from the glass in the TC, an amplification of aberrations from the lens, an ~30 loss of field of view, a stop loss of aperture so pushing up the iso or slowing down the shutter speed, and slowing down of the AF. My own experience for telephoto work is that increasing the pixel density is a better alternative to using a TC (though I do use both for much of the time!).

A 1.4 TC is the same like DOUBLING the pixel count - it lifts a 600mm lens to a 960mm lens. Because the image area is blown up by a factor of 1.4² = 2 you loose 50% of the light per area = 1 stop. So a 1.4x TC is a 40% increase compared to the 16% increase from 24->32 MPixels.
 
Upvote 0
I am sure there are a lot of people out there happily shooting their 24 mp aps-c cameras stopped down to f8.0 and beyond, not noticing the effect of diffraction limitations.

Yes - for the web and e.g. 20x30cm prints. With the EF-M 32 you will see strong differences between f/5.6 and f/16. While with f/5.6 the lens is pixel perfect the resolution degrades to ~ 6MPix at f/16. Which is well enough for fine 20x30cm prints and maybe good enough for 30x40 prints.
If you want to print 100x150cm f/5.6 is the better choice with a 24MPix sensor and the above mentioned lens. But who prints 100x150cm ? I would like to do that but it is expensive and wall area is limited :)
EDIT: And the number of photos who are worth artistically is much more limited ...
 
Last edited:
Upvote 0

AlanF

Desperately seeking birds
CR Pro
Aug 16, 2012
12,346
22,521
A 1.4 TC is the same like DOUBLING the pixel count - it lifts a 600mm lens to a 960mm lens. Because the image area is blown up by a factor of 1.4² = 2 you loose 50% of the light per area = 1 stop. So a 1.4x TC is a 40% increase compared to the 16% increase from 24->32 MPixels.
You have misunderstood what I wrote: a 1.4x is a 20% increase over a 1.16x increase, by which I meant 1.4/1.16 = 1.2, in other words a 1.4 gives 20% more resolution than a 1.16!
 
Upvote 0
Jul 28, 2015
3,368
570
Yes - for the web and e.g. 20x30cm prints. With the EF-M 32 you will see strong differences between f/5.6 and f/16. While with f/5.6 the lens is pixel perfect the resolution degrades to ~ 6MPix at f/16. Which is well enough for fine 20x30cm prints and maybe good enough for 30x40 prints.
If you want to print 100x150cm f/5.6 is the better choice with a 24MPix sensor and the above mentioned lens. But who prints 100x150cm ? I would like to do that but it is expensive and wall area is limited :)
EDIT: And the number of photos who are worth artistically is much more limited ...

You shoot whatyou have to. If you are doing landscape f5.6 will probably give too little DOF - sometimes these diffraction comments ignore the necessities of the art and end up being purely philosophical.
Can you show me an image where a high-res sensor 'diffration limited' photo has less resolution than a lower-res sensor image that (in theory) is less 'diffraction limited'??
 
Upvote 0
AlanF: Sony IMX 586, 1/2", f/2, 48Mpx. That makes it 1100Mpx for FF sensor area. That is pretty close to that point. I am not exactly going to argue the precise position of that point. It laso depends on what do you mean by the limit. I was arguing FOR more resolution, as dedicated cameras are nowhere near of being diffraction limited.

mb66energy: Teleconverters always have more losses than more pixels. There is glass in bitween, which "damages the image" to a point... Therefore if you had 1,4x more square amount of pixels, and just cutting the area, the outcome would be better than with TC.
 
Upvote 0
Jul 28, 2015
3,368
570
mb66energy: Teleconverters always have more losses than more pixels. There is glass in bitween, which "damages the image" to a point... Therefore if you had 1,4x more square amount of pixels, and just cutting the area, the outcome would be better than with TC.

But that's not how photography works (or rather how photographers think) , is it? you could treble the pixels on a sensor and people would still put a teleconverter on it because they will want even more reach. Otherwise people like me who started with a 8MP 30D would never use a teleconverter because sensors are now 30MP.
 
Upvote 0
If your having storage problems with 24mp crop sensor files you need to upgrade that 1990’s computer hardware your using

If you shooter with the basic 1-bit per pixel JPEG setting it’s like a 2-3MB file, but push it to the HQ setting and add RAW, and your looking at 30MB plus per picture. That 10X increase is what most people who shoot for quality and flexibility are referring to when they talk about storage sizes getting out of hand. There’s also a tax on the processing bandwidth. If you’re not using a fast SSD system, all the handing off of from one program or process to the next plus DAM system requirements can easily slow down a decent modern laptop, tablet or PC/Mac.

If everyone got what they wanted, e.g., RAW 14-bit video and 80MP lossless RAW stills, only the über-anal with exacting data management and strict, regular culling and backup procedures would be left standing. The majority of users who want these high data-rate features would be not just running out of room, but they’d also be misplacing files, failing to back everything up, leaving files on memory cards, and totally plugging the toilet.

It’s easy to give in to the allure (and maybe it’s a totally justifiable endeavor) but a lot of people with 40MP to 50MP cameras and professional video cameras have dealt with it, and though it gets almost zero attention in gear reviews and the groping pat downs of new equipment, it’s just as important to msnGe as any other part of the process.

So I fully agree with a better 24MP sensor on APS-C than a “new and improved” 33MP version. I think 20-24MP is the format’s reasonable upper limit anyway. Fujifilm’s new 26MP sensor doesn’t appear to provide any IQ improvements over the 24MP version—in fact it seems to take a small hit.

To me 32.5MP looks like the result of achieving an 8K video spec more than it does optimizing still photo quality. And given Canon’s refusal to give their photo cameras real full sensor 4K, this appears to be destined for a Cinema EOS camera anyway, so if this comes to a photo camera I’d be weary of the hype—since you might be taking one for the pro video team rather than scoring yourself a real upgrade.
 
Upvote 0
Mikehit: Sure! That wasn´t the point though. There was a claim with this meaning: "You are better off putting on the TC, instead of more pixels". No. More pixels are better, and more pixels with good TC are even better. But we already have teleconverters. We do not have pixels available to us :-D
You have a good point with photographers thinking though. It is not that easy to pick the right amount of zoom,magnification and pixels to "general random photography". You never know how wide angle of view you need to catch the right image. Sometimes it is too little, sometimes it is too much. That´s where more pixels help to a certain point, because you will end up with "more pixels on the bird" so to speak.
 
Upvote 0
AlanF: Sony IMX 586, 1/2", f/2, 48Mpx. That makes it 1100Mpx for FF sensor area. That is pretty close to that point. I am not exactly going to argue the precise position of that point. It laso depends on what do you mean by the limit. I was arguing FOR more resolution, as dedicated cameras are nowhere near of being diffraction limited.

mb66energy: Teleconverters always have more losses than more pixels. There is glass in bitween, which "damages the image" to a point... Therefore if you had 1,4x more square amount of pixels, and just cutting the area, the outcome would be better than with TC.

What do you mean by “dedicated cameras are nowhere near of being diffraction limited.”?

On APS-C at 24MP f/8, sometimes lower, is where diffraction starts to creep in, and on FF the same’s true for f/11 and 36MP. So if you’re talking about not being limited yet it seems like you’re talking about shooting at bright apertures and giving up high quality results with image that requires deep focus?

As far as I know, between Nikon, Canon and Sony there still hasn’t been an APS-C BSI sensor, let alone a stacked sensor. Those are improvements I think crop shooters would rather have in their day-to-day photography than giant MP counts that seem to be in the rightful domain of FF (unless that’s pushed to over 80MP too—in which case you can forget anything stacked and likely a top burst rate of 3FPS barring a ultra top secret decade long innovation that Canon’s about to release). Or finally buying modern sensors from Sony or maybe Samsung soon.
 
Upvote 0
FairlyKors: Exactly. The diffraction limit needs more specification to it, to discuss it properly.
Yes, diffraction starts to creep in with higher apertures. That can be taken as diffraction limitation. but at f/8, if you put 100Mpx sensor behind it, you still would get more resolved. That way, one could say you are not diffraction limited. My daily driver was Canon EOS M6 + 22mm f/2 lens. Most shots were taken at f/2, and the rest 99% at f/2,8-f/4. So the camera was not diffraction limited. The image was "sensor resolution limited" in cases the resolution was needed.
Sensor resolution definitely is cheaper than lens resolution at this moment. Yet it is put as some kind of niche, even by manufacturers. I do not understand that.
The war is over, image is good. Just put as many pixels on the sensor as technology allows, and that´s it. We´re done with this topic (now talking to manufacturers and marketing)...

AFAIK there was BSI FF sensor available, and I´d guess Sony has some BSI APS-C?
The more pixels request is not here as a sacrifice to other image quality aspects. It has been proven time and time again.
There is speed and storage issue always, but that is not problem for everyone. There is market for hi-res....
 
Upvote 0

AlanF

Desperately seeking birds
CR Pro
Aug 16, 2012
12,346
22,521
AlanF: Sony IMX 586, 1/2", f/2, 48Mpx. That makes it 1100Mpx for FF sensor area. That is pretty close to that point. I am not exactly going to argue the precise position of that point. It laso depends on what do you mean by the limit. I was arguing FOR more resolution, as dedicated cameras are nowhere near of being diffraction limited.

mb66energy: Teleconverters always have more losses than more pixels. There is glass in bitween, which "damages the image" to a point... Therefore if you had 1,4x more square amount of pixels, and just cutting the area, the outcome would be better than with TC.
I am also in favour or more resolution, so we are in agreement. But, I was pointing out the that the good cellphone cameras are not diffraction limited - they are about the same as an f/5.6 lens on a 5DSR or 7DII, a popular combination on this site. In contrast, the Nikon P1000 is about 3x over the DLA!
 
Upvote 0

Joules

doom
CR Pro
Jul 16, 2017
1,801
2,247
Hamburg, Germany
32.5 MP on APS-C? I am not sure, if this is too much.

When you realize that 24 MP already delivers a DLA (Diffraction Limited Aperture) of f/6.0, according to TDP., as mentioned in the EOS 80D test.
What's the use in more resolution when you lose it again at an apeerture somewhere between f/5.0 or f/6.0?
With all this talk about diffraction, I decided to have a little play with it myself.

I set up my 24 MP 80D with the sharpest lens I own, the Sigma 35mm 1.4. Put it on a tripod in the garden and shot some newspaper text behind a small bird feather in full sunlight. Attached is a compilation of the results. The upper row is made up of unprocessed (just whitebalance) crops from the center of the frame with all apertures I'd consider interesting. Below each is the same crop, deconvoluted in Photoshop CC using smart sharpen with the radius written below the aperture values. The radius where set by eye, looking at the contrast of the i in the "mit" to the right of the feather. I increased it, until tiny halos became visible, and dialed it back again slightly, to get it natural. Using some more advanced method for deconvolution may yield nicer results, but for now I get along with smart sharpen best.

Seeing as these are really hefty crops (about 350px wide each, from the 6000px full width), I'd say all the processed crops would pass as sharp at less magnification. At this magnification, you can start to see detail getting lost due to diffraction in the tip of the feather at f/11. And at f/16 it's quite noticable - but it's not really blurry. And thanks to the larger depth of field, it may actually look sharper over all, if the subject included any depth (This is just laying on the ground). It just lacks the finest of detail. And as for the 2.8 and 5.6 image - with more pixels you could pull out even more detail from those.

Also, I think with 32.5 MP, the pixel density is enough that even Canon won't hesitate to eliminate the AA filter, making the images all a bit more detailed yet again.

I think it may not be necessary for the most cases to go above 24 MP, but it certainly doesn't seem useless to me. And diffraction is not a thing that causes any image at higher apertures to become automatically useless or blury.
 

Attachments

  • Diffraction_Deconvolution-001.jpg
    Diffraction_Deconvolution-001.jpg
    1.6 MB · Views: 357
  • Like
Reactions: 4 users
Upvote 0