Will Canon Withdraw from the Megapixel War?

[WarStreet]

There are posts talking about diffraction limitation. We should also look at it from a different angle. The diffraction limit is cause by the lens, not the sensor. At lower resolution sensor, we just never see it. With high resolution sensor, the diffraction limitation just reduce the high resolution sensor to a lower resolution sensor. Example: 18 Mp diffraction limit is f 6.3 while 10 mp diffraction limit is f 10.6 (??) . If you set the lens at f11 with a 18 mp sensor, you just reduce the actuall resolution of the 18 Mp to be 10 Mp, assuming a good lens is used.

Diffraction is not that evil, we can go above diffraction limit without caring much about it.

Using 18-55 IS II lens at the 18mm on the center in DXO (I hate their new site design, and it is not working well with chrome)

400D @ f11 gives 39 lp/mm
7D @ f11 gives 47lp/mm 20% better then the 400D

400D @ f22 gives 31 lp/mm
7D @ f22 gives 34 lp/mm 10% better.

Using 100mm macro IS lens on the center

400D @ f11 gives 38 lp/mm
7D @ f11 gives 46 lp/mm 21% better

400D @ f32 gives 24 lp/mm
7D @ f32 gives 24 lp/mm Here we hit a performance limit.

 

[WarStreet]

I ran a set of exposures where the clear part of the step wedge (a Stouffer 21-step wedge, where each step is 0.5 stops) was just shy of clipping, and the bare backlight was clipped. The DR tests were done with the goosenecks off, so the only light was the backlight for the wedge. 'Analysis' was qualitative, looking pixel values under the dropper in ACR, seeing where the steps could no longer be distinguished.

Maybe a stupid question, but when you are using the dropper in ACR, are you using this with the native exposure only, or also after under and overexposing the image in ACR trying to get the most from the RAW file ?

Also, do you think there are disadvantages when using Highlight tone priority ? I think I do manage to get something more with HTP, and I don't care of the negligible extra noise.

 

[Loswr]

...when you are using the dropper in ACR, are you using this with the native exposure only, or also after under and overexposing the image in ACR trying to get the most from the RAW file ?

Also, do you think there are disadvantages when using Highlight tone priority ? I think I do manage to get something more with HTP, and I don't care of the negligible extra noise.

Just the native exposure, no pushing or pulling to expand the exposure.

The disadvantage to HTP is the one you named and don't care about. The extra stop of highlight headroom results from the camera using an ISO one stop lower than the one you select (which is why ISO starts at 200 with HTP enabled), but exposing at the indicated aperture/shutter speed. The intentional underexposure allows the file to be pushed in-camera for more highlight headroom. The tradeoff is one stop of extra noise in the shadows, especially at the lower ISOs where we tend to shoot. Personally, I don't find the increased shadow noise negligible - it's actually differentially greater than what the noise would be for a proper exposure.

 

[Eagle Eye]

It's not really a megapixel war as much as it is an effort to push sensor technology to the limit. Canon has no competition; they've just been pushing forward on their own to design sensors that reach up the upper limits of what a traditional SLR lens can resolve. Keep in mind that recent Canon cameras have offered the sRAW format for those who are file size conscious. In the next few years, I would anticipate Canon focusing more on issues you mention, such as dynamic range. For all the whining about Canon's lack of ingenuity, they have set the bar for resolution and can only improve from there. If Canon is designing a new small sensor for a mirrorless, I'd expect to see some of the gifts of that R&D showing up in the D!GIC 5 and 6 processors.

 

[WarStreet]

Just the native exposure, no pushing or pulling to expand the exposure.

Do you think that the pushing/pulling expansion of exposure is just an ACR trick or true native DR gain hidden in the raw file ? I don't think it is a trick, since after the pushing/pulling, patches of 255 and 0 RGB values turn to different and distinguishable values both by the dropper and also visually.

The extra stop of highlight headroom results from the camera using an ISO one stop lower than the one you select (which is why ISO starts at 200 with HTP enabled), but exposing at the indicated aperture/shutter speed. The intentional underexposure allows the file to be pushed in-camera for more highlight headroom.

I asked you about the disadvantages since you did not used HTP in your test, and since I use it frequently I was curious about your opinion.

I did had a vague idea that HTP works along the lines of what you have explained, but never put much thought about it. I believed that HTP happens before the creation of the raw file, and that it really does increase DR and that it cannot be reproduced by post processing the raw with the pushing/pulling mentioned before. But maybe all this is true only if I compare two shots with similar exposures with and without HTP. But what about if I compare a shot with ISO 200 with HTP, and a shot with ISO 100 with one stop underexposure without HTP together with raw PP, will I get the same results ? For this comparison we have to assume we are shooting something with a DR at least 1 stop higher than the camera capability.

Personally, I don't find the increased shadow noise negligible - it's actually differentially greater than what the noise would be for a proper exposure.

A long time ago, I was trying to shoot low light sports and I really tried to get all I can from the camera trying to find the best compromises. I need to use fraction ISO and tried to simulate this with underexposure. With some tests I did, and also from DXO full SNR, I concluded that the difference is very small for my intended needs. The results varies depending the ISO and grey scale % compared. For my 500D, comparing ISO 3200 @ 18% grey scale with ISO 1600 @ 9% grey scale gives similar results. On the other hand using ISO 100 & 200 shows that correct exposure gives the best results. I hope I am interpreting this data correctly!

 

[moreorless]

Nikon's way, as I understand it, is smarter.

It released the D3X with high resolution for those who need / want it, and a full year later the D3S with half the resolution for those who don't.

Nikon are only offering that choice right at the top of the range though were as Canon are offering something similar(obviously half crop) at 1/3rd the price(and almost 2/3rds the weight) with the 5D mk2/7D.

Personally I feel that the MP race on crops may well be reaching its limate, I'm sure things will go past 20 as the numbers just too strong a draw but beyond that I see either ISO/Video or something new being a factor.

On FF though I'd say the race is far from over, at the very top end I'm sure both Canon and Nikon will want to close the gap with the 645D. Lower down we may see a spilt with a seperate Canon ISO/speed 5D/3D but I think there are alot of amature users who'd welcome 40 megapixels, espeically with the cost of large prints dropping all the time

 

[UncleFester]

On FF though I'd say the race is far from over, at the very top end I'm sure both Canon and Nikon will want to close the gap with the 645D.

They can want, but I'm sure they know they'll have to increase sensor size as well. Different beasts all together.

 

[macgregor mathers]

Nikon's way, as I understand it, is smarter.

It released the D3X with high resolution for those who need / want it, and a full year later the D3S with half the resolution for those who don't.

Nikon are only offering that choice right at the top of the range though were as Canon are offering something similar(obviously half crop) at 1/3rd the price(and almost 2/3rds the weight) with the 5D mk2/7D.

The Nikon D3S has 12MP, half that of the D3X. The Canon 7D has 18MP, 85.7% that of the 5Dmk2.

Beside the rather small difference in resolution, the 7D is (as you noted) a crop camera, with all the baggage that comes with it.

 

[Lawliet]

On FF though I'd say the race is far from over, at the very top end I'm sure both Canon and Nikon will want to close the gap with the 645D. Lower down we may see a spilt with a seperate Canon ISO/speed 5D/3D but I think there are alot of amature users who'd welcome 40 megapixels, espeically with the cost of large prints dropping all the time

If you want MF quality you have to deal with the AA filter first. Whats the point in recording frequencies that have been mangled?

On a related line I don't see many amateur users how could use 40MP. Prints are cheap(have been for quite a while, 24" or 36" printers are in the same price bracket as a single lens for that resolution.), but good tripods aren't, and they aren't only a financial commitment, but also in terms of handling. Affordable flashes that actually stop motion? IGBT switching doesn't help if you need power, so we're back at bank breaking.
How many amateurs have&use that kind of equipment? Gives you a good idea of reaping actual benefit vs. numbers game.

 

[Edwin Herdman]

Whats the point in recording frequencies that have been mangled?

What's the point in having an AA filter in the first place, then? Why do all the manufacturers use them on all sub-MF bodies?

Luminous-Landscape has an interview with a Pentax rep from the trade show in Japan where the 645D was introduced. Moire was already a concern for them with the 645D. For many sorts of photographs, lacking an AA filter would simply be a horrible idea (the usual "Wall of Shame" and cityscape test photos more urban-based photographers do would have some obvious moire problems). MF backs usually had less well-developed technology and so AA filters were one less thing to worry about - for the most part they overpowered the problem of false details and false patterns with more resolution.

Anyhow, the whole "megapixel war is bad" issue is overstated. With the significant exceptions of bigger monitors needed for 100% display sizes, and all the data bloat related problems that come with bigger pixel counts, increasing pixel counts has been generally a good phenomenon. Nikon's one big weakness so far has been that they are reliant on others for sensors to a degree that Canon and Sony (for example) aren't - so comments from Nikon like that they are looking for "a better balance" between resolution and ISO is sheer marketing deception at worst when they are reliant on Sony (who has, remember, the Alpha line) for sensors.

 

[Lawliet]

Whats the point in recording frequencies that have been mangled?

What's the point in having an AA filter in the first place, then? Why do all the manufacturers use them on all sub-MF bodies?

Cutting the spectrum at nyquist's limit. Not below it. The latter is where the mangled mess comes from.
Look at files from an E-5 for example...better then the artifacts introduced by raising local contrast back up.

 

[jseliger]

I'm surprised no one (to my knowledge) has linked to this Ars Technica article on the supposed end of the megapixel war. Some people obviously need more, but I think the larger point is that those people comprise a smaller and smaller part of the overall camera-buying population. I've seen the various articles about how the iPhone is now the most used "camera" on Flickr, Facebook, and so forth.

I have a 3GS that I used to take the picture that appears in this links post simply because I happened to have the camera on me. Sure, I usually have a Canon SD 800 in my pocket most of the time, and a gifted Rebel XTi that comes out for occasions and sometimes for dirty pictures, but for a lot of purposes any of the three is "good enough" in terms of megapixel count.

I'd rather have the dynamic range, aperture, and so forth, than more megapixels. That's not to say I'd gratuitously turn down more, but, say, 10 as opposed to 15 megapixels doesn't bother me. If you said, "You can have pretty much any camera you want today," I'd probably choose an s95 because it's easy to carry and I'm likely to have it with me.

 

[Lawliet]

I'd rather have the dynamic range, aperture, and so forth, than more megapixels. That's not to say I'd gratuitously turn down more, but, say, 10 as opposed to 15 megapixels doesn't bother me.

Thats where I like double blind tests - e.g. I printed pictures from the same shooting but taken with different cameras in large format. Ask unsuspecting (but not inexperienced) Audience to judge image quality. Conclusion: 12 or 21 MP, it doesn't matter.
For dynamic range, try processing video shot with a 5D2 right after your worked with footage from say an arri alexa. I want the same DR for photography. 8)

 

[WarStreet]

I'm surprised no one (to my knowledge) has linked to this Ars Technica article on the supposed end of the megapixel war.

That article is a "copy and paste" you read all over the web in forums from guys expressing their opinion and not necessary knowing the technical aspect of the subject.

The description of the author on that website:

Chris Foresman
"Karaoke Wizard"
et Contributing Writer
Chris is a high-functioning high school and college dropout. A modern-day Renaissance man, he has spent time releasing indie records, selling shoes, making copies, designing t-shirts, taking photographs, teaching graphic design, providing tech support for Linux webhosts, and generally raising a ruckus. Now a full-time tech-writing idiot savant, Chris has written about music, photography, vegetarian dining, and of course Apple. In his free time he enjoys watching movies, shopping at Target and IKEA, singing karaoke, eating brunch, and drinking beermosas.


As I have said previously in this thread, it is true that the more technology improves, the more it reaches or surpasses the needs of users and it is applicable for every spec, not just for MP count. That's why good companies like Canon improve on all aspects of the technology so to accommodating different needs and avoiding bottlenecks.

 

[moreorless]

On FF though I'd say the race is far from over, at the very top end I'm sure both Canon and Nikon will want to close the gap with the 645D.

They can want, but I'm sure they know they'll have to increase sensor size as well. Different beasts all together.

Theres never going to offer the same thing I agree but if say the Canon IDs mk 4 has 30+ megapixels that is likely to mean the draw towards a 645 is a little less strong than if it only has 25.

 

[Lawliet]

Potential 645 customers know what they are doing, should be beyond falling for catalog numbers.

 

[moreorless]

Potential 645 customers know what they are doing, should be beyond falling for catalog numbers.

For a sizeble number of people though I'd guess its not going to be cut and dry whether to move up to MF and more Megapixels on FF is one reason not to.

I don't think you can rule out the effect of the growth in MF on the amature markets view of whats desireble either, even if its not actually an option for people.

 

[WarStreet]

I won't rule out the possibility of a 1dsIV with 40mp, but still it won't resolve the same detail of a 40mp MF, using bigger lenses.

 

[Rocky]

Potential 645 customers know what they are doing, should be beyond falling for catalog numbers.

For a sizeble number of people though I'd guess its not going to be cut and dry whether to move up to MF and more Megapixels on FF is one reason not to.

I don't think you can rule out the effect of the growth in MF on the amature markets view of whats desireble either, even if its not actually an option for people.

I hope we are not starting the same debate about Leica VS Hasselblad in the film days again.
MF in digital will give better picture, especially in large enlargement due to the larger size of individual pixel and hence lowe rnoise, better dynamic range. However, the size weight and cost will make most people, including professionals think twice before jumping into it. Just look at the price of the lenses and digital body of Hasselblad.
On the other hand, FF will give you much better mobility and cost saving. So MF is for amature with super deep pocket and with an assistant.

 

[moreorless]

I hope we are not starting the same debate about Leica VS Hasselblad in the film days again.
MF in digital will give better picture, especially in large enlargement due to the larger size of individual pixel and hence lowe rnoise, better dynamic range. however, the size weight and cost will make moat people, including proffessionals think twice before jumping into it. just look at the price of the lens and digital body of Hasselblad.
On the other hand FF will give you much better mobility and cost saving. So MF is for amature with super deep pocket and with assistant.

I'm not really debating the merits of each system so much as speculating the impact a growth in digital MF might have on FF. A high end Hasselblad is obviously something totally different but the 645D is getting into the same kind of ballpark as a 1Ds so on some level its competision.