Pages: 1 ... 5 6 [7] 8 9   Go Down

Author Topic: Canon 5D Mark III official...  (Read 53544 times)

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13983
    • http://www.flickr.com/photos/bernardlanguillier/sets/
Re: Canon 5D Mark III : apparently no support for AF f5.6-f8.0?
« Reply #120 on: March 03, 2012, 05:50:41 pm »

Canon has not rulled out fixing the F8 on the D1X so it could follow the same tweak could be made to the new 5D3

Canon has not confirmed it will no longer make 1.3 crop sensors. It would not surprise me if the 7D2 was moved up from 1.6 to 1.3 Crop with improved AF etc.

That would be one great way to piss off big time the millions who have invested in APS-C lenses!

There are similar talk on the Nikon side about the D400 being a D800 with a D4 sensor, but I don't see this happening either. This would show a total lack of respect for the customers who have been supporting you all along.

Cheers,
Bernard

deejjjaaaa

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1170
Re: Canon 5D Mark III official...
« Reply #121 on: March 03, 2012, 05:53:58 pm »

One aspect of the higher resolution of the D800 is diffraction. On my 1Ds mkIII I see a clear difference in 100% details going from f/8 to f/11. The D800 would need to be stopped down to f/5.6 to get all details.

This means for landscape shooters that they either will not utilize the full resolution of the D800 or will have to do focus stacking. I'm certainly not saying that all landscape pictures need to to be stopped down to f/8 but a large percentage of how I shoot need to be stopped down to f/8 or f/11 and sometimes f/16 where a real serious loss of resolution is seen. Less so on a print up to a super A2 which the largest I can print on my Epson 3880. I have tested focus stacking using Helicon Focus and I probably have not quite fine tuned it yet. I see some types of pictures work just fine, but others have serious halo problems.

I'm not trying to defend Canon or saying that the D800 is the wrong product for landscape, but for sure the resolution advantage will not always be seen in reality.

ok, let us see... APS-C sensor, F11, K10D/10mp vs K5/16mp, 40/2.8 fix... shall be a good idea what FF @ F16 will do

F11 @ 10 mp = 2011 / 1950 / 1865 (center/border/extreme corner, lw/ph)
F11 @ 16 mp = 2372 / 2315 / 2284 (center/border/extreme corner, lw/ph)
Logged

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13983
    • http://www.flickr.com/photos/bernardlanguillier/sets/
Re: Canon 5D Mark III official...
« Reply #122 on: March 03, 2012, 06:10:14 pm »

One aspect of the higher resolution of the D800 is diffraction. On my 1Ds mkIII I see a clear difference in 100% details going from f/8 to f/11. The D800 would need to be stopped down to f/5.6 to get all details.

Yes... that has been the case for many years. You would see even better detail with your 1ds3 at f7.1. In fact f8 is already very far from the aperture you need to have good front to back DoF on many subjects, so is f11 in fact.

It really boils down to what you call critical sharpness. Deciding that 22mp is the perfect compromise is arbitrary at best.

MF is not better, very far from it. The key metrics affecting diffraction is pixel pitch. The truth is that the MF lenses used on an IQ180 start to suffer from diffraction at the same time D3x lenses start to suffer from diffraction, meaning at around f8. So what about 39mp backs? It is correct to say that the aperture starting to be affected by diffraction is higher since they have a larger pixel pitch... but this effect is more than compensated by the fact that you need to use a longer lens to reach the same composition everything else being equal.

Anyway you look at it you will find that the smaller format is superior to achieve more DoF.

Anyway, there are 4 well known solutions for this that have been used by photographers for years:
- focus on landscape styles that do not require everything to be in focus,
- compose so as to avoid having close subjects in your composition. Hansel Adams grand landscape comes to mind as a perfect example,
- use T/S lenses or movements
- use DoF stacking

I'm not trying to defend Canon or saying that the D800 is the wrong product for landscape, but for sure the resolution advantage will not always be seen in reality.

True, just like the potential of the 22/24mp cameras was already very challenging to tap into from a DoF standpoint. Yet, the areas that is sharp will benefit significantly from the added details for a given print size.

Cheers,
Bernard

dreed

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1715
Re: Canon 5D Mark III official...
« Reply #123 on: March 03, 2012, 06:43:52 pm »

Yes... that has been the case for many years. You would see even better detail with your 1ds3 at f7.1. In fact f8 is already very far from the aperture you need to have good front to back DoF on many subjects, so is f11 in fact.

This contradicts with my real experience but...

Quote
It really boils down to what you call critical sharpness. Deciding that 22mp is the perfect compromise is arbitrary at best.

I believe that the sensor in the 5D Mark III has been designed primarily for HD video performance.

The images that come out of the 5D Mark III are 5760 pixels wide at full resolution.

5760 = 1920 x 3   (1920 is the horizontal resolution of 1080p or HD video on BluRay/TV)

There is no way that is a fluke or arbitrary.

My money is on that sensor size allowing for cleaner/better/faster video to be extracted from it.

Unfortunately I don't know enough about how video is encoded from the sensor to know if this makes 3 to 1 pixel binning easy, allows moire to be almost reduced or something else.
« Last Edit: March 03, 2012, 07:25:14 pm by dreed »
Logged

Bernard ODonovan

  • Full Member
  • ***
  • Offline Offline
  • Posts: 220
Re: Canon 5D Mark III : apparently no support for AF f5.6-f8.0?
« Reply #124 on: March 03, 2012, 06:55:54 pm »

That would be one great way to piss off big time the millions who have invested in APS-C lenses!

There are similar talk on the Nikon side about the D400 being a D800 with a D4 sensor, but I don't see this happening either. This would show a total lack of respect for the customers who have been supporting you all along.

Cheers,
Bernard


I think Canon have already upset some 1.3 users. A 1.6 crop on a 1.3 sensor is easy to do, if the Canon crop lenses could also be made fit that would ease worries. Moving the 7D2 to 1.3 does not end 1.6 Camera's in any case. A Full frame upgrade for the old 1 series 1.3 sports Cams was always on the cards. Time will tell what Canon does next. They will respond where market demand takes them (hence their recent hint they would make a higher MP Cam to match Nikon if the market wants it)...
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Re: Canon 5D Mark III official...
« Reply #125 on: March 04, 2012, 01:57:12 am »

ok, let us see... APS-C sensor, F11, K10D/10mp vs K5/16mp, 40/2.8 fix... shall be a good idea what FF @ F16 will do

F11 @ 10 mp = 2011 / 1950 / 1865 (center/border/extreme corner, lw/ph)
F11 @ 16 mp = 2372 / 2315 / 2284 (center/border/extreme corner, lw/ph)


That's a good indicator. I just went through the exercise of comparing the same models of Canon lenses tested at Photozone where they have sometimes duplicated their results using two different bodies, the 8mp Canon 350D and the 15mp Canon 50D.

8mp on cropped format is equivalent to a little under the 22.3mp of the 5D3, in terms of pixel density or number of pixels per unit area of sensor. The 15mp of the 50D would be equivalent to a 38 or 39mp FF sensor that used the same pixels.

In all the examples I compared at Photozone, the same models of lenses consistently produced higher resolution at all aperture tested up to F11, in the centre of the image as well as the borders of the cropped frame, using the 50D body. I didn't notice any tests at F16 at Photozone, but my own tests have indicated there is still some resolution advantage at F16 using a 50D, although small. The cut-off point where there would be no resolution advantage to, say, a 39mp full-frame sensor, is beyond F16, probably around F20 and certainly F22.

Of course, it goes without saying that the greatest increases in resolution are always apparent at the apertures where the lenses are sharpest, usually around F4 to F5.6. Most lenses are pretty much equal at F11 and F16.

When a manufacturer produces an upgraded lens which is sharper than the previous model, one generally does not hear complaints that the additional sharpness is only apparent at F2.8 to F5.6 and therefore not of much use because DoF is so shallow at such apertures.  ;D
Logged

Walter Schulz

  • Full Member
  • ***
  • Offline Offline
  • Posts: 105
Re: Canon 5D Mark III official...
« Reply #126 on: March 04, 2012, 02:15:14 am »

That's a good indicator. I just went through the exercise of comparing the same models of Canon lenses tested at Photozone where they have sometimes duplicated their results using two different bodies, the 8mp Canon 350D and the 15mp Canon 50D.

Klaus is adamant about one point and always has been: The results may not be used for comparisions between different camera systems.

"Please note that the tests results are not comparable across the different systems. This does also apply for the new EOS tests based on the EOS 50D because of differences in the sensor system (e.g. AA-filter) as well as different RAW-converters."

Therefore your deductions are based on false assumptions and therefore worthless.

Ciao, Walter

PS: EF 100/2.8 USM Macro on 350D was tested to f/22
http://www.photozone.de/canon-eos/167-canon-ef-100mm-f28-usm-macro-test-report--review?start=1
« Last Edit: March 04, 2012, 02:22:19 am by Walter Schulz »
Logged

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051
Re: Canon 5D Mark III official...
« Reply #127 on: March 04, 2012, 03:03:12 am »

Therefore your deductions are based on false assumptions and therefore worthless.
Even though the tests may not be comparable in a 1:1 fashion, I think it is hard to conclude that such a comparision has exactly no value.

If MTFx0 is interpreted as an absolute measure, estimated in some test A, then a different test B that tries to estimate the same should not be too far off from each other if each test is reasonably accurate?

-h
Logged

Walter Schulz

  • Full Member
  • ***
  • Offline Offline
  • Posts: 105
Re: Canon 5D Mark III official...
« Reply #128 on: March 04, 2012, 03:21:50 am »

http://forum.photozone.de/
and ask Klaus himself, please. These are his data sets which are messed up here and I think he is the very person whose opinion about this messup is relevant.

As stated in Klaus' quote you have different data sets with a different processing as input therefore the results are not compareable without putting those differences into context.

Ciao, Walter
« Last Edit: March 04, 2012, 03:26:31 am by Walter Schulz »
Logged

deejjjaaaa

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1170
Re: Canon 5D Mark III official...
« Reply #129 on: March 04, 2012, 03:40:05 am »

Klaus is adamant about one point and always has been: The results may not be used for comparisions between different camera systems.
yawn... some how all tests (slrgear, lenstip, photozone, feel free to suggest more) consistently show that increase in megapixels (yes, on different systems - exactly the case we have here) always lead to increase in resolution until the very extreme stopping down... so there is no reason to believe that FF 36mp in camera w/ AA compensated (D800E) will deliver the same (and not more) resolution as FF 22mp on 5DmkIII w/ AA filter @ F11 or F16... feel free to show a 3rd party test (we assume tripod, MLU, remote shutter release - please no hand holding of consumer level, double trombone, 28-300 zoom @ 300 w/ VR/IS switched off) where that is not the case and yes, please feel free to show that on different systems with 1.5+ increase in megapixels between both
Logged

Walter Schulz

  • Full Member
  • ***
  • Offline Offline
  • Posts: 105
Re: Canon 5D Mark III official...
« Reply #130 on: March 04, 2012, 03:49:55 am »

yawn... some how all tests (slrgear, lenstip, photozone, feel free to suggest more) consistently show that increase in megapixels (yes, on different systems - exactly the case we have here) always lead to increase in resolution until the very extreme stopping down

I always gave those "this new sensor with higher pixel count will outresolve lenses, therefore image quality will decrease"-posters a hard time. You are addressing the wrong person, I think.

Ciao, Walter
Logged

deejjjaaaa

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1170
Re: Canon 5D Mark III official...
« Reply #131 on: March 04, 2012, 03:53:07 am »

PS: EF 100/2.8 USM Macro on 350D was tested to f/22
http://www.photozone.de/canon-eos/167-canon-ef-100mm-f28-usm-macro-test-report--review?start=1

great... so let us see how 8mp ->15mp delivers on 1.6x APS-C sensor w/ Canon EF 100mm f/2.8 USM macro :

8mp @ F16 = 1663/1640
15mp @ F16 = 1973/1956

F16 @ 1.6x cropped APS-C...

so you think F16 on FF w/ 22mp w/ AA -> 36mp w/ AA compensated will yield no difference  ::)
Logged

deejjjaaaa

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1170
Re: Canon 5D Mark III official...
« Reply #132 on: March 04, 2012, 03:54:54 am »

I always gave those "this new sensor with higher pixel count will outresolve lenses, therefore image quality will decrease"-posters a hard time. You are addressing the wrong person, I think.

Ciao, Walter

I see no hard time in your postings... I saw no tests (again - tripod, MLU, remote shutter release and no consumer grade lenses) that backup your point.
Logged

Walter Schulz

  • Full Member
  • ***
  • Offline Offline
  • Posts: 105
Re: Canon 5D Mark III official...
« Reply #133 on: March 04, 2012, 04:03:21 am »

Do we have a personal problem?
Those guys and postings I mentioned are/where at a german speaking forum and not here.

And "my point" is about messing with numbers/test results not indended to be used this way.

Ciao, Walter
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Re: Canon 5D Mark III official...
« Reply #134 on: March 04, 2012, 06:16:34 am »

Klaus is adamant about one point and always has been: The results may not be used for comparisions between different camera systems.

"Please note that the tests results are not comparable across the different systems. This does also apply for the new EOS tests based on the EOS 50D because of differences in the sensor system (e.g. AA-filter) as well as different RAW-converters."

Therefore your deductions are based on false assumptions and therefore worthless.

Ciao, Walter

PS: EF 100/2.8 USM Macro on 350D was tested to f/22
http://www.photozone.de/canon-eos/167-canon-ef-100mm-f28-usm-macro-test-report--review?start=1

I think you've misunderstood those warnings. Photozone primarily tests lenses for comparison purposes. It is reasonable to suppose that such warnings apply only to the evaluation and comparison of lens quality. For example if one wanted to compare a particular Nikon lens with a Canon equivalent, one couldn't because the results at Photozone are always a product of both lens performance and sensor performance, and one wouldn't be able to separate the performance of the sensor from the performance of the lens. Likewise, if one wanted to compare a lens that had been tested on the 8mp 350D, with a newer or upgraded lens that had only been tested on the 50D, one would have great difficulty interpreting the results because the 50D is likely to make an inferior lens appear as good as a better quality lens on the 350D.

What I'm doing cannot be described as comparing different camera systems in the sense the warnings apply. The lens is the same, the camera body is different but from the same manufacturer, as opposed to the body being the same but the lenses different.

When comparing lenses, the camera body should be the same but the lenses are obviously always different. In both cases the camera systems are different in their own way. If the Photozone results are valid for comparison of different lenses on the same body, then they are equally valid for comparison of different bodies on the same lens. But they are not valid for comparison of different lenses on different bodies. Got it?

The fact that different sensors have different strengths of AA filters is unavoidable. The AA filter is part of the camera, like a lens coating is part of a lens. So what! One presumes that the higher the pixel count, the weaker the AA filter.

What we are concerned about here is, if it's technologically possible for a high pixel-density sensor to continue to deliver more resolution at f stops that we know are in part diffraction limited. If part of that process consists of using a weaker AA filter (or no AA filter at all in the case of the D800E), and better RAW converters, then that's fine by me.

I'm hardly likely to complain when I see a surprisingly sharp and detailed result at F16 from my new D800, "Boo! What a cheat! Nikon has used a weak AA filter and a new, more sophisticated RAW converter."  ;D

By the way, thanks for bringing to my attention those results down to F22. What we see here is that an 8mp sensor at F16 delivers more resolution than a 15mp sensor at F22. However, what's interesting also is that the 15mp sensor continues to deliver more resolution at F22 than the 8mp sensor at F22. That is a susprising result. I wouldn't have thought it likely. My own tests have demonstrated that the resolution of the 10mp 40D and 15mp 50D are visually the same on the monitor, as regards detail at F22, using the same lens.

Logged

Hans Kruse

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2106
    • Hans Kruse Photography
Re: Canon 5D Mark III official...
« Reply #135 on: March 04, 2012, 02:33:41 pm »

yawn... some how all tests (slrgear, lenstip, photozone, feel free to suggest more) consistently show that increase in megapixels (yes, on different systems - exactly the case we have here) always lead to increase in resolution until the very extreme stopping down... so there is no reason to believe that FF 36mp in camera w/ AA compensated (D800E) will deliver the same (and not more) resolution as FF 22mp on 5DmkIII w/ AA filter @ F11 or F16... feel free to show a 3rd party test (we assume tripod, MLU, remote shutter release - please no hand holding of consumer level, double trombone, 28-300 zoom @ 300 w/ VR/IS switched off) where that is not the case and yes, please feel free to show that on different systems with 1.5+ increase in megapixels between both

Interesting to see that a Nikon 50mm lens at f/8 on D3 resolves exactly as f/16 on D3X. So basically doubling pixel count on the D3X is eliminated by stopping down from f/8 to f/16.

Hans Kruse

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2106
    • Hans Kruse Photography
Re: Canon 5D Mark III official...
« Reply #136 on: March 04, 2012, 02:50:36 pm »

and in Canon land the 1Ds mkIII 50mm f/1.4 lens at f/22 is about the same resolution as 50mm on 40D at f/8
As comparison the 85mm f/1.8 on 1Ds mkIII at f/8 (gives the same resolution as the 50mm f/1.4 on the 1Ds mkIII at f/8).

The 40D is 10MP and the 1Ds mkIII is 21MP. So a similar patterns as expected from the Nikon resolution tests by DxO.
« Last Edit: March 04, 2012, 02:53:02 pm by Hans Kruse »
Logged

Hans Kruse

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2106
    • Hans Kruse Photography
Re: Canon 5D Mark III official...
« Reply #137 on: March 04, 2012, 02:59:29 pm »

However stopping down the D40 to f/22 as well as the 1Ds mkIII gives a lower resolution on the D40 than on the 1Ds mkIII.
According to tabel 3 in http://www.luminous-landscape.com/tutorials/resolution.shtml they should have about the same resolution 4MP. This is clearly not the case so the assumptions in the article are not correct.



Slobodan Blagojevic

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 18090
  • When everyone thinks the same, nobody thinks
    • My website
Re: Canon 5D Mark III official...
« Reply #138 on: March 04, 2012, 03:02:31 pm »

Not sure what you guys are trying to prove: that more megapixels is pointless because when shot at f/22 the resolution will not be better than at f/8 at less megapixels?

The way I read it is that up to f/22 there will be resolution advantage of more megapixels. Not to mention that no self-respecting digital photographer will shoot at f/22 if he can get away with shooting with f/8 or in-between.

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051
Re: Canon 5D Mark III official...
« Reply #139 on: March 04, 2012, 03:04:13 pm »

Comparing MTF50 values when varying sensor resolution, lense quality and diffraction limits might not be a perfect indicator on the likely degree of "sharpness" obtainable by proper sharpening and deconvolution. I think that the PSF of each of those can have distinctly differing shape for a given e.g. MTF50.

-h
Logged
Pages: 1 ... 5 6 [7] 8 9   Go Up