Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Lens resolution vs. image sensor resolution
#11
Shortly said I have appreciated a lot these Photozone test reports and have read them with great interest because it is not a single lens test but test of a combination of body+lens with all the pros and cons of both the lens and body.

Linear pixel density is the main factor, so far I have understood, to give reliable resolution results of the lens.

It is interesting that the linear pixel density is the same both in 8mb/1,6 crop Canon 20D/350 and in 21mb/FF 1Ds Mk III / 5D mk II bodies. So if the resolution is good with 350D it good also with 1Ds Mk III.

The same equivalence can be applied also to my 1dMK3 and 1Ds Mk2 bodies, which also have same pixel density.
#12
Already said but, is there any reason for upgrading to fancy high MP cameras for testing purposes? A Nikon D200 with 10mp will tell you all that you need when it comes to comparing the various lenses by means of contrast, color reproduction, CA etc... The key word is "comparable" here. Actually I've startled when Klaus said "we're going to switch to new Nikon bodies for testing soon". Ok, this may be a must concerning various reasons but what will happen then if we'd like to compare 3 lenses of 24-70mm range and let's say the 3rd will be tested with new body and the rest was already tested in D200? Then we'll need to take other technical issues into consideration, which does not make life easier for a proper lens choosing.



I believe the strength of PZ is that you can find soooo many lenses reported with a very decent reliability. This is nothing but a high performance.



Regards,



Serkan
#13
Hi Serkan,[quote name='PuxaVida' timestamp='1281095388' post='1542']

Already said but, is there any reason for upgrading to fancy high MP cameras for testing purposes? A Nikon D200 with 10mp will tell you all that you need when it comes to comparing the various lenses by means of contrast, color reproduction, CA etc... [/quote]

I hope you don't mind me replying here, but I certainly do think so, that it makes a difference that is.



Lower MP cameras by their AA-filters put an artificial cap on resolution and a bunch of other features of a lens: you often get flat top MTF charts if you like, where edges to borders to centre of a lens , from a certain f-stop, often only one or two stops from wide open, behave more or less the same. This is not typical for most lenses at all, so in order to see what a lens really does or can do, the use of high MP cameras is a necessity. Either that or an optical bench. However, the latter is beyond the means of many a company, let alone PZ <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='Big Grin' />.

Quote:The key word is "comparable" here. Actually I've startled when Klaus said "we're going to switch to new Nikon bodies for testing soon". Ok, this may be a must concerning various reasons but what will happen then if we'd like to compare 3 lenses of 24-70mm range and let's say the 3rd will be tested with new body and the rest was already tested in D200? Then we'll need to take other technical issues into consideration, which does not make life easier for a proper lens choosing.



I believe the strength of PZ is that you can find soooo many lenses reported with a very decent reliability. This is nothing but a high performance.



Regards,



Serkan

IMO, the better one knows the characteristics and quirks of a lens, the better one can use it to its advantage when actually out shooting.



And I totally agree with your last statement!



Kind regards, Wim
Gear: Canon EOS R with 3 primes and 2 zooms, 4 EF-R adapters, Canon EOS 5 (analog), 9 Canon EF primes, a lone Canon EF zoom, 2 extenders, 2 converters, tubes; Olympus OM-D 1 Mk II & Pen F with 12 primes, 6 zooms, and 3 Metabones EF-MFT adapters ....
#14
Hi Wim,



of course I cannot deny that the new lenses are better to be tested on new high MP sensors to get a broader range of results. I'm not against this. But what I personally believe is that the comparatibility is more important than this issue. To be more precise, let's say you're looking for a mid range zoom. If you can find the results of three 24 ~ 85mm lenses by Nikon, Tamron & Sigma tested with the same body, than this package really means something. Of course it would be better if all these three lenses were tested by the same high MP sensored body, but OTOH it would be definetly the worst case if 2 of them were tested with D200 and the last one with D3X. How could you choose between those three?



Maybe I'm missing the point... I don't have so much technical knowledge as yours but isn't it like rating the lap times of 3 different drivers with exactly the same race car?



Best regards,



Serkan
#15
Hi Serkan,

[quote name='PuxaVida' timestamp='1281113865' post='1554']

Hi Wim,



of course I cannot deny that the new lenses are better to be tested on new high MP sensors to get a broader range of results. I'm not against this. But what I personally believe is that the comparatibility is more important than this issue. To be more precise, let's say you're looking for a mid range zoom. If you can find the results of three 24 ~ 85mm lenses by Nikon, Tamron & Sigma tested with the same body, than this package really means something. Of course it would be better if all these three lenses were tested by the same high MP sensored body, but OTOH it would be definetly the worst case if 2 of them were tested with D200 and the last one with D3X. How could you choose between those three?



Maybe I'm missing the point... I don't have so much technical knowledge as yours but isn't it like rating the lap times of 3 different drivers with exactly the same race car?



Best regards,



Serkan

[/quote]

Ah, I see now what you mean.



The only point I tried to make really, is that the test on lower resolution MP cameras don't really tell you all, as most lenses will do quite well on those. The differences start really to become noticeable on the high resolution sensors. IOW, even if all three lenses were tested on the D200, it wouldn't tell you an awful lot, as they would very likely be quite close due to the limitations introduced by the sensor assembly. The differences would likely be only quite small from an optical POV. And similar constraints are true for FF lenses tested on APS-C - they either overperform as a whole, because you only see approximately 40 % of the centre of the image, or they seem to underperform because the size and arrrangement of sitewells isn't optimal for that particular lens (e.g., a 17-40L on a crop body).



However, rendering would likely be very different, and this unfortunately is something you would need to check for yourself, whether you like the results from one lens as compared to another.



With regard to choosing between two lenses on a lower MP APS-C body vs a high MP FF body, well, that is nigh impossible. Best thing to do, roughly, is up the overall rating of the lens tested on the D3X by 1/2 a rating (I mean here good - very good -excellent, not the actual numbers), average the border measurement and centre measurement for the corner measurement on APS-C, and see what you end up with with regards to the rating then. Roughly this is equivalent, I would say.



I would consider the ratings similar as three different race cars under the same conditions on the same circuit driven by the same driver, if the camera is the same that is. A better driver (better camera) will get different results, as will different classes of cars (standard production racing cars vs F1 racing cars, consumer lenses vs professional lenses). Generally, a better driver will get better results than a less qualified driver, etc., up to a limit of course, but in that case you have to look at the absolute terms (lap times or absolute MTF values - 3500 LWPH is better than 2200 LWPH in absolute terms as you need less magnification to get the same result in print, making any aberrations or faults physically smaller as well).



I reckon the thing to do really is determine what qualities you find most important, and rate those, adjusted more or less as outlined higher up, just don't put any meaningful value to the absolute MF value, only to the adjusted ratings. You have to remember that +/- 10 % or even 15 % is neither here nor there, because even standard production samples may vary as much from one specimen of a lens to the next, while still being within specification. BTW, this is the same with cars: some of the cars from the production line may do, e.g., 215 km/h tops, while the next specimen may do 240 km/h (real life experience, BTW). Of course, mechanical defects like decentering are not applicable in this case - I return decentered lenses or have them (re)calibrated.



Now, coming back to rendering: this is something that is very personal, and makes other optical qualities a secondary point. F.e., some people do really like Zeiss glass. They tend to say it has something, 3D pop, high microcontrast, whatever. But that is very personal, and all about rendering. So far, personally, I hated the rendering of all the Zeiss glass I tried. However, I do like the rendering of certain pieces of Canon glass a lot. Hence I opt for those pieces of Canon glass. A lens like the 50L for example, which doesn't fare very well at all in tests. However, its rendering is magical to me, I just love it. And funnily enough, it is darn sharp despite the weak points showing up in those tests. So, all of this is very personal in the end. And it does mean that although the tests here are a good starting point to make sure you get a certain minimum level of optical quality, you still need to check that you would actually be happy with that lens yourself. I do find that just taking a few shots in less than ideal conditions generally shows whether you will be happy with it or not, however, so if you can, that is always something to try out first, prior to committing to a purchase.



HTH, kind regards, Wim
Gear: Canon EOS R with 3 primes and 2 zooms, 4 EF-R adapters, Canon EOS 5 (analog), 9 Canon EF primes, a lone Canon EF zoom, 2 extenders, 2 converters, tubes; Olympus OM-D 1 Mk II & Pen F with 12 primes, 6 zooms, and 3 Metabones EF-MFT adapters ....
#16
Argh... I should check these forums more regularly!



I have to give credit to the OP for seeing what's right... although it might be unrealistic given the financial circumstances. And as others have suggested, with the current method, we can at least evaluate how a lens performs relative to other lenses that we're familiar with.



[quote name='Klaus' timestamp='1280404245' post='1381']

Beyond 10mp (APS-C) and 20mp (full format) an camera upgrade has more to do with "political correctness" rather than sense.

[/quote]



Don't mind if I comment on this... while it may not be possible to get any more detail out of the lenses on sensors with more pixels than the ones you mentioned, higher resolution Bayer sensors do reduce the spatial size of the digital noise and demosaicing artifacts, for a fixed reproduction magnification (not 100% viewing on the screen).



Of course, I don't mean simply increasing the megapixel count without improving the efficiency of the sensors is good. Particularly, the fill factor shouldn't be decreased to accommodate the extra pixels. But recently we've been seeing improvements in this department, with the help of Canon's gapless micro lenses, etc.



GTW
  


Forum Jump:


Users browsing this thread:
1 Guest(s)