Reading resolution charts comparing MFT lenses to FX lenses

Started Mar 29, 2013 | Questions thread
Anders W
Forum ProPosts: 19,026Gear list
Like?
Re: Reading resolution charts comparing MFT lenses to FX lenses
In reply to KenBalbari, Mar 31, 2013

I think we are pretty much in agreement on most of what you say below Ken. As to DxOMark and their P-MPix, however, the problem is precisely the one you mention. You say you don't know the technical details. Neither do I. And until I do, I don't know what to make of the measure.

Until recently, DxO still reported MTF values, and did so more extensively than anyone else. You could see the entire MTF curves at each focal length and f-stop at four different points on the sensor (center, one third out, two thirds out, and extreme corner). As long as they reported their results in that form, I used to make use of their data and the results were in many cases closely aligned with those you would get on the basis of results from Lenstip or LensRentals. See here for example:

http://www.dpreview.com/forums/post/41639871

KenBalbari wrote:

Anders W wrote:

If you start comparing with small-sensor compacts, then it is obvious that MFT and larger sensors do better for resolution. But how do you distinguish between the possible reasons and single out noise (or noise reduction) as the culprit? There are also potential differences in lens performance and the impact of diffraction (much stronger for small-sensor compacts when shot at the same f-stop) to be reckoned with. Not saying that noise is irrelevant when you start comparing with small-sensor compacts, only that your claim is stronger than the evidence at our disposal permits.And when it comes to comparing current MFT with current FF equipment in base-ISO shooting, I find no reason at all to think that noise appreciably hampers the MTF performance of the MFT equipment.

True, diffraction also likely plays a large role.  Noise may be a limit in some cases, but at these resolutions only with these very small sensors.  I confess I may actually have been thinking more of dynamic range, which in theory really is limited by noise (which limits the ability to resolve very close tones in dark areas).  I think resolution as we seem to mean it here, of very different tones (like a typical resolution test chart), is probably not generally influenced by noise in larger sensor cameras.


Hmm. If you or anyone else can tell me how DxO defines and measures P-MPix (on a technical level, I know how they try to explain it in layman terms), we can discuss the matter further. At present, the DxO P-MPix measure is just mumbo-jumbo to me.

I don't know the technical details, but DxO seems to be pretty good on the technical end, and seems to be putting good measures of things like MTF and accutance into this.  And the results seem to make sense to me.  It's defined simply enough as "the pixel count of a sensor that would give the same sharpness if tested with a perfect theoretical optics".   And the point of this translation is to make it easier to make fair comparisons.

As pointed out above, I don't think reasoning on the basis of such general principles tells us much about the way current MFT equipment compares with FF as far as MTF values are concerned. If FF sensors do better, it is because they have more pixels. But even when you start comparing MFT 16 MP cameras with the 36 MP D800, the difference is rather marginal. For example, the 20/1.7 on an E-M5 manages 1050/875 at f/2.8 on an E-M5 versus 1073/889 for the 50/1.4G at f/5.6 on a D800.

Sources:

http://www.lensrentals.com/blog/2012/03/d800-lens-selection

http://www.lensrentals.com/blog/2012/05/wide-angle-micro-43-imatest-results

Good point.  And it's not an isolated case, either.  DP Review tested those same lenses, on the D3 and G1 respectively, and got a fairly similar result.  In this case the FF wins narrowly overall, but the MFT is actually sharper in the center (link).

I'm not sure why, whether AA filters play a part, or diffraction (maybe my prime suspect), or perhaps there simply aren't currently full frame lenses which are sharp enough to benefit from more than 20 MPIX of resolution.   But the resolution advantage of full frame does seem to be small.  Not nonexistent though.  I should point out that the 20mm lens you mentioned was the best performing at f2.8 of the m4/3 lenses on that page, while the Nikon page lists a half dozen lenses (including 3 Zeiss) which broke 1000 lp/ph in lensrentals tests.  But still, that puts the resolution advantage typically only at maybe 15%.

I think that's backed up pretty well by DxO tests.  Of 98 Nikon FX lens/camera combinations they've tested, the top 2 scored only 17 P-Mpix.  An additional 4 scored 16 P-MPix.  All of those were on the D800.  The best the Nikon D4 has done so far is only 12 P-MPix.  They've yet to test any lenses yet on the current m4/3 sensors, but the best lenses the previous generation GH2 and GX1 were delivering 9-11 P-MPix.

And for good measure, I'll throw in some tests from Popular Photogtraphy:


"The D800 has enough resolution for an Excellent rating through all but its top ISO of 25,600. At ISO 50, it served up 3510 lines per picture height."

"The D600’s imaging proved well above economy class, with an Excellent rating in overall image quality in our tests from ISO 50 through ISO 3200. Its resolution was particularly impressive: 2930 lines per picture height at its lowest sensitivity of ISO 50."

"Despite the Mark III having about one-third fewer pixels than the Nikon D800, it still easily scored an Excellent rating in our resolution test, with 2750 lines per picture height at its lowest (expanded) sensitivity of ISO 50"

"The D4 captured 2530 lines per picture height at ISO 50 and didn’t drop below it until ISO 1600; the 1D Mark IV fell just barely short of an Excellent rating with 2490 lines at ISO 50. Meanwhile, Nikon’s D3s managed to resolve 2330 lines at ISO 100. (To its credit, that’s quite a good score, considering that it has only a 12.1MP sensor.)"

"With just enough resolution (2530 lines per picture height) to achieve top honors in this measure, plus remarkably accurate color recreation and well-controlled noise at lower ISOs, the GX1 earned an Excellent rating in overall image quality from its lowest sensitivity of ISO 160 through ISO 400.
Though it is considerably behind Sony’s NEX-7 in resolving power, the GX1 delivers proportionally more resolution from its 16MP Micro Four Thirds sensor than the Sony delivers from its considerably larger APS-C-sized sensor.
And while the GX1’s resolution fell just below the cutoff for an Excellent rating by ISO 800, with 2480 lines, it held onto an impressive amount of resolving power even at its top two sensitivities of ISO 6400 and ISO 12,800, where it turned in 2320 and 2180 lines, respectively."

"The main limiting factor on the PL3’s overall Image Quality rating is its resolution. That’s not to say that it’s bad; with 12.3 megapixels, the PL3’s LiveMOS sensor turned in 2230 lines per picture height at its lowest sensitivity (ISO 200) in our resolution test. Since our rating scale is wide enough to include cameras that have twice as many pixels, the fact that the PL3 landed toward the top of the Very High rating band (which runs from 2000 to 2249 lines) says good things about this camera.
Since the PL3 drops to only 2190 lines of resolution at ISO 1600, with a noise rating of Low, its balance of noise control and resolving power earn it a rating of Very High in overall image quality from ISO 200 through ISO 160"


They also haven't published results yet for the GH3, E-M5, E-PL5 generation.   I wouldn't be shocked though if they can at least get somewhat close to the D800 or D600.

And as you can see, even the GX1 was pretty much a match for the D4 at ISO 800 and below, and beat the 1D Mark IV.  And the older 12MP Pens were performing near as well as the 12MP Nikon D3s.

So I don't think you can really judge by either sensor size or MP.  You really have to look at the results of individual cameras and the available lenses.  And while I think there is likely still a resolution advantage possible with full frame with the right choices there, I do think it is still likely not great at lower ISO/with good light.

 Anders W's gear list:Anders W's gear list
Panasonic Lumix DMC-G1 Olympus OM-D E-M5 Olympus E-M1 Panasonic Lumix G Vario 14-45mm F3.5-5.6 ASPH OIS Panasonic Lumix G Vario 7-14mm F4 ASPH +21 more
Reply   Reply with quote   Complain
Post (hide subjects)Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow