Crop Factor, Low Light and Aperture with m4/3 lenses? Part 2

Started 6 months ago | Discussions thread
Anders W
Forum ProPosts: 19,064Gear list
Like?
Re: Crop Factor, Low Light and Aperture with m4/3 lenses? Part 2
In reply to Jack Hogan, 6 months ago

Jack Hogan wrote:

bobn2 wrote:

Anders W wrote:

bobn2 wrote:

Anders W wrote:

bobn2 wrote: Then you'll find that lenses ranked 1-368 are FF lenses, then we get the Panasonic Leica DG Nocticron 42.5mm F1.2 ASPH at 369. There is some duplication there down to same lens different camera, but there's little point spending a lot of time cleaning up an 'analysis' like that.

What is the relationship between those lens rankings and the factor we are trying to isolate in this discussion: Lens resolution.

Resolution is a factpr in those rankings.

Yes. But not the only and as already indicated, the others introduce systematic bias.

That is why there is absolutely no point wasting time producing 'evidence' for you, Anders. Whatever it is it will not be persuasive enough to budge you from endless argument in support of your own concocted theories.

No there isn't, the (shrinking) target is always moving for some folks

Jack! Long time, no see! What a surprise to see you here.

I would rather have expected to encounter you in another thread, the one where we discussed these things six months ago and where you promised to answer all my outstanding questions but never did although I dutifully took the time to respond to all your queries, in detail, and several times around (since you insisted). See here and the posts in the same subthread that follow:

http://www.dpreview.com/forums/post/52662160

How come? Is it a habit of yours not to deliver on such promises?

As has been said correctly in this and several other threads mFT lenses typically need to be at least twice as 'sharp' in order to match FF linear spatial resolution in equivalent photographs - precisely for the reason mentioned by the OP:

Who has ever denied that an MFT lens has to be twice as sharp per millimeter on the sensor and equally sharp per image diagonal in order to go equal with FF for resolution? As far as I know, that much is not subject to debate.

even mFT's admittedly state-of-the-art lens design is typically not enough to compensate for the format's size penalty. There are a few exceptions to this rule, but they are just that, exceptional exceptions.

As it turns out, that was not the OP's point at all. What he wanted to say was merely that an FF lens is likely to do better on an FF sensor than on an MFT sensor, which is hardly a fact disputed by anyone either. See here:

http://www.dpreview.com/forums/post/53922201

As has also been said correctly in this thread by noirdesir amongst others, although ideally system tolerances would scale perfectly with size, in practice they typically don't. They don't because the difference in size we are discussing here is not orders of magnitude away: if an improved process becomes available to one size it will soon trickle down/up to the other negating its advantage, as Bob mentioned; and because as the system shrinks at these sizes there are more things that get harder than easier to do as far as 'sharpness' is concerned, making it difficult to scale down perfectly as parts get smaller. This is how The_Suede put it recently:

'And then you have optimization theory and deviation statistics in manufacturing that quite accurately predicts that it's hard to halve tolerances in centering, glass surface imperfections and element-to-element distances in the lens just because you halved the scale. Often you get maybe 2/3 tolerances in a 1/2 scale production scenario'

This doubly applies to tolerances necessary to achieve equivalent focus precision in the real world. So where's the beef?

Why is the case that larger sensors have higher production costs than smaller sensors? And why is it possible to successfully mold rather than grind certain lens elements if they are small rather than large?

It's very difficult to compare lens performance on different cameras because there typically are too many variables to keep under control, too few tests from too many sources, each with their own methods. Dxomark.com has put together the best effort to date attempting to control the variables, test multiple cameras and provide meaningful 'sharpness' data as perceived by the average photographer as uniformly as they could - in so doing they created one of the largest and well thought out, if obscure, databases in photography. One may disagree with some of their choices or methods but nobody questions the fact that the folks there are doing scientific work.

You are wrong about that. I certainly question that DxOMark is doing scientific work in the area of we are discussing here: that of "sharpness".

You see, what distinguishes scientific from non-scientific knowledge is that the former rests on data obtained by methods described with sufficient rigor to be reproducible as well as fully assessable. Since DxOMark's data on sharpness are not of that kind, they are by definition non-scientific. You said they were obscure and that is indeed the word here. This in turn means that none of us are even in a position to say whether and on what grounds we agree or disagree with their methods inasmuch as those methods are in important regards unknown.

This is not the only way in which DxO fails to adhere to scientific standards. For example, it is not in line with scientific practice to knowingly report the wrong data ("for web display reasons"), as they recognized they did here:

http://forum.dxomark.com/index.php?topic=2414.0

And it is even less in line with scientific practice not to correct the error once it is pointed out (as they still haven't, nearly half a year later).

Nor is it in line with scientific practice to promise a full description of their methods but not deliver it or to use a biased standard for comparing sharpness across formats that doesn't take differences in aspect ratios into account (enlarged to same width rather than same diagonal length):

http://forum.dxomark.com/index.php?topic=1757.0

In case you don't remember, I am hardly the only one who questions the DxO data on sharpness. Here is a passage from my reply to you here half a year ago:

The_Suede said (among other things):

Inconsistencies and a total disconnect from visual inspection results makes the P-MPix metric quite useless.

Falk Lumo (falconeyes) seconded what The_Suede said and added (among other things) the following:

FIRST advice: Don't compare lenses across different bodies.

John Sheehy said (among other things):

For comparing [lenses across] sensors, it is a worthless, deceptive metric.

Their data confirms* the OP's point.

As already indicated above, you completely misunderstood the point the OP actually tried to make.

As of today mFT cameras are typically not able to make up for their smaller sensing size with better lenses. The pudding is everywhere, all one has to do is open their eyes to see it.

Jack

*Selecting 'sharpness' under Optical Metric Scores as the ranking criterion in Bob's link to their database (limted to sensors between 12-18MP) shows the first mFT camera/lens combo on page 5 (Zeiss 42.5mm f/1.2 on 16MP EM5), a $1500 lens easily bested by its FF equivalent, a $500 lens on a 16MP body (Df+Nikkor 85mmf/1.8G). The Df on the other hand is well represented on page one with a host of lenses across the range.

Interesting example. According to your own MTF data, the Df with the Nikkor 85/1.8G manages about 1750 lwph as against 1850 to 1950 lwph for the much smaller, lighter, and less expensive 45/1.8 on a GM1 or EM1. Why is that and why are your results out of line with those from DxOMark?

 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