On 1/12/2017 9:17 AM, Bob Benson wrote:
I've been puzzling over m4/3 vs. FF for a while. I'm hoping Moose/AG, etal
can help me.
I shoot both, a 24 MP A7 for some stuff, and 16 MP µ4/3 bodies for other. The twist may be that I shoot the smaller
format for sharp, stuff, fine detail, etc., and the large format for soft focus, abstracts, etc.
It also depends on what one is going to do with the resulting image files. I have the Ctein demo print of moon and the
Oakland-SF bay bridge. Taken with the very first µ4/3 camera, the 12 MP E-P1, and printed at 17x22", the detail and
color subtlety are stunning.
As shown below, the HR Mode of the E-M5 II resolves quite a bit more fine image detail than the straight, 16 MP mode
when pixel peeping @ 100%. Yet, I've stood in Ctein's printing studio, looking at cropped prints of carefully made
images of the same subject, same camera, within seconds of each other, printed at what would be huge sizes for the image
sizes.
Yes, there is a difference, but my oh my it is subtle, more a matter of surface texture in unpainted wood than real
detail differences. Yeah, those are there too, but nobody but nuts like us are going to see them, and then only if we
decide to look hard.
My question now is stimulated by a recent camera review that stated "FF
sensors collect 4 times as much information as M4/3" ... presumably based
on the difference in sensor size. My initial response: "really" ?
Presumably based on some wild thing happening in the reviewer's brain. If the web has taught us anything, it is that the
authority, knowledge, experience and expertise we may have assumed/assigned to the few reviewers of magazine days is
misplaced in the era when almost anyone may write authoritative sounding reviews on the web.
Not saying that he's necessarily wrong, but that he may not be completely right.
My question has two parts.
1. For FF and M/43 sensors with the same pixel count, say 20meg, is the
actual amount of information collected significantly different? (I think I
understand the issues of dynamic range and LOF and high ISO noise, but I'd
thought these are artifacts of the circuits and lenses used, not the amount
of information per se.)
It seems you are defining "information" to be resolution of detail in the subject within the focal plane. If we ignore
that greater DoF delivers more linear separation of subject detail across a larger part of an image of a 3D subject,
which amounts to more information, then no, in simplistic terms, there is not a significant different in the amount of
resolved details.
I suppose this is the same question: for FF and M/43 with the same pixel count,
would the RAW files be the same size?
Everything about this subject is multidimensional and tricky. They are not the same question. The answer to the second
is a slightly provisional yes; if the cameras do not use lossless compression in writing Raw files, they would be he
same size. They would be the same size before and after compression/decompression.
All things are not equal. AG has already raised issues like AoV and lens resolution. Example: I'd already been doing
some testing of lens resolution when Brian's questions about lens sharpness came up, so I added some shots. Shooting
from the same distance using OMZ 500/8 and 300/4.5, more detail is resolved in the 500mm image. Moving the 300 mm
closer, so the image size on the sensor is the same as with the 500 mm at the greater distance, and the 300 mm resolves
more detail.
FF and µ4/3 sensors shooting with the same AoV and subject distance are necessarily using different lenses. All things
are Never equal in this sort of comparison. The Panny 20/1.7 lens resolves less detail on a 16 MP sensor than the Oly
25/1.8. The 25/1.8 resolves more detail in HR Mode than at simple 16 MP. Can we then logically say that the 20/1.7 will
resolve no more detail in HR Mode? Nope, all this stuff is interactive. Put another way, more familiar from audio, the
frequency response of the system is a complex combination of all the stages, not simply the response of the narrowest
component.
Ag has written in some detail about this some time ago.
AND, it is possible to sample beyond the amount of information the lens/sensor system can provide. In HR Mode, the E-M5
II produces three output files:
A Raw, ORF file of 9280 x 6932 = 64 MP
A JPEG of 7296 x 5472 = 40 MP
A non-HR ORF of 4640 x 3472 = 16 MP
Pretty much everyone who has tested these files agrees that the actual resolution is significantly less than 64 MP, but
definitely much more than 16 MP. Oly's choice of 40 MP for the JPEGs sort of implies that's the actual resolution they
think is captured. A lot actually depends on the subject. In IR's tests, this HR mode resolves no more detail than the
36 MP Nikon D810 with some subjects, while stomping it in others. The difference comes down to factors you likely
haven't considered, color resolution and moire.
Because the Oly HR Mode samples each pixel location with sensels of each color, it has higher color resolution than
conventional Bayer Array demosaicing. Combined with no moire, it does a MUCH better job with things like colored fabric.
<http://www.imaging-resource.com/PRODS/olympus-e-m5-ii/olympus-e-m5-iiTECH2.HTM>
Look further down, and see the Oly HR even gives the MF Pentax 645Z more than a
run for its money on the fabrics.
So, a camera/sensor system can out resolve much larger sensors with some subjects, and not with others. How confusing is
that?:-)
2. As FF sensors increase in pixel count, say to 60meg, will they begin to
exhibit similar issues of dynamic range and high ISO noise because of the
density of the circuits approaching that of the m4/3 sensors?
Yes, but there are other problems which effectively limit pixel density for general purpose photography. Most cameras
use Anti Aliasing filters to avoid moire in fine, repeating detail. To go over about 24-30 MP on FF, and realize the
resolution benefit, very weak or no AA filters are necessary. Oly gets away with no AA filters on the E-M5 II and E-M1
II because of the relatively modest sensor sizes. Then they don't have either an artificial limit to resolution in HR
Modes from a filter, nor moire problems, as they don't have to demosaic. (Which may be a reason the E-M1 II is 20, not
24 MP.)
This is a real problem with some subjects. Take a look at the D750, 24 MP, AA filter, vs. the D810, 36 MP, no AA filter,
and the HR E-M5 II here. <http://www.moosemystic.net/Gallery/tech/X-T1_E-M1_D750/FON.htm>
Like those rainbows? So I don't think you are going to see FF sensors of 60 MP, or µ4/3 sensors over about 20 MP, at
least until there is some other technical advance.
Just sayin' It's far from a simple question.
Meandering Moose
--
What if the Hokey Pokey *IS* what it's all about?
--
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/
|