Olympus-OM
[Top] [All Lists]

Re: [OM] Like-a the Leica?

Subject: Re: [OM] Like-a the Leica?
From: Moose <olymoose@xxxxxxxxx>
Date: Thu, 28 Apr 2016 20:57:28 -0700
On 4/28/2016 4:41 PM, Chuck Norcutt wrote:
Ignoring High Res mode, is there something I don't understand about DxOMark's "Sharpness" scores here? <http://www.dxomark.com/lenses/brand-olympus/mounted_on-Olympus_OM-D_E-M1-909/launched-between-1987-and-2016/mount_type-Micro_4_3/sensor_brand-Olympus#hideAdvancedOptions=false&viewMode=list&yDataType=global>

I have tried to understand them; maybe I did once, but it didn't last.

As I understand it their sharpness score is the number of MP resolved by a given lens/aperture on this sensor. The highest score for any lens is the 75/1.8 no doubt due largely to the f/1.8 aperture. The more common lenses with widest apertures running from f/3.5 to f/6.7 only manage to resolve 4-8MP. At such small apertures the Airy disks on these lenses well exceed the pixel pitch.

Bayer array sensors are only about 50% efficient at rendering the data from sensor sensels into RGB pixel resolution. Put another way, the process of interpolating color data for each originally one color sensel loses somewhere around/up to half the assumed resolution. Shoot a highly detailed subject twice, once normal, once High Res, process both, downsample the HR image to the same dimensions as the normal and it has much more clear detail.* Having found this to be true, I simply (continue to) ignore the DxO ratings as indications of lens resolution.

Remembering that resolution is linear, half from a 16 MP sensor when squared would be 4 MP. Then the range from 4 to 8 MP would be a result of some combination of lens resolution (and aperture?) interacting with the inefficiency of Bayer interpolation.

None of the above is meant to be anywhere near exact, just indicative of what is going on. I wouldn't write it, weak as I am on the theory and math, but for the experimental confirmation.

M. P. Moose

* HR Mode doesn't just move the sensor half sensel widths to improve resolution. It also moves by full widths so that each sensel location is sensed by separate R, G & B sensels. So, no resolution loss from color interpolation, and better color.

--
What if the Hokey Pokey *IS* what it's all about?

<Prev in Thread] Current Thread [Next in Thread>
Sponsored by Tako
Impressum | Datenschutz