ZEISS FE 24-70 gets DxO marked !!!

Started Mar 7, 2014 | Discussions thread
LeicaBOSS Regular Member • Posts: 432
Re: ZEISS FE 24-70 gets DxO marked !!!

Kirino wrote:

LeicaBOSS wrote:

You do realize that it's almost certain that they didn't actually put the lens on a NEX-7. Those scores are just "predicted" based on the sensor data and the lens data from whatever camera they did the actual tests on.

You really cannot use DxO scores like they are reliable and repeatable.


Not really "replace" but it's close enough that many people can pass on the FE 35mm. Bokeh not as good. Corners not as good. Not as compact. BUT... at 35mm the 24-70mm is really solid and without those odd concentric color rings that the FE 35mm has.

-- hide signature --

From time to time, I point my camera at the right things. This is generally when I forget everything I've learned.

LOL. How do you generate those field maps without putting the lens on to the NEX 7. If what you say is true, can you explain why lens are testing better on the A3000 20 megapixel sensor?

First, nobody but the engineers at DxO can actually explain the results of their tests - because the granular details of their data and computational approach are almost completely obscured.

As to the first question - given that their field maps are computed based on various data points (it's not like it's an actual image of an image field) - it's not unreasonable to expect that if they test a lens on one known sensor, and have sensor data for other cameras, that the results can be extrapolated to other sensors without bench testing. I cannot say for sure this is what they do...

But take this thought exercise:

1. When a new camera body is tested, results on that body appear for every lens ever tested on that system. Instantly.The 70D review came out, and every lens ever tested for the EF mount appears as "tested" on the 70D

2. DxO notes that they 'borrow' lenses to do their lens analyses.

3. To believe they physically mount each lens to each camera to generate their lens data - we must believe that A) They have every lens ever tested in inventory AND do a bench test on each lens for every camera that comes in  B) They go back and re-borrow every lens ever tested when a new camera comes in AND do a bench test on each lens  or C) They have a mathematical way of interpolating this data based on a limited number of actual lens tests and the archival sensor data

3a. In the case of Canon EF (full frame) - they report data on 109 lenses. They claim to take "thousands" of frames for each lens test. Let's say it takes 2 hours (conservative) to test a lens/camera combo. This means 218 hours of labor goes into lens tests for every Canon camera that is tested. This is 6.2 weeks of labor in France.

3b. If, in fact, they just have all those lenses sitting there - waiting to be tested on each body that comes in... They have $166,558 in Canon EF full-frame mount lenses alone. Just sitting there, waiting to be tested.

3c, The reverse of this is with bodies. This also means, when a new lens comes in, they have every body ever tested for the system sitting there in inventory as well.

Food for thought.

-- hide signature --

From time to time, I point my camera at the right things. This is generally when I forget everything I've learned.

 LeicaBOSS's gear list:LeicaBOSS's gear list
Leica M9 Sony SLT-A77 Canon EOS M Sony Alpha a7R Canon EF 50mm f/1.8 II +17 more
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow