[Scythels]

FWIW my username was Scythels until I changed it to something a bit friendlier =)

 

 

I'm pretty sure you never saw a Sigma testing unit based on Foveon sensor in action

 

I study optics and specialize in metrology and lens design.  In the research lab I work in one of my current roles is to design and develop testing methods for advanced optical systems.  One project involves the intersection of aberration theory and freeform surfaces.  Freeform surfaces may be thought of as "aspheres+."  It will be 10-30 years before freeform surfaces see use in mass-market optics due to the assembly and fabrication difficulties, but they will provide the same "revolution" that aspheres began so many years ago.  Unfortunately, because there is no symmetry they are remarkably sensitive to assembly and manufacturing errors and aligning involves monitoring the aberration field until the correct alignment is achieved.  This requires access to information about what aberrations are present in the image, thus interferometry or MTF bench testing (which enlarges the spot from the lens 20x or more and then re-images it) are the only way to analyze the system. 

 

The A1 unit is basically a variable-flange APS-C camera.  From sigma:

 

 

The lenses to be measured are hooked up to the  image capture device, and the special charts are captured and analyzed using new, proprietary algorithms designed to work in conjunction with the high-resolution sensor.

tl;dr it's a modified camera and modified ISO 122333 chart, with a new, proprietary demosaicing algorithm.  There is no patent protecting this machine, it is not a new, "novel" method of testing.  It's the same as a PZ or other review website chart-based test. 

 

 

But contrary to you, I use Foveon sensors and see what they do and see how they outresolve the Sony sensors.

Why the vitriol JoJu?  A great deal of what I do involves working with sensors with "even higher" higher (that doesn't actually say much) spacial resolution than the Foveon chips because they are B+W, high-density sensors.  The client's lens I am working with is specced at 104lp/mm corrected nyquist frequency.  Even the 5Ds is only about 61lp/mm.  This is a common small format sony HAD chip.  If you would pardon my french, it "outresolves" the shit out of anything in your average camera. 

 

 

What counts to me: each Sigma lens design since they started their global vision thing was great - at minimum, and usually better than that.

Quite a few are "meh." 

 

18-200, 18-300, 17-70, 30/1.4, 24-105 to name a few.  In my personal opinion they are all rather poor, as are most camera lenses from a lens designer's stand point.  They all merely "meet spec," as we say - smaller format optics and research-grade optics are corrected to about a full order of magnitude smaller spot size. 

 

 

Neither you nor I do know the numbers of lenses sent back.

Well that's not explicitly true.  I have "dud" numbers from the largest buyer of camera lenses in the US.  They say that the # of bad sigma lenses has not significantly decreased since 2009 when they began logging data. 

 

 

The last Nikkor I got is in the workshop because of a faulty VR - the one Sigma "copied".

A sample size of 1 (or 2, 3, 4, 5, and so on) does not speak for the population. 

 

 

And can be found on many pictures left on a SD card when you get back the body from service, throughout all manufacturers.

As far as I am aware, there is only one manufacture that does this... and I have data from all the big players.

 

 

Sure, you're shopping at those kind of companies, aren't you. I'm not. Actually they can check it with a crystal ball, I don't care, as long as the lens is helping me doing great pictures.

Nope, I don't have anywhere near the $4500 to drop on one of their lenses.  I just have a friend who spent last summer doing QC for them.  I also know that the particular lens I am thinking of is regarded as one of the very best SLR lenses available at the moment.