01-12-2016, 03:31 AM
Quote:Well, Brandon, a couple of things in your post deserve another point of view. To me, some of your posts I simply don't understand because you're talking laboratory instead of shop That makes you sound cocky from time to time (and I use the verb "sound" on purpose, as I don't know if you are cocky). Not that I understand "shop" much better than "laboratory" I'm only one of those consumers spending the money to make lens designer's salary.Any of the technologies I'm talking about are "industry-ready" and utilized. Freeform has been around since the 1980s (polaroid SX-70) commercially. Older-generation designers unable to utilize advanced technologies aren't trash, but they also are not the best designers around and I have no intention of mincing my words.
Putting the old guys more or less completely into the science trash bin is another thing taking some sympathies - but quite normal for well educated dudes in your age. At first, the 28 year old pioneer at Zeiss might be young but benefits of all the experience and knowledge of hundreds of old guys, the persons who learnt a life long simply because there were no internet researches possible when Zeiss started to make his microscopes better by try and error, no programs available. The guy might be a genius by designing freeform elements, but change his computer against a chalkboard or notepad and see what happens. That works the other way around, too - give the old guys the modern computer and see what happens. Not very much, I assume.
All the today less cool people were necessary to make the cool work of few possible - and even if they only were working against them, resistance is also a motivator. It's not a crime to remember this from time to time, it's also no weakness to be humble. No offense meant, I guess you get it the right way.
Also, lens design is not just optical design and I suspect you're talking more about the optics, the mathematical miracles delivering all this contrast and resolution record numbers. However, what happens if the mechanics can't stand the environment those lenses are used in? If it's so obviously simple to design outstanding stuff, why don't we see more of it at affordable prices?
I totally agree with your statement about Nikon although I'm not entirely sure if it's grammatically flawless : Too late and not daring enough. But please put against a more daring and more up-to-date design: they mess up with some things and apparently don't run a good prototype testing lab. They get some great photog names, give them the lens and show what it can do. The VR fault of the 300PF is telling something how much optical designers talk with electronical and mechanical design colleagues. Not testing a lens with all available bodies and at all shutter speeds is inexcusable. There were other callbacks as well and I'm afraid, not everything can be fixed in Nikon's repair shops.
Freeforms aren't really designed by hitting "go" in the software either. In producing a tilted and decentered (folded) system the performance immediately dies. A 50/1.8 like the 1.8 STM has about 1.5-2 waves rms wavefront error at full aperture. A freshly titled system has about 20-40 waves rms wavefront error. To recover performance you need to do a very detailed study of the new aberration fields of the system - some geometries are uncorrectable (e.g. those in which the coma is in the wrong orientation) while others can be more than recovered and outdo the original symmetric form. An example would be a recent TMA telescope we designed - nominal was 1/5 wave worst point in the field, after folding it up and turning it into a freeform the performance was 1/100 wave worst point in the field.
As of yet the tools for this aren't totally supported in the design software, so the math is done in Matlab in our case, someone else might use python or something. Code V has been around since the 80s though, and the first optical design codes (FLAIR, etc) were designed in the 1950s and 1960s by Robert. E Hopkins. I think Zemax was started in the 90s, but I don't use it so I am not sure.
3rd order aberrations can be corrected by hand calculation - the maths is not so brutal. They are sufficient up to about a 5 degree full field of view at f/10. The higher order aberrations are 'impossible' by hand (for a 6-element double gauss each coefficient would require over 10 pages of math, the probability of making no errors in the calculation is ~0). This is where CAD comes in - perfect calculation of the formulas, and since the 90s or so real ray tracing instead of paraxial approximations.
The mechanics of consumer lenses are interesting. They offer many more features than your average industrial lens, so they are a small miracle. On the other hand, they offer much less precise construction, are less robust, etc.
Here's an industrial projection lens [url=http://i.imgur.com/SqRoGEc.jpg.%C2][/url] The mount is 1/8" steel, as are many of the barrel pieces. Not even a supertelephoto has this level of build quality.
Regarding outstanding at good prices - optical manufacturing is extremely expensive. If you want a single piece of the front element for a double gauss 50/1.4 you will pay $550+ for it. If you want a hundred pieces you may pay $25/ea for them. At a hundred thousand pieces, I don't know - perhaps only $15-$10/ea. At that point it depends almost entirely on what glass it is - the most common is $7.20/lb, but the most expensive ones are several hundred dollars per pound.
A full-frame sensor is also actually extremely large. Most designs are made for perhaps a 5mm detector, 10mm at the largest. distortion and petzval become very difficult to sufficiently control on such a large image area; they are also the most difficult aberrations to work with in general, as there is no zero condition for them where an aplanatic surface produces 0 spherical, 0 coma, and 0 astigmatism. Clever tricks can be used for the other aberrations, but to fix petzval and distortion the solution is pretty much make it bigger. Miracles of price can be worked on smaller chips (see: 10-18 for $300, the brevy of f/0.95 and faster lenses for m4/3, etc) but full-frame is difficult.