01-22-2015, 03:48 PM
Quote:I am wondering why the mirrorless lenses are generally less "harmonic" than the DSLR designs.I would imagine that it is partly forced due to the register distance and digital medium, in lenses which are either symmetric or ultra-wide angle designs they essentially must make the lens rearside telecentric - meaning the chief ray travels roughly parallel to the optical axis instead of hitting the sensor at an angle. Here you see a 12mm f/1.2 lens for roughly 1" chips:
The green ray bundle originates off-axis and travels to the edge of the sensor - clearly it is the chief ray (I would know since I optimized this lens, but I am trying to clarify as I often reference the chief ray and I am not sure people know what it is). The center ray of the bundle is the 'actual' chief ray.
See how the angle at which light approaches the sensor is slight? that means this lens is 'nearly' or 'somewhat' telecentric. If the center green ray were a straight line parallel to the dot-dashed line it would be a truly telecentric lens.
Because the rear element diameter will rival the size of the sensor, you are forced into a nearly telecentric design space. Due to the troubles with digital chips and obtuse angles, it would be natural to just go all the way.
But a telecentric design forces several potentially nasty conditions on the design, so you lose degrees of freedom to optimize with.
There is also the question of size- take the 12mm f/2 from olympus for example. If you scaled that lens to FF you would genuinely have a 24mm f/2 without changing anything but the size of everything. However, the increase in rho by a factor of 2 would enormously increase aberrations with dependency on rho and you would likely become staunchly limited by 5th order aberrations.
Anyway, that lens is quite small. In an academic or research sense it is often easily possible to produce a diffraction limited design (I would ignore most people using the phrase diffraction limited online - photographic lenses are very rarely DL). The trouble is the many many restrictions imposed physically. You want the rear element +=2mm from the flange distance to avoid an excessively long lens, but it physically cannot get too close. You have unfair restrictions placed on you for lens diameter and length. You are also limited in the materials you may use by price or availability. Here is an abbe diagram:
If you find BK7, that is usually the left termination point. BK7 happens to be super easy to make and is very common - it is the cheapest materials and most materials are listed in the catalog as a multiple of the price of BK7. When designing lenses, the left cap for glasses that are allowed is usually BK7 - beyond that is the "low dispersion" realm where glasses cost 60x+ what BK7 does.
On the other side there is LaSF. LaF materials are fine generally. LaSF are hard to make and hard to work with. They are fragile and universally 100x+ the cost of BK7. If you use them they quite literally drive the discussion regarding price of the lens.
Restrictions like these physical and material ones cause photographic lenses to often be not as good as we'd like. Cinema lenses have many of these problems relieved through much larger bodies and much higher prices.
I will reply more later - class soon.
Still, the distortion is very disappointing. Canon's 16-35 is equivalent but has values of rho much larger and a larger field while having a better correction without costing much more.
edited to add more:
the tl;dr of above is that there are more constraints in mirrorless lens design (which is quite ironic, as the removal of the mirror removes one constraint but the mechanics and packaging add more than the one taken away).
The additional constraints hurt the design by limiting what can be corrected.
The sensor could contribute to the vignetting and definitely does contribute some, as sensors always will, but I am skeptical that the vignetting is not mostly optical. Vignetting is a 'valid' corection mechanism for corner performance if the customer or designer is okay with a loss of energy at the edge of the detector.