Samuli Vahonen
Offline
• • • •
Upload & Sell: Off
|
 p.38 #11 · Zeiss "Batis" AF lens line to be released for FF E-mount | |
Matt Grum wrote:
If done correctly, correction for barrel distortion makes the centre of the image sharper. I've not seen any software that does it optimally, however, but that doesn't make the concept of software distortion correction flawed, only the implementation.
Making center sharper or corners less sharp - the issue remains, artificial weird looking transformation how small textures appear (or like theSuede says, it's mostly artifacts...so not claiming it's real detail, more question of how it appears to human eye). Also all pixels have been interpolated when doing distortion correction, so like theSuede described, it's not actually getting sharper in center, just losing less sharpness than in corners.
(of course is barrel distortion is like 35% then compacting larger number of pixels to smaller area may actually improve center - but on normal application it's not if softness caused by interpolation is taken into account)
Matt Grum wrote:
I have a real hardwood floor in my house that was quite expensive. There are now some really high end laminate floors that are in many ways superior (e.g. more durable and scratch resistant) and approach the same prices. However I would still prefer to know I have real wood.
I haven't seen, heard or felt laminate which is as comfortable walk and feels as good&warm (without any floor heating stuff involved) under the feet as proper wooden floor. I don't think those two are comparable, except in aesthetics - at least not in cold countries like in Finland. Good choice you made for your house!
Matt Grum wrote:
So I can divide things up according to whether I care about authenticity and the process, or whether only the final result matters. For me camera systems go in the same bucket as shoes - I don't care what the process is as long as the final result is good.
I think this is the hard part - "good" final result is subjective as we all have different requirements for our images. Also it's subjective if "good" is enough - for example I have to make compromises at work on the usual enterprise life every day. But on photography I can't accept just good, and I always aim for best. If I haven't done best I can, I'm disappointed and try to learn from mistake and avoid same issue in future to do my best in future.
As a practical example, I had Zeiss ZE-series 2/25 with me in my last holiday, which was mainly about shooting mountain landscapes, and some of the landscapes photos I took can't be enlarged above typical size shown here in forum because of the field curvature on extreme corners. When I got home and did see those photos I was disappointed. I had ruined my changes of even doing my best back at home when selecting what lenses to pack with me to the holiday. As a corrective measure, it's clear to me that for a trip, which is mainly about landcapes, I'll take Leica Elmarit-R 28mm v2 or Batis 25 [when available] or Otus wide [when available]. Of course there are drawbacks on taking Leica; the "sun in the frame shots" would not be possible, and it's not a lens for closeup boke containing shots either - however it would have been best compromise for this location and holiday.
Same way I would be disappointed for every photo, if I would need to use distortion correction to make them usable. Maybe ignorance is a bless, and we all just should process so that automatic lens corrections would be always "ON", so that we would not even be aware of what our equipment could do at their best. Hmmmmm, that didn't sound right... 
Matt Grum wrote:
Ok, so how about a slightly more relevant example. If you have two $1000 25mm f/2 lenses and they produced the same level of distortion in the image, and one was corrected optically, and one was corrected digitially, and the results are identical except that the digitally corrected lens has a flatter field, less spherical aberration and slightly less longitudinal CA (axial colour), which would you chose?
In order to produce identical results by your definition it means that on small textures (microcontrast) your software corrected lens would need to have extreme high 40lp/mm in MTF to compensate required interpolations so that appearance of small textures stays same as with the other 2/25 lens (unless it's really crappy and has 40lp/mm MTF <30% or something else weird). In order to maintain uniform microcontrast over image also MTF curves would need to mirror the microcontrast loss due to distortion correction, which varies a little based on distance from image center and due to every pixel being interpolated would also overall be reduced. I don't think it's feasible to design this kind of lens for $1000, sounds more like $10k for the >90% contrast requirement in 40lp/mm.
Also one thing to consider - if someone would make this kind of superlens that it has 40lp/mm contrast to compensate the interpolations: due to diffraction it would drop below 90% contrast already on f/4, and due to diffraction both lenses would be as bad BEFORE SOFTWARE correction at f/11 - then you apply the software correction to software corrected lens and results would be really bad for microcontrast.
If the goal is to produce small photos e.g. 1280px wide to be posted on forum, where resizing will cause small detail to be lost in any case, then software correction is superior. So the answer lies on your requirements for final image and it's intended usage.
For my usage I would prefer not to pick either of your lenses, maybe the software corrected lens if distortion would be low enough to not correct it for nature photos. But idea of having LoCA sounds bad as it cannot be corrected in software except on focal plane, and even then only in some conditions, and as my photography usually contains boke the LoCA lenses are really not welcome.
Part of the argument here is also that we are no longer in "Alternative Forum". For example this thread is visible on Alt and Sony forums. We have two very different group of people; "the Sony people", who wish to get nice Sony system and don't care about alternative stuff (=same as typical Canon or Nikon forum user). And then the "Alternative people" who are used to adapt lenses for best results or value - and don't care so much about is the camera body Sony or Xxxx as long as it works for them (currently Sony A7-series seems to work best for most of us at 2015, but over the years we have seen people jump to Canon 5D for 1st full frame, then Nikon D800's for better fullframe sensor etc.), and via their experience they have learned to appreciate wider usability possibilities of lenses. I have hard time seeing consensus on this digital correction topic.
PS. I don't fundamentally avoid and hate distortion correction. When I shoot panoramas the photos are anyways interpolated when 2D-projection is calculated in a way that smallest detail is lost (or artifacts according to theSuede...). As part of the process panorama software always fixes the lens distortions (it's measured based on how key points on overlapping images differ). In order to maintain quality despite extra distortion&perspective corrections, panoramas should be shoot with plenty of extra resolution for intended image requirements and usage. Typically 15000px wide panorama looks ok resized to 4K display or similar print width, which 24Mpix camera is able to produce. So maybe 15000x10000 [150Mpix] (if maintaining 3:2 ratio) would be enough for me to use distortion corrections - but then I could see how much could be achieved when using full image...
PS2. If somebody is reading this at Zeiss: please speed up getting these Batis lenses to market, before your potential customers get bored on discussing theoretical stuff about your coming lenses
Samuli
|
? If yes then I would wear the genuine leather. If no, then I wouldn't care because I wouldn't know anyway.