[Scythels]

JoJu,

 

Canon INC is a $40B company, yes?  They are the single largest player in the camera market by about 2x.  Optics is a $15T USD industry worldwide.  Cameras are a very small part of optics which is already a small industry.  The majority of optics are very low volume, very high price units and are tested in methods similar to what I described. 

 

E.g one of the ref. surfaces we have is this: http://www.astro-electronic.de/IMG_2054x.jpg - it is a 37mm f/0.68 lens.  It is about a $20k usd part, and the "performance" of the lens is many orders of magnitude higher than a camera lens in its given role (as what allows ref. surfaces to be so high performance is that they only have to work for monochromatic light) while working at an aperture that has never been achieved in a camera lens. 

 

 

 

because you're not able to do better at this price

 

 

Testing on production level and being good enough

Good enough varies from individual to individual.  I once did some variance analysis for LensRentals on about 25 copies of each the 35/1.4A and the 35/1.4L with obvious duds pre-rejected.  The 35L varied 7.8% between copies, the 35A about 13.03%.  From that data, sigma allows approx. twice as much variance as canon.  Perhaps they have improved in the 8 or so months since I did that, I don't know - but I doubt it.

 

You may argue that 25 copies isn't enough to speak to the entire population - this is true, but then 25 copies is far more than any single, average user will ever use and the gap between the two is nearly 50% so the result is still significant. 

 

So I would say "which is better than others" is incorrect with regards to QC.  They do have some better designs, but these designs are coming up on nearly 20 years newer than their competitors and the superiority is not quite as glaring as some make it out to be.  Take the 35A vs the 35L for example - sigma's lens clearly has less spherical aberration, less oblique spherical, and a bit less coma, but the difference in the center of the two lenses is "just" perceptible and the corners aren't actually visually that much worse than the 35A from the 35L.  If you look at the comparison here - http://www.the-digital-picture.com/Revie...&APIComp=0 - you can see that.  The tested 35A seems to be a "bad" copy with a tilted tangential field but you may not have noticed that if I didn't point it out[...]  I would say the "general" picture from the lens is okay anyway.  Or there's Roger's data, which indicates that the 35A is only marginally better than the 35L for the most part.

 

 

YOU are not able to make the design

I am a bit confused.  Very very few lens reviewers do anything related to building, designing, etc, optical systems.  That doesn't mean they can't criticize things.  In the same capacity that I can say Kubrick did a terrible visual rendition of A Clockwork Orange, I can speak about sigma's lens designs and constructions. 

 

Besides, in a couple years I will be designing lenses for a living. 

 

 

But pulling exotic lens design against photo-industry stuff

The $4500 lens I have referred to is a photo-industry lens.  It even covers a full-frame sensor which is a bit unusual for lenses in general. 

 

Here's a short list of lenses that are cheaper that any SLR or MILC lens you own, and simultaneously better, optically:

 

* the lens in your smartphone

 

* the lens in any "premium" compact camera

 

* Small-sensor "photo industry" lenses like the ones used for the google street car.

 

A lens that you wish to out-resolve 20-24mp class FF sensors with requires MTF of 50% at just 30lp/mm.  This is extremely low in the grand scheme of things, and nearly all ILC camera lenses are pretty poor optically.  From all manufactures. Exceptions are things like canon's 300/2.8L II, 400/2.8L II, the 135mm f/2 APO-Sonnar, etc.  Those are merely "good" lenses, compared to all other lenses. 

 

----

 

If you wish to give me advise, I will offer you some as well.  The ILC camera/lens industry in total is less than five percent of the total optics industry.  Real World Lens Business is so, so, so much more than what fits EF/F/FE/X/u43/etc mount.

[JJ_SO]

AiryDiscus, when using bold, italic and underline to emphasize your point of view, why not going the extra mile and 

 

yell really loud?



 

Just asking.

 

Can't help, just can't take you serious, no offense, kid.

 

as for your list:

 

Have no dumbsmartphone

 

Funny, those premium compact cameras are not much in use, but I'm sure they kick butt like hell.

 

google street view or street car which rides on his own to places no one wants to see?

 

20-24 MP FF? Come on, you want to bother with minor bread crumbs? Aren't you aiming to higher spheres?

 

Full frame covering lenses? wow, another mind-blowing thing just entering my horizon yet, thank you soooooo much.

 

The way you criticize, young man, just happens to be without any respect to manufacturers. Prove to do better.

 

As for 35/1.4A vs 35/1.4L - ever cared to take a look at the price tag? You're right, 25 copies are a drop in the ocean and after years of having their high priced lens in the market, Canon should be able to keep sample variations low. But now, what tells that number? If the worst Sigma is as good as the best Canon, who cares about sample variation?

 

You're probably referring to that blogpost: http://www.lensrentals.com/blog/2014/09/...35mm-f1-4s

 

So, why aren't you quoting Roger properly? Let me do that for you, sunshine:

 

"The optical bench results for the Sigma 35mm f/1.4 lens show that it does has the best MTF, as most testing suggests. All of these lenses are quite good, however, and the Canon and Zeiss lenses aren't far behind the Sigma.

<p style="font-size:13px;color:rgb(74,74,74);font-family:'Helvetica Neue', Helvetica, Arial, sans-serif;">The field curvature may be a bit surprising to many of you, and may actually be of more practical value than the MTF results. We'll be discussing the field curvature of these lenses a bit more in the next post.

<p style="font-size:13px;color:rgb(74,74,74);font-family:'Helvetica Neue', Helvetica, Arial, sans-serif;">The variation between best and worst areas on each lens is amazingly good for the Canon and Nikon lenses. The Sigma and Rokinon 35mm f/1.4s have a bit more variation, enough so that on most copies, if you really, really tested carefully, you could probably detect one corner a bit softer than the other.  I doubt you'd see it in a photograph for the "average" lens, but since the variance is higher, there's probably a higher chance of getting a copy with a corner soft enough to notice in photographs.

Over all, though, all five of these lenses are very good from an MTF standpoint. Some are better than others, obviously, but price range, bokeh, and a number of other factors will make a bigger difference than the MTF for a lot of photographers."

 

And this at 40% less money… and since my bodies are made by Nikon, to me it's 53% less. Do I need to tell you what I give on your "sample variation"? 

[Scythels]

Dear JoJu,

 

I believe you misunderstand what I am trying to say. 

 

 

20-24 MP FF? Come on, you want to bother with minor bread crumbs? Aren't you aiming to higher spheres?

I'm referring to design criteria for many camera lenses.  With camera lenses, it is good practice to design the lens for the sensor it will be mated to.  Obviously this is complicated for ILC lenses, but the 20-24mp FF chip pixel size is fairly typical for camera lens specs. 

 

Here is a 60mm f/1.4 lens design in progress at f/1.4. 

 

http://i.imgur.com/oSMvhFX.png

 

The 30-40lp/mm region is what is "of interest" regarding sharpness for a 20-24mp FF chip.  The dotted line is the diffraction limit.  "Good" lenses are within about 50% of this at decently high spacial frequencies (80 - 120lp/mm is my personal ballpark for f numbers between 1.4 and 4).  Here is the lens again at f/8:

 

http://i.imgur.com/FrkmdqO.png

 

Over most of the sensor, this appears as sharp as is possible. 

 

Here is a very high performance lens working at f/0.9:

 

http://i.imgur.com/URuhQeB.png

 

In case you can't see that this is a real, aberrated lens because of the performance, here is out to 500lp/mm:

 

http://i.imgur.com/sHWPgLV.png

 

This is top-tier performance. 

 

Here instead is something more "common" for "good" lenses:

 

http://i.imgur.com/YPmpwBo.png

 

This piece cost $1500 USD equivalent when it was sold. 

 

So you can see, "general camera lenses" perform rather poorly and there is enormous room to improve.  "Giants" like the 135 aposonnar, or canon 300/2.8L II, or the 400/2.8L II are maybe 60-75% of the way to the performance of that f/2.7 lens I showed. 

 

I am very frank with my opinions.  If this offends you, tough luck =)  I need not be "nice" to any manufacture, I'm not on any of their payrolls and see no reason to treat camera lenses as if they are better than they actually are. 

 

 

As for 35/1.4A vs 35/1.4L - ever cared to take a look at the price tag?

Does it matter?  Obviously in the grand scheme of things it does, but they are not priced in different classes and the extra for the 35L doesn't place it in a class such that it will have tighter tolerances because of the expected customer base (i.e cinema, scientific, or other market with high expectations).  The 35A has twice the variation of the 35L.  If you really want I can dig up the old data and show it to you numerically, but I suspect you have already plugged your ears.

 

 

So, why aren't you quoting Roger properly?

I'm avoiding quoting Roger because there are some things I have been told under NDA and some things I have read on a blog post, the same as you.  It is a defense mechanism to avoid giving out information under NDA.  This is why I also will not disclose the name of the manufacture of the $4500 lens above, or which mfg uses a bookshelf to do QC, and so on. 

 

----

 

bonus factoid: the automotive sector of the photonics industry is currently approximately equal in size to the entire camera industry:

 

http://www.osa-opn.org/home/industry/201...QyjqOHHrL9

[popo]

A quick question, is that "f/0.9" lens designed intended to be used over the visible spectrum? Or is it optimised for a smaller region to be used at any time? Or may it be restricted in other ways, such as focus distances? e.g. in a similar way astronomical telescopes only need to work well at infinity focus. I guess I'm asking, what are realistic limits for a photographic lens which ideally would be corrected for use over the whole visible light spectrum, and cope with focus ranges close to infinity.

[Scythels]

Hi popo,

 

The f/0.9 lens is designed over the standard visible spectrum, just as most photographic lenses are.  It is designed only for a close focus distance and has no moving elements, but its entire assembly could be moved to refocus, I just don't know what the performance would be like.  At infinity it is actually f/0.76 or so according to Code V.

 

Realistic limits are set by price and physical constraints.  For example, a 50mm f/1.0 lens can be designed with 300/2.8L II performance or better, it would just be very very large and heavy, and the rear element probably would need to be much bigger than the throat diameter of F or EF mount.  At more reasonably apertures e.g f/2 or even f/1.4, the sky's the limit on performance if you can pay for it.

 

IL Camera lenses are heavily limited by price, size, and weight limitations so they perform relatively poorly.  When the price and size limitations are loosened some, designs such as the 300/2.8L II or its peers emerge.

[JJ_SO]

Dear JoJu,

 

I believe you misunderstand what I am trying to say.

No, you didn't get my sarcasm

 

So you can see, "general camera lenses" perform rather poorly and there is enormous room to improve.  "Giants" like the 135 aposonnar, or canon 300/2.8L II, or the 400/2.8L II are maybe 60-75% of the way to the performance of that f/2.7 lens I showed. 

 

I am very frank with my opinions.  If this offends you, tough luck =)  I need not be "nice" to any manufacture, I'm not on any of their payrolls and see no reason to treat camera lenses as if they are better than they actually are.

That's my point as well. I'm not related to you and your behavior as if you could design better camera lenses anytime is just infantile. I guess, none of your as "better" or "superior" displayed lens designs is available on the market? Course not, what stupid question am I asking. Next to things like AF, aperture drive or OS/VR there won't be a common camera mount included in your cheered designs and a couple of raindrops could ruin the whole experimental setup as well. I know, this all will just cost nothing more after you rambled about and the lenses will sell itself, no need to make advertisements for them.

 

Does it matter?  Obviously in the grand scheme of things it does, but they are not priced in different classes and the extra for the 35L doesn't place it in a class such that it will have tighter tolerances because of the expected customer base (i.e cinema, scientific, or other market with high expectations).  The 35A has twice the variation of the 35L.  If you really want I can dig up the old data and show it to you numerically, but I suspect you have already plugged your ears.

The price does matter, at least to me. And since the L lens is in no way optical better than the Art lens I don't benefit in buying a worse lens with tighter tolerances to stay underneath the cheaper lens. My plugged ears are just a defense mechanism to big mouth people.

 

I'm avoiding quoting Roger because there are some things I have been told under NDA and some things I have read on a blog post, the same as you.  It is a defense mechanism to avoid giving out information under NDA.  This is why I also will not disclose the name of the manufacture of the $4500 lens above, or which mfg uses a bookshelf to do QC, and so on.

No difference between your presentation and one of a fraud who claims, "I could tell you things but then I would have to shoot you" - if you're not allowed to go in details neither to show a ready lens with those superior resolution curves, why not just shut up? The way you do it, I can't see a difference between possible truth and also possible lie. But once you get a job in the optical industry you can show us how good a lens can be. This will be fun. I'm just afraid I will not live long enough to see it happen.

 

Maybe nobody told you: Pretending to be able to make better lenses and at the same time judging the whole or majority of photo lens industry as idiots only giving us lousy stuff, it's opposite of cool. You'd need to have a charisma like Steve Jobs to go on with that attitude.  

 

Now, how about a contribution to the opening question of the thread instead of hijacking it to show off how much you know about lens design? Any idea about the VR / OS / IS, how they work or what they do, when switched off? 

[Scythels]

 

That's my point as well. I'm not related to you and your behavior as if you could design better camera lenses anytime is just infantile.

It's actually readily possible to design better lenses.  The sky is the limit in lens design, the big problems are making it manufacturable and making it cost effective.  ILC lenses are generally pretty poor performers, any lens designer worth their salt can design lenses several times better with relative ease.  Typically the requirements when you take lens design courses are actually to produce higher performance lenses than them by a rather large margin. 

 

I believe you are missing the point of what I am saying - ILC camera lenses are highly aberrated.  All of them.  Even the "greats" are really "not so greats but kinda goods."  Again, from all manufactures.  I would estimate that 85% of imaging optics are better corrected than ILC camera lenses. 

 

 

I guess, none of your as "better" or "superior" displayed lens designs is available on the market?

The f/2.7 design is about 35 years old I think.  Robert E Hopkins designed it ages ago for the nikon F mount.  You used to be able to buy it for ~$750USD equivalent today.  Bob is often considered the "father of optical engineering," one of the reasons I chose to show you a design of his is to prove that it isn't just research optics that are this well corrected. 

 

 

a couple of raindrops could ruin the whole experimental setup as well

I am a bit confused why you speak as if I am a fool and shouldn't talk about the A1 testing tool sigma created, but are willing to speculate as to the durability of these lenses.  The f/0.9 lens is extremely large and the complete assembly weighs several hundred (if not thousand) pounds.  It's not a type of lens designed for environmental use, so I don't know exactly how durable it is - but is extremely highly engineered mechanically as well as optically so I doubt even hitting the barrel with a sledgehammer would disturb the optics.  The f/2.7 lens is built as well as other premium MF lenses from its era.  By most photographer's standards that is very very well and "indestructible" but I personally wouldn't exactly describe that construction style that way. 

 

 

I know, this all will just cost nothing more after you rambled about and the lenses will sell itself, no need to make advertisements for them.

Advertising is the Achilles' Heel of the entire optics industry.

 

 

No difference between your presentation and one of a fraud who claims[...]

I am not a liar. http://i.imgur.com/y0gMhKQ.jpg

 

 

if you're not allowed to go in details neither to show a ready lens with those superior resolution curves, why not just shut up?

This will come, actually sometime in the near future, but not for free as it is actually legally complicated to do so.  The f/0.9 lens I showed data from I made a call to the designer to ask if I could show you the MTF curves, and the f/2.7 lens's patent has expired.  I can show you an f/1.2 lens bob designed that you can buy used for $100 if you want as well.  It was actually the first commercial lens to use an asphere. 

 

 

But once you get a job in the optical industry you can show us how good a lens can be.

I already have one (two, actually).  The f/0.9 example I showed is close to the "maximum capacity" of what is possible because it is so well corrected and so fast.  They really don't get all that much better than that.  The fastest lenses of that class are currently about f/0.6 and 3-8% higher resolution at 500lp/mm.  There are actually serious problems in that area of lens design now that we are approaching the f/0.5 limit and reflector (vs refractor) systems will begin to have to be explored, which presents issues as reflector systems require a great deal of optical "folding" to remain compact.  I suspect freeform surfaces will come into play extensively, but I am not sure they are manufacturable at the required diameters. 

 

With regard to the camera industry, they do generally give "lousy stuff" in that (IL) camera lenses tend to be "very highly" aberrated.  Smaller format lenses are much better e.g a customer's lens I am doing early prototype QC/mtf testing on which has an MTF of 68-48% at 104lp/mm depending where on the sensor you look.  By comparison, the 135mm f/2 APO-Sonnar has an MTF of .48-.03 depending where in the sensor you look at the same frequency.  The customer's lens is also cheaper per unit than the APO-Sonnar.  Like most lenses, it cannot simply be purchased as it was designed and is being built under contract for a specific customer of theirs.

 

 

Now, how about a contribution to the opening question of the thread instead of hijacking it to show off how much you know about lens design? Any idea about the VR / OS / IS, how they work or what they do, when switched off?

I don't know everything JoJu (far from it, really) but I've said all I know there already. 

 

http://forum.photozone.de/index.php?/top...entry29828

 

Perhaps there is an opto-mechanical engineering hanging around this forum that knows better than I do.

[Brightcolours]

Which lens was that first commercial lens  f1.2 lens with aspherical element? Was that that small film lens?

[Scythels]

Navitar 12mm f/1.2.  Designed circa 1954 before the first computer aided optical design software was released in 1961.  You can find it online as "Elgeet 12mm f/1.2" since it was manufactured by the Elgeet company.  In reality it was more like f/1.3 - f/1.4 but that would be seen by the photography as "T/1.4" instead of an actual rounding of the F number, though that is extremely common (e.g nikon 24/1.4 is f/1.45). 

 

Here's the MTF as manufactured:

 

http://i.imgur.com/RwlI9EB.jpg

 

It was actually nearly perfectly corrected for all third order aberrations which were all that was very well understood at the time.  I did an optimization of it as part of a case study to see what could be done without changing the design from what was possible in its day and age and managed to produce a manufacturable and toleranced lens with this nominal MTF:

 

http://i.imgur.com/g5IUsOe.jpg

 

IIRC the tolerance was MTF of 0.4 at 30lp/mm for any image height.

[Brightcolours]

That was a nice little lens, how big is 1/2" video tube actually, to understand its FOV it was designed for?