[Brightcolours]

Not sure why the IS would be better...

[Brightcolours]

Personally, I couldn't care less about the complexity of a lens. In fact, given two lenses, a simple and a complex one, featuring the same IQ, I'd take the simpler one any day.

 

Also: why is that that a f2.8-4.0 lens is easier to design than a constant f4 lens? It doesn't seem to make much sense to me given the 2.8-4.0 will be brighter at best and the same at worst.

The reason is because the apparent aperture needs to stay constant when wide open. At 16mm it needs to be 16/4=4mm, at 70mm it needs to be 70/4=17.5mm. To achieve that one needs to use a few tricks which give one less freedom in design. 

[thxbb12]

The reason is because the apparent aperture needs to stay constant when wide open. At 16mm it needs to be 16/4=4mm, at 70mm it needs to be 70/4=17.5mm. To achieve that one needs to use a few tricks which give one less freedom in design. 

 

So what you're saying is that a 16-70 f2.8-4 is easier to design than a 16-70 f4?

Following your example above: 16/2.8 = 5.7mm, 70/4 = 17.5mm. In which way does it make it more difficult than 4mm and 17.5mm?

I don't see who would actually rather have a constant f4 lens over a f2.8-4 one. That seems very silly IMO.

[Scythels]

Zoom lens design is extraordinarily complicated and there will actually be an article on it sometime down the road on photozone, but here's some basics.

 

Zoom lenses are designed with a set of "modules."  These are groups, but not in the way lens specs list groups (where a "group" is simply one set of lens(es) that only have two air-to-glass surfaces). 


Typically or classically a zoom uses 4 groups.  Fewer or more is possible, but it is very difficult optically to do it with fewer, and difficult mechanically to do it with more. 

 

The size of the entrance pupil is what determines the f number.  The size of the physical aperture does not matter.  If the zoom motion occurs only in front of the aperture stop, the lens will have a constant f number.  This forces the design into a very narrow solution space since there is extremely limited room to work with.  You have essentially thrown away a third (or more) of your degrees of freedom for correction in order to have a constant aperture. 


If you wish to allow zoom motions to happen on the rear side of the aperture as well, the focal length will be changed by movements which do not affect the entrance pupil, thus the f number does not stay constant with focal length change. 

 

The entrance pupil is the image of the physical aperture as seen from the front of the lens.  It is virtual but does have a location and apparent size.

[thxbb12]

Zoom lens design is extraordinarily complicated and there will actually be an article on it sometime down the road on photozone, but here's some basics.

 

Zoom lenses are designed with a set of "modules."  These are groups, but not in the way lens specs list groups (where a "group" is simply one set of lens(es) that only have two air-to-glass surfaces). 


Typically or classically a zoom uses 4 groups.  Fewer or more is possible, but it is very difficult optically to do it with fewer, and difficult mechanically to do it with more. 

 

The size of the entrance pupil is what determines the f number.  The size of the physical aperture does not matter.  If the zoom motion occurs only in front of the aperture stop, the lens will have a constant f number.  This forces the design into a very narrow solution space since there is extremely limited room to work with.  You have essentially thrown away a third (or more) of your degrees of freedom for correction in order to have a constant aperture. 


If you wish to allow zoom motions to happen on the rear side of the aperture as well, the focal length will be changed by movements which do not affect the entrance pupil, thus the f number does not stay constant with focal length change. 

 

The entrance pupil is the image of the physical aperture as seen from the front of the lens.  It is virtual but does have a location and apparent size.

 

<p style="color:rgb(40,40,40);font-family:helvetica, arial, sans-serif;">To paraphrase what I wrote to BC: in light of what you just wrote, designing a 16-70 f2.8-4 is easier than designing a 16-70 f4?

<p style="color:rgb(40,40,40);font-family:helvetica, arial, sans-serif;">If I follow the same logic, is designing a 16-70 f1.4-4 also easier than a 16-70 f4?

<p style="color:rgb(40,40,40);font-family:helvetica, arial, sans-serif;">That doesn't seem to make sense.

<p style="color:rgb(40,40,40);font-family:helvetica, arial, sans-serif;">Additionally: who would actually rather have a constant f4 lens over a f2.8-4 one?

<p style="color:rgb(40,40,40);font-family:helvetica, arial, sans-serif;">Likewise, who would rather have a 100-400 f4 vs a 100-400 f2.8-4?

<p style="color:rgb(40,40,40);font-family:helvetica, arial, sans-serif;">I know what I'd like better and I think I'm not the only one...

[davidmanze]

 It was not may weeks previous that we were in "awe" of the sharpness of the Otus 50 and 85mm F1.4s and their technical and optically  brilliant  offerings,

 

 

   now it seems, we have to satisfy ourselves with technical and optical excuses.

 

  If I were Mr. Sony,  I'd be taking out Mr.Sigma for a slap up sushi dinner and it would completely slip my mind to invite Mr. Zeiss!

[Klaus]

Regarding astigmatism once more - the 16-70mm is hardly the only lens affected by this. Essentially all lenses till 50mm are. 

[Brightcolours]

Regarding astigmatism once more - the 16-70mm is hardly the only lens affected by this. Essentially all lenses till 50mm are. 

Not many lenses show such a poor border performance at f4 in your MTF tests, and yet seem to do a bit better in the images shown. Hence my suggestion to shoot a few MFD shots to see if the astigmatism is visible. 

My example lens (the 20mm f3.5) shows what I mean.

[Scythels]

Designing a 16-70 f/2.8-4 may be easier, it may not.  It depends how you arrange the groups of lenses and the general form of them.  If for example a group begins with a negative element, you would not want it behind the aperture stop in a wide-angle lens because the beam will vary in position enormously as you move away from the center of the lens.  This will force the lens to have enormously large diameter lenses to avoid vignetting.  Thus, a "P" group behind the stop is the most advisable form for a WA zoom.

 

Increasing the aperture while maintaining quality is difficult.  Whether it is more or less difficult than designing a constant aperture zoom - I cannot say absolutely.  What I can say is that f/2.8 is not so ridiculously difficult an aperture to reach, nor is it exceptionally sensitive.  F/2 and faster is where things start to become difficult and the difficulty/cost rises rapidly. 


Up to f/2.8 or so is fairly "easy," depending on the FoV of the lens.  Faster than that and you run into big vignetting issues and in general limit the design even more.

[thxbb12]

Designing a 16-70 f/2.8-4 may be easier, it may not.

 

All right. We are back to square one ;-)

From everything that has been said above, the only conclusion one can draw is that Sony/Zeiss did a bad job at designing their 16-70 lens.