Nikon 80-200 f/2.8 Vs Nikon 70-200 f/2.8 Vr?????

[OreoCookie]

It's not a consumer lens, the D2H was released in 2003, and not long after, many Nikon executives were saying that Nikon was going to go full steam on APS-C.

It's a pro lens, no argument there, but also very popular among enthusiasts.
That doesn't change that most of its customer base was still using film, including a sizeable part of the pro market. The D70 was a prosumer/ambitious amateur camera at the time (I remember, the kit sold for about 170,000 ¥ when I first saw it together with the E-1, the Canon Kiss (that's what the 300D was called there) and the Nikon D2H (aka the photo machine gun)).

Besides, all you need to do for a lens to be very, very, very good on crop sensor is to have a damn good lens on 35 mm film to begin with. Weight and size give no indication that Nikon has attempted to make this lens smaller in favor of DX sensors. It's hard to believe Nikon would try to cripple their lenses intentionally. All Nikon had to do to make this lens work extraordinarily well on APS-C-sized sensors is to make a very good lens that works on 35x34 mm^2.

The sample images that Hmac has posted are nothing out of the ordinary from tests back in the film days, although (from the top of my head) the vignette seems more pronounced than on film (which also makes sense, because sensors are more sensitive to incident light from a large angle).

In fact, if the market hadn't wanted FX so bad, Nikon would probably have left it alone. In 1999, with the introduction of the D1, Nikon said they were committed to replacing film with digital, and while they always left the door open in terms of statements from Nikon Japan and 35mm sized sensors, many of the country execs were totally focused on DX and parroting the line that Nikon thought DX was the way to go (if you dig back, you can actually find 2003-2004 statements from Nikon Japan execs with the DX plan being a 5 year plan at which point they'd maybe do something FF- but in 2003-2006 most Nikon execs, especially Nikon Europe and Nikon USA were saying DX, DX, DX.

Since Nikon is a Japanese company and their products are (to my knowledge) developed in Japan, I guess what counts are the statements from the mother company.
Plus the fact that Nikon has yet to introduce a pro-grade 50-135 zoom clearly points to the direction that Nikon wanted all their lenses to be FX-ready (the only pro lenses that aren't able to fill FF sensors are to my knowledge the 2.8/17-55 and a fisheye; even if I'm missing one or two, point is that essentially all pro lenses that Nikon has released afterwards are FF-compatible).

At best, they were keeping their options open -- but then, they'd have to make sure their lenses work very well on all types of sensors.

As I said before, one of the reasons why the 70-200 does not play as nicely with FX sensors is not because it's 35x24 mm^2, but because it's a sensor and not film -- both of which have different characteristics. In the meantime, people have been spoiled by the sweet spot advantage and seem to have forgotten what things were like in the film days.

 

[Digital Skunk]

Plus the fact that Nikon has yet to introduce a pro-grade 50-135 zoom clearly points to the direction that Nikon wanted all their lenses to be FX-ready (the only pro lenses that aren't able to fill FF sensors are to my knowledge the 2.8/17-55 and a fisheye; even if I'm missing one or two, point is that essentially all pro lenses that Nikon has released afterwards are FF-compatible).

Agreed. And you're right about the DX lens issue. I read that the 12-24 DX performed reasonably well on the current film bodies of the time.

people have been spoiled by the sweet spot advantage and seem to have forgotten what things were like in the film days.

I can attest to that.

 

[Chip NoVaMac]

For those on a budget may want to look at the new Tamron 70-200/2.8 @ $699 in my quick shots seems to offer a good bang for the buck.

 

[compuwar]

As I said before, one of the reasons why the 70-200 does not play as nicely with FX sensors is not because it's 35x24 mm^2, but because it's a sensor and not film -- both of which have different characteristics.

That's still a function of the lens design. With lenses, it's not just the image circle that's important, but the angle of coverage. On Nikon's large format lenses, they're about the most conservative you can possibly be in- stating image circle they seem to actually publish the angle of coverage. Part of Nikon's "answer" to problems with the sensor-flange distance and poor angles of coverage is vignette control on the D3, obviously but make no mistake that the smaller angle of coverage is an artifact of the lens design. That design may be constrained by the fixed flange to focal plane distance, the diameter of the flange, price, the fact it's a zoom, or other trade-offs- but any major vignetting from that small a distance is an artifact of the particular lens design.

The fact that DX sensors don't have the same issues show that it's not the fact that it's a sensor, it's the fact that it's a sensor behind a lens with an AoC that isn't of an optimal size for that sensor. For that matter, scanning backs would have the same issues if it were simply a sensor vs. film one, and they quite obviously don't. But at approximately 3x4" of "effective" size, they fall inside the AoC of any 4x5 lens. I'd happily wager that many LF lenses with a larger AoC would perform just as effectively on a scanning back with an effective 4x5 size though, because their designs work at much larger AoCs, not because they're shooting film or digital.

 

[OldCorpse]

That's still a function of the lens design. With lenses, it's not just the image circle that's important, but the angle of coverage. On Nikon's large format lenses, they're about the most conservative you can possibly be in- stating image circle they seem to actually publish the angle of coverage. Part of Nikon's "answer" to problems with the sensor-flange distance and poor angles of coverage is vignette control on the D3, obviously but make no mistake that the smaller angle of coverage is an artifact of the lens design. That design may be constrained by the fixed flange to focal plane distance, the diameter of the flange, price, the fact it's a zoom, or other trade-offs- but any major vignetting from that small a distance is an artifact of the particular lens design.

The fact that DX sensors don't have the same issues show that it's not the fact that it's a sensor, it's the fact that it's a sensor behind a lens with an AoC that isn't of an optimal size for that sensor. For that matter, scanning backs would have the same issues if it were simply a sensor vs. film one, and they quite obviously don't. But at approximately 3x4" of "effective" size, they fall inside the AoC of any 4x5 lens. I'd happily wager that many LF lenses with a larger AoC would perform just as effectively on a scanning back with an effective 4x5 size though, because their designs work at much larger AoCs, not because they're shooting film or digital.

Finally someone makes sense!

Please learn something about lens design, before blaming "the laws of physics" for the atrocious performance of the 70-200 on FX sensors.

 

[Digital Skunk]

Finally someone makes sense!

Please learn something about lens design, before blaming "the laws of physics" for the atrocious performance of the 70-200 on FX sensors.

The issue is that it's not atrocious, and frankly only a few shooters in the world of photography would say that.

Also, it performs the same way as many other lenses on FX sensors and film. It seems that the few that would say the lens is "atrocious" are ignoring that fact, and either don't shoot a wide range of glass and bodies to know this, or are over analyzing small imperfections that are apparent in a majority of lenses made for sensors.

Dare I say that some shooters aren't operating their equipment correctly. And by reading all of the posts, compuwar is stating the same thing that has already been said: "it's lens design" but some are ignoring the fact that it's a design that is in MANY lenses from both Canon and Nikon.

 

[OreoCookie]

Please learn something about lens design, before blaming "the laws of physics" for the atrocious performance of the 70-200 on FX sensors.

Wow, somebody is on a particularly high horse today

The problem is not the size of the image circle or the mount or that Nikon somehow designed it for DX sensors or anything along these lines. The reason why vignetting, for instance, is more pronounced is due to the `laws of physics' (what other laws would lenses (glass) obey?).

Sensors and film react differently to light that are incident at a larger angle. Yes, a substantially new mount would alleviate that, but then all tele lenses would look more like the Bigma (and I mean the big Bigma).

And yes, all of this is physics. With film, the angle of the light rays that hit the material are not as important, because film has a certain thickness and there are many, many layers of light-sensitive crystals that can be hit.

With CCDs and CMOS sensors, the probability to capture the photons depends much more strongly on the angle (relative to the surface normal): the larger the angle, the smaller the probability the photon is absorbed (that's why you want to rotate solar cells towards the sun if possible). As far as I know, this effect is known as pixel vignetting. Newer sensors have microlenses on them so that the light is bent towards the surface normal and thus more photons register.

Lenses that have been developed for digital sensors have light that is more `parallel' to the surface normal even at the edges of the sensor. Hence lenses that were designed with film in mind will produce more vignetting on digital sensors than on film than the latest generation of lenses.

Pixel vignetting is obviously in addition to the usual vignetting that gets worse with the sensor size, too.

If you look into the Canon camp, all of this is really old news. The last link also compares other artefacts between full frame sensors and APS-C-sized sensors.

 

[Westside guy]

The problem is not the size of the image circle or the mount or that Nikon somehow designed it for DX sensors or anything along these lines. The reason why vignetting, for instance, is more pronounced is due to the `laws of physics' (what other laws would lenses (glass) obey?).

Sensors and film react differently to light that are incident at a larger angle. Yes, a substantially new mount would alleviate that, but then all tele lenses would look more like the Bigma (and I mean the big Bigma).

When it comes to vignetting, I don't think the mount is the issue so much as the relative narrowness of that 8.5" lens overall. Maybe I'm misinterpreting what you're trying to say; but if it were due to the angle at which the light is hitting the sensor at the corners, rather than simply the result of less light overall hitting the corner (physical vignetting in other words), I'd expect the vignetting to be somewhat more linear.

Of course this could be easily resolved by someone willing to go put this lens on a Nikon film camera, shoot some frames, and pay for the developing.

 

[OreoCookie]

When it comes to vignetting, I don't think the mount is the issue so much as the relative narrowness of that 8.5" lens overall.

Well, no, having a bigger mount just allows you to put in bigger lenses, so the full frame sensor would be in the sweet spot of a lens made for some imaginary larger format.

Maybe I'm misinterpreting what you're trying to say; but if it were due to the angle at which the light is hitting the sensor at the corners, rather than simply the result of less light overall hitting the corner (physical vignetting in other words), I'd expect the vignetting to be somewhat more linear.

I cannot give you a simple angular dependence, but it's definitely not a linear effect. Neither one of them is linear. The absorption probability depends on the intensity on the surface (it depends on the cos of the angle). I can give you a more detailed physical explanation (although I'm not an expert in this field), but

 

[Westside guy]

I cannot give you a simple angular dependence, but it's definitely not a linear effect. Neither one of them is linear. The absorption probability depends on the intensity on the surface (it depends on the cos of the angle). I can give you a more detailed physical explanation (although I'm not an expert in this field), but

Feel free - I have a degree in physics.

 

[Digital Skunk]

Of course this could be easily resolved by someone willing to go put this lens on a Nikon film camera, shoot some frames, and pay for the developing.

Oh darn it! I just sold my N80 now that film was officially dead...

 

[ChrisA]

For those on a budget may want to look at the new Tamron 70-200/2.8 @ $699 in my quick shots seems to offer a good bang for the buck.

You can buy a good used Nikon 80-200 for less then $699. Depending on which version, the Nikon lens sells for between $400 and $650. Of course I'm comparing used vs. new prices here. My opinion is that it hardly matters. I have amy lenses I bought new and some used. They are all used now and It's hard for me to remember how I got each lens

 

[OreoCookie]

Feel free - I have a degree in physics.

A fellow physicist
As you know, electromagnetic waves are purely transversal, the E and B field oscillate perpendicular to the direction of propagation. If the angle (always relative to the surface normal) is 0, the plane of oscillation coincides with the surface of the sensor and the probability that the photon is absorbed is maximal.

If the electromagnetic wave does not hit the sensor head on, then the surface feels only a fraction of the effect, namely the projection of the electromagnetic field onto the surface. If you draw a little diagram and think about it a little, you find that the component of the E field (and B field) that is acting on the surface has to be multiplied with cos alpha where alpha is the angle relative to the surface normal. If alpha = 0, cos alpha = 1 and the maximum effect is felt. The intensity is proportional to the absolute value square of E and B, so the intensity of the surface scales as cos^2 alpha = 1/2 ( 1 + cos 2 alpha ). If for simplicity, you assume that the probability that you absorb a photon is proportional (linear) to the intensity on the surface, you see that it would scale as cos^2 alpha as well.

You can expect that the probability that a photon is absorbed is not just proportional to the intensity on the surface, but also depends on the angle relative to surface normal and whether the photon hits the sensor pixel `diagonally' (e. g. on one of the corners) or `straight' (e. g. on the middle of the left side of the chip).

Film is a bit different: the part of the electromagnetic wave that is not reflected enters the film substrate and has not just one possibility to `hit' a suitable molecule, but many times as the film is thick. With sensors, the active layer is much, much thinner.

If you have a lens, then alpha increases towards the corners (the light rays need to be bent to reach them).

 

[Hmac]

Photography sensors (unlike film), however, typically use a series of microlenses at the periphery to try to account for the increased angle of incidence, no? So theoretically, a properly designed sensor should be able to accomodate the increased angle of incidence no matter what it is? Yes, of course there are limits, but can't those limits be extended with a bigger image circle? The argument has always been that the smaller-than-Canon diameter of the F-mount shouldn't matter relative to full-frame because of that microlens system (of course, Canon folk say that that won't work and that Nikon's F-mount is irretrievably flawed because it's too small for proper full-frame coverage).

Does that mean, then, that Nikon's D3 sensor needs a better system of microlenses to accomodate the higher-than-DX angle of incidence, or does it mean that the image circle from some of their camera lenses is just too small? I see that Nikon is moving toward some lens re-designs, presumably to address the latter.

 

[Abstract]

With CCDs and CMOS sensors, the probability to capture the photons depends much more strongly on the angle (relative to the surface normal): the larger the angle, the smaller the probability the photon is absorbed (that's why you want to rotate solar cells towards the sun if possible).

I don't think that's true, particularly for CMOS. I'm holding a book right now called Handbook of Radiation Effects, and from my experience with x-rays and MOSFETs, you get greater radiation absorption when you hit it at an angle, not less. Just take a ruler and measure height of a television, and do the same thing by measuring the diagonal. We all know the diagonal is longer, and so photons that hit the CMOS at an angle will have more "stuff" to absorb it. Actually, I'm not 100% sure because I really don't know about light photons hitting a sensor. I assumed it was the same as x-rays, despite the fact that x-rays tend to pass through material while light doesn't really do that.

Oh, I'm getting a PhD in the subject of "Blasting MOSFETs with X-rays While Placed In a Man's Ass". It doesn't mean I know anything about photography, though. I don't.

 

[Digital Skunk]

This thread just got serious....!

We never learned this much in art/media/journalism school.

I thought Video Directing was hard.

 

[OreoCookie]

Photography sensors (unlike film), however, typically use a series of microlenses at the periphery to try to account for the increased angle of incidence, no?

Yes:

Newer sensors have microlenses on them so that the light is bent towards the surface normal and thus more photons register.

So theoretically, a properly designed sensor should be able to accomodate the increased angle of incidence no matter what it is?

No, because you have yet another surface and at every surface you have reflections, especially if the incident angle is getting larger.

Yes, of course there are limits, but can't those limits be extended with a bigger image circle?

No, what counts for this type of vignetting is the angle at which the light hits the corners of the sensor.

The argument has always been that the smaller-than-Canon diameter of the F-mount shouldn't matter relative to full-frame because of that microlens system (of course, Canon folk say that that won't work and that Nikon's F-mount is irretrievably flawed because it's too small for proper full-frame coverage).

If you follow the two links I've provided, you see that Canon folks have been discussing this for a long time now (because they have FF digital bodies for a long time (I'm forgetting the Kodak 14n here!)).

Does that mean, then, that Nikon's D3 sensor needs a better system of microlenses to accomodate the higher-than-DX angle of incidence, or does it mean that the image circle from some of their camera lenses is just too small?

No, that's just one way to compensate. I don't think micro lenses can fully compensate these effects.

I see that Nikon is moving toward some lens re-designs, presumably to address the latter.

Yes, this is the most effective way to deal with the problem. The problems are particularly noticeably with (i) FF sensors and (ii) lenses that have been designed for film. If Nikon redesigns the 70-200 with sensors in mind, I'm sure we'll see a killer lens. On the other hand, it'll be so expensive, I'll have to keep my 80-200

Fortunately, vignetting is rather harmless, because it's easy to correct, but loss of sharpness is more serious.

 

[OreoCookie]

I don't think that's true, particularly for CMOS. I'm holding a book right now called Handbook of Radiation Effects, and from my experience with x-rays and MOSFETs, you get greater radiation absorption when you hit it at an angle, not less.

The absorption properties of xrays and visible light are totally different. You will not be able to infer from one to the other.

Just take a ruler and measure height of a television, and do the same thing by measuring the diagonal. We all know the diagonal is longer, and so photons that hit the CMOS at an angle will have more "stuff" to absorb it.

Yes, you have more pixels in the way in the diagonal, but they are just a bunch of different pixels. If you focus on one single pixels, one of the most important parameters is the angle relative to the surface normal. Reality is usually more complicated than a simple model, though, but simple models help us understand reality.

 

[Hmac]

Yes, of course there are limits, but can't those limits be extended with a bigger image circle?

No, what counts for this type of vignetting is the angle at which the light hits the corners of the sensor.

But, a bigger image circle means decreased angle of incidence of the light at the edges of the sensor, doesn't it?

If you follow the two links I've provided, you see that Canon folks have been discussing this for a long time now (because they have FF digital bodies for a long time (I'm forgetting the Kodak 14n here!)).

Yeh, I've been reading that stuff from them for years.

 

[OreoCookie]

But, a bigger image circle means decreased angle of incidence of the light at the edges of the sensor, doesn't it?

No, unless you change the camera mount as well.

If the lens is constructed for a larger image circle, then a smaller sensor sits in the sweet spot of the lens which will reduce regular vignetting, loss of sharpness but does not address sensor-specific issues. But it will not necessarily alter the incident angle at the corners (unless the lens is built for digital sensors). Think of the Hasselblad XPan, for instance. This camera uses medium format lenses to cover panorama shots on 35 mm. Here, the larger image circle of the lens does not help with a fictitious full-frame digital sensor compared to film other than that a superb medium format lens will be even better if you restrict yourself to 35x24 mm^2.

That's the sweet spot advantage DX sensors have, repeated with a lens designed for an even bigger image circle. Since the 70-200 has been designed with film in mind, the image circle is `large enough' and certainly `not to blame'.

An indication that Nikon has not reduced the image circle is weight: lenses with smaller image circles are lighter, much lighter. The equivalent of a 70-200 on crop sensors, e. g. Tokina's 50-135, weighs about half, although it's built like a tank. If you compare the 70-200's weight to its predecessors, you will find out that it hasn't changed much, it's still about 1.5 kg. If Nikon had focussed on DX sensors, the weight would very likely be reduced (smaller lenses, smaller tubes, less weight in total).

 

[Digital Skunk]

Yes, this is the most effective way to deal with the problem. The problems are particularly noticeably with (i) FF sensors and (ii) lenses that have been designed for film. If Nikon redesigns the 70-200 with sensors in mind, I'm sure we'll see a killer lens. On the other hand, it'll be so expensive, I'll have to keep my 80-200

Fortunately, vignetting is rather harmless, because it's easy to correct, but loss of sharpness is more serious.

Now this is the post and the person that is making sense. Everyone has, but this is as clear as it can get without sounding as redundant as I have.

When Nikon introduced the 12-24 and 24-70 f/2.8 I knew they would be about the cost of a D300 ($1799) each. The PC Shift was way over my $2200 expected price, it hit $2799 I think. The only new lens to stay at the same price was the 60 mm Macro which just added AF-S and now states that it's made for sharpness across the frame with FX, like the other new lenses.

 

[compuwar]

Yes, this is the most effective way to deal with the problem. The problems are particularly noticeably with (i) FF sensors and (ii) lenses that have been designed for film. If Nikon redesigns the 70-200 with sensors in mind, I'm sure we'll see a killer lens. On the other hand, it'll be so expensive, I'll have to keep my 80-200

This seems to contradict your earlier statement that it's not the design of this particular lens.

 

[compuwar]

An indication that Nikon has not reduced the image circle is weight: lenses with smaller image circles are lighter, much lighter.

Doesn't this depend heavily on the lens design, and where the nodal point is? I've got 300mm LF lenses that will cover 8x10 that are a lot smaller than even my "little" 300/4 for 35mm both in terms of elements, lens mount, weight and overall length. We're talking orders of magnitude difference in image circle size- the LF lens will cover the entire 35mm camera body with room left over.

 

[OreoCookie]

This seems to contradict your earlier statement that it's not the design of this particular lens.

No, not really.
Initially, I've said that it's no surprise that the 70-200 performs worse in some respects on full frame sensors, but that this has nothing to do with this particular lens. Even though newer lenses would fare better (because they take the peculiarities of sensors better into account), a large part of the effect can be attributed to size effects. Vignetting on film has been a lot more pronounced than on APS-C-sized sensors and IMO people have been spoiled by the sweet spot advantage -- which has nothing to do with the lens you use.

If you really go down to detail, things are more complicated, because there are ways to improve IQ on dslrs that take the sensor into account. Also, even on film, newer lenses (usually) have better IQ than their predecessors.

 

[Westside guy]

Here's the problem as I see it, though. If the vignetting problem is based on the angle of incidence of the light hitting the sensor's edge photosites, it would seem to me that a) all full-frame lenses that cover a focal length of 200mm should display identical behavior at 200mm, since the angle of incidence at the sensor's edges should be the same; and b) shouldn't the problem get worse as focal length gets shorter, since the angle of incidence gets more extreme with shorter focal length? As far as I know, this problem is mainly manifesting itself with the 70-200 f/2.8 VR lens, and not, say, the 200mm f/2 - and certainly not the 50mm f/1.8 (as far as I've heard).

Also, if the actual F-mount design were the problem - which I think you alluded to in earlier posts - wouldn't that also be a bigger issue with shorter focal length lenses, such as the 50mm?

When a problem is restricted to just one particular lens, it seems more realistic to suspect it's the lenses design. And when that lens is (according to other people, since I don't own it) noticeably narrower than its predecessors, it just shouts "physical vignetting" to me.

Since I don't own a D3, I'll admit I might not be aware of problem reports regarding other lenses on this camera. But if it were a widespread problem, I'd think we'd be hearing about it from all corners.