Using teleconverters

[EricV]

Here is another way to think about DoF with teleconverters.  Suppose your camera sensor was perfect, with infinitely small pixels.  Suppose also that your lens was perfect, limited only by diffraction.  Then there would be no reason to use a TC at all -- you could achieve exactly the same result by cropping the image from the original lens.  This includes all DoF considerations.

[Ray]

EDIT: The PF lens may well be the equal of the Leica 280mm f/4........... 

Surely not! Didn't you claim earlier that the Leica 280/F4 was close to being a perfect lens? 

The issue I was bringing up is that Ray hasn't used the 280/4, nor the 300mm PF so it seems unlikely he's used a really good lens with a really good TC so IMHO his blanket statement that a lens plus TC will produce mediocre image quality is likely based on experience with lenses that aren't as good.

I think you've misunderstood my blanket statement. There are a limited number of common words to describe general lens performance. Words and phrases such as excellent, very good, good, quite good, average, mediocre, poor, very poor and so on, are approximate and imprecise. It's the relativity I was trying to get across. In other words, my blanket statement that a good lens becomes a mediocre lens of longer focal length, when a teleconverter is attached, was merely a way of describing the fact that whatever the quality of the lens used with a teleconverter, the longer focal length that results will effectively be a reduced quality lens of longer focal length compared with the quality of the shorter lens, when each lens is used within its focal length range. In other words, the effectively longer lens will have a reduced MTF response, compared with the actual, shorter lens.

Now, if this is not true, then I would be one of the first to be overjoyed at such news. Who wants to carry around more weight and more lenses than they need, in order to achieve their desired image quality?? 

Whenever I've investigated this issue on the internet, the consensus of opinion seems to be, to quote just one example, that the 70-200/F2.8 lenses, whether of the Canon or Nikon variety, when used with the latest 2x converter from Canon or Nikon, do not produce as good an image quality as the latest Canon 100-400, or Nikon 80-400 zooms used at 400mm.

If this situation has now changed with recent technological advances, and teleconverters for use with the 70-200/F2.8 have become so good that image quality now exceeds, or even equals, that from the latest Nikon AF-S 80-400 G, then I have made a big blunder in buying the new Nikkor 80-400. I would have preferred to have spent a bit more for the 70-200/F2.8 plus 2x converter. 

But that's not the only issue to consider. The point has been made that a really good lens with a good converter will produce better results. (What a surprise! Who would have thought that!    )

The point I would make in responses is that a really good lens will also produce better results without a converter. Before I use any converter on a regular basis, I would want to know how much sharper or more detailed the image which has been enlarged by the converter is, compared with the crop of the same scene without the converter, after interpolation and sharpening.

I don't dispute that the image from the converter should be at least marginally sharper and more detailed, under ideal conditions that favour the use of tripod and the same ISO setting; but marginally sharper is not good enough for me because most of my shots are not taken under ideal conditions. A one-stop disadvantage in either shutter speed or noise, using a 1.4x converter, will likely obliterate any marginal advantage seen under ideal conditions.

[Ray]

Here is another way to think about DoF with teleconverters.  Suppose your camera sensor was perfect, with infinitely small pixels.  Suppose also that your lens was perfect, limited only by diffraction.  Then there would be no reason to use a TC at all -- you could achieve exactly the same result by cropping the image from the original lens.  This includes all DoF considerations.

I agree with that analogy completely, Eric.

[dwswager]

Whenever I've investigated this issue on the internet, the consensus of opinion seems to be, to quote just one example, that the 70-200/F2.8 lenses, whether of the Canon or Nikon variety, when used with the latest 2x converter from Canon or Nikon, do not produce as good an image quality as the latest Canon 100-400, or Nikon 80-400 zooms used at 400mm.

At least with the Nikkors, that is correct.  The 70-200mm f/2.8 with 1.4x TC-14EIII is the equivalent at equiavlant focal lengths, but with the 2x TC-20EII it is not at 400mm.

If this situation has now changed with recent technological advances, and teleconverters for use with the 70-200/F2.8 have become so good that image quality now exceeds, or even equals, that from the latest Nikon AF-S 80-400 G, then I have made a big blunder in buying the new Nikkor 80-400. I would have preferred to have spent a bit more for the 70-200/F2.8 plus 2x converter. 

I waited 5 years for Nikon to update the 80-400mm to AFS and change it from precise focus to fast focus.  I ended up selling the spectacular 80-200mm f/2.8D and buying the 70-200mm f/2.8G VRII and TC-14EIII.  But there are advantages and disadvantages to both setups.  From 70-200mm, the 70-200mm f/2.8 is much better than the 80-400mm.   And to 280mm it is the equivalent.  Only getting to 400mm does the 80-400mm outperform.   (I have no experience with the 1.7x TC-17EII).

[Ray]

I ended up selling the spectacular 80-200mm f/2.8D and buying the 70-200mm f/2.8G VRII and TC-14EIII.  But there are advantages and disadvantages to both setups.  From 70-200mm, the 70-200mm f/2.8 is much better than the 80-400mm.   And to 280mm it is the equivalent.  Only getting to 400mm does the 80-400mm outperform.   (I have no experience with the 1.7x TC-17EII).

Thanks. That is as I imagined. It goes without saying that the 70-200/F2.8 (without converter) would outperform the 80-400 between 80 and 200mm. That to 280mm it is the equivalent, needs clarifying.

I can imagine that the 70-200/F2.8 used at 200mm with 1.4x converter might equal the 80-400 used at 280mm, although it's not clear if this is only under ideal conditions where the light and shutter speed disadvantages of the 80-400 lens don't apply.

But what about other focal lengths between 200 and 280? For example, to achieve a focal length of 240mm with the 70-200, with 1.4x converter, one would use the lens at 172mm, approximately. Surely it would be better to remove the converter, take the shot at 200mm, then crop slightly to get the equivalent 240mm focal length.

[telyt]

Surely not! Didn't you claim earlier that the Leica 280/F4 was close to being a perfect lens?  

Just trying to eliminate brand bias by hypothetically supposing the 300 PF was the equal of the 280/4 APO.  

I think you've misunderstood my blanket statement. There are a limited number of common words to describe general lens performance. Words and phrases such as excellent, very good, good, quite good, average, mediocre, poor, very poor and so on, are approximate and imprecise. It's the relativity I was trying to get across. In other words, my blanket statement that a good lens becomes a mediocre lens of longer focal length, when a teleconverter is attached, was merely a way of describing the fact that whatever the quality of the lens used with a teleconverter, the longer focal length that results will effectively be a reduced quality lens of longer focal length compared with the quality of the shorter lens, when each lens is used within its focal length range. In other words, the effectively longer lens will have a reduced MTF response, compared with the actual, shorter lens.

OK my apologies for misunderstanding.

Now, if this is not true, then I would be one of the first to be overjoyed at such news. Who wants to carry around more weight and more lenses than they need, in order to achieve their desired image quality??  

Exactly why I carry the 280/4 and 1.4x extender.  I sold my 400mm lens.

Whenever I've investigated this issue on the internet, the consensus of opinion seems to be, to quote just one example, that the 70-200/F2.8 lenses, whether of the Canon or Nikon variety, when used with the latest 2x converter from Canon or Nikon, do not produce as good an image quality as the latest Canon 100-400, or Nikon 80-400 zooms used at 400mm.

If you're limiting yourself to CaNikon…

Before I use any converter on a regular basis, I would want to know how much sharper or more detailed the image which has been enlarged by the converter is, compared with the crop of the same scene without the converter, after interpolation and sharpening.

I think this is something you'll have to try for yourself with the specific combinations of lens and extender you have in mind.  Results that may be completely acceptable to someone else may be unacceptable to you or vice versa… and there's more to it than measurebating.

[Ray]

I think this is something you'll have to try for yourself with the specific combinations of lens and extender you have in mind.  Results that may be completely acceptable to someone else may be unacceptable to you or vice versa… and there's more to it than measurebating.

Of course Doug, I have tested this for myself, but not comparing high quality primes which match when one of them has an extender attached. My tests have always compared images from the same lens, with and without converter, and I've always found that the converters (an older Canon 1.4x and a newer Nikon 1.4x) provided slightly improved detail. The problem is, the improvement has been too marginal and tends to be cancelled by the f/stop disadvantage, and sometimes a noticeable resolution fall-off at the edges, which one doesn't get when cropping the image from the shorter lens.

Exactly why I carry the 280/4 and 1.4x extender. I sold my 400mm lens.

But haven't you deprived yourself of a 560/F5.6 by selling that lens? As a 'birder' surely you would find a 560mm lens very useful; or was that 400mm lens not really good enough for use with an extender? 

[BJL]

Here is another way to think about DoF with teleconverters.  Suppose your camera sensor was perfect, with infinitely small pixels.  Suppose also that your lens was perfect, limited only by diffraction.  Then there would be no reason to use a TC at all -- you could achieve exactly the same result by cropping the image from the original lens.  This includes all DoF considerations.

On one hand I agree, and will even go a bit further: a TC can be discarded in favor of cropping to the same FOV (or using a different body with a smaller sensor) with essentially the same results including DOF, _if_ that crop:
1) delivers enough resolution
2) delivers enough dynamic range: at the lowest shutter speed allowed by consideration of camera and subject motion, the photo-sites have enough highlight headroom; that is, protecting highlights does not force use of a higher exposure index ["ISO speed"] and thus gathering less light from the subject.

On the other hand,
1) Resolution is likely to be higher with a TC, unless it is so bad that it degrades absolute resolution in "lines per mm at the sensor" by more than the TC magnification factor
2) Using all the sensor area via the TC potentially improves dynamic range when abundant light can be gathered, by gathering and counting more photons by using all the photo-sites instead of only one half (1.4x TC) or one quarter (2x TC) of the sensor area.

[dwswager]

But what about other focal lengths between 200 and 280? For example, to achieve a focal length of 240mm with the 70-200, with 1.4x converter, one would use the lens at 172mm, approximately. Surely it would be better to remove the converter, take the shot at 200mm, then crop slightly to get the equivalent 240mm focal length.

This all depends on what and how you are shooting.  Of course, the base lens is better from 98-200mm.  But if I'm shooting soccer outside, then I trade that for the 98-280mm focal range.  Indoor Futsol soccer, I don't need use the TC and keep the better optical performance and 1 stop of light.  Shooting baseball and softball I switch back and forth depending on what I'm shooting and from where.  Typically shooting batters, I'm base lens.  Shooting outfielders, I might even use the 2x TC. 

Bottom line is that every decision is a case of making trades.  One of the reasons I bought the D810 over the D750 is because shooting at the 1.2x and 1.5x crops still leaves a decent size file.  I traded $1000, in part, to get that. 

The whole DOF issue needs some clarification.  The reason to use a TC is because you need more reach.  The reason you need more reach is because you can't get any closer.  If you could shoot from a closer position, then that is what you should do.  Hence, all DOF questions should be answered with the camera subject distance fixed at that common spot for all lens/aperture combinations.

The next question is are you better off shooting with the TC or shooting at a smaller focal length and cropping digitally.  The answer will vary depending on your equipment, shooting style and output requirements. 

[jjj]

Bottom line is that every decision is a case of making trades. 

Indeed. Plus, what one thinks worthwhile to trade is a very individual choice. One which also depends, as you nicely illustrated on the individual shooting circumstances.

[telyt]

My tests have always compared images from the same lens, with and without converter, and I've always found that the converters (an older Canon 1.4x and a newer Nikon 1.4x) provided slightly improved detail. The problem is, the improvement has been too marginal and tends to be cancelled by the f/stop disadvantage, and sometimes a noticeable resolution fall-off at the edges, which one doesn't get when cropping the image from the shorter lens.

If you're limiting yourself to CaNikon…

But haven't you deprived yourself of a 560/F5.6 by selling that lens? As a 'birder' surely you would find a 560mm lens very useful; or was that 400mm lens not really good enough for use with an extender?  

280mm APO + 2x APO = 560mm APO

I prefer to get closer to my subjects so 560mm isn't a high priority for me.

[Ray]

2) Using all the sensor area via the TC potentially improves dynamic range when abundant light can be gathered, by gathering and counting more photons by using all the photo-sites instead of only one half (1.4x TC) or one quarter (2x TC) of the sensor area.

BJL,
I think this point needs clarifying. I might have given the impression that the image from a lens with converter always has the disadvantage of one stop higher noise and 1 stop lower DR, compared with the cropped and interpolated image from the lens without converter.

In fact, it seems reasonable to suppose that the disadvantage in this respect is at least partially, if not completely counteracted by the utilisation of the entire sensor to record the enlarged image, provided one uses the same shutter speed for the images being compared.
In other words, one loses one stop of SNR and DR only if one thinks it appropriate to use double the shutter speed as a result of the effectively longer focal length. This shouldn't apply when using a tripod to shoot static subjects, or when the light is so bright that the shutter speed is faster than it need be at base ISO, in conjunction with a fairly wide aperture.

However, the situation is different when comparing two separate lenses, as in Doug's example of the Leica 280/F4 with 1.4x extender, compared with the Leica 400/F4. He claims that the 280 with extender is slightly sharper than the 400/F4 at F5.6.
In both cases, the full sensor is used to record the images. The disadvantage is that Doug has deprived himself of the benefits of the 400/F4 used at full aperture, which will allow either a lowering of ISO, and the corresponding increase in SNR and DR, or the use of a faster shutter speed without any SNR disadvantage, and possibly a sharper image to boot. 

[telyt]

However, the situation is different when comparing two separate lenses, as in Doug's example of the Leica 280/F4 with 1.4x extender, compared with the Leica 400/F4. He claims that the 280 with extender is slightly sharper than the 400/F4 at F5.6.
In both cases, the full sensor is used to record the images. The disadvantage is that Doug has deprived himself of the benefits of the 400/F4 used at full aperture, which will allow either a lowering of ISO, and the corresponding increase in SNR and DR, or the use of a faster shutter speed without any SNR disadvantage, and possibly a sharper image to boot.  

Other features of the 400mm f/4 APO I'm missing out on include much greater weight, the need to use a tripod (vs. the shoulder stock/monopod rig I often use) and the impossibly shallow DOF at f/4

Ray I think you can resolve many of your questions by just going outside (or wherever you typically make photographs) and take some pictures.

[Ray]

Other features of the 400mm f/4 APO I'm missing out on include much greater weight, the need to use a tripod (vs. the shoulder stock/monopod rig I often use) and the impossibly shallow DOF at f/4

Ah! I see! Then all your problems are solved. The new Canon 100-400 USM L IS II is about the same weight as your Leica 280/F4 but potentially much more versatile and useful. When this lens is used with the new 50mp, 5DS (or 5DSR), I would expect the results to be stunning.

You might turn up your nose and think that no Canon lens could match the quality of a Leica prime, especially a Canon zoom, but you should take into consideration the following 3 major factors; (1) Your Leica prime is a lower quality lens when extender is attached, (2) The new Canon 100-400 is much improved, according to reviews that I've read, (3) The impressive 50mp of the new Can 5DS(R) will effectively raise the quality of any lens attached to it.

I'm seriously thinking of buying this combination myself, so I've provided in the attached image the results of my research on Photozone, comparing the new Canon zoom with other Canon primes at particular focal lengths within the 100-400 range, ie. 100mm prime, 200mm, 300mm and 400mm prime.

All the Photozone tests of these lenses have used the Canon 5D Mk2, so the comparison should be relevant. None of the primes can match the overall sharpness of the new zoom at the same focal length. Wow! That's the lens for me! 

Ray I think you can resolve many of your questions by just going outside (or wherever you typically make photographs) and take some pictures.

Don't you worry about that, Doug. I take huge quantities of photos every year; literally hundreds of thousands in total since the time I got my first DSLR in 2003. Organising them is now a major chore.

[BJL]

When comparing the use of a TC to the alternative of cropping to the same final FOV,  the shutter speed needed to control image blurring due to motion of either subject or camera is the same, when judged by final images displayed at the same size, etc.  The "1/f" guideline goes out the window when one is going to crop and thus enlarge more, just as it does when one changes to a smaller format (which is what cropping is, effectively).  So when comparing use of a 1.4x TC to a 1.4x crop, the f-stop change with the TC dictates a _doubling_ of exposure index (not quadrupling).

Except that in the (perhaps rare) situation when the TC option can get adequate shutter speed at the sensor's base-ISO speed setting; then the non-TC option is stuck with that same setting, and so avoiding blown highlights potentially requires halving the exposure time; that is the case where less light is gathered, and so shadow handling could be worse without the TC than with.  (Equal per pixel DR with fewer pixels means worse overall IQ and DR, as has been often discussed!)

[dwswager]

Except that in the (perhaps rare) situation when the TC option can get adequate shutter speed at the sensor's base-ISO speed setting; then the non-TC option is stuck with that same setting, and so avoiding blown highlights potentially requires halving the exposure time; that is the case where less light is gathered, and so shadow handling could be worse without the TC than with.  (Equal per pixel DR with fewer pixels means worse overall IQ and DR, as has been often discussed!)

Cameras have particular performance ranges in different areas.  Not all shooting situations require the use of the entire range.  If I come off base ISO and lose some DR, the question to be answered is "Does it matter"?  If it doesn't matter, no harm, no foul.  If it does, then you need to decide is it worth trading away some DR to get something else.  If I am intending to shoot at f/4 anyway for DOF reasons, then TC or no TC I'm at the same exposure, but I give up a little quality with the TC if I need the focal length.

This is rather funny.  My youngest plays high school and club soccer.  Her club team is very good, but has problems scoring.  They pass up too many opportunities.  We are always trying to get them to understand that "You can't score if you don't shoot.  You don't want to waste opportunities taking bad shots, but the good shot you take is ALWAYS better than the great shot you didn't"

It is the same with photography.  You want to get the shot with the best possible quality you can by choosing between all the available, viable, options.  But the good shot you get is ALWAYS better than the great shot you didn't!

[telyt]

You might turn up your nose and think that no Canon lens could match the quality of a Leica prime

Don't presume to know what I think.

Seriously Ray, you're over-thinking this whole thing.  Photography isn't about measurebating.  Go take pictures.

[BJL]

... We are always trying to get them to understand that "You can't score if you don't shoot.  You don't want to waste opportunities taking bad shots, but the good shot you take is ALWAYS better than the great shot you didn't"

It is the same with photography.  You want to get the shot with the best possible quality you can by choosing between all the available, viable, options.  But the good shot you get is ALWAYS better than the great shot you didn't!

I totally agree (also with the bits I edited out); I was pedantically talking about a probably rare edge case, and in fact I am a big advocate of using "excessive pixel counts" to eliminate the need for TC's.

With film and processing costs eliminated, I long ago embraced the policy of "if the subject is interesting, take your best shot and worry later about how good it is", especially with long telephoto wildlife shots where the opportunity is fleeting.  For a worst-case example, I have some shots of birds with a 200mm lens where I would have needed over 400mm to frame it as desired, and on a previous camera with a mere 5MP sensor.  So I did a massive crop to little over 1MP, and still got something that is a nice record of what I saw, and worth sharing with friends and family, even if it will not make me rich or famous.

[Ray]

When comparing the use of a TC to the alternative of cropping to the same final FOV,  the shutter speed needed to control image blurring due to motion of either subject or camera is the same, when judged by final images displayed at the same size, etc.  The "1/f" guideline goes out the window when one is going to crop and thus enlarge more, just as it does when one changes to a smaller format (which is what cropping is, effectively).  So when comparing use of a 1.4x TC to a 1.4x crop, the f-stop change with the TC dictates a _doubling_ of exposure index (not quadrupling).

Now this is getting interesting. You've clarified something I questioned back in post #5 where I wrote:

"However, if one is comparing the technical quality of images taken with and without converter, the image without converter is cropped and enlarged through a different process of interpolation, but enlarged nevertheless.

In order to achieve the maximum 'freezing of movement' in the image without converter, I would speculate that one should use the same shutter speed that is appropriate for the image produced with the converter, but I'm not certain about this."

I could see the logic here, but wasn't sure if there were other factors involved which might create the requirement for a faster shutter speed when using the converter.

If it really is true that there is no requirement for a faster shutter speed when using a converter, then this fact alone could provide an explanation as to why some people find that a converter provides a useful improvement in image quality, for same size prints.

There's a natural tendency to use a faster shutter speed with the longer focal length, whether the lens is  longer in 'equivalence' terms or longer by design.  If it's the case that many people when using a converter  automatically use a faster shutter speed, then it's quite likely if they occasionally decide to make a comparison of the same scene shot without a converter, the image shot without the converter might have had a suboptimal shutter speed, whereas the image shot with the converter might have been optimal, due to a doubling of shutter speed.

So far, according to the points you have raised or agreed with, BJL. it's looking very favourable for the converter.
(1) No loss or change in DoF despite the drop in f/stop number.
(2) No requirement for a faster shutter speed than what you would use without a converter, for a given size image/print of same FoV.
(3) No increase in noise, or reduction in DR, as a result of increasing ISO to accommodate the drop in f/stop, because the larger area of the recorded image offsets such increase in noise.

Wow! Everyone should get a teleconverter. These are amazing devices. 

However, (sorry to introduce a bit of negativity), I'm still not sure about point #3. The logic is not holding up.

For example, let's consider the effects of a 1.4x converter. The converter doesn't add any light. It enlarges the image with a corresponding reduction in light per pixel. Each pixel receives just half of the number of photons it would have received without converter, comparing images of the same FoV.

Let's suppose that the image information reaching a certain number of pixels in the cropped image without converter, is almost drowned by system noise and read-out noise, but not quite. By raising  the deepest shadows in processing, one can still discern some detail. albeit degraded detail which requires significant noise reduction.

Now imagine what happens if those very low level signals are cut in half. They might be literally drowned in system noise, thermal noise, read-out noise etc. They are below the threshold. I can't see how the use of 2 pixels in place of one pixel can alter that fact. Correct me if I'm wrong. 

[spidermike]

For clarity I'm gong to break this down to the different phrases

For example, let's consider the effects of a 1.4x converter. The converter doesn't add any light.

Nope

It enlarges the image with a corresponding reduction in light per pixel.

Reduction in light intensity, yes

Each pixel receives just half of the number of photons it would have received without converter,

Only if you force the camera to use the same shutter speed. If you are using auto-metering the shutter speed lengthens to accommodate the reduced aperture

comparing images of the same FoV.

This is where it could get interesting anmd is something I had not considered before: if you crop an image and blow it up, you are taking a fixed amount of light information and spreqding it over a larger area. Is this analogous to the teleconverter taking a smaller amount of light from the 'lens output' and spreading it over a larger area? It makes sense but I hardly see my images fading on cropping but maybe the computer program accounts for it? I guess one way to look at it is how did it used to work in the days of film processing - I never processed my own stuff so don't know.
vs teleconverter then when you crop and blow that crop to the same size as the teleconverter image you are also desaturing the amount of light available?






Let's suppose that the image information reaching a certain number of pixels in the cropped image without converter, is almost drowned by system noise and read-out noise, but not quite. By raising  the deepest shadows in processing, one can still discern some detail. albeit degraded detail which requires significant noise reduction.

Now imagine what happens if those very low level signals are cut in half. They might be literally drowned in system noise, thermal noise, read-out noise etc. They are below the threshold. I can't see how the use of 2 pixels in place of one pixel can alter that fact. Correct me if I'm wrong. 
[/quote]