teleconverters and resolving power

[Don Haines]

What does a teleconverter do for you if you wish to resolve fine details on a distant object?

The point that I am trying to make is that unless you have a really sharp lens, you are better without teleconverters.

My past experience is that using a crop body, if you have an ultra-sharp lens (like the series 2 big whites) a teleconverter will improve your resolving power, if you have a sharp lens (quality of 70-200F4) it is neutral, and if you have a soft lens it will hurt you.

The test is to capture photos of a small bird from 25 feet away. Since I can not get a bird to pose for me as I change lenses, I substituted a bird sized stuffed animal. All shots were cropped to the same portion of the bird's head and all images resampled to 500 pixels wide. unfortunately, I do not have a "big white" at home so that part of the test is left out... so the sharp lens used is a 70-200F4 and the soft lens is a Sigma 120-400F5.6. To make things interesting, a SX50 was thrown in as well.

The first pair of pictures is the SX50 at 50X zoom and then at 200X zoom (digital zoom enabled).
The second trio of pictures is of the 70-200 at 200mm, with a 1.4X teleconverter, and then with a 2X teleconverter. The third trio of pictures is of the 120-400 at 400mm, with a 1.4X teleconverter, and then with a 2X teleconverter.

In the first set, we find the image with digital zoom turned on is better than without. This came as a surprise to me as I had always assumed that digital zoom was an evil to be avoided......

In the second set, we see that the addition of the 1.4X teleconverter makes the image slightly better and that the 2X teleconverter makes things worse.

In the third set the teleconverters just degrade the images.

Something particularly important here is that a sharp 200mm lens provides more detail of that distant object than the soft 400mm lens.... but the most shocking conclusion is that a p/s camera (the SX50) can out-resolve both.

The last picture is the SX50 at 200X on the left and the 70-200 at 200mm on the right....

 

[dgatwood]

My experience has been that at higher magnifications, because the algorithms have less to work with, they don't do as well as optical glass, whereas at lower magnifications (<2), they can interpolate well enough to do as well as or better than a TC. YMMV.

Here's an eagle, shot with a 70-300L lens at 300mm, digitally blown up by 3x, and cropped to the resulting actual pixels.

Here's the same eagle, shot from the same spot, using the same 70-300L, but with a Kenko 3x TC, similarly cropped to the actual pixels.

I find the 3x TC shot to be significantly sharper.

 

[Mt Spokane Photography]

This is a hot button topic, each of us has their own perceptions and experience. Since various lenses and TC's have differing amounts of sharpness, the results individuals get will vary widely, and add into that the other factors that make each shot different. I haven't seen any authoratitive studies done, but the carefully done ones tend to support the view that cropping can be effective up to a point.
I tried to compare what I thought were carefully taken distant shots and found the results to be inconsistent, sometimes a TC was better, sometimes a crop.

 

[candc]

The last picture is the SX50 at 200X on the left and the 70-200 at 200mm on the right....

So the sx50 is 1200mm or 2400mm equivalent at 200x zoom?

 

[Don Haines]

This is a hot button topic, each of us has their own perceptions and experience. Since various lenses and TC's have differing amounts of sharpness, the results individuals get will vary widely, and add into that the other factors that make each shot different. I haven't seen any authoratitive studies done, but the carefully done ones tend to support the view that cropping can be effective up to a point.
I tried to compare what I thought were carefully taken distant shots and found the results to be inconsistent, sometimes a TC was better, sometimes a crop.

Agreed.

My conclusion was that for MY lenses that the sigma 120-400 was for all essential purposes, worthless. A smaller/lighter/faster lens (the 70-200F4) is wider and will resolve further than the sigma. It focuses faster and is easier to use. I can't think of a single way or scenario in which the 120-400 is better.

It's a good thing to test your gear and find out what works for you...

I posted this because I thought that some might find it interesting and it would be nice to hear about other combinations that work or that fail.

The biggest surprise is how well the SX-50 works.... but be warned, it only works like that on objects that are not moving... it has glacially slow auto-focus and shutter lag so bad that sometimes you wonder if it is still working.

 

[jrista]

@Don: Before I comment any further...is it safe to assume that you maintained subject framing for each and every photo, or did you keep the camera at the same distance and only change the focal length?

The reason I ask is, if you normalize subject framing, then in every single case, fewer optical elements will be better. That does not, realistically, demonstrate the real-world use case for a teleconverter, though...which is to increase focal length to get you more reach from the same camera distance (and, therefor, enlarge the subject relative to the frame).

I am not saying your analysis is wrong...just that if you maintained framing, it doesn't actually demonstrate the strengths of a TC in the situations they were intended to be used. It would be very cool if you could re-do the test, but keep the camera at the same physical location, and scale up the crops to the same dimensions as the images with the TC's attached. (Would be best, in that case, to start with the subject filling the frame at the longest possible focal length, then work backwards to the shortest focal length from there, and crop to maintain framing after the fact.)

 

[Don Haines]

@Don: Before I comment any further...is it safe to assume that you maintained subject framing for each and every photo, or did you keep the camera at the same distance and only change the focal length?

The reason I ask is, if you normalize subject framing, then in every single case, fewer optical elements will be better. That does not, realistically, demonstrate the real-world use case for a teleconverter, though...which is to increase focal length to get you more reach from the same camera distance (and, therefor, enlarge the subject relative to the frame).

I am not saying your analysis is wrong...just that if you maintained framing, it doesn't actually demonstrate the strengths of a TC in the situations they were intended to be used. It would be very cool if you could re-do the test, but keep the camera at the same physical location, and scale up the crops to the same dimensions as the images with the TC's attached. (Would be best, in that case, to start with the subject filling the frame at the longest possible focal length, then work backwards to the shortest focal length from there, and crop to maintain framing after the fact.)

Everything was shot from a tripod 25 feet away from the target. The camera did not move.

 

[jrista]

@Don: Before I comment any further...is it safe to assume that you maintained subject framing for each and every photo, or did you keep the camera at the same distance and only change the focal length?

The reason I ask is, if you normalize subject framing, then in every single case, fewer optical elements will be better. That does not, realistically, demonstrate the real-world use case for a teleconverter, though...which is to increase focal length to get you more reach from the same camera distance (and, therefor, enlarge the subject relative to the frame).

I am not saying your analysis is wrong...just that if you maintained framing, it doesn't actually demonstrate the strengths of a TC in the situations they were intended to be used. It would be very cool if you could re-do the test, but keep the camera at the same physical location, and scale up the crops to the same dimensions as the images with the TC's attached. (Would be best, in that case, to start with the subject filling the frame at the longest possible focal length, then work backwards to the shortest focal length from there, and crop to maintain framing after the fact.)

Everything was shot from a tripod 25 feet away from the target. The camera did not move.

Hmm. I'm rather surprised by the results, then. I'd have expected the 1.4x to put more pixels on subject, and produce better normalized results in pretty much every case. With the 2x, there is bound to be more degradation, so tough to predict.

I understand the SX50...it's pixel pitch is incredibly small, and with such a small sensor, creating lenses with high resolving power is a lot easier. So the fact that it resolves more detail with digital zoom (cropping) than the 70-200 @ 200mm isn't really all that surprising.

 

[Don Haines]

@Don: Before I comment any further...is it safe to assume that you maintained subject framing for each and every photo, or did you keep the camera at the same distance and only change the focal length?

The reason I ask is, if you normalize subject framing, then in every single case, fewer optical elements will be better. That does not, realistically, demonstrate the real-world use case for a teleconverter, though...which is to increase focal length to get you more reach from the same camera distance (and, therefor, enlarge the subject relative to the frame).

I am not saying your analysis is wrong...just that if you maintained framing, it doesn't actually demonstrate the strengths of a TC in the situations they were intended to be used. It would be very cool if you could re-do the test, but keep the camera at the same physical location, and scale up the crops to the same dimensions as the images with the TC's attached. (Would be best, in that case, to start with the subject filling the frame at the longest possible focal length, then work backwards to the shortest focal length from there, and crop to maintain framing after the fact.)

Everything was shot from a tripod 25 feet away from the target. The camera did not move.

Hmm. I'm rather surprised by the results, then. I'd have expected the 1.4x to put more pixels on subject, and produce better normalized results in pretty much every case. With the 2x, there is bound to be more degradation, so tough to predict.

I understand the SX50...it's pixel pitch is incredibly small, and with such a small sensor, creating lenses with high resolving power is a lot easier. So the fact that it resolves more detail with digital zoom (cropping) than the 70-200 @ 200mm isn't really all that surprising.

Remember... it's a crop camera. I tried the same test with a 5D2, the 70-200, and the teleconverters and found that both the 1.4X and the 2X increased resolving power.... I think that over Christmas I will try my friend's 5D2 and my 60D with and without teleconverters on my bird target and see what happens... I will try them both at low ISO and at high ISO..

The Sigma 120-400 is a really soft lens in comparison to the 70-200. It doesn't surprise me in the least that it does not play well with teleconverters... my experience has been that unless you have a sharp lens, to ignore them.

 

[Quasimodo]

I love the idea for this thread! I have a 2xIII TC and have had excellent results with the 70-200 II, 135L, 200/2L, 300/2.8L II, and the 500/4L II.

I would like if you or anyone else could do a real test of this, given FOV variations and such.

 

[AlanF]

Don
I'm with you here on these issues. I have found and posted examples that the 2xTC III on the 300mm f/2.8 II really does increase its resolving power, and lots of use that combination. The 1.4xTC doesn't enhance my 100-400L at all and spoils its performance. The SX50 is great and I started a thread showing that it could outperform the 100-400L on a 5DIII, and got a lot of flak. You must use RAW on the SX50.
Alan

 

[jrista]

Remember... it's a crop camera. I tried the same test with a 5D2, the 70-200, and the teleconverters and found that both the 1.4X and the 2X increased resolving power....

Ah, sorry. I guess I misunderstood that. The wording "past experience" somehow made me think you were testing with FF.

I think that over Christmas I will try my friend's 5D2 and my 60D with and without teleconverters on my bird target and see what happens... I will try them both at low ISO and at high ISO..

I would really love to test a 1D X in comparison to a 7D. The intriguing thing about such a test is that both cameras have the same megapixel count...so scaling is in a convenient ratio. Maybe next year, when I have some money to rent a 1D X, I'll give it a try.

The Sigma 120-400 is a really soft lens in comparison to the 70-200. It doesn't surprise me in the least that it does not play well with teleconverters... my experience has been that unless you have a sharp lens, to ignore them.

True. If the Sigma is that soft, it is probably best to avoid TCs. Even the EF 100-400, even though it has relatively good IQ, doesn't play well with TCs.

 

[Quasimodo]

But seriously, this is a great topic, and it would be interesting to see the results on what which lenses does, or performs, with a different type of teleconverter (and the threshold from where you are punished by using it, to where you gain by using it). If there are differences on the results based on the scale of sensors, well then; it should be included too. The 50x zoom cameras and such is imho a sidetrack, as the real questions was the effect of using a TC.

As much as I think this is a great question, I find it hard to believe that this has not been explored before as a comparison, lens by lens.

G.

 

[Loswr]

As much as I think this is a great question, I find it hard to believe that this has not been explored before as a comparison, lens by lens.

Many such comparisons are done, but often they're done using test charts, e.g., the ISO 12233 crops on TDP. Typically, those charts are shot filling the frame (there are markings at the edges for various aspect ratios to facilitate that), and as jrista points out, that fails to recapitulate the situation where you're focal length limited.

I did some testing after getting my 600 II. I have several sizes of the ISO 12233-type chart that TDP uses, and I took the smallest one (4x6", not too different from a small bird), framed it a bit loosely with the narrowest AoV - 7D + 600 II + 2xIII. I then shot from the same distance with the 1.4x, bare lens, and the same combos with the 1D X, and cropped them all to the FoV of the 7D/2x/600 combo. Then I sold the 7D...

I may repeat with the EOS M, since it seems sharper than the 7D.

 

[candc]

I have done some experimenting with this myself and found that there is a point where the sensor out resolves the lens. Adding magnification from a tc above that point is just going to make it look worse. I like to use tc's because you get a better look at what you are shooting but you have to find the point that works best for the camera, lens, and tc combo. You have more room with a ff camera but you need to start with a really sharp Lens on a crop camera because the sensor is already looking really close at the lens. I think as a general rule: the lower the pixel density of the sensor, the better teleconverters work.

 

[jrista]

... I then shot from the same distance with the 1.4x, bare lens, and the same combos with the 1D X, and cropped them all to the FoV of the 7D/2x/600 combo. Then I sold the 7D...

LOL

I highly suspect a similar experience in my not so distant future.

(Although, the 7D will probably find a home in a box, to be hauled out whenever I have to demonstrate the value of an appropriately strong AA filter or perhaps visually demonstrate the benefit of greater pixel density in reach-limited scenarios... ;P)

 

[TexPhoto]

I have a 400mm f2.8 IS and both the 1.4X and 2X converters. I wish I could say it the tele converter always improved the photo, or that not using teleconverer was better, or at least as good. I can't. The best I can say is the teleconverters resolve more detail, but the photo looses some of the crisp look that is what makes this lens so special.

 

[jrista]

I have a 400mm f2.8 IS and both the 1.4X and 2X converters. I wish I could say it the tele converter always improved the photo, or that not using teleconverer was better, or at least as good. I can't. The best I can say is the teleconverters resolve more detail, but the photo looses some of the crisp look that is what makes this lens so special.

Aye, you definitely lose some microcontrast with TCs. I rented a 300/2.8 L II last year, and used it with both TCs. At 420/4 with the 1.4x TC, the image remained nice and crisp, very sharp. Was still better than my 100-400. At 2x, though, I definitely felt the loss of contrast, even though the subject was so much larger in the frame.

Similarly, with the Kenko 1.4x, while the subject is larger, it feels like detail is lost (even warped). I am not sure if that is just how the Kenko is, or if it might actually have a decentered element, but overall IQ between a bare lens and the 1.4x Kenko (Pro 300 DGX, to be exact) downsampled to the same dimensions as the bare lens is minimal, and in more cases than not, the bare lens still maintains the IQ edge.

The Canon 1.4x TC III apparently has slightly less center sharpness than either the range of Kenkos (such as the MC4 and Pro 300 DGX) and the 1.4x TC II, however it's midframe and corner sharpness have been improved considerably. When you hear about "center frame", it really means DEAD CENTER of the frame, basically within the spot meter circle. Anything outside that area, and IQ quickly falls off. I've noticed this with the Kenko 1.4x...IQ in the dead center of the frame is OK, but that funky warped loss of detail rapidly kicks in outside of the center. (So, if you need a good TC, the Canon Mark IIIs are actually the best deal, even though some TCs, including the 1.4x TC II, have better center sharpness.)

 

[Quasimodo]

As much as I think this is a great question, I find it hard to believe that this has not been explored before as a comparison, lens by lens.

Many such comparisons are done, but often they're done using test charts, e.g., the ISO 12233 crops on TDP. Typically, those charts are shot filling the frame (there are markings at the edges for various aspect ratios to facilitate that), and as jrista points out, that fails to recapitulate the situation where you're focal length limited.

I did some testing after getting my 600 II. I have several sizes of the ISO 12233-type chart that TDP uses, and I took the smallest one (4x6", not too different from a small bird), framed it a bit loosely with the narrowest AoV - 7D + 600 II + 2xIII. I then shot from the same distance with the 1.4x, bare lens, and the same combos with the 1D X, and cropped them all to the FoV of the 7D/2x/600 combo. Then I sold the 7D...

I may repeat with the EOS M, since it seems sharper than the 7D.

Thank you.

I have tried to look at the ISO lens comparison charts on the TDP, but I don´t really understand what I am seeing there. In Don Haines way of posting it makes perfectly sense from the get go. In that way I can make up my own mind about the differences in sharpness, contrast, and overall IQ. I think it would be swell if someone who has access and time would make a comparison page with all the combos of TC vs. Lenses, vs. sensors