[rdenney]

Series Contents

Teleconverters are so seductive, but they always magnify flaws, and sometimes introduce a few of their own. Is it worth taking a teleconverter to avoid having to carry a longer lens? That's the question I sought to answer in this review.

The real question is: Should one use a teleconverter, or merely crop the photo from the shorter lens?

Here is a 1:1 center crop from an image made with a 645 A* 300/4, at f/11:


Dang that looks good. That A* 300/4 earns its reputation.

And here is a crop from an image made with the 1.4 converter, downsampled to the same resolution as the image above:


No question: Downsampling the longer lens combination is not better than using the shorter lens. But considering these crops (at least the first one) is part of a 7-foot-wide print viewed at 100 pixels/inch, they are both really good. I'd use the 1.4 when I needed all fifty million of my pixels, but otherwise the A* 300 just has nowhere to go but down. It didn't go very far down here, but the converter still degraded the image more than enlargment.

Another question is whether a shorter lens plus a converter is better than carrying both the shorter and the longer lens.

Here is a 1:1 crop made using the 400/5.6 ED(IF), at f/11:


And here is a crop at 1:1 (without downsampling) from the 300+1.4 at f/11:


Not much difference here, but I would say that the 400 looks a bit better. Clearly, the A* 300 is a (slightly) better lens, but the 400 is worth bringing if one needs to optimize for that focal length.

So, the advice is: If you use the 1.4, start with a great lens. Stop it down a bit, and then don't worry about it. The converter has visible effects at high magnification, but it is still a very capable teleconverter.

Rick "always room for a converter" Denney

---

Last edited by rdenney; 06-30-2018 at 12:10 AM.

[mikeSF]

Rick, thanks for this but can you elucidate you process a little more re: your Question#1?

Wouldn't you need to be upsampling the 300mm image to match the larger image size obtained by the 300mm+TC configuration? In a real world situation requiring a teleconverter, you would be discarding half of the pixels of the 300mm file and would need to upsample that to match the resulting image square of the 300+TC, no?

I have done this testing as well and arrived at a very different result. Though it is surely rig-specific, I have concluded that optics win over cropping and magnifying pixels. Help me understand. thanks!

[desertscape]

My field experience with Pentax 1.4x converters shows that they do better with long lenses that are not telephoto designs. The ED lenses are not telephoto designs and do well with the 1.4x. The older telephoto designs already magnify the image from the front group, so the addition of a 1.4x to this only magnifies the aberrations more. Using a 1.4x on the telephoto designs, one must use the lens at its sweet spot, otherwise softness results. In the case of the 600 Takumar (67) used at its sweet spot, a 1.4x converter will do very well. Actually, the use of the 1.4 reduces the longitudinal chromatic, as it intercepts the incoming rays long before reaching the film.

[rdenney]

Rick, thanks for this but can you elucidate you process a little more re: your Question#1?

Wouldn't you need to be upsampling the 300mm image to match the larger image size obtained by the 300mm+TC configuration? In a real world situation requiring a teleconverter, you would be discarding half of the pixels of the 300mm file and would need to upsample that to match the resulting image square of the 300+TC, no?

I have done this testing as well and arrived at a very different result. Though it is surely rig-specific, I have concluded that optics win over cropping and magnifying pixels. Help me understand. thanks!

In the old days, we'd just crank the enlarger up another several inches. But that isn't possible here. I didn't want people saying--hey, pixelation is undermining the image because you upsampled it. I am looking at these images at 100 pixels/inch, so the pixels are big.

Most don't print that big. Even if people make 20x27 prints, the image resolution will be 300 pixels per inch from the 645z. Enlarging it to make the resolution 214 pixels/inch would have the same effect as using the converter. It would take a pretty discriminating viewer to see the digital effects of that--in that print we have not had to upsample and the pixel effects are still too small to see.

I wanted to avoid the digital effect. So, I'm turning it around. If I print the standard image at 300ppi, what resolution would I have to print the teleconverted image to make items in the image the same size? The answer would be 420 pixel/inch. Now, if that printer only prints at 300, it will downsample that 420ppi file to 300ppi before printing it, even if it's done in the printer driver. We expect that downsampling to improves an image, so if the downsampled image appears degraded, it's nearly certain the downsampling isn't the cause. If the differences were subtle, we might be willing to argue that the downsampling caused the effect (though at this stage of my experience, I can say I've never seen an image get worse by being downsampled, if it's still printed with small enough pixels). But here the differences were quite apparent. At least they were apparent when viewed at 1:1 on a low-resolution monitor. But that isn't to say that the difference would be important when printed at real sizes.

Rick "the effect you are seeing is the teleconverter" Denney

---------- Post added 05-22-18 at 11:01 AM ----------

My field experience with Pentax 1.4x converters shows that they do better with long lenses that are not telephoto designs. The ED lenses are not telephoto designs and do well with the 1.4x. The older telephoto designs already magnify the image from the front group, so the addition of a 1.4x to this only magnifies the aberrations more. Using a 1.4x on the telephoto designs, one must use the lens at its sweet spot, otherwise softness results. In the case of the 600 Takumar (67) used at its sweet spot, a 1.4x converter will do very well. Actually, the use of the 1.4 reduces the longitudinal chromatic, as it intercepts the incoming rays long before reaching the film.

Agreed. Traditional telephoto designs have, in effect, built in converters, and usually pretty simple ones at that. Both the lenses that appeared in this review are modern lenses with ED elements. In any case, the A* 300 is a slightly better lens than the FA 400.

I also got pretty good results using this converter on the FA 200/4, though the converted image is not as good as from the 300 by itself. That's to be expected. I included that comparison in my review of the 200.

Rick "not reluctant to use the 1.4, but reluctant to believe it doesn't have an effect even on the best lenses" Denney

[mikeSF]

In the old days, we'd just crank the enlarger up another several inches. But that isn't possible here. I didn't want people saying--hey, pixelation is undermining the image because you upsampled it. I am looking at these images at 100 pixels/inch, so the pixels are big.

Most don't print that big. Even if people make 20x27 prints, the image resolution will be 300 pixels per inch from the 645z. Enlarging it to make the resolution 214 pixels/inch would have the same effect as using the converter. It would take a pretty discriminating viewer to see the digital effects of that--in that print we have not had to upsample and the pixel effects are still too small to see.

I wanted to avoid the digital effect. So, I'm turning it around. If I print the standard image at 300ppi, what resolution would I have to print the teleconverted image to make items in the image the same size? The answer would be 420 pixel/inch. Now, if that printer only prints at 300, it will downsample that 420ppi file to 300ppi before printing it, even if it's done in the printer driver. We expect that downsampling to improves an image, so if the downsampled image appears degraded, it's nearly certain the downsampling isn't the cause. If the differences were subtle, we might be willing to argue that the downsampling caused the effect (though at this stage of my experience, I can say I've never seen an image get worse by being downsampled, if it's still printed with small enough pixels). But here the differences were quite apparent. At least they were apparent when viewed at 1:1 on a low-resolution monitor. But that isn't to say that the difference would be important when printed at real sizes.

Rick "the effect you are seeing is the teleconverter" Denney

printing? old days? enlarger?

OK, you are seeking to answer a different question altogether. I thought you were attempting to test whether it is better to use a TC versus cropping pixels on the SAME lens. Re-reading this, that is not your intent here.

If I read this right, you are asking whether a hypothetical 215mm lens fitted with a 1.4x TC compares favorably to a 300mm lens. Why?

[normhead]

My own take on the same question.
Understanding folks saying a TC doesn't ad more detail. - PentaxForums.com

The HD DA 1.4 TC is a piece of modern glass, designed to produce a 37% increase in subject detail. 40% magnification minus an estimated 3% loss in resolution. You can find a lot of TCs out there that weren't manufactured anywhere close to that spec. And you can find a lot of lenses out there that aren't manufactured anywhere near the specs of the DA*200. I almost use the TC as source of testing lens quality. If it's a good quality piece of glass with low distortion and CA, it will look just as good with a high quality TC on its it does without.

On the other hand, even with a super high quality TC like the HD DA 1.4 my older lenses, A-400, F 70-210 come to mind off the top of my head, there just isn't enough over-kill in the amount the lenses out resolve modern sensors, if they in fact out resolve them at all, to see the benefit of a TC making any difference over just using software to enlarge the image and making a bigger image without the TC.

Based on the images I view day to day however, my Tamron 300 2.8 with the F 1.7x AF Adapter does just as good as and 500 ƒ4.5 lens and considerable better than any 50-500, 150-600 type lens I've seen images from to the point where I'd never consider another lens under 600mm. The 300 and 1.7x are just as good.

[rdenney]

printing? old days? enlarger?

OK, you are seeking to answer a different question altogether. I thought you were attempting to test whether it is better to use a TC versus cropping pixels on the SAME lens. Re-reading this, that is not your intent here.

If I read this right, you are asking whether a hypothetical 215mm lens fitted with a 1.4x TC compares favorably to a 300mm lens. Why?

No, that was 215 pixels/inch, not mm, and I'm only using the 300mm lens for this question--obviously I wasn't very clear in my explanation. The question you are asking is A versus B:

A: Use, say, a 300, and print at 215 pixels/inch to make a print where the subject of the photo is at the required size.

B: Use that 300 with its teleconverter and make a print at 300 pixels/inch where the subject of the photo is the same size.

So, which degrades the image more? The lower print resolution of option A, or the teleconverter in option B? In my test, I determined it to be the teleconverter. But that only works if the resolutions in question are all adequate. At 1:1 on the screen, they are already topped out--more enlargement would require upsampling of the original image with inherent digital effects. So, I downsampled the larger image to match, so that we could look at them at the same scale.

Rick "the difference you see is caused by the teleconverter" Denney

[mikeSF]

No, that was 215 pixels/inch, not mm, and I'm only using the 300mm lens for this question--obviously I wasn't very clear in my explanation. The question you are asking is A versus B:

A: Use, say, a 300, and print at 215 pixels/inch to make a print where the subject of the photo is at the required size.

B: Use that 300 with its teleconverter and make a print at 300 pixels/inch where the subject of the photo is the same size.

So, which degrades the image more? The lower print resolution of option A, ...

Had you merely mentioned the words "on film" in your original post, I would not have been confused. Got it.
In this film-based investigation, you are limited by what your eyes can see on a paper print.

I would suggest a better crop vs TC comparison would occur by inspecting digital images at the pixel level. I understand now that was not your intent.
no worries.

[rdenney]

My own take on the same question.
Understanding folks saying a TC doesn't ad more detail. - PentaxForums.com

The HD DA 1.4 TC is a piece of modern glass, designed to produce a 37% increase in subject detail. 40% magnification minus an estimated 3% loss in resolution. You can find a lot of TCs out there that weren't manufactured anywhere close to that spec. And you can find a lot of lenses out there that aren't manufactured anywhere near the specs of the DA*200. I almost use the TC as source of testing lens quality. If it's a good quality piece of glass with low distortion and CA, it will look just as good with a high quality TC on its it does without.

On the other hand, even with a super high quality TC like the HD DA 1.4 my older lenses, A-400, F 70-210 come to mind off the top of my head, there just isn't enough over-kill in the amount the lenses out resolve modern sensors, if they in fact out resolve them at all, to see the benefit of a TC making any difference over just using software to enlarge the image and making a bigger image without the TC.

Based on the images I view day to day however, my Tamron 300 2.8 with the F 1.7x AF Adapter does just as good as and 500 ƒ4.5 lens and considerable better than any 50-500, 150-600 type lens I've seen images from to the point where I'd never consider another lens under 600mm. The 300 and 1.7x are just as good.

Yes. Implied in your statement is the notion that lenses should out-resolve sensors. I take the opposite view. I want a sensor that out-resolves my lenses. That way, I know I'm getting all the lens has to offer, and my tolerable degree of enlargement is lens-limited rather than sensor-limited. It has take me a long time to reach that point.

These lenses were excellent when we made real prints from real negatives. We almost never make prints the size of the 1:1 display on our computer monitor (about 7 feet wide), but we can certainly make all the prints we might have once made using film, and then some. And none of these images would look terrible on a 7-foot print, viewed from maybe a bit more than six inches. I would never have made a print that big from film--the film itself wouldn't have allowed it.

(But if Pentax goes to a 100MP sensor, expect bleating from the masses, who evaluate every shot at 1:1 on their monitor.)

What this means to me is that I don't have any concern about using the 1.4 extender even when it creates a noticeable effect, if that's what it takes to get the shot. It just means I might not print it as large. I don't own the 645 A 600mm lens, but the extender with the 400 is good enough for images I might not need to print so large, but still large enough to need to preserve sensor resolution.

Rick "lens-limited with the 645z, but not by much" Denney

---------- Post added 05-22-18 at 01:06 PM ----------

Had you merely mentioned the words "on film" in your original post, I would not have been confused. Got it.
In this film-based investigation, you are limited by what your eyes can see on a paper print.

I would suggest a better crop vs TC comparison would occur by inspecting digital images at the pixel level. I understand now that was not your intent.
no worries.

In fact, it's the opposite of my intent.

My first digital camera was a Canon 10D. 6 megapixels. The largest print that maintains 300 pixels per inch was 7x10. I might get 12x18 at a tolerable 166 pixels per inch, but then I'd have to perform some trickery to make it look okay. For that camera, what happens at the pixel level matters, because the pixels are big enough to limit my print size. Testing at full pixel resolution made sense.

My second DSLR was a Canon 5D, at 13MP. It could make an 9.5x14" print at 300 pixels/inch, but look pretty good at 12x18. The biggest printer I had at the time had a 13" platen, so it was okay. But the pixels were still big enough to limit print size. Testing at full resolution still made sense.

Then, I bought an Epson 3800 and started printing 16x20 prints from film. A print that size (that maintains my requirement of never losing the illusion of endless detail even viewed up close) from 4x5 or 6x7 is easy-peasy. I scan the 6x7 in my Nikon 9000, which gives me an abundance of pixels (about 90MP). As long as what the film has on it can stand the enlargement, the pixels are not limiting at all. Likewise 4x5 film scanned on an Epson flatbed, which gives me about the same 90MP but with each pixel having more information in it. Printing those scans large is limited by the optics, the scanner, and the film, not by the pixel resolution. I once again tested my system at print size, as I had always done with film.

The 645z, for the first time in my own odyssey, removes the pixel limitation. It is extremely unlikely that I or most people will ever make a print bigger than the sensor of the 645z will print at 300, which is 20x27. In fact, most of the time I will be printing no larger than 16x20, which is 410 pixels/inch on that sensor. Even at 240, I'm printing 36" wide. That means I can look at the lenses based on how they will perform on a likely print, rather than how they perform in the unrealistic case of 1:1 pixels on a low-resolution monitor. It's a philosophy, but I think with modern sensors, expecting lenses to be sharp to the pixel is going to be very expensive and impractical. And all that extra expense will serve nothing but bragging rights on the Internet; very few will ever make prints that big.

Here's another way to look at it: If I want a print to resolve five line pairs per millimeter (about the limited of human perception), I'll need lenses capable of that times the enlargement factor. For a full-frame camera, a 20" print requires lenses that resolve 100 line pairs/mm. That's very demanding and lenses that can do it are as expensive as cars. A 645z requires only 57 line pairs/mm, which is still demanding but quite doable even with traditional glass, carefully used. But that is a punishingly high standard of sharpness in a print. The sensor provides no limitation here--the system becomes lens limited. Putting a test on the internet means 1:1 samples, but proper interpretation of those samples should take into account actual desired display (print) size. in a 20" print, that standard is only 60% of the Nyquist limit on the 645z sensor. So, I don't need lenses to be flawless at 1:1.

Rick "who will still use the large-format camera if more is needed" Denney

[normhead]

For a full-frame camera, a 20" print requires lenses that resolve 100 line pairs/mm

A typical k-5 image with a measured lw/ph of about 2100 lw/ph will give you a 20" print at 100 lines per inch. Despite what folks here might say, most people don't care if they have that kind of resolution. Even with a direct comparison, many just don't care to examine the image close enough to see the difference. They are interested in the artistic merit of the photo, not the technical.

My biggest gripe on the forum is people who evaluate lenses as if everything they print will be printed really big. Years ago I did a poll where about 20% of the respondents favoured images from the old FA 35-80 ar 35mm reduced to web size. But only about 3% preferred it pixel peeping. The corollary statement here, is love of very high resolution lenses may decrease your enjoyment of your images at the size you usually view them.

People always want to say more resolution is always better if you print larger. There is no evidence to support that notion. Every time someone has tried to prove this theory they've come up empty. The simple fact in my mind is some theoretically poor resolution images make fine prints, especially when upsized to make best use of the printer.

I've tried to make the point that rendering is more important than straight up resolution. But the point that an image can have too much resolution also needs to be made. The resolution needs to augment the image. The notion that there can be too much has exactly the same level of empirical support as the notion that you can never have too much. Yet one is repeated constantly and the former is ignored.

A diffraction limited image from the 1.2 sensor in my XG-1. I love this effect. But I'm not going to tell you how quickly the techies on this site turn up their noses and claim it's an inferior image, just based on it's lack of sharpness.

Well, many people like this kind of image. Sharpness is a bourgeois concept. I have a similar image taken with my K-1 and DFA 28-105 and I (and many other non-photographer people will like this one. Understanding the difference between artitistc merit and technical evaluation, is the difference between photographer, and a photographic lab tech.

[rdenney]

A typical k-5 image with a measured lw/ph of about 2100 lw/ph will give you a 20" print at 100 lines per inch. Despite what folks here might say, most people don't care if they have that kind of resolution.

A camera sensor that can do 2100 line pairs in the height of the picture doesn’t mean the lens attains it.

As to what people do or don’t care about, that is not my concern. As the artist, I get to declare what I think viewers ought to care about by delivering photos that embody those attributes.

Not much argument from me on the general warning about not being consumed by 1:1 image evaluation on a computer monitor, but also not really relevant to a post about lens testing.

The question of craft versus art comes up in all art forms. I once asked my tuba teacher, a renowned symphony musician, which was more important, technique or musicality? His answer: “Yes.” Musicality cannot exist without technique, and technique serves no purpose without musicality.

Painters insist on developing their drafting and drawing skills. Fulfilling the requirements of architecture school required drafting and drawing classes before design classes. Even if an artist works in the abstract, they do so as a response to good technique, not as a replacement for it.

Ansel Adams and his group reacted against the mode of the day by favoring sharp photos. They had poor lenses so they shot in large formats to minimize how magnified those faults would become. They felt that sharp images are uniquely photographic, and the defining characteristic of photography that should not be ignored.

That’s not to say that they weren’t also interested in their artistic expression. It was Adams who disparaged “sharp pictures of fuzzy concepts”.

But even the Pictorialists against whom the members of Group f/64 were rebelling knew how to make sharp images. They just chose not to, and their reason (to emulate paintings) was what annoyed the members of f/64.

Go back to my teaser for this series. I said there was not a dog in the bunch. Every one of the lenses I evaluated is capable of great art.

But when comparing lenses, it’s the lenses I want to compare, not the photographer. If I have any artistic powers, I don’t want them undermined by under-performing equipment, so I look to see what is best. Online testing makes that possible for people who are considering a lens.

Now, to big prints. Let me ask: What is the appeal of a 50mp medium-format camera if not for making big prints? It’s not a practical camera for making photos only for web display. My iPhone is already pretty good at doing that. My smaller DSLR makes great small prints. I can’t get just as thin a depth of field with a full-frame DSLR. There is greater tonality in medium format images at most print sizes, but I’m not just interested in that.

My personal standard is this: I never want to lose the illusion of infinite detail. I want prints that invite people to come close. And when they do, I don’t want them to be disappointed. Attaining sharpness is no more a bourgeois concept that being able to play all 48 scales. The biggest print I can make is 16x20, but I might someday own a 24” printer. My small DSLR does not provide a sense of endless detail even at 16x20.

To what purpose I put that sharpness? That is a different question and one that can’t be resolved with lens testing. But the natural scene presents endless detail, and that’s part of what I want to capture.

Rick “people can draw their own conclusions” Denney

[jpzk]

I am late, as usual, (trying to) following this series of 1.4x teleconverter(s).

Do you have a section combining the DA*300/4 with such TC ?
Thanks !

[rdenney]

I am late, as usual, (trying to) following this series of 1.4x teleconverter(s).

Do you have a section combining the DA*300/4 with such TC ?
Thanks !

I don’t. I only own the manual-focus 645 A* 300. I don’t think there’s a D anywhere in there.

I have included samples of images made using the teleconverter in my reviews of the FA 200/4 and the FA 400/5.6.

Rick “no series, just one converter” Denney

[normhead]

A camera sensor that can do 2100 line pairs in the height of the picture doesn’t mean the lens attains it.

Well actually, the Imagina Resources measured the 2100 lw/ph with a Sigma 70 macro, so there is a pretty good chance the lens sensor combination will be in the ball park.

Based on the images provided, the loss of micro-contrast is noticeable. I probably wouldn't use that lens + TC combination. I'll try and put up some images later with different combos.

[RGlasel]

I don’t think there’s a D anywhere in there.

I'm wondering if everyone posting on this thread noticed that your are testing 645 format lenses and I have to assume that your teleconvertor is also for 645 format.