Macrophotography and image quality - ideals vs reality

[Karim Ghantous]

I was reading some comments in a thread about the 90mm Macro on 43rumors.com. And this one stuck out:

Quote

 

If the test images from Lenstip were actually taken using a tripod -- they don't say, but I assume they are -- then the TC is a 'no-go'.
My solution to that issue simple - although I know it horrifies the technical purists - if you need greater magnification, simply use the much maligned Digital Teleconverter (DTC). You get the 2X magnification at the same aperture, and your image is cropped to the center of the frame which is the sharpest part.

Now before anyone has an outright stroke, I have used it for a lot of macro work, especially with my Oly 30mm f/3.5 (before I got my 60mm f/2.8) -- and it simply cranks out amazing results. It works every bit as well as cropping after the fact in post, only you can enjoy composing in camera rather than wasting time later.

 

Unsurprisingly, the 2x TC doesn't give you the best image quality. The solution: crop the image (although I think the DTC is 1.5x, not 2x). It sounds so... dirty, right? However, you have two choices: more pixels but a softer image, or fewer pixels and a sharper one. Of course this is just a rule-of-thumb, and not a law.

When it comes to extremes, you have to pick your poison. You can crop MFT a little, but not too much. But if the result is better than a TC, maybe consider it. And it means you don't have to spend money on an accessory you don't need or want. OTOH, maybe the 1.4x TC plus a little cropping is the better option. You just have to test.

Also, remember that camera debayering is not as good as a RAW converter, so shoot RAW when it matters.

[Ken Katz]

Unless my math is in error (entirely possible these days), then cropping an image so that it gives you the coverage of lens with 2x the focal length should result in reducing the image to about 1/4 of the original size and therefore loose about 3/4 of your pixels.   With the current OM sensor, that would result in a 5 mp final image, which would be enough for a reasonable 8 x12" print.  In the old days, a pro telephoto lens or pro zoom lens plus pro 2x extender was better than cropping, but may be worse than a longer consumer grade lens.  Probably makes sense to rent and test before buying.

Frankly, most of the comments being written about the new OM 90mm macro are mostly nonsense and a lot of BS by folks who sound like they are optical engineers who work for Zeiss, second guessing OM's design parameters for that lens.  The lens has built in 2:1 capabilities, which should provide a frame filling, 5184 pixel wide image of 9mm sized subject (4x FF equivalent).  I would expect that most users would only consider an extender to increase their working distance from the subject, not for additional magnification beyond 2:1.  Probably useful for photographing Murder Hornets, though for me, my working distance would about a continent or 2 away.  

Edited February 15, 2023 by Ken Katz

[rodeo_joe1]

Just my twopenn'orth:

An optical teleconverter cannot and does not improve the quality of the lens it's attached to. In fact it usually makes it worse. 

All an optical TC does is to effectively magnify the image from the original lens. 

Therefore it follows that if a "lossless" magnification can be achieved by another method, then that method is to be preferred. And often a crop can be made from a high-pixel count sensor that introduces no adverse effect on perceived image quality. In fact it has the advantage of not reducing the effective aperture (less diffraction) and improves depth-of-field over getting the same final magnification at the same numerical aperture by increasing the lens focal length. 

Obviously there's a trade-off and each method should be taken on its merits. Possibly through empirical means (I.E. trial and error). However, the availability of new AI interpolation software now throws another variable into the mix. Or yet another option is that of using pixel-shift resolution doubling - provided that both the subject and camera make that possible. 

On 2/15/2023 at 4:45 PM, Ken Katz said:

Unless my math is in error (entirely possible these days), then cropping an image so that it gives you the coverage of lens with 2x the focal length should result in reducing the image to about 1/4 of the original size and therefore loose about 3/4 of your pixels. 

Ah, the old chestnut of image area and ergo pixel numbers, being any criterion of image quality. A total red herring I'm afraid. Magnification is one-dimensional, as is any other parameter by which image quality might be objectively measured; such as MTF figures, resolution, or circles-of-confusion.

That area increases according to the product of linear dimensions is a mathematical inevitability, but really doesn't have any bearing on the objective measure of image quality. That's ultimately down to the angular acuity of the viewer's eye, and again that angle is a one-dimensional number. Halving or doubling it doesn't make the viewer's eyesight 4 times or one-quarter as effective. 

Final viewing magnification is the only important factor. 

Plus, nobody really expects that viewing an object at 100 times its real size is going to be pin-sharp.

Edited February 17, 2023 by rodeo_joe1

[Ken Katz]

Don't disagree with Rodeo, but one should note that the current OM system has only a 20mp sensor, so that a significant crop (as described above) leaves you with only 5mp which utilizes a sensor area of around 56 sq mm, which is not much bigger than the main camera on many current top of the line smart phones.  I do this regularly on photos of critters using my cheap/lightweight 40-150mm kit lens, but I am not trying to market such images, and don't expect miracles in terms of possible print size.  It does keeps my camera/lens kit at around 21 oz (600 grams for those of you in the more advanced places in the world), which helps my deteriorating vertebrae. 

I used to shoot with a Canon D60, with a 6 mp sensor / 1000 ISO max.  I was well aware of its limitations and strengths compared with the alternative (film).  Had a 1 GB IBM microdrive since it cost about $200 and a solid state card was almost $1K.

Edited February 17, 2023 by Ken Katz

[rodeo_joe1]

There was some similar debate about the merits of a TC in another forum, and I did a quick experiment using the 2 TCs that I had to hand - a Nikon TC-200 and a 7 element 2x 3rd party job. 

I fitted the teleconverter + prime lens to a 36 megapixel D800, while the prime lens alone was fitted to my 24 Mp D7200 and cropped afterwards to give a similar field of view. In theory this gives the D800 the advantage of more pixels and more optical magnification.

Except that the image degradation due to either TC was far more objectionable than the lack of pixels. 

This is a small crop from the D7200 with a prime lens fitted. The specular LoCa is a bit distracting at this crop level - 

Not as bad as the added fuzz and shallower depth-of-field of the TC-200 though, extra pixels notwithstanding - 

The 3rd party TC is no better - 

OK. I'm sure that there are better teleconverters on the market these days, but at a high price. And no matter how expensive a TC is, it cannot actually improve the basic quality of the lens it's attached to. 

Edited February 18, 2023 by rodeo_joe1