Thinking about capturing/increasing image detail... what does/doesn't work.

[sk66]

One thing I have noticed over the years is that absolutely nothing increases recorded image detail as much as reducing the subject distance does (this assumes adequate/appropriate light and no diffraction limiting). And I've been thinking about "why" that might be.

Enlargement doesn't increase detail, it only makes what detail there is larger... i.e. printing/displaying larger or cropping an image and printing/displaying at the same size.

Similarly using a crop sensor to record a smaller portion of the lens' FOV does not necessarily increase the recorded detail. This is essentially the same as cropping/enlarging, thus the term "crop factor." The limit of detail will be controlled by how much the lens transmits (limited by aperture/aberrations) and the resolution of the sensor. *IF* the lens is able to transmit fine enough details (wider apertures) *AND* the smaller sensor has a smaller pixel pitch (same/higher MP count) then it will record more fine detail... The limitation here is typically the lens/aperture and using a smaller sensor is only recording the same amount of detail (w/ reduced contrast).

This leads me to TC's... TC's are fairly complex lens groups, but essentially they simply crop/enlarge a smaller potion of the lens' FOV much as using a crop sensor of the same MP count does. And again the limitation will typically be the lens (aperture/aberrations). Additionally a TC will frequently add some additional aberrations of it's own, and it definitely magnifies the effects of the lens' aberrations. The result is typically that the lens has to be stopped down farther, which causes the lens to (potentially) transmit even less fine detail.

And this leads me to the lens itself. A lens is, again, just magnification/enlargement. The limits here will be the amount of detail that actually exists in the scene/subject, the relative size of those details (distance), and the lens itself (aperture/aberrations). Using a longer lens (or TC) does not increase the amount of detail that exists, nor does it change the physical distance. Additionally, due to manufacture/design limitations longer lenses/greater magnification often have greater aberrations and/or smaller apertures, both of which limit the max amount of fine detail that can be transmitted (and are VERY expensive to overcome).

In practice, I have found that there is the *potential* to record more fine detail by increasing magnification/enlargement. With a TC/longer lens I believe it is due to a reduction in what I call "the FL subject distance," the point in space where the lens' FOV intersects the subject FOV. But the effect seems to be much less significant than a reduction in physical subject distance, and that potential is offset by the demands/limitations of greater magnification in the first place.

All of this just shows that you can't really buy your way to getting more detail... well, you can w/in limits, but it will probably cost a whole lot of money. And it will be less effective than craft/technique/access will be (at least no more effective).
I can't say that any of this is studied/verified fact... I haven't done any empirical measurements and I wouldn't know how to go about doing so. But it sure seems to be the way things work. And it goes some way towards explaining my dissatisfaction w/ the results I've gotten from a lens/camera combination that costs nearly $20k... which is what got me thinking about all of this in the first place.


*Note that we should not confuse an increase/decrease in fine detail w/ perceived sharpness, they are related but not the same thing.

 

[Craig_85]

Atmospheric haze definitely plays a part Steven. I've done a bit of long lens work over the years and the longer FL I had, which encouraged me to take longer shots would normally be softer than if I'd used a shorter lens and put myself closer to the subject.

 

[holty]

interesting reading , i never thought about it like that

 

[sk66]

Atmospheric haze definitely plays a part Steven. I've done a bit of long lens work over the years and the longer FL I had, which encouraged me to take longer shots would normally be softer than if I'd used a shorter lens and put myself closer to the subject.

Agreed... haze and heat waves. But even when conditions are ideal/better, getting closer will always generate better results IME.
I didn't go into the exponentially increasing demands on technique that comes with working at high magnification, but that's part of it as well.

 

[frod]

This is obvious, to an extent. it's worth bearing in mind, however that since lenses can get soft at close focal distances. The X100 lens springs to mind.

 

[wave01]

Not sure what the op wants to hear. Tc I don't like them that's just me. I try to use a lens and look at what I am shooting at I need the right light and the right distance to the subject. This is the science part of photography. If you then have the right speed and aperture and iso then you should get a reasonable shot

 

[HoppyUK]

What's the question?

With wildlife and inevitably long lenses, I don't think there's any doubt that getting closer is better in terms of image quality than simply bolting on extra focal length. Everything gets exponentially harder, and the higher your reference standard of quality (and I suspect yours is quite high) then the more noticeable the incremental impact on fine detail will be. In no particular order, I would say the key issues are likely to be shutter speeds, atmospheric pollution, and tele-converters. They are all potential wreckers of image quality, they will always have a detrimental impact to some extent, and the compound effect is potentially severe.

TBH I'm not seeing any great mystery here.

 

[Cagey75]

Get a D800 and crop the bejaysis out of images, all good. Better than a TC on lesser sensors.

 

[sk66]

TBH I'm not seeing any great mystery here.

Once I spelled it all out (for myself) it makes complete sense. But you see all sorts of comparisons/discussions about the sharpness of long lenses, and these often include impressive images w/ enviable fine detail. I've owned 800/5.6, 500/4, and I use a 400/2.8 w/ TC's. I've done bench comparisons for sharpness of the 400/2.8 w/ 2x against the 800/5.6. To my eyes there is a loss of IQ for the 400 + 2x wide open (800/5.6) compared to the 800/5.6 prime, but stopped down a bit the difference is negligible (I *might* be able to tell w/ pixel peeping). And at short distances all can deliver very impressive fine detail. But at longer distances all come up lacking in fine detail, at least in my experience (and yes, my standards are rather high... hopefully trying to improve).

Short distances are actually more demanding of technique/settings because relative speeds increase significantly (camera shake/subject) and DOF decreases. Longer distance reduces the technique/settings requirements, but introduces environmental factors.

Maybe I'm dense, but I always expected greater magnification to increase recorded detail when reducing distance wasn't an option... but in fact it doesn't (or rather very seldom does IME).

 

[sk66]

Get a D800 and crop the bejaysis out of images, all good. Better than a TC on lesser sensors.

Uhm, maybe... not necessarily.
The primary benefit is the same as from all FF/larger sensors, greater intrinsic "sharpness" for any lens/aperture/situation.
I own the D810...

 

[sk66]

I need the right light and the right distance to the subject.

I'm at that point as well... I just don't take the picture if the distance is too far... FL is just for composition/filling the frame.

 

[Cagey75]

Uhm, maybe... not necessarily.
The primary benefit is the same as from all FF/larger sensors, greater intrinsic "sharpness" for any lens/aperture/situation.
I own the D810...

TC beyond 1.4x often degrade IQ. I have tested this personally, I had a 1.7x TC, and compared images cropped and ones using the TC. Sad to say, the cropped ones won out for me almost every time. That was using the D800E and 300mm F4, with and without the TC. I did like the 1.4x TC though, there's a happy balance somewhere

 

[chris malcolm]

Interesting discussion.

Photographic image making depends on light travelling in straight lines. Even the best air isn't perfect. Even on the clearest day the sky is still blue. My longest lens is a 500mm, and not the sharpest -- an f8 reflex. But for detail on distant shots the biggest limiting factor is air quality. At distances of 100 yards or more it sometimes spoils shots. At distances of a few miles it often does. At distances of dozens of miles I'm often lucky to be able to see the thing at all.

Then of course there's glass, which isn't perfect either, plus the all important quality of the glass surface. The less glass and the fewer lens components in a compound lens the better. That's why expensive zooms struggle to compete with cheaper primes.

As a gadget geek I'm tempted by the idea of a 1,000mm lens. But being realistic about air quality I wonder how often it would actually be able to resolve more detail than my 500mm. I already know that when the air is not good there's not much detail resolving difference at a mile between 300mm and 500mm.

Does anyone shoot for detail at distance using an orange filter? That should help.

 

[Darwinean_JohnUK]

All other things being equal, maximising the size of image capture media is a significant factor.
Large format with best lens captures more detail than Medium format with best lens.
Medium Format with best lens captures more detail than tiny FF with best lens.

 

[chris malcolm]

All other things being equal, maximising the size of image capture media is a significant factor.
Large format with best lens captures more detail than Medium format with best lens.
Medium Format with best lens captures more detail than tiny FF with best lens.

That's an interesting idea. So if I upgraded to Hasselblad, for example, I'd capture more detail with the Hasselblad 1200mm lens. [Google Google.] Oops. Hasselblad's longest lens is only 300mm. Other things aren't equal. What a pity.

 

[sk66]

TC beyond 1.4x often degrade IQ. I have tested this personally, I had a 1.7x TC, and compared images cropped and ones using the TC. Sad to say, the cropped ones won out for me almost every time. That was using the D800E and 300mm F4, with and without the TC. I did like the 1.4x TC though, there's a happy balance somewhere

I agree. I have all three TC's which I use w/ the 400/2.8. The 1.7 is my least used, I find the latest 2x to be at least as sharp.

 

[sk66]

Does anyone shoot for detail at distance using an orange filter? That should help.

Wy should that help?

 

[sk66]

All other things being equal, maximising the size of image capture media is a significant factor.
Large format with best lens captures more detail than Medium format with best lens.
Medium Format with best lens captures more detail than tiny FF with best lens.

This is true. Larger formats have intrinsically greater sharpness due to less physical enlargement requirements.

 

[HoppyUK]

This is true. Larger formats have intrinsically greater sharpness due to less physical enlargement requirements.

And by the same token, a telecon will always reduce sharpness. You're pulling more resolution from the lens, so contrast goes down.

 

[dizidav]

That's an interesting idea. So if I upgraded to Hasselblad, for example, I'd capture more detail with the Hasselblad 1200mm lens. [Google Google.] Oops. Hasselblad's longest lens is only 300mm. Other things aren't equal. What a pity.

Could always go with Pentax they do a 400mm.

 

[sk66]

And by the same token, a telecon will always reduce sharpness. You're pulling more resolution from the lens, so contrast goes down.

IDK... there must be some kind of tradeoff in the system, or rather a lot of tradeoffs. But physical subject distance is king.

This is a rather extreme crop from a D810 image taken with the 400/2.8 w/ 2x TC almost wide open, but near MFD. 800/6.3, 1/250, 560, w/ flash.


SGK_0433-Edit.jpg by Steven Kersting, on Flickr

Full image before cropping

I would never get that level of detail at a significantly longer distance.
If I had used no TC there would probably be even more fine detail, but it would be "smaller," perhaps to the point of being irrelevant/invisible in normal viewing. Or it would require much greater cropping/enlargement and the result would be very similar.
Basically, the choice of using crop factor/cropping, TC's, or longer lenses seems to be a circle of tradeoffs with the net result typically being very similar images. There is no clear/easy answer, other than reducing subject distance.

 

[chris malcolm]

Wy should that help?

An orange filter should help for the same reason that the sky appears blue and the sun yellow, yet the sun is really white and the sky black. It's Rayleigh scattering of of the higher visual wavelengths by the air molecules. As distance increases more blue and green is scattered beyond image forming capabilities and dilutes image contrast. Hence orange sunsets when atmospheric distance is at its maximum. Seabirds tend to have the most need for very long distance vision, and tend to have colour vision which optimises the red and yellow parts of the spectrum.

 

[sk66]

As distance increases more blue and green is scattered beyond image forming capabilities and dilutes image contrast. Hence orange sunsets when atmospheric distance is at its maximum.

Yellow will be more effective at filtering blue... with the obvious influence on other image colors/WB. But at extreme distances the blue wavelengths doesn't even reach the viewer/camera... that's why it appears orange. I suppose at moderate distances it could be a factor, but with the built in UV/IR filtration at the sensor I'm not sure how significant it might be. I think haze (particles) and air disturbances (heat waves) are probably typically more significant.

 

[HoppyUK]

<snip> There is no clear/easy answer, other than reducing subject distance.

Seems fairly obvious at the kind of distances long telephotos are often used at. I often wonder when long lens users say "it's soft at 600mm" whether what they're seeing is just atmospheric pollution. Easy to check if you could maybe print out two identical targets, one half the size of the other, then shoot them with the same lens at say 50m and 100m to compare.

As a general comment on your example pic, birds are popular for showing fine detail but actually they're not always very testing. The detail of the hummingbird looks amazing because we never get close enough to birds to see anything like that, but even the finest feathers there demand way less resolution than say the eyelashes in a head and shoulders portrait.

Orange and red filters are often used for cutting haze in black & white film photography and can be very effective, but they distort tonal relationships like green grass. I would expect that today, an extra dollop of the clarity slider in Lightroom would have a similar effect, and the latest version of Lightroom CC has a new haze reduction feature.

 

[sk66]

I often wonder when long lens users say "it's soft at 600mm" whether what they're seeing is just atmospheric pollution.

IDK... I am rather certain that a lot of the "disappointing results" with long FL's is due to technique... i.e. the ROT for SS is completely useless.

As a general comment on your example pic, birds are popular for showing fine detail but actually they're not always very testing. The detail of the hummingbird looks amazing because we never get close enough to birds to see anything like that, but even the finest feathers there demand way less resolution than say the eyelashes in a head and shoulders portrait.

I get your point, but I'm not sure I can agree when it's a hummingbird... the bird is about 8cm from tip of bill to tip of tail. And there are some of those nearly invisible thin spider web strands in the image as well.

This is a 100% crop of the eye (~4mm)... You can almost make out my silhouette sitting at the table under the umbrella. But, there was certainly even more detail that I did not capture... (I don't think he'd sit still for a macro shot).

 

[HoppyUK]

You're missing the point Steven - the size of the subject is irrelevant, it's the scale of reproduction that is misleading. That hummingbird is a lovely capture for sure, but in critical optical terms it's not actually that demanding. I can also see flare around the head, and CA on the bill that is taking the edge off ultimate image quality that would be more noticeable in a normal subject. Those characteristics are typical of a telecon when you get to look really close.

 

[Jayst84]

This is why you just skip the whole crappy lens problem, and just make shadowgrams. Then all you have to worry about is finding finer grain films, or smoother glass plates.

 

[sk66]

You're missing the point Steven - the size of the subject is irrelevant, it's the scale of reproduction that is misleading.

You're right, I'm not understanding... If the subject is smaller and has smaller/finer details, then it should be "more demanding."

That hummingbird is a lovely capture for sure, but in critical optical terms it's not actually that demanding. I can also see flare around the head, and CA on the bill that is taking the edge off ultimate image quality that would be more noticeable in a normal subject. Those characteristics are typical of a telecon when you get to look really close.

TBH, I'm not sure if the flare/CA is a characteristic of the TC, or rather a characteristic of a prime lens nearly wide open simply magnified by the TC... I suspect it is more the latter. (not sure why I didn't at least deal w/ the CA...must not have been looking this close).

I don't usually photograph things besides birds with such long FL's. But I did find this image of a spider/web taken in early morning fog while hoping to photograph Elk.


And a 100% crop (slightly reduced quality to allow upload)


It's not macro quality, but IMO it's pretty impressive detail when you consider the subject and situation (1/40 SS, ~ 15ft). But it also goes to my point... reducing the distance and using a shorter FL would almost certainly have been better than any other option.

 

[HoppyUK]

You're right, I'm not understanding... If the subject is smaller and has smaller/finer details, then it should be "more demanding."
<snip>

The size of the detail in the subject is not relevant, it's the size of the reproduction on the sensor - in pixels or lines-per-mm terms. But that's by the way, an aside comment I made earlier that isn't directly related to the point in question.

Given that I'm pretty sure you've got the basic technique issues pretty well nailed, I think we agree on the probable causes here - ie, atmospheric pollution, and you may be underestimating the impact of the telecon. Looking at Canon's MTF graphs for the 400/2.8L, adding a 2x telecon knocks 92% MTF contrast at 30-lpmm down to 74%, and that's a very noticeable difference.

In practise, with a closer subject and the bare lens, image quality will be very high, but if you then turn to a more distant subject and bolt on the telecon, sharpness takes a double hit.

 

[sk66]

The size of the detail in the subject is not relevant, it's the size of the reproduction on the sensor - in pixels or lines-per-mm terms.

In which case a shorter FL should (almost) always win... because they have a greater reproduction ratio (other than macro designs).

Given that I'm pretty sure you've got the basic technique issues pretty well nailed, I think we agree on the probable causes here - ie, atmospheric pollution, and you may be underestimating the impact of the telecon. Looking at Canon's MTF graphs for the 400/2.8L, adding a 2x telecon knocks 92% MTF contrast at 30-lpmm down to 74%, and that's a very noticeable difference.

In the first image I don't think the environmental conditions had any impact (other than backlighting/CA). In the second I was shooting through some fog even at 15ft... but I suspect the SS/DOF/technique are the main issues.

I may very well be underestimating the impact of the TC in absolute "bench test" measures... but I don't really care about that. It's not about how much it reduces the IQ, it's about how the remaining IQ compares with other options available. I did bench test (subjectively evaluated) my 800/5.6 against my 400/2.8 w/ TC's when deciding if I was going to keep the 800. Wide open at f/5.6the 800 had an advantage, but it was minimal by f/8. Given that my use generally dictates smaller apertures for DOF I feel that gives the 400/2.8 some distinct advantages (i.e. the ability to go to 400/2.8 and even greater IQ). And given that the practical limitations of shooting at 800mm (moving subjects/SS requirements/etc) tend to mitigate/eliminate any bench test advantages, I found no practical advantage to the 800/5.6... plus it was even bigger. So for me, while the TC does reduce the 400's IQ, it is not a notable loss of IQ comparatively/practically.

I have, of course, also compared using the D8xx in DX mode with lesser/no TC (and other crop bodies as well). And again the realities of using a smaller sensor area with a high pixel density and smaller apertures *generally* delivers subjectively equivalent results in practice. In certain conditions the D810 *can deliver better IQ at the pixel level. But IMHO, it is primarily due to the lower demands on technique/settings that the shorter FL requires. But it's also a smaller image area, and enlargement/display tends to equalize the results. However, if the subject/situation allows more optimal use of the D810 w/ shorter FL's then that is my best choice IMO... but that is very seldom, or other limitations outweigh (i.e. FPS).

I can generally look at an image/settings and know what limited the IQ, but that doesn't mean that I could necessarily have done any better in practical terms...i.e. stopping down farther would introduce other issues, increasing ISO would add noise offsetting any potential IQ increase, etc, etc. And I've spent the money on the best equipment to give me the most advantages... that's what has led me to this post.

When you see really impressive images there is a 99.9% chance that a very significant factor is the image was taken from as short of a distance possible(suitable), with the shortest FL possible, that could deliver the required image/display size. Some of that is technique (fieldcraft/time/patience), some of that is access/situational, and the very least of that is gear... i.e. it would make very nearly zero difference if I used my 80-400/5.6 at 400/8 or my 400/2.8 at 400/8 (or the 800/5.6 vs 400/TC).

I think we can largely leave environmental conditions out of the equation because that is a huge variable that is entirely out of one's control... and, it's the same issue of subject distance.

In a way, none of this is new nor a revelation... I've said/known all of this individually (i.e. lens selection/options, lens vs DX, D810 vs D750, etc, etc). But in a way it is, at least for me. The idea that "FL is only for composition" is a bit of a twist. It basically means that if you can't get close enough to meet your demands/requirements, then don't bother... there's nothing else you can do/buy that will make a significant/practical difference. It means that, while I won't probably stop taking the shots, I should just temper my expectations... and quit spending money on a problem equipment can't fix.

(not that I will, but now I know better )

BTW, there was no "question" in my mind posed in the original post; unless someone can convince me that my observations/conclusions are wrong... I was just sharing for the benefit of others going down this road. Although I *almost* wish someone can tell me where I'm mistaken.