Optical diffraction

[Digifrog]

Hello

Ok following on from the D800 thread. This isn't about the D800 but concerns lens diffraction which it seems has been amplified by the D800's 36mp sensor. Some estimate the diffraction limit will be about f/9.So this thread is about that not the camera.

Anyway, not to be put off by this I have all sorts of information and numbers swirling around in my head and tbh it's beginning to hurt a little. I would really appreciate some clarification and advice. It's probably been asked before but may I continue.

See, I want the D800 and more importantly I have been looking at what glass to get for it. There are presently three choices. They are...

1. Nikon 16-35mm f/4 VR
2. Zeiss 18mm Distagon T*
3. Zeiss 21mm Distagon T*

I came across this review video for the 18mm which is hugely positive...
http://www.youtube.com/watch?v=Fz1hXZpuP3U

.. the video states somehere that all you have to do is select f/8, set f/8 to infinity on the focus ring and everything will be sharp, especially as f/8 is well inside the D800's suspected f/8 diffraction limit. OK, this is standard hyperfocal procedure but the DoF calculators confuse me with subject distance blah blah blah, I shoot landscapes so the whole view is the subject.

What I need to know is and what I would like is to just know what distance from A to B (infinity) will be sharp for the above 3 lenses. Could I ask for someone to replace the x's with numbers please? This way I will be far more at ease.

1. Zeiss 18mm, aperture f/8 & f/8 set to Infinity = x.xx ft (x.xx M) - Infinity will be sharp.

2. Zeiss 21mm, aperture f/8 & f/8 set to Infinity = x.xx ft (x.xx M) - Infinity will be sharp.

3. Nikon 16-35mm (16mm), aperture f/8 & f/8 set to Infinity = x.xx ft (x.xx M) - Infinity will be sharp.

4. Nikon 16-35mm (24mm), aperture f/8 & f/8 set to Infinity = x.xx ft (x.xx M) - Infinity will be sharp.

(How do you set f/8 to infinity on a zoom anyow?)


I'm sorry for my ramblings. I've never used manual focus before and I'm not a techy kind of person.

Thank you.

Scott

 

[tdodd]

You seem to be quite confused. f/8 is the aperture and f/8 is f/8. It has nothing at all to do with the distance at which you choose to focus. If you want to focus at infinity then focus at infinity. A zoom lens won't stop you or make it any easier or harder than a prime lens. Focal length also makes no difference. Infinity is still infinity, and it is a distance, not an aperture value.

Taking your first example, what I think you mean to say is....

1. Zeiss 18mm, aperture f/8 & FOCUS DISTANCE set to Infinity = x.xx ft (x.xx M) - Infinity will be sharp.

As for filling in your x.xx answers, visit this website - http://www.dofmaster.com/dofjs.html - and input the values you know, in order to be left with the x.xx answers you seek. For the focal lengths discussed, and with aperture at f/8, you will find that setting a focus distance of 1,000m will serve perfectly well as a figure for "infinity", but you can make it 10,000m or 1000,000m (or feet) if you prefer.

Finally, in case you didn't already know, while f/9 might be the DLA for the D800, in practice that will only begin to become apparent when viewing at 100% on a monitor. For normal sized prints or display sizes the D800 will perform just the same as any other full frame body with respect to diffraction at any given size of enlargement (print). You really shouldn't be afraid to use apertures up to f/16 or even f/22 if that's what you need - for "normal" usage scenarios.

 

[Digifrog]

Thank you Tim

I know f/8 is the aperture (thanks, I hope I'm not that naive). I am referring to (and sorry if I haven't made it clear enough) is hyperfocal distancing on a manual lens with f/8 selected and lining up (f/)8 to the Infinity symbol on the focus distancing ring (as explained in this first step here http://www.youtube.com/watch?v=hy-7aGSDGh8).

I this not the correct procedure for hyperfocal distancing?

Regards

 

[Suvv]

Use the link that Tim gave. Input your camera, the focal length and the aperture (f8), ignore the Subject Distance. Read the hyperfocal distance at the bottom right - everything at half this distance to infinity will be in focus if you focus at the hyperfocal distance. By lining up your infinity mark to the f8 mark you are automatically setting the focusing to the hyperfocal distance.

Or the easy way - if you can set infinity on your focusing ring to an f8 mark there should be another f8 mark, just read off what distance is shown opposite that mark to get the near limit

 

[tdodd]

Ah, OK, sorry. None of my nine Canon lenses have DOF markings whether prime, zoom, L or non-L, so I'm not familiar with the technique. I had no idea that you were talking about lining up f/8 with DOF markings on the lens. I'm afraid I didn't watch the video. I misunderstood what you were saying in your calculation examples. My bad.

I do have some very old third party lenses for my 35mm Pentax film camera. They do have such markings, even on the zooms, but I haven't used them for over 20 years. The markings are similar to those shown here - http://www.northlight-images.co.uk/article_pages/hyperfocal_distance.html - and illustrate how DOF alters over the focal length range.

I hope you get the answers you need.

EDIT : Oh yes, and now I've actually looked at the markings properly I agree with Kev. The problem is that you probably won't get a lot of precision in the answer to x.xx from looking at the lens barrel. Maybe that will be important. Maybe not.

 

[Digifrog]

Use the link that Tim gave. Input your camera, the focal length and the aperture (f8), ignore the Subject Distance. Read the hyperfocal distance at the bottom right - everything at half this distance to infinity will be in focus if you focus at the hyperfocal distance. By lining up your infinity mark to the f8 mark you are automatically setting the focusing to the hyperfocal distance.

Or the easy way - if you can set infinity on your focusing ring to an f8 mark there should be another f8 mark, just read off what distance is shown opposite that mark to get the near limit

Ah, OK, sorry. None of my nine Canon lenses have DOF markings whether prime, zoom, L or non-L, so I'm not familiar with the technique. I had no idea that you were talking about lining up f/8 with DOF markings on the lens. I'm afraid I didn't watch the video. I misunderstood what you were saying in your calculation examples. My bad.

I do have some very old third party lenses for my 35mm Pentax film camera. They do have such markings, even on the zooms, but I haven't used them for over 20 years. The markings are similar to those shown here - http://www.northlight-images.co.uk/article_pages/hyperfocal_distance.html - and illustrate how DOF alters over the focal length range.

I hope you get the answers you need.

EDIT : Oh yes, and now I've actually looked at the markings properly I agree with Kev. The problem is that you probably won't get a lot of precision in the answer to x.xx from looking at the lens barrel. Maybe that will be important. Maybe not.

Thanks Kev, Tim

Aye I've seen the DoF calculator.

I want to potentially marry the D800 with a Zeiss WA which does have those markings. For me it's a case of just wanting to line up the say the 16 on the focussing ring (if my chosen aperture is f/16) to the infinity symbol and just shoot away knowing that everything from 4ft to infinity will be in focus (on a 21mm lens), as in no need to focus on any particular point in the frame.

I'll give the calculator a study again, though the camera in question isn't listed yet.

 

[tdodd]

You don't need the camera to be listed. You just need to pick a full frame body, as it's the sensor size that drives the maths, not the pixel count. A D700 would be a perfect substitute. But equally a D3, D3s or 5D2 will get the job done.

 

[HoppyUK]

Hello

Ok following on from the D800 thread. This isn't about the D800 but concerns lens diffraction which it seems has been amplified by the D800's 36mp sensor. Some estimate the diffraction limit will be about f/9.So this thread is about that not the camera.

Anyway, not to be put off by this I have all sorts of information and numbers swirling around in my head and tbh it's beginning to hurt a little. I would really appreciate some clarification and advice. It's probably been asked before but may I continue.

See, I want the D800 and more importantly I have been looking at what glass to get for it. There are presently three choices. They are...

1. Nikon 16-35mm f/4 VR
2. Zeiss 18mm Distagon T*
3. Zeiss 21mm Distagon T*

I came across this review video for the 18mm which is hugely positive...
http://www.youtube.com/watch?v=Fz1hXZpuP3U

.. the video states somehere that all you have to do is select f/8, set f/8 to infinity on the focus ring and everything will be sharp, especially as f/8 is well inside the D800's suspected f/8 diffraction limit. OK, this is standard hyperfocal procedure but the DoF calculators confuse me with subject distance blah blah blah, I shoot landscapes so the whole view is the subject.

What I need to know is and what I would like is to just know what distance from A to B (infinity) will be sharp for the above 3 lenses. Could I ask for someone to replace the x's with numbers please? This way I will be far more at ease.

1. Zeiss 18mm, aperture f/8 & f/8 set to Infinity = x.xx ft (x.xx M) - Infinity will be sharp.

2. Zeiss 21mm, aperture f/8 & f/8 set to Infinity = x.xx ft (x.xx M) - Infinity will be sharp.

3. Nikon 16-35mm (16mm), aperture f/8 & f/8 set to Infinity = x.xx ft (x.xx M) - Infinity will be sharp.

4. Nikon 16-35mm (24mm), aperture f/8 & f/8 set to Infinity = x.xx ft (x.xx M) - Infinity will be sharp.

(How do you set f/8 to infinity on a zoom anyow?)


I'm sorry for my ramblings. I've never used manual focus before and I'm not a techy kind of person.

Thank you.

Scott

Scott, this won't stop your head spinning but you are confusing several things here.

Lens diffraction is not 'amplified' by the D800's 36mp sensor, because it all happens before the image even gets there. It just means that if the sharpness level is capped by diffraction at a level below the D800's theoretical maximum resolution, you will not see the benefit of 36mp.

Hyperfocal distance and depth of field (some recent threads on this if you seach against my user name) work to a entirely different standard, and it's way below 36mp. Which kind of makes the point about how hard it will be to get the most out of a D800.

I don't think there is a camera ever been made that can't comfortably exceed the sharpness standard for DoF - which is 0.2mm on a 10in print, when viewed from 12in. To transalte that into pixels, it equates to only 1.1mp. And if you are using HFD technique, then by definition you are limiting the sharpness of foreground and infinity subjects to that level of resolution.

Sharpness is not just about resolution, it's about resolution and contrast. In fact, contrast is the major contributor to visual acuity when viewing any image in any kind of normal way. That's what you're paying for with nice Zeiss glass and L-grade lenses or whatever, and the extra resolution you get at the same time is more or less a handy by-product. You can only see it in huge enlargements viewed very close, like 100% on screen.

Check this link and scroll down any further than the two bar charts near the top. They both have the same resolution, but the one on the left looks far sharper because it has much higher contrast. http://www.cambridgeincolour.com/tutorials/lens-quality-mtf-resolution.htm

If you want the sharpest images possible this side of medium format, get a D800. Use the best lenses you can find, such as the tilt & shift option mentioned in my PMs, preferably around f/8, though the extreme corners will probably be getting better even at f/11, regardless of what diffraction is doing to the rest of the image. 18mm is too extreme for optical perfection.

 

[Digifrog]

Scott, this won't stop your head spinning but you are confusing several things here.

Lens diffraction is not 'amplified' by the D800's 36mp sensor, because it all happens before the image even gets there. It just means that if the sharpness level is capped by diffraction at a level below the D800's theoretical maximum resolution, you will not see the benefit of 36mp.

Hyperfocal distance and depth of field (some recent threads on this if you seach against my user name) work to a entirely different standard, and it's way below 36mp. Which kind of makes the point about how hard it will be to get the most out of a D800.

I don't think there is a camera ever been made that can't comfortably exceed the sharpness standard for DoF - which is 0.2mm on a 10in print, when viewed from 12in. To transalte that into pixels, it equates to only 1.1mp. And if you are using HFD technique, then by definition you are limiting the sharpness of foreground and infinity subjects to that level of resolution.

Sharpness is not just about resolution, it's about resolution and contrast. In fact, contrast is the major contributor to visual acuity when viewing any image in any kind of normal way. That's what you're paying for with nice Zeiss glass and L-grade lenses or whatever, and the extra resolution you get at the same time is more or less a handy by-product. You can only see it in huge enlargements viewed very close, like 100% on screen.

Check this link and scroll down any further than the two bar charts near the top. They both have the same resolution, but the one on the left looks far sharper because it has much higher contrast. http://www.cambridgeincolour.com/tutorials/lens-quality-mtf-resolution.htm

If you want the sharpest images possible this side of medium format, get a D800. Use the best lenses you can find, such as the tilt & shift option mentioned in my PMs, preferably around f/8, though the extreme corners will probably be getting better even at f/11, regardless of what diffraction is doing to the rest of the image. 18mm is too extreme for optical perfection.

Hi Rich

Ah sorry, forget I used the word "lens" when referring to diffraction. I don't know why I typed "lens" there in my first sentence. I am of course referring to diffraction limitation of a camera given its sensor size & pixel density and in the D800's case approx. f/9 is its diffraction limit.

The thread was started after reading DPNow's blurb on the subject...
http://dpnow.com/forum2/blog.php?b=296
...I didn't know before reading this that all cameras have their own diffraction limits (my D80 for example is f/11). So the D800 is more diffraction limiting than other cameras due to it's 36.3MP...
http://dpnow.com/7644.html


My OP was really to gauge distances when using hyperfocal distancing given a specific lens & aperture (I used f/8 as an example due to peak sharpness but of course limiting DoF), my main premise for HD.

What, apart from obvious distortion, would 18mm affect optical perfection? I'd have thought being decent enough glass (Zeiss) & shooting landscapes/seascapes it wouldn't matter so much (slightly curved horizons are easily fixed) plus it gives extra DoF over the 21mm. I used to shoot with the Sigma 10-20 and I loved its perspective creating view. Not so great for keeping verticals granted but I love wide. As long as the image is sharp front to back I don't mind.

 

[HoppyUK]

Scott, I'll not be bookmarking the site you linked

Pixel related diffraction is a theory that, as I have said, I don't have much time for. It doesn't stack up in practise, not to any significant effect anyway. The fact is that if you have more pixels, and a good enough lens, you do at least have a chance of getting more detailed pictures.

Optical diffraction is different. It exists, and it's real. On the site you linked, the example given is taken on an Olympus Four Thirds camera at f/22, which is equivalent to f/45 in full-frame terms, so no wonder it looks poor!

Try this example. Let's say you have a typical f/2.8 lens, and at that aperture diffraction limits its potential sharpness to which we'll assign a value of 10. But lenses are never at their best at lowest f/number, and at f/2.8 the lens is only capable of a sharpness value of 4 due to other uncorrected aberrations.

The lens gets sharper as the f/number is raised, and by f/4 it has risen to 5. Meanwhile, the diffraction limit has simultaneously lowered, but it's still high at a value of 9, so overall the lens is still limited to 5.

Closing down to f/5.6, and the diffraction ceiling is getting lower, it's now at 8, while other aberrations continue to get better, now at 6, so the cross-over point is getting close.

At f/8, diffraction has dropped potential sharpness to 7, while other aberrations are improved and they are now also at 7 - the cross-over point, maximum sharpness peaks at f/8, with a value of 7.

Thereafter, even though other apsects of lens performance are still getting better, diffraction continues to increase and now dominates, so at f/11 diffraction will cap sharpness at maybe 6, and at f/16 it will drop to say 5. All this is independent of the sensor.

Now assume we have an exceptionally good lens. At f/2.8 it is already up to a sharpness value of 6, with diffraction at 10. At f/4, it's jumped to 7 and diffraction has only capped sharpness at 9. At f/5.6, other lens aberration are up to 8, and diffraction is now down to 8. So this better lens now peaks earlier, and at a higher level - at f/5.6, with a sharpness value of 8.

So you can see that diffraction always puts an overriding cap on maximum sharpness, which means that if you have a very high resolution sensor that is capable of resolving more detail than the lens can muster, you're not going to see it. And if you are going to make best use of the D800's 36mp, then it will only become visible with an exceptionally good lens, at lower f/numbers before diffraction becomes a limiting factor.

As a rough guess, in reality you will probably get the best result out of it at f/5.6, lens willing, whereas maybe a D700 might peak a smidge later at f/8, but the peak will be slightly lower. At f/8, the D800 will have fallen to the same level, because both are now diffraction capped to the same standard.

In other words, you won't lose anything because of the D800's high pixel count, but you won't gain anything either unless you bear in mind the diffraction limit and stay below it.

Now, that 18mm Zeiss lens. Zeiss lenses are made by Cosina, and very good they are too (mostly) but I mention it because of the rose-tinted spectacles effect of certain West German brands.

18mm is extremely wide on full frame, also retrofocus design, and that's a tough call. It will be significantly sharper in the centre at all apertures, and the extreme corners might not be properly sharp until f/11 or so. Super-wides are just like that, and the only lenses that get truly super-sharp from corner to corner are much longer, more like 200mm-plus and stuffed full of exotic glass like fluorite, and cost thousands.

The best wide lenses for landscape are T&S, because they are very highly corrected (and expensive) and you can use them at lower f/numbers and still get enough depth of field, while staying below the diffraction threshold.

 

[vizzair]

Also note the D7000 has a pixel density slighty greater than the D800 ( 38Mb FX equivalent) , and that has been in service enough for people to notice if it made that big a difference.

 

[woof woof]

Ok following on from the D800 thread. This isn't about the D800 but concerns lens diffraction which it seems has been amplified by the D800's 36mp sensor. Some estimate the diffraction limit will be about f/9.So this thread is about that not the camera.

I'd just forget about diffraction. Well, put it to the back of your mind anyway.

I believe it's a property of light passing an obstruction and I have difficulty factoring in how a high mp count sensor can therefore be dramatically worse than a low mp count camera for most peoples shots either in print or on screen.

I believe that if you look close enough you'll see diffraction at wider apertures when using higher mp count cameras but I believe this is because the higher mp count allows you to see it whereas a lower mp count may hide it. At equal print and screen sizes I doubt that diffraction will be a dramatically more serious issue with 36mp than it would be with 24mp. I could be wrong but I personally would ignore it until proved wrong to do so.

I generally use wider apertures than smaller ones and tend to use the widest I think will do the job but I was shooting some model cars the other day with a G1 (micro four thirds) and even at f16 when viewed at 100% they look good to me. So I ignore it

 

[hollis_f]

I'd just forget about diffraction. Well, put it to the back of your mind anyway.

I believe it's a property of light passing an obstruction and I have difficulty factoring in how a high mp count sensor can therefore be dramatically worse than a low mp count camera for most peoples shots either in print or on screen.

It can't produce a worse image. It can show up the effects of diffraction before a low MP sensor.

Diffraction happens whenever light passes through a hole, causing a single point of light to get spread out into a blob (a complicated blob). The smaller the hole the bigger the blob.

If you've got huge big pixels then, with a certain sized hole, the blob will just cover a single pixel. But if you've got lots of teeny-tiny pixels then the blob will cover many pixels.

If you print the image then the blob will be the same size on both images (assuming the sensors are the same size). But if you zoom in to look at individual pixels then you'll see the blob covering many pixels on the sensor with little pixels - so you might (if you hadn't thought about it) think that it was giving a worse result.

 

[woof woof]

It can't produce a worse image. It can show up the effects of diffraction before a low MP sensor..

Yup. I think that's what I said. But viewed at the same size/distance the higher mp count image will probably look ok. This is why looking at images at 100% or more is probably not a brilliant idea. Look at a 6mp image at 100% sensible print resolution and size and you're looking at a good sized image, but you'll be able to take it all in. Look at a 36mp image at 100% sensible print resolution and size and you're looking at a much bigger picture. Now imagine a camera with millions and millions and millions of mp and imagine that we had perfect lenses that could match the resolution and for a moment imagine that diffraction isn't an issue, point this wonderful piece of kit at someone and look at the resultant image at 100% (imagine that we've got mega screens and printers that can display this image) and you'll see individual cells. This is obviously a silly thing to do as what you wanted to see was a model in a bathing suit. My point is that as the mp count and resolution rises we should think more about how an image is going to be printed and viewed and if we are going to produce a whole image at A3 it's really not helping things if we examine it as if it's going to be the size of a house and then agonise over a few fuzzy pixels.

Diffraction happens whenever light passes through a hole, causing a single point of light to get spread out into a blob (a complicated blob). The smaller the hole the bigger the blob.

Is what I said. Light has a duality, sometimes it acts like a particle and sometimes it acts like a wave. Diffraction affects light when it acts like a wave although to be fair some would say that all "particles" have wave like properties. Other people would argue that there are no particles. For now though and for simplicities sake it's easier to think of diffraction as a property of light when acting as a wave. It's a sort of interference pattern.

If you've got huge big pixels then, with a certain sized hole, the blob will just cover a single pixel. But if you've got lots of teeny-tiny pixels then the blob will cover many pixels.

If you print the image then the blob will be the same size on both images (assuming the sensors are the same size). But if you zoom in to look at individual pixels then you'll see the blob covering many pixels on the sensor with little pixels - so you might (if you hadn't thought about it) think that it was giving a worse result.

That's pretty much what I meant. A high mp count won't cause diffraction but it will make it more visible if you look closely enough. Although at extremes you wont have to look too closely, you'll see a lack of contrast and some "fuzziness" but this could be mitigated to some extent in post capture processing. The issue of pixels being close together and light hitting two or more instead of one and thus causing a "fuzzy" image is something for designers to think about but I'm not sure that this issue should be called diffraction as such. The diffraction of light as it hits the sensor would be exactly the same regardless of the mp count but if the pixels are closer together you'll see it because the sensor is capable of showing it, if you look closely. Look at it sensibly and will you still see it? That's up to each of us to decide.

I don't know what magnification is sensible when looking at a 30+mp image. Lets imagine that 50% is the new 100%. At 100% it may well look naff but at 50% it might just look lovely.

Model car image, G1 (micro four thirds) @f16, 40 second exposure, 100% crop. Although this is only a crappy web image via photobucket I hope it's possible to see that there's lots of detail and diffraction isn't a real world issue in this image although some would expect that @16 on MFT will be awful it's good enough for me.

If I was in the market for a new Nikon I don't think that diffraction would be a worry for me. I would however shoot some test shots and view them to decide what aperture was an issue at what image size and what happens once it's been processed. Once armed... on with the shooting.

 

[hollis_f]

I'm on my iPad,which hasn't got an Exif viewer, so these may not be quite correct.

f/5.6

f/32

Yes, there is a discernible difference in these 100% crops. Would I let it worry me? Nope.


Edit: To correct aperture for first shot.