Ad
Ad
Ad
Pages: [1] 2 3 4 »   Bottom of Page
Print
Author Topic: Lens Diffraction Test, with Hasselblad HC lenses  (Read 15115 times)
gwhitf
Sr. Member
****
Offline Offline

Posts: 820


« on: June 05, 2009, 06:20:34 PM »
ReplyReply

Today, I tested three Hasselblad HC lenses for this witchcraft term "diffraction" that I've heard thrown around on this forum. I thought it was hogwash.

It is not. Maybe it's more critical with digital than film. I don't know. I'm not a scientist. But I know sharp from soft.

I tested the 35mm HC, the 50mm HC, and the 80mm HC, on an H2 body with a P45+. I tested it tethered, inside, using strobe to eliminate camera shake. Shutter at 800th. Profoto head adjusted each time, dialed down one stop so the light would not have to move.

The range was, in general f8 to f22, in full stops. The files are too large to post here; no use in posting an artifacted JPG, when we're discussing subtleties like this.

But in general, my advice is to NEVER shoot an HC lens any smaller than f11. Even f16 shows noticeable softness, and 22 is a joke, and let's not even talk about 32 or 45.

So boys, instead of just "stopping down and carrying focus", you'd better be learning to shoot focus brackets, and build the layers. No, not good news at all.

If Mr. Grover reads this, meet me on the corner of Main and Sorrow, at high noon, and I'll turn over the bad news files.

My hunch is that this factor affects every brand, although I'm sure there are some differences between brands. My advice: TEST YOUR OWN GEAR. Do not rely on my information, or any other hearsay. Just take this as a warning. Just because it says Zeiss or Planar or whatever on the front, it only means that some of the fstops are sharp. Test your own stuff so that you know your own gear.
« Last Edit: June 05, 2009, 06:23:16 PM by gwhitf » Logged
brianc1959
Jr. Member
**
Offline Offline

Posts: 53


WWW
« Reply #1 on: June 05, 2009, 06:34:32 PM »
ReplyReply

Quote from: gwhitf
But in general, my advice is to NEVER shoot an HC lens any smaller than f11. Even f16 shows noticeable softness, and 22 is a joke, and let's not even talk about 32 or 45.

Your standards might be too high here.  For example, virtually all of the softness you see at f/16 when using sensors with 6 or 6.8 micron pixels can be recovered by proper sharpening.  The reason is that even at f/16 you still have some MTF at the Nyquist limit of the sensor.  So in this case all of the information that the sensor is capable of capturing is in the image, but you need post-processing to make the most of it.
Logged
Ray
Sr. Member
****
Offline Offline

Posts: 8939


« Reply #2 on: June 05, 2009, 07:31:34 PM »
ReplyReply

I'd like to see and 'equal DoF' comparison between the P45+ and the Canon G10. Such a comparison would entail using the G10 at it's sharpest aperture of around F3.5 or F4, and the P45+ at F22 or F25.
Logged
ThierryH
Sr. Member
****
Offline Offline

Posts: 409


« Reply #3 on: June 05, 2009, 09:45:49 PM »
ReplyReply

Yes "George", this witch effect called diffraction affects all brands equally.

Under same f-stop/sensor/pixel size or pitch, there is absolutely no difference concerning diffraction, from a lens to another: once you have reached a certain limit-aperture, diffraction is there and affecting equally all lenses. So there are no differences between brands, concerning this optical effect.

I wonder also why this seems to be new, when one of the main reasons for using a view camera IS diffraction: effectively, a view camera is used to be able to set the sharpness plane the best possible way to reduce the stopping-down (Scheimpflug rule), respectively to change the extension direction of the DOF, and THEREFORE to be able to use the lens at its best f-stop: what does it bring to buy expensive glass, when one simply stops down and enters diffraction in the equation? Better in this case to buy a cheap glass, the effect will be the same. Alternatively the view camera allows to tilt/swing the sharpness plane for special sharpness effects, but one often forgets that diffraction is the main reason of using a view camera, together with perspective corrections.

I have said it many times, f16 is for me the limit when stopping down with a 33/39 MPx sensor, without any noticeable alteration of the image.

Best regards,
Thierry

Quote from: gwhitf
My hunch is that this factor affects every brand, although I'm sure there are some differences between brands. My advice: TEST YOUR OWN GEAR. Do not rely on my information, or any other hearsay. Just take this as a warning. Just because it says Zeiss or Planar or whatever on the front, it only means that some of the fstops are sharp. Test your own stuff so that you know your own gear.
« Last Edit: June 05, 2009, 10:08:28 PM by ThierryH » Logged

Dustbak
Sr. Member
****
Offline Offline

Posts: 2372


« Reply #4 on: June 06, 2009, 01:20:10 AM »
ReplyReply

Naturally you should be careful with stopping down beyond f11/f16. Nothing new you are telling here.

I can tell you that with Schneider Digitars it is even worse. I wonder how that is possible since according Thierry all lenses should be affected equally? I don't think that is entirely the case. These show diffraction starting from f11 and f16 is already total garbage. Much more so than my HC lenses. Even between those there are differences.

Thierry is right this is where movements are for. The DoF is already very thin (we seem to have become much more critical with digital, acceptable sharpness needs to be redefined), even f16 doesn't give you a whole lot so if you need DoF you really need to have movement.

Even with my view camera movements are sometimes not enough. Try a cubus-like item from close-up. You will find you have to compromise somewhere (if you don't want to focus-bracket your way out).
« Last Edit: June 06, 2009, 01:41:44 AM by Dustbak » Logged
Carsten W
Sr. Member
****
Offline Offline

Posts: 599



WWW
« Reply #5 on: June 06, 2009, 02:28:32 AM »
ReplyReply

Quote from: Dustbak
I can tell you that with Schneider Digitars it is even worse. I wonder how that is possible since according Thierry all lenses should be affected equally? I don't think that is entirely the case. These show diffraction starting from f11 and f16 is already total garbage. Much more so than my HC lenses. Even between those there are differences.

All lenses of equal focal length set to an equal aperture are affected equally by the laws of physics. Are you talking about the same focal lengths? The determining factor is the size of the hole, whose diameter is focal_length/aperture. View cameras generally use lenses of greater focal length and so the hole is larger, i.e. less diffraction. The shape of the aperture blade edge might make a difference though, so one lens might differ a touch from another of identical focal length at the same aperture. There shouldn't be huge differences though.
Logged

Carsten W - Recent Photos
tho_mas
Sr. Member
****
Offline Offline

Posts: 1696


« Reply #6 on: June 06, 2009, 03:18:52 AM »
ReplyReply

Quote from: gwhitf
So boys [...] No, not good news at all.
So boy: what's the point for you? You never noticed it. Either you never used f stops smaller than f16 or you don't care about sharpness. Keep shooting as usual - that simple.
And take a look at Helicon Focus...

Quote
If Mr. Grover reads this...
Your dealer didn't comment anything about diffraction?
The jump start of my dealer was the advice to use mirror lock down from 1/60'' and to avoid apertures smaller than f16 (and a side note on moirée)... and to figure out the rest by myself...
« Last Edit: June 06, 2009, 03:54:32 AM by tho_mas » Logged
Dustbak
Sr. Member
****
Offline Offline

Posts: 2372


« Reply #7 on: June 06, 2009, 03:29:54 AM »
ReplyReply

Quote from: carstenw
All lenses of equal focal length set to an equal aperture are affected equally by the laws of physics. Are you talking about the same focal lengths? The determining factor is the size of the hole, whose diameter is focal_length/aperture. View cameras generally use lenses of greater focal length and so the hole is larger, i.e. less diffraction. The shape of the aperture blade edge might make a difference though, so one lens might differ a touch from another of identical focal length at the same aperture. There shouldn't be huge differences though.

Kind of difficult to compare I guess. I have the 120/5.6macro & 90/4.0 of the Schneiders and I mostly use the HC100/2.2. I noticed that the Schneiders I can only use up to f11 max but that is already pushing it. The HC100/2.2 I often use at that aperture and performs much better than either one of the Schneiders at that range of apertures. Anything below f11 the HC lens is no match for the Schneiders with the exception of (software corrected) CA.
« Last Edit: June 06, 2009, 03:31:54 AM by Dustbak » Logged
tho_mas
Sr. Member
****
Offline Offline

Posts: 1696


« Reply #8 on: June 06, 2009, 03:35:14 AM »
ReplyReply

Quote from: brianc1959
virtually all of the softness you see at f/16 when using sensors with 6 or 6.8 micron pixels can be recovered by proper sharpening.
I agree.
Sometimes it's useful to use f16 because the image gets more even. At f11 and especially at f8 the focal plane is incredibly sharp - looks like there is a thin layer of extra sharpening. But only in the center of the image! In the post you push the corner sharpness much more than the center. With f16 the image is not really less sharp at the edges than at f11 - it's just the center that has less contrast. So with f16 you set a more even sharpness and end up with roughly the same as with f11 (but you have more DOF). Well... at leat that's my experience...
« Last Edit: June 06, 2009, 03:39:27 AM by tho_mas » Logged
Carsten W
Sr. Member
****
Offline Offline

Posts: 599



WWW
« Reply #9 on: June 06, 2009, 03:42:06 AM »
ReplyReply

Quote from: Dustbak
Kind of difficult to compare I guess. I have the 120/5.6macro & 90/4.0 of the Schneiders and I mostly use the HC100/2.2. I noticed that the Schneiders I can only use up to f11 max but that is already pushing it. The HC100/2.2 I often use at that aperture and performs much better than either one of the Schneiders at that range of apertures. Anything below f11 the HC lens is no match for the Schneiders with the exception of (software corrected) CA.

Ah, I forgot something: not all lenses are diffraction-limited. In other words, some designs show weakness before real diffraction sets in. It may be that Schneider has made some compromise at small apertures in favour of more sharpness at large apertures. This is a bit beyond my knowledge though.

A second point is that, IIRC from my physics classes, diffraction depends on the shape of the edge, and its "texture". It could be that the Hasselblad HC lenses use better materials and design of aperture blades.
« Last Edit: June 06, 2009, 10:05:51 AM by carstenw » Logged

Carsten W - Recent Photos
ThierryH
Sr. Member
****
Offline Offline

Posts: 409


« Reply #10 on: June 06, 2009, 06:17:01 AM »
ReplyReply

hi Dustback,

I wouldn't know why an aperture of f16 on a Schneider lens (in your case a Digitar) would give more diffraction than any other brand of lens at the same f16. It's simple, the aperture is equal for all lenses, at a certain given f-stop, letting through the same amount of light for a given exposure time, provided that all other parameters are equal. One important parameter, one possibility in your discovery, could be the reproduction scale: one often forgets that the effect of diffraction gets worse with the reproduction scale becoming bigger. In this case, the distance lens to sensor plane increases, respectively the diffraction "circles" created become bigger and more noticeable. In other words, a diffraction effect with a certain f-stop should (is) always be given at a certain reproduction scale, in data sheets, usually at infinity.

When I was speaking about f16 being a limit for 33/39 MPx sensors, I was refering to a reproduction scale of infinity. Taking a scale of e.g. 1:1 (subject = reproduction size), the critical f-stop is MUCH less, respectively 2 f-stops less, and this is in total accordance with the distance lens to sensor (or film), which is then twice the distance as at infinity. Twice the distance means 4 times less light reaching the sensor (film): remember, light decreases by the square of the distance. Which in return brings us to square one: an engraved f-stop of 16 at a reproduction scale of 1:1 is nothing else but a REAL f-stop of 32. And that is certainly "total garbage" as you put it, with a 33/39 MPx sensor.

Now, I am not saying that you were experiencing this at R = 1:1, but it might be the reason.

Best regards,
Thierry

Quote from: Dustbak
I can tell you that with Schneider Digitars it is even worse. I wonder how that is possible since according Thierry all lenses should be affected equally? I don't think that is entirely the case. These show diffraction starting from f11 and f16 is already total garbage. Much more so than my HC lenses. Even between those there are differences.
Logged

ThierryH
Sr. Member
****
Offline Offline

Posts: 409


« Reply #11 on: June 06, 2009, 06:18:58 AM »
ReplyReply

Carsten,

when we are speaking about the focal length, then we automatically introduce the other important factor in diffraction effects: the reproduction scale/distance lens-sensor.

Best regards,
Thierry

Quote from: carstenw
All lenses of equal focal length set to an equal aperture are affected equally by the laws of physics. Are you talking about the same focal lengths? The determining factor is the size of the hole, whose diameter is focal_length/aperture. View cameras generally use lenses of greater focal length and so the hole is larger, i.e. less diffraction. The shape of the aperture blade edge might make a difference though, so one lens might differ a touch from another of identical focal length at the same aperture. There shouldn't be huge differences though.
« Last Edit: June 06, 2009, 06:23:32 AM by ThierryH » Logged

gwhitf
Sr. Member
****
Offline Offline

Posts: 820


« Reply #12 on: June 06, 2009, 06:21:50 AM »
ReplyReply

What's true is, I have no idea about other brands of lenses and whether they'd produce the same surprising results either. My message is for everyone to simply take the time and test their own gear, so they'd KNOW, and not assume. To me, there's no upside to trying to apply USM to an image to try to bring it back sharp again. Just like when you'd throw a loupe on a piece of E6 in the old days, "it's either tack sharp or it's not", and if it's not, it bugs me.

I would hope that the Contax line or Mamiya line would not show this diffraction past f11. What I was surprised at was the AMOUNT of deteriation past f16. I did the test with Sharpening disabled completely in 3.79, and no other USM was applied in Photoshop later. I wanted to see the real data.

Maybe I just feel like an idiot for not doing this detailed test sooner. I plead guilty to seeing the Hasselblad logo and trusting that it'll be sharp. Color me embarrassed. I'd love to do the test with the Canon line of lenses, and on the other end, with lenses on a technical camera like an Alpa or ArTec. Just to see for myself.

There are times, shooting people, or working fast, where it's simply not possible to shoot "focus brackets" in layers, and then also to devote the time for post, to build every image. The whole test was a real wake-up call for me. That's all I'm saying.

Test away.
Logged
ThierryH
Sr. Member
****
Offline Offline

Posts: 409


« Reply #13 on: June 06, 2009, 06:23:04 AM »
ReplyReply

That is absolutely true, though I am tempted to think that most manujfacturers are using the same shape of blades and that even a little difference in this shape should make a difference of 1 stop or more when diffraction is visible.

However, I have no clue here as well, and speaking out of my own experience.

Thierry

Quote from: carstenw
A second point is that, IIRC from my physics classes, diffraction depends on the shape of the edge, and its "texture". It could be that the Hasselblad HC lenses use better materials and design of aperture blades.
Logged

tho_mas
Sr. Member
****
Offline Offline

Posts: 1696


« Reply #14 on: June 06, 2009, 06:35:50 AM »
ReplyReply

Quote from: gwhitf
The whole test was a real wake-up call for me.
So that's a good thing - just be happy!
Quote from: gwhitf
I did the test with Sharpening disabled completely in 3.79, and no other USM was applied in Photoshop later. I wanted to see the real data.
Did you set the sharpening slider to zero or did you disable sharpening in the preferences/menu?
« Last Edit: June 06, 2009, 06:52:00 AM by tho_mas » Logged
eleanorbrown
Sr. Member
****
Offline Offline

Posts: 632


WWW
« Reply #15 on: June 06, 2009, 08:28:40 AM »
ReplyReply

I agree here.  I never ever go above f16 on any of my Hasselblad H prime lenses on my 45+ back or even a lesser resolution back for that matter. Eleanor

Quote from: ThierryH
I have said it many times, f16 is for me the limit when stopping down with a 33/39 MPx sensor, without any noticeable alteration of the image.

Best regards,
Thierry
Logged

Dustbak
Sr. Member
****
Offline Offline

Posts: 2372


« Reply #16 on: June 06, 2009, 08:30:07 AM »
ReplyReply

Quote from: ThierryH
hi Dustback,

I wouldn't know why an aperture of f16 on a Schneider lens (in your case a Digitar) would give more diffraction than any other brand of lens at the same f16. It's simple, the aperture is equal for all lenses, at a certain given f-stop, letting through the same amount of light for a given exposure time, provided that all other parameters are equal. One important parameter, one possibility in your discovery, could be the reproduction scale: one often forgets that the effect of diffraction gets worse with the reproduction scale becoming bigger. In this case, the distance lens to sensor plane increases, respectively the diffraction "circles" created become bigger and more noticeable. In other words, a diffraction effect with a certain f-stop should (is) always be given at a certain reproduction scale, in data sheets, usually at infinity.

When I was speaking about f16 being a limit for 33/39 MPx sensors, I was refering to a reproduction scale of infinity. Taking a scale of e.g. 1:1 (subject = reproduction size), the critical f-stop is MUCH less, respectively 2 f-stops less, and this is in total accordance with the distance lens to sensor (or film), which is then twice the distance as at infinity. Twice the distance means 4 times less light reaching the sensor (film): remember, light decreases by the square of the distance. Which in return brings us to square one: an engraved f-stop of 16 at a reproduction scale of 1:1 is nothing else but a REAL f-stop of 32. And that is certainly "total garbage" as you put it, with a 33/39 MPx sensor.

Now, I am not saying that you were experiencing this at R = 1:1, but it might be the reason.

Best regards,
Thierry

You might be right here. Most of the time I use the Digitars closer than the HC lenses, a lot closer actually. The reason I grab the X-act2 is because I am so close my DoF is not sufficient enough anymore and I need movements. Figures how much closer I am  
Logged
Dick Roadnight
Sr. Member
****
Offline Offline

Posts: 1730


« Reply #17 on: June 06, 2009, 09:18:50 AM »
ReplyReply

Quote
All lenses of equal focal length set to an equal aperture are affected equally by the laws of physics. Are you talking about the same focal lengths? The determining factor is the size of the hole, whose diameter is focal_length/aperture. View cameras generally use lenses of greater focal length and so the hole is larger, i.e. less diffraction.

I can tell you that with Schneider Digitars it is even worse. I wonder how that is possible since according Thierry all lenses should be affected equally? I don't think that is entirely the case. These show diffraction starting from f11 and f16 is already total garbage. Much more so than my HC lenses. Even between those there are differences.

Even with my view camera movements are sometimes not enough. Try a cubus-like item from close-up. You will find you have to compromise somewhere (if you don't want to focus-bracket your way out).

Sharpness is relative, and when using short Schneider Apo-Digitars with 5 micron pixels at f4 to f8, you are looking sharpness beyond what is possible with (most) other lenses: cheap lenses seem to have more DOF because, by apo-digitar standards, they are never sharp.

All the relevant formulae are in Harold Merklinger's "Focussing the View Camera".

DOF merge is often the only way to get good depth of field: digital provides the solutions for the problems it creates (or makes apparent - diffraction was not a problem when the resolution of film was the limiting factor).



Logged

Hasselblad H4, Sinar P3 monorail view camera, Schneider Apo-digitar lenses
Jack Flesher
Sr. Member
****
Offline Offline

Posts: 2595



WWW
« Reply #18 on: June 06, 2009, 09:47:57 AM »
ReplyReply

Quote from: ThierryH
I wouldn't know why an aperture of f16 on a Schneider lens (in your case a Digitar) would give more diffraction than any other brand of lens at the same f16. It's simple, the aperture is equal for all lenses, at a certain given f-stop, letting through the same amount of light for a given exposure time, provided that all other parameters are equal.

I think you actually answered why with the last 7 words above, "provided that all other parameters are equal;"  in most cases, all of the other parameters are not generally going to be equal...

In actuality, the best lenses will show diffraction more readily than lesser lenses; the sharper the lens is, the easier it is to see when and where diffraction anomalies kick in.  (As to why it's more of a problem with digital than film, I suspect it's because we can easily view our direct digital captures at 100%, where with film one would have to have a perfect scan or perfect analog enlargement first to view it, and perfect scans are difficult enough where perfect analog enlargements are rare to non-existent.)  So in this case, I would submit the Schneider Digitars (or Rodenstock HR digitals) being among the best glass available for digital sensors, will show diffraction issues more readily than even the best MF lenses, regardless of brand.  

The other issue is MF lenses have to deal with a fixed point flange-focal for the camera's mounting system, where the Schneider or Rodenstock are not thusly design-constrained.  The single exception may be lenses in the 80mm focal length for MF cameras, as these lenses usually have flange focals very close to the lens focal to begin with, so the best are probably nearly as good as the Schneider/Rodenstock equivalent.  In reality, even 50mm through 120mm MF lenses may be close enough to the system's flange focal that the design constraints are not overly detrimental to the final image quality.

Cheers,
« Last Edit: June 06, 2009, 09:55:08 AM by Jack Flesher » Logged

gwhitf
Sr. Member
****
Offline Offline

Posts: 820


« Reply #19 on: June 06, 2009, 05:17:42 PM »
ReplyReply

I just feel like I've learned a good lesson by doing these tests. I just never in a million years thought it would affect the golden brand, , (I mean Fuji). I thought that bad stuff like this only affected "those other guys that live across town", like Sigma, or Tamron, or the like.

(Actually, I have a Sigma 70mm 2.8 for the Canon, and it's the sharpest lens in my Canon case, so go figure).

Multiple tests so far, and the sweet spot with the 35, 50, and 80 HC are between f5.6 and f11. If resolution and absolute sharpness is the goal. At least these are my findings. Scary, scary territory past f16; only venture into those woods with extreme caution.
« Last Edit: June 06, 2009, 05:18:49 PM by gwhitf » Logged
Pages: [1] 2 3 4 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad