Reading mode: Light Dark

F-stop blues - being duped into fast lenses

Go to last post
I

inasir1971

Guest
Apologies for vague title but wanted to share an article that I came across with everyone. Have seen this issue referenced but just got the full article. Since the D800, I know at least I am using fast primes more than I have before. One of the questions that often comes up is which one - the 1.4 or the 1.8?

An article by DXO labs (Nov 2010) sheds some light (pardon the pun) on how modern sensors do not benefit as much as we think from fast lenses. To compensate for this (so we keep buying expensive fast lenses) manufacturers have been boosting the gain (ISO) at larger apertures so fast lenses seem faster. This gain is unknown to the user - you might think that you are shooting at ISO 100 but the camera sees f1.4 and has cranked up the ISO to 140 to give the illusion that the sensor is getting more light.

Here is a graph (from the article) showing a few models and the gain implemented:





The most egregious offender seems to Sony, but everyone would seem to be doing it. The applied gain is very pronounced faster than f2 for 3 of the above, and by even f2.8 for Sony.

Loss in efficiency seems correlated to pixel size - smaller pixels (higher resolution), benefit less from fast glass. Since that is the direction we are headed (see D3200) this is of relevance.

This issue is not academic and it directly affects shooting and glass choice.
  • selecting faster apertures may be counter productive leading to noisier images (as the additional gain applied would be unknown to the user).
  • faster lenses may not give you the gains that you think you are getting.
This is not arguing against fast lenses - the benefits of better optics, DOF control, and build will still be there, the sensors are simply not able to collect much of the additional light.

The full article: F-stop blues
http://www.dxomark.com/index.php/Publications/DxOMark-Insights/F-stop-blues
 
That's quite interesting.

The other "con" of faster lenses is that often as you get wider and wider apertures, vignetting gets worse. So actually, the difference between f2 with no vignetting and f1.4 with heavy vignetting is probably less than one stop when you average out the whole frame.

--
My travel photography blog - http://www.frescoglobe.com
 
Sorry, but I don't see much value in this article. None of the latest Nikons are represented in the graphs that they offer. Indeed, the graph that you embedded uses the d200 as Nikon's sample. The d200 is 7 years old...... Seems like they could have at least used the d7000 in that graph. What's completely missing is the data on the d3200, d800, d7000 and so on.

Even so, it seems obvious that the newer sensors aren't as bad as the old ones. The data on the d3x seems to demonstrate that trend.

Kerry

--
my gallery of so-so photos
http://www.pbase.com/kerrypierce/root
 
Member said:
"..... fast lenses - the benefits of better optics, DOF control, and build will still be there"
Click to expand...
Member said:
"the sensors are simply not able to collect much of the additional light."
Click to expand...
As pointed out, that article does not include any contemporary Nikon DSLR.
The sensors on the the D$ and D800 are quite different than the D200!

I do know that I like working with "fast glass":

http://forums.dpreview.com/forums/read.asp?forum=1030&message=41736125

Best Regards,

RB

http://www.dpreview.com/members/2305099006/challenges
http://www.pbase.com/rbfresno/profile
 
well maybe another myth ... on the other hand, who knows

Just set your cam to manual mode.

Shoot with f2. Then set f1.4, set the exposure time to 1/2 of the value and voila ...
you should get an equally exposed image.
If not gotcha or they do some very tricky stuff.

Even better use an old non-cpu lens. Then the cam has no idea what's going on and can't compensate with dirty tricks.
 
The main reason I want fast glass is for subject isolation and selection within a chosen DOF.

That's it. And that's why I pay for it.

Robert
 
Someone said test it - so done.

D800 AF-S 35 1.4G; black frame (lens cap on), viewfinder closed
exposure 20"; High ISO NR off; LE NR off
Large JPEGS downsized in View NX

The question is does the camera increase gain without you knowing when large apertures are selected. The test is a simple black frame to look at noise levels, only difference between shots is Aperture is changed between f1.4 and f4.

The black frame noise should be the same as the ISO and exposure are unchanged and it is a black frame - however, they are not.

The f1.4 shots show considerably more noise than the f4 shots. Here are the overlayed histograms for ISO 6400 and ISO 12800.





So the D800 is (like other cameras described in the article) compensating for loss in efficiency at wider apertures by increasing gain to falsely give you the impression that wider apertures are allowing in more light. Test is easily repeatable as above.

The (downsized) ISO 6400 black frames









The (downsized) ISO 12800 black frames









EDIT: typo corrected
 
I can believe that auto-ISO for large apertures, but I don't get this:
Member said:
Loss in efficiency seems correlated to pixel size - smaller pixels (higher resolution), benefit less from fast glass.
Click to expand...
I don't see how the pixel size matters, do you have some academic physics-info about this?
 
artnaz said:
artnaz said:
Loss in efficiency seems correlated to pixel size - smaller pixels (higher resolution), benefit less from fast glass.
Click to expand...
I don't see how the pixel size matters, do you have some academic physics-info about this?
Click to expand...
Digital sensors are not as good at accepting light from angles as film is. They work better when the light comes in more straight on. Large aperture lenses at their max aperture deliver some of the light at a much larger angle (less perpendicular to the sensor). It is, in fact, this steeper angle that gives them a shallower DOF.

Thus, large aperture lenses work less efficiently with digital sensors when used at their max aperture.

Since this angle issue is one of the edges of each pixel and/or the microlenses blocking some of the light that comes in at an angle, it makes reasonable sense that the smaller pixels will block a larger proportion of the light at these angles because they have more edges and a greater proportion of the cell near an edge.

--
John
Gallery: http://jfriend.smugmug.com
Popular: http://jfriend.smugmug.com/popular
Portfolio: http://jfriend.smugmug.com/portfolio
 
gl2k said:
Even better use an old non-cpu lens. Then the cam has no idea what's going on and can't compensate with dirty tricks.
Click to expand...
The camera still knows what's going on. If it didn't how would you possibly be able to meter and auto-expose your image? You're still setting an aperture on the lens and that aperture is fed back to the camera via mechanical coupling. So the camera is certainly capable of applying this gain at wide apertures.
--
Mike Dawson
 
inasir1971 said:
Apologies for vague title but wanted to share an article that I came across with everyone. Have seen this issue referenced but just got the full article. Since the D800, I know at least I am using fast primes more than I have before. One of the questions that often comes up is which one - the 1.4 or the 1.8?
Click to expand...
I am not saying that you are wrong, the phenomena you describe is fairly well known. Luminous Landscape had a lengthy and much debated article on the subject a few years back.

I just want to point out that often is the difference between the f1.8 and the f1.4 versions of a lens also about a lot of other things then the aperture itself. Or to phrase it differently: The price premium you pay for a f1.4 version over a f1.8 version is only partially about the larger aperture in itself. Also, larger apertures are not only about light, they also affect DOF and how out of focus areas are rendered.

But I would strongly recommend reading the article you linked to for all those people arguing Nikon should build f1.2 versions of their 50 and 85 mm lenses like Canon have :)

--
-----------------------------------------------------------
I don't suffer from insanity, I enjoy every moment of it!

By the way, film is not dead.
It just smell funny
 
I'd like to agree with the OP here with my own experience. I was a fan of the 85mm f1.4 D lens back in the days of Tri-X and Kodachrome, and purchased one for my Nikon N90s film camera. I found that even at moderate distances of 10 to 12 feet, the DOF was so shallow that I couldn't reliably nail the focus to shoot at f1.4. But on the occasion that I nailed the focus, the backgrounds were totaly destroyed!

Fast forward to the digital era. That same lens easitly gets the focus correct on my D700. But I feel that the backgrounds are not as smoothed out by "the cream machine"! I've always felt that the pixel wells were so deep that the silicone at the bottom was not seeing the full periphery of that large aperture! I seemed to be getting f/2.8 DOF when my lens was set to f/1.4!!! I've since sold all my film cameras, and consequently cannot do an A-B comparison of DOF between film and digital at f/1.4 on this lens! Can someone on this forum that owns both film and digital bodies do a quick test for me? I'd love to see if I'm truly missing out on the gorgeous bokeh of this fine optic!!! Thank you - Jeff Barlowe
 
Just having a lens cap on is not going to create a black frame if you expose at ISO 6400 + 20s + f/1.4. There is light leakage in such a situation.

Why not expose at 1/8000s and see if you still get an increase in noise?
 
Don't you think that we buy our f/1.4's based on the real-world results they create rather than specs.
 
The only aspect of fast lenses that we are not getting the full benefit from is their greater light gathering ability - the DOF control, optical quality, build quality, remain unaffected.

This isn't my work, I was sharing an article I found that I have found to be at least true with regards to increased gain applied to ISO which I found surprising.

The article linked to says that this issue is related to the microlenses and their reduced efficiency at capturing light from an oblique angle.

I take no issue with this - the fact that sensors might be less efficient at capturing light from fast lenses doesn't bother me, it's part and parcel of technological change.

However, it is important to note that ISO is being affected without our knowledge - you might not want to shoot above a certain ISO because of noise or DR characteristics perhaps. If you're shooting at ISO 6400 and change to f1.4, behind the scene the ISO is increased to somewhere between 8000 and 10000 (I'm guessing).

Perhaps it is done to compensate for the loss in efficiency so 1s @ f1.4 remains equal to 8s @ f4? No idea, not researching this beyond this post - just noting that it does happen.

You can see the change in exposure parameters with the camera as Av is changed. Set 'A' mode, lens cap on, ISO 6400, viewfinder eyepiece closed, LCD illumination off (so no light leak, we're not Canon but still in case), and simply rotate the dial to change Av. Put some tape around the lens cap, viewfinder to eliminate any light from getting in if you want. The exposure does change. With the posted combo, I get:

Av Tv
1.4 0.1s
2.0 0.2s
2.8 0.4s
4.0 0.76s
5.6 1.3s
8.0 2.5s
11 5s
16 10s

Which roughly translates to a doubling of applied gain for each aperture stop - or a linear relationship between the applied gain and aperture area.

This is just an FYI post so we know what the camera is doing, and in case anyone wants to or needs to take this into account when selecting aperture values for shooting.

Fwiw, my old Canon does it too and roughly seems to follow the same relationship as the above - gain applied is proportional to aperture area.
 
inasir1971 said:
The only aspect of fast lenses that we are not getting the full benefit from is their greater light gathering ability - the DOF control, optical quality, build quality, remain unaffected.

This isn't my work, I was sharing an article I found that I have found to be at least true with regards to increased gain applied to ISO which I found surprising.

The article linked to says that this issue is related to the microlenses and their reduced efficiency at capturing light from an oblique angle.

I take no issue with this - the fact that sensors might be less efficient at capturing light from fast lenses doesn't bother me, it's part and parcel of technological change.

However, it is important to note that ISO is being affected without our knowledge - you might not want to shoot above a certain ISO because of noise or DR characteristics perhaps. If you're shooting at ISO 6400 and change to f1.4, behind the scene the ISO is increased to somewhere between 8000 and 10000 (I'm guessing).

Perhaps it is done to compensate for the loss in efficiency so 1s @ f1.4 remains equal to 8s @ f4? No idea, not researching this beyond this post - just noting that it does happen.

You can see the change in exposure parameters with the camera as Av is changed. Set 'A' mode, lens cap on, ISO 6400, viewfinder eyepiece closed, LCD illumination off (so no light leak, we're not Canon but still in case), and simply rotate the dial to change Av. Put some tape around the lens cap, viewfinder to eliminate any light from getting in if you want. The exposure does change. With the posted combo, I get:

Av Tv
1.4 0.1s
2.0 0.2s
2.8 0.4s
4.0 0.76s
5.6 1.3s
8.0 2.5s
11 5s
16 10s

Which roughly translates to a doubling of applied gain for each aperture stop - or a linear relationship between the applied gain and aperture area.

This is just an FYI post so we know what the camera is doing, and in case anyone wants to or needs to take this into account when selecting aperture values for shooting.

Fwiw, my old Canon does it too and roughly seems to follow the same relationship as the above - gain applied is proportional to aperture area.
Click to expand...
It should also be noted that with the 1.4 it renders a higher shutter speed than if you held up a 1.8 lens at the same iso, same scene. So clearly that, along with the dof and rendering benefits, mean 1.4 glass is worthwhlie if required.
 
inasir1971 said:
The only aspect of fast lenses that we are not getting the full benefit from is their greater light gathering ability - the DOF control, optical quality, build quality, remain unaffected.
Click to expand...
Actually, if it is the oblique angle light that is not captured by the sensor, then DOF would be affected and f/1.4 wouldn't render DOF like f/1.4, but probably more like f/1.8 or f/2.0. It is the obliqueness of the angle that creates the narrow DOF. When you don't capture the most oblique light rays, it's similar to stopping down the lens and using a smaller aperture. You should end up with more DOF because of the filtering of the most oblique light rays.
--
John
Gallery: http://jfriend.smugmug.com
Popular: http://jfriend.smugmug.com/popular
Portfolio: http://jfriend.smugmug.com/portfolio
 
You must log in or register to reply here.

Keyboard shortcuts

f
Forum
n
Next
w
Next unread
m
My threads
Mobile site
About
Editorial content
Cameras & Lenses
Community
All content, design, and layout are Copyright © 1998–2025 Digital Photography Review All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
When you use DPReview links to buy products, the site may earn a commission.
©GPS Media - Guides, Products, Services.
Back
Top