D.P. Explain please :)

CGrieves

CGrieves

CGrieves

Associate
Joined
25 Oct 2002
Posts
2,377
Location
Sarf Lahndahn
From another thread-

Furthermore digital sensors really don't respond wel to ultra fast apertures. there is a strong decline in efficiency as lenses get faster than around f/2.2. So much so that an f1.2 lens doesn't really provide any more light than a f1.4 lens and an f1.8 lens is not really noticible lower.
Click to expand...

Could you shed some more light ( :) ) on this? As I enjoy low-light stuff, I was planning to go for the 50mm 1.2L over the 1.4, thinking it was worth the money for the extra aperture....
 
I would look at the Sigma 50 personally. If not get the L for the reliability.
Wide aperture lenses tend to vignette allot which robs allot of their low light performance.
You notice this when correcting them in Lightroom.

Maybe DP is also referring to the T-stop value of lenses?
All lenses absorb some light instead of passing it on, I haven't checked but maybe wide aperture lenses absorb more.
Still I prefer the extra DOF control.
 
Last edited:
Yep, vignetting, DOF and make/model aside, I'm just intrigued by the suggestion that a f1.4 doesn't let in roughly twice the light of an f1.8... I'd like to know why, otherwise I'll just get another plastic fantastic :)
 
As AE says, just get the Sigma 1.4 and save yourself a bunch of money.

When I was deciding on a 50, I went into WeX and tried them all out (excluding plastic fantastic).

The Sigma edged it over the Canon 1.4 in terms of performance, and to be honest, the L really wasn't worth the slightly better IQ compared to the Sigma IMO because of it's focussing speed (slow)

kd
 
Hehee, thanks chaps, but you're missing the point, my fault for being vague :)

Not bothered about the actual model of lens, I'm interested in D.P.'s assertion that higher apertures don't actually provide measurably more light or better performance. If so then extra money for bigger apertures might be a waste in my case. I'm interested in the reasoning and physics behind it.
 
While a little bit of DP's usual dark magic of optics voodoo science, there is an element of truth, particularly as AE said, due to vignetting. Afaik it has almost nothing to do with digital sensors but there are diminishing returns at this sort of aperture

The F/stop only refers to the ratio between the focal length of the lens and the diameter of the aperture. It doesn't account for how much light is let in at all other than the generally correct rule of thumb that more glass means more light. The accurate measure, or T-stop, as used in cinematography, does a better job of accounting for how much light gets let through. The huge aperture lenses will vignette a lot wide open and as such will be darker - i.e. they don't let as much light through. However you won't see any fast prime that doesn't vignette wide open, so it's a tricky one. An f/1.2 may well let in twice the light of an f/1.8 lens when both are wide open, but the 50 1.2 is brighter at f/1.8 than the 50mm f/1.8 because it's not vignetting as much.

Ultimately the f/stop is more practical for me as it tells you how the focus will be rendered while the t/stop (transmittance) is only useful if you want to be really precise with light levels - something that was only only really useful for the early days of digital (and of course for cinema where DR is a huge factor and concern.
 
http://www.dxomark.com/index.php/Publications/DxOMark-Insights/F-stop-blues

Dxo mark explains it quite well. What happens is the rays of light hitting the sensor at wide apertures are far from perpendicular form the sensor plain. The micro-lens array over the sensor is OK at capturing the light that is slightly off-a is but fails to capture strongly skewed light paths, unlike film. Therefore there is a loss of up to 1 stop of light compared to what the f/1.2 lens should be capable of capturing. Furthermore, the cameras automatically increase the sensor gain (I.e. the ISO) with reporting this to try to mask the behavior. The fore you might think you fancy 85mm f1.2L lens is capturing so much more light but you have really just increased the ISO and under exposed the photo a little to reduce shutter speeds. You can do the same at f1.4 or f1.8 and get much the same benefit.
 
Ksanti said:
While a little bit of DP's usual dark magic of optics voodoo science, there is an element of truth, particularly as AE said, due to vignetting. Afaik it has almost nothing to do with digital sensors but there are diminishing returns at this sort of aperture

The F/stop only refers to the ratio between the focal length of the lens and the diameter of the aperture. It doesn't account for how much light is let in at all other than the generally correct rule of thumb that more glass means more light. The accurate measure, or T-stop, as used in cinematography, does a better job of accounting for how much light gets let through. The huge aperture lenses will vignette a lot wide open and as such will be darker - i.e. they don't let as much light through. However you won't see any fast prime that doesn't vignette wide open, so it's a tricky one. An f/1.2 may well let in twice the light of an f/1.8 lens when both are wide open, but the 50 1.2 is brighter at f/1.8 than the 50mm f/1.8 because it's not vignetting as much.

Ultimately the f/stop is more practical for me as it tells you how the focus will be rendered while the t/stop (transmittance) is only useful if you want to be really precise with light levels - something that was only only really useful for the early days of digital (and of course for cinema where DR is a huge factor and concern.
Click to expand...

It has nothing to do with vignetting, it is to do with the inability the micro lens array to capture strongly off-axis light. The chemical surface on film reacts to any incoming light path, with digital sensors unless the light path is close to perpendicular then the light is lost through specular reflection and other events (photos blocked form entering the electron wells etc)
 
Last edited:
Forget the theory, I am curious going to test it out in my living room in the real world.

Tripod, Manual mode, artificial light (low) so it is constant.

Same settings except the apertures.
 
Last edited:
Raymond Lin said:
Forget the theory, I am curious going to test it out in my living room in the real world.

Tripod, Manual mode, artificial light (low) so it is constant.

Same settings except the apertures.
Click to expand...

Beware that you camera will automatically increase the sensor gain without telling you, so the reported ISO will remain the same but the actual utilized ISO will increase in order to give you the illusion that the shutter speed has decreased. You will have to look at the noise to verify that the ISO has been secretly increased. You should also find slight underexposure when used wide open like that, again resulting in decreased shutter speed.


The results reported by DXO mark are well known and reported elsewhere. Things were much worse with earlier DSLRs who had fairly basic micro-lens arrays. In the DXOmark link you can see camera like the Nikon D70 performing particularity bad because the sensors just couldn't capture a lot of the light form the lens wide open.

It also looks to be strongly affected by the pixel pitch so the 12MP D3/D700 FF cameras are relatively well performing, the 18MP Canon 1.6X crop camera are the worse(in that test, i expect the Nikon/Sony 24Mp sensors to be even worse).
 
Raymond Lin said:
Like I said, I'll see for myself.
Click to expand...

Raymond I take it you didn't actually read the supporting information from the link DP posted?

From what I can gather from the article; camera manufactures program the firmware to simulate perfect scaling of the aperture at it's widest settings.

How the camera does this (again from what I can gather), is by upping the camera's ISO on the sly. You think you are shooting ISO100 F1.2, but in fact you are shooting ISO 160 F1.2 yet you're RAW files still report ISO100.

The point I'm making is if you plan to do a 'real world' test, you won't see a difference in exposure (except due to vignetting) even using manual setting as the camera still compensates for this phenomenon, but there will be more noise.

Will there be noticeably more noise?
Maybe if you are using a sensor and lens combination that loses a full stop.
From ISO100 to 160 on a 5D3? probably not.

To more obviously see this in the real world, perhaps pick an ISO just before the RAW file falls apart. For instance I remember the D700 used to be ok at ISO12800, yet ISO 25000 would be disproportionately degraded.

Edit:

When you think about it. If this phenomenon does occur then camera makers would have to compensate. Otherwise users would complain about their camera's being broken, similar to how people complain about the Tamron 24-70 F2.8 to F4 bug in some affected samples.
 
Last edited:
Surely if the camera compensates without actually showing it in ISO then the images will look to have the same even exposure.

Not one is a stop darker yet more grainy. Then what was the point of upping the ISO on the sly?

And no, I seldom read the link regarding technical stuff you and DP links to, especially DXO. No offence, saw too much of that site last year, putting those 3 letters in the same order makes me want to puke. Lol

What I want to find out is if there really is no difference (I am hoping there is) between F/1.2 and F/1.8, up to F/2.2 (as he stated earlier). That is over 1 stop of light.

If there is no difference then your ISO bumping on the sly will partly explain it, I get that. The DXO link would explain it even further, I don't need to read it, I'd take your word on it.

However, if there is visually difference not only between 1.2 to 2.2, but each of the increment all then all this ISO bumping is irrelevant since why would it be darker yet higher ISO?

So I am not looking at how grainy it is, that part is not the point I am trying to prove, nor will it matter if each photo gets darker as I up the aperture.
 
Last edited:
I wonder if you could use some kind of doctored lens or a lens with some kind of adapter where the aperture could be set manually on the lens and the camera would have no idea what it was. Suppose that's the only way you could perform a totally objective test, without any risk of the camera "fiddling" things on the sly.
 
Raymond Lin said:
Surely if the camera compensates without actually showing it in ISO then the images will look to have the same even exposure.

Not one is a stop darker yet more grainy. Then what was the point of upping the ISO on the sly?

And no, I seldom read the link regarding technical stuff you and DP links to, especially DXO. No offence, saw too much of that site last year, putting those 3 letters in the same order makes me want to puke. Lol

What I want to find out is if there really is no difference (I am hoping there is) between F/1.2 and F/1.8, up to F/2.2 (as he stated earlier). That is over 1 stop of light.

If there is no difference then your ISO bumping on the sly will partly explain it, I get that. The DXO link would explain it even further, I don't need to read it, I'd take your word on it.

However, if there is visually difference not only between 1.2 to 2.2, but each of the increment all then all this ISO bumping is irrelevant since why would it be darker yet higher ISO?

So I am not looking at how grainy it is, that part is not the point I am trying to prove, nor will it matter if each photo gets darker as I up the aperture.
Click to expand...

Perhaps use the vignette profile in Lightroom for both aperture settings. Then compare the exposure. I'm assuming they will be very close if not the same.
 
I would let the pictures do the talking and let you guys make your minds up. EXIF intact.

1.2
x2hHO0E.jpg.png

1.4
VElbZJm.jpg

1.6
VNfTHU4.jpg.png

1.8
FMv88gn.jpg.png

2.0
QtSdrtd.jpg.png

2.2
1syKZgn.jpg.png


Again

1.2
vdobWz8.jpg.png

1.4
LVb1DLr.jpg.png

1.6
JjqOYMk.jpg.png

1.8
goZtybs.jpg.png

2.0
U3XHi4R.jpg.png

2.2
EaGtc6Y.jpg.png
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom