T-Stop, F-Stop and Vignette

[StudentOfLight]

I've been evaluating my new copy of the Tamron 35mm f/1.8 VC and one of the points of differentiation from other lenses in it's class is the peripheral falloff performance. The Tamron promises very even illumination across the image frame.

I decided to compare it to a lens which I know to have very strong vignette, the Canon EF 40mm STM pancake lens. I put the lenses in manual focus set to minimum focus distance and shot images of a plain wall (completely out of focus) The wall was lit by window light from the upper right hand side. I know these are not precise scientific tests but I believe these shots are sufficient for some productive discussion.

I've attached a few comparisons to show the difference in illumination across the image frame. The 40STM appears to be brighter in the centre of frame but only slightly darker in the periphery given the same amount of (theoretically calculated) light while the Tamron produces a very flat illumination across the frame.

The Tamron appears to deliver a similarly bright frame with 0.3EV more light so perhaps has a transmission of about T2.0-2.1. According to DXO-Mark published data, (http://www.dxomark.com/Lenses/Canon/Canon-EF-40mm-F28-STM-mounted-on-Canon-EOS-5D-Mark-II__483) the 40mm STM is T2.8 so does the 40STM overexpose the middle of frame in order to achieve a T-Stop that matches it's F-Stop.

 

[takesome1]

Why not just go to TDP website and look at the comparison that Bryan has put together.

http://www.the-digital-picture.com/Reviews/Lens-Vignetting-Test-Results.aspx?Lens=1003&Camera=453&FLIComp=0&APIComp=0&LensComp=810&CameraComp=453&FLI=0&API=2

 

[StudentOfLight]

Why not just go to TDP website and look at the comparison that Bryan has put together.

http://www.the-digital-picture.com/Reviews/Lens-Vignetting-Test-Results.aspx?Lens=1003&Camera=453&FLIComp=0&APIComp=0&LensComp=810&CameraComp=453&FLI=0&API=2

Thanks, I didn't know he'd done a vignette tool as well, however does anyone know the answer to my question? Do lenses that have a matching the T-stop and F-stop rating overexpose the centre in order to make up for the dark corners or is the F-Stop understated or perhaps some other alternative?

 

[chrysoberyl]

Why not just go to TDP website and look at the comparison that Bryan has put together.

http://www.the-digital-picture.com/Reviews/Lens-Vignetting-Test-Results.aspx?Lens=1003&Camera=453&FLIComp=0&APIComp=0&LensComp=810&CameraComp=453&FLI=0&API=2

Thanks, I didn't know he'd done a vignette tool as well, however does anyone know the answer to my question? Do lenses that have a matching the T-stop and F-stop rating overexpose the centre in order to make up for the dark corners or is the F-Stop understated or perhaps some other alternative?

It is interesting that TDP indicates the opposite of what you have determined. I am much more convinced by your test.

 

[old-pr-pix]

Don't know the direct answer to your question; however, I suspect DxO measurement anomalies may be a part of the suspected problem. Notice that the T-stop rating DxO shows varies based on the body to which the 40 mm STM is mounted. DxO indicates the lens is T 2.8 on a 7D and 5DII, but is T2.9 on 6D, and T3.0 on the 7DII? Maybe someone can explain that. I also checked the 300 f2.8 and it shows similar results: T-3.2 on 7D and 5DII/III but only T-3.5 on 7DII? How does changing the body (same sensor size between 7D & 7DII) affect the amount of light transmitted through a lens? I could understand if the peripheral illumination correction was part of the measurement, but that would really distort what is supposed to be a lens characteristic.

 

[Bdube]

The f stop (and correspondingly, t stop) of a lens is measured on axis. IIRC the formula for T stop is something like F# / sqrt(avg transmission) across a given wavelength bandwidth.

If a lens is f/2 with 2 stops of vignetting, it is genuinely f/4 in the corners. The T/# is not measured in the corner, but the center.

Variance from DxO has to do with different efficiency of the sensors, among other things. The variability and nonrepeatability of the transmissions measurements are a side effect of trying to infer things about the lens from the lens+camera.

 

[old-pr-pix]

Are you saying that the 7DII sensor is actually ~7% less efficient than the 7D (i.e. difference between T-2.8 and T-3.0)? I thought the 7DII was measured to be more efficient? Or, are you saying that however DxO measures the sensor output - rather than optically measuring the lens - and then imputing the T-stop rating is what induces error?

 

[neuroanatomist]

How does changing the body (same sensor size between 7D & 7DII) affect the amount of light transmitted through a lens?

It doesn't. But DxO doesn't directly measure the amount of light transmitted through a lens, they measure the amount of light recorded by a camera sensor. Different cameras have color filter arrays (Bayer masks) with different light transmission characteristics, and those differences affect the way DxO measures transmission.

 

[old-pr-pix]

Got it! Another example of DxO science at work... down rate a lens based on the performance of some other component in the system. Are we still waiting for them to up rate all Canon lenses based on 5DsR performance?