Is it T Stops not F Stops we should care about?

[MonsMeg]

I should start by saying if this has been discussed here elsewhere please send me to the thread, but an initial search didn't throw anything up. My question is really, is anyone aware of any broad standardised test of T Stops across lenses? It's not meant to be an idle question, I've been to the DxO site which gives the scores for quite a few lenses (although not obviously mapped to EV unless it's a 1:1 mapping?). I'll attach a screen grab which shows (I think) that the Canon EF 24-70mm f2.8L ii USM and the Canon EF 24-70mm f2.8L USM seem to have 1/2EV of different amounts of light hitting the sensor (which seems like quite a lot). In other words (I think) one is letting in 1/2 a stop of light "more" to the sensor even though they have the same (and constant) F2.8 aperture. (DxO have also helpfully attached the same camera to both lenses). The second question relates to any real world experiences people may have had. For any given aperture primes are supposed to be better for light transmission than zooms for any given F Stop (less glass), and I've noted how long zooms seem to struggle with low light more than they "should" given their stated aperture.

 

[Joules]

I guess you are talking about stills. If so, unless you are deciding between two purchasing options and all other factors (focal length, rendering, weight, price,...) are not enough to make a decision, I don't know what would be the point of stressing over T stop.

Say it were a full stop of light your missing out on... That's still just one stop. Would it be nice to have that stop? Surely. But a stop of light doesn't make or break an image. And the difference in T stop is far less than a whole stop typically as far as I'm aware.

 

[privatebydesign]

T-stops really don’t make any difference in stills captures, they end up being a side show that gets people here all worked up every now and again.

The really important things for still lenses are focal length and aperture, then probably resolution/sharpness, focus speed and accuracy, image stabilization etc etc.

T stops are very relevant for people shooting video especially when they are shooting the same scene with multiple cameras and for ease of editing and look want the exposures to be the same across those cameras. However given the power of post processing and the inevitable grading the differences are nowhere near as important as they were when ‘movies’ were shot on film.

But back to stills, when comparing lenses for still photography the key issue is depth of field control, t stops don’t tell you anything about that, you could have a very bad transmission lens with a narrow depth of field or a very good transmission lens with a smaller aperture both with the same t value.

 

[AlanF]

There is a little bit of information in the T-stops, but not enough to get het up about. Here is a comparison of zooms and primes at 400mm, with the difference in stops calculated from the T numbers and f-stop:
400mm DO II T=5 = - 0.64 stops
100-400mm II (400mm) T= 6.3 = - 0.34 stops
200-400mm (400mm) T= 4.7 = -0.46 stops
400mm f/2.8 II T= 3.1 = -0.3 stops

A couple of take away points are:
1. The 400mm DO II is in practice only 2/3 stops faster than the 100-400mm II and not 1 stop as the DO optics are not quite as efficient as refractive.
2. The long zooms have similar losses in stops to the prime.

I took point 1 into consideration when deciding to sell my 400mm DO II and keep the 100-400mm II.

 

[MonsMeg]

I guess you are talking about stills. If so, unless you are deciding between two purchasing options and all other factors (focal length, rendering, weight, price,...) are not enough to make a decision, I don't know what would be the point of stressing over T stop.

Say it were a full stop of light your missing out on... That's still just one stop. Would it be nice to have that stop? Surely. But a stop of light doesn't make or break an image. And the difference in T stop is far less than a whole stop typically as far as I'm aware.

Thanks Joules, appreciated. Yep, stills. I'm asking for a small group of wildlife photographers one of which has a Sigma 150-600mm f5.6-6.3 and another who is considering this lens vs an older used Canon 600mm f4. The debate has ranged over all the usual topics (not least price), but the one interesting thing we noticed when out shooting at low light which is when we do most wildlife was that the Sigma behaved less like a f6.3 and more like a f7 or f8 (in terms of the iso that was being used by the camera vs mine with the prime). DxO says the Sigma has a T6.7, and that's an average apparently over the range of the zoom according to DxO methodology? Possibly when she's at the full 600mm the T is higher again, maybe? And she's mostly at 600mm. I agree that consensus is it's marginal, just it was a bit of a surprise as it didn't feel that marginal to her.

 

[Sporgon]

I’d say not for stills. A few years ago I noticed that one of my favourite lenses had a T stop quite a bit slower than its fully open f/1.8 stop according to DXO. Another one, the EF 35 f/2 IS had a T stop the same as it’s f stop. So I thought this should be very easy to see with both lenses set to f/2 and the same exposure in raw with no cooking. Result: no difference in output exposure at all.

 

[MonsMeg]

There is a little bit of information in the T-stops, but not enough to get het up about. Here is a comparison of zooms and primes at 400mm, with the difference in stops calculated from the T numbers and f-stop:
400mm DO II T=5 = - 0.64 stops
100-400mm II (400mm) T= 6.3 = - 0.34 stops
200-400mm (400mm) T= 4.7 = -0.46 stops
400mm f/2.8 II T= 3.1 = -0.3 stops

A couple of take away points are:
1. The 400mm DO II is in practice only 2/3 stops faster than the 100-400mm II and not 1 stop as the DO optics are not quite as efficient as refractive.
2. The long zooms have similar losses in stops to the prime.

I took point 1 into consideration when deciding to sell my 400mm DO II and keep the 100-400mm II.

Cheers Alan. Handy stats you have there! I'm probably looking at your 100-400mm -0.34 stops drop, vs 200-400mm -0.46 stops drop and the prime 400mm at -0.64 stops drop and speculating those T stop drops are averages over the zoom range? So maybe at the long end of the zoom the drop in light is greater than the average suggests? Maybe? That said though, it's still looking like a second order issue. Plus you've obviously used both the prime and zoom, so experienced both in the real world, and based your decision off that. PS, just noticed you've already stated it's at 400mm for the zooms, so scrub my earlier average comment. Thanks.

 

[MonsMeg]

I’d say not for stills. A few years ago I noticed that one of my favourite lenses had a T stop quite a bit slower than its fully open f/1.8 stop according to DXO. Another one, the EF 35 f/2 IS had a T stop the same as it’s f stop. So I thought this should be very easy to see with both lenses set to f/2 and the same exposure in raw with no cooking. Result: no difference in output exposure at all.

Nice one. Thanks.

 

[MonsMeg]

T-stops really don’t make any difference in stills captures, they end up being a side show that gets people here all worked up every now and again.

The really important things for still lenses are focal length and aperture, then probably resolution/sharpness, focus speed and accuracy, image stabilization etc etc.

T stops are very relevant for people shooting video especially when they are shooting the same scene with multiple cameras and for ease of editing and look want the exposures to be the same across those cameras. However given the power of post processing and the inevitable grading the differences are nowhere near as important as they were when ‘movies’ were shot on film.

But back to stills, when comparing lenses for still photography the key issue is depth of field control, t stops don’t tell you anything about that, you could have a very bad transmission lens with a narrow depth of field or a very good transmission lens with a smaller aperture both with the same t value.

Interesting point on DoF and T stops. Hadn't considered that. I'll pass that on.

 

[AlanF]

Thanks Joules, appreciated. Yep, stills. I'm asking for a small group of wildlife photographers one of which has a Sigma 150-600mm f5.6-6.3 and another who is considering this lens vs an older used Canon 600mm f4. The debate has ranged over all the usual topics (not least price), but the one interesting thing we noticed when out shooting at low light which is when we do most wildlife was that the Sigma behaved less like a f6.3 and more like a f7 or f8 (in terms of the iso that was being used by the camera vs mine with the prime). DxO says the Sigma has a T6.7, and that's an average apparently over the range of the zoom according to DxO methodology? Possibly when she's at the full 600mm the T is higher again, maybe? And she's mostly at 600mm. I agree that consensus is it's marginal, just it was a bit of a surprise as it didn't feel that marginal to her.

You have to go into the DxO site and you will find it gives the Sigma at different lengths. I couldn't find the 150-600mm, At 500mm for the 150-500mm, the T-stop is 7.3 (click on transmission) in:

DXOMARK

www.dxomark.com

They have the Tamron 150-600mm, and that is T= 7.5 at 600mm.

DXOMARK

www.dxomark.com

 

[MonsMeg]

You have to go into the DxO site and you will find it gives the Sigma at different lengths. I couldn't find the 150-600mm, At 500mm for the 150-500mm, the T-stop is 7.3 (click on transmission) in:

DXOMARK

www.dxomark.com

They have the Tamron 150-600mm, and that is T= 7.5 at 600mm.

DXOMARK

www.dxomark.com

OK, that is seriously helpful. Well played Sir. Using the interim data on the zoom range I can build a decent comparison across lenses even if zoom ranges differ, (if ultimately needed). You've saved me days on behalf of my mates. Thank you.

 

[privatebydesign]

OK, that is seriously helpful. Well played Sir. Using the interim data on the zoom range I can build a decent comparison across lenses even if zoom ranges differ, (if ultimately needed). You've saved me days on behalf of my mates. Thank you.

I have to ask, to what end?

 

[neuroanatomist]

T-stops really don’t make any difference in stills captures, they end up being a side show that gets people here all worked up every now and again.

The really important things for still lenses are focal length and aperture, then probably resolution/sharpness, focus speed and accuracy, image stabilization etc etc.

Well, unless you're DxO Mark. If you are, then T-stops make a very big difference in your ranking of lenses.

@MonsMeg if you use lens data from DxOMark, be aware of two main points. 1) Use only the measurements, not the Score. 2) They only test one copy of each lens, and all lenses have copy variation.

DxO Mark's Lens Score is based on the use case of shooting in 150 lux, the light level in a dimly-lit warehouse. That's why a lens like the EF 50/1.8 gets a higher score than a lens like the EF 500/4 II, despite the latter being sharper and having less of all the aberrations they test. All the 50mm lens has is a better T-Stop. Granted, if I were going to shoot in a dimly-lit warehouse, the 50/1.8 would likely be a better choice...but it's really not a 'better' lens.



Moreover, their lens score isn't really a lens score at all, but a lens + camera score. The base ISO dynamic range of the sensor behind the lens is a significant component of the Lens Score, so the same lens (e.g. Zeiss, Sigma) scores higher on Nikon than on Canon bodies (and used to score much higher, when the base ISO DR difference was greater). Yes, DxO Mark...the only review/ranking site that considers dynamic range to be a critical metric for a lens.

 

[AlanF]

Well, unless you're DxO Mark. If you are, then T-stops make a very big difference in your ranking of lenses.

@MonsMeg if you use lens data from DxOMark, be aware of two main points. 1) Use only the measurements, not the Score. 2) They only test one copy of each lens, and all lenses have copy variation.

DxO Mark's Lens Score is based on the use case of shooting in 150 lux, the light level in a dimly-lit warehouse. That's why a lens like the EF 50/1.8 gets a higher score than a lens like the EF 500/4 II, despite the latter being sharper and having less of all the aberrations they test. All the 50mm lens has is a better T-Stop. Granted, if I were going to shoot in a dimly-lit warehouse, the 50/1.8 would likely be a better choice...but it's really not a 'better' lens.

View attachment 199297

Moreover, their lens score isn't really a lens score at all, but a lens + camera score. The base ISO dynamic range of the sensor behind the lens is a significant component of the Lens Score, so the same lens (e.g. Zeiss, Sigma) scores higher on Nikon than on Canon bodies (and used to score much higher, when the base ISO DR difference was greater). Yes, DxO Mark...the only review/ranking site that considers dynamic range to be a critical metric for a lens.

Their overall scores are indeed close to worthless but individual measurements on MTF values and T-stops that do not depend on sensor and light level are probably valid to varying extents. The MTF values are subject to copy variation but T-stops less so.

 

[MonsMeg]

I have to ask, to what end?

Ha! Yes, no problem. I think we'll agree this is going to be a second order issue, from a purely technical perspective. That said depending on our own personal financial circumstances buying a 600m lens whether Sigma zoom or used Canon prime can be a significant financial outlay and an emotional purchase. I was asked to investigate this issue, and since AlanF allows me to look at T stops at the top end of these big zooms (rather than an average) I can give my friend a decent quantified answer on relative loss of light at 600mm whether she goes Sigma or Canon prime. That will (I hope) help her to decide which way to go in terms of purchase. It will certainly help her at her local camera club when some smarty pants says "well what about T stops?", because there's always "that" person who sows doubt (particularly in her club, particularly around significant purchases apparently) but won't do the leg-work to quantify the size of the issue. Did I answer what you asked?

 

[MonsMeg]

Their overall scores are indeed close to worthless but individual measurements on MTF values and T-stops that do not depend on sensor and light level are probably valid to varying extents. The MTF values are subject to copy variation but T-stops less so.

Good to know. Yes, I'll just be using T scores as opposed to an overall score.

 

[neuroanatomist]

Their overall scores are indeed close to worthless but individual measurements on MTF values and T-stops that do not depend on sensor and light level are probably valid to varying extents. The MTF values are subject to copy variation but T-stops less so.

All of their measurements are made from RAW images, they do not use an optical bench (unlike, for example, Roger Cicala at Lens Rentals). Thus if you take a lens like the excellent Sigma 85/1.4, it is T=1.6 on a 1D X II, T=1.7 on a 5DIV, T=1.8 on a 5DsR, and T=1.9 on a 7DII.

 

[AlanF]

Good to know. Yes, I'll just be using T scores as opposed to an overall score.

There is no substitute to actually trying out the lens. As PBD wrote there are so many parameters. And importantly, can she physically handle the lens.

 

[MonsMeg]

There is no substitute to actually trying out the lens. As PBD wrote there are so many parameters. And importantly, can she physically handle the lens.

Fortunately a mutual friend has the Sigma zoom and I have a Canon prime, so she can try both fairly painlessly on her camera and keep the camera/sensor as a constant. In retrospect when we were last out and this topic raised its head my sensor was probably 'boosting' the apparent light gathering power of the prime on my camera (via iso), vs our mutual friend who had an older camera with the Sigma. As ever many variables.

 

[AlanF]

All of their measurements are made from RAW images, they do not use an optical bench (unlike, for example, Roger Cicala at Lens Rentals). Thus if you take a lens like the excellent Sigma 85/1.4, it is T=1.6 on a 1D X II, T=1.7 on a 5DIV, T=1.8 on a 5DsR, and T=1.9 on a 7DII.

Good - they have done repeat measurements to give T = 1.750±0.064 (standard error). We should, therefore, for each lens do the same calculations for multiple bodies to get a more reliable estimate.