Olympus 25mm f/1.2 PRO: Finding T#

Started Aug 3, 2018 | Discussions
MEDISN
MEDISN Senior Member • Posts: 1,086
Olympus 25mm f/1.2 PRO: Finding T#
15

Following Jim’s thread, and ensuing geek discussion, in the PST forum last month, I reached out to AiryDiscus to see if we could provide some measurements of the mZD 25mm f/1.2 PRO. Since the DxO “transmission” values were released (T=1.6, 1.7, 1.8 depending on the camera), there has been ongoing speculation concerning the impact of 19 elements, micro-lens shading, super-secret gain adjustments and so on. Let’s dive in!

EPD / EFL => F/#

F/# / sqrt(T_avg) => T#

The relevant standard here is ISO 517

Entrance pupil diameter (EPD) was measured on my end based on AiryDiscus instruction below. Effective focal length (EFL) is measured by measuring the lateral displacement in the image plane of two slits angularly spaced 0.01 degrees at a distance of infinity. The spectra is measured in a way that is in accord with this – Link

The EPD procedure is as follows:

  1. Take the lens to be measured (25/1.2), remove front and rear caps, ensure aperture is fully open, and place it in a v-block or other fixture to hold it in a stable fashion.
  2. With a camera and your macro lens, focus such that the limiting feature of the aperture is in crisp focus and take a picture.
  3. Remove the lens being measured and replace it with a ruler; the camera may move provided the focus ring is not touched.
  4. Capture a picture of the ruler at precisely the same focus; both meaning it is well-focused, and the focus ring was not adjusted.
  5. In photoshop or some other, use the ruler tool to measure some known length on the ruler, e.g. 1cm => 706.24px. This yields the "plate scale," which for the prior values would be 1.4159um/px.
  6. Repeat the prior process with the lens aperture. If you measure, say, 1462.87px, then 1462.87 * 1.4159 / 1e3 = 20.713mm.

This gives the entrance pupil diameter.

Screenshots of my measurements in Photoshop:

The effective focal length and transmittance were measured by AiryDiscus. The EFL is 25.03mm

T_avg[390-670] = 0.789016

Results Summary:

EPD: 19.40mm => f/1.29

EFL: 25.03mm

T#: 1.36

A BIG thanks to AiryDiscus for his guidance and help with this.

LeDatas Forum Member • Posts: 76
Re: Olympus 25mm f/1.2 PRO: Finding T#
3

Well, a T# of 1.36 is much better than 1.8

I really like how that lens renders images.

 LeDatas's gear list:LeDatas's gear list
Olympus E-M1 II Panasonic Lumix G Vario 7-14mm F4 ASPH Panasonic Leica Summilux DG 25mm F1.4 Olympus M.Zuiko Digital 45mm F1.8 Olympus M.Zuiko Digital 17mm F1.8
knickerhawk Veteran Member • Posts: 6,574
Re: Olympus 25mm f/1.2 PRO: Finding T#
2

MEDISN wrote:

Following Jim’s thread, and ensuing geek discussion, in the PST forum last month, I reached out to AiryDiscus to see if we could provide some measurements of the mZD 25mm f/1.2 PRO. Since the DxO “transmission” values were released (T=1.6, 1.7, 1.8 depending on the camera), there has been ongoing speculation concerning the impact of 19 elements, micro-lens shading, super-secret gain adjustments and so on. Let’s dive in!

EPD / EFL => F/#

F/# / sqrt(T_avg) => T#

The relevant standard here is ISO 517

Entrance pupil diameter (EPD) was measured on my end based on AiryDiscus instruction below. Effective focal length (EFL) is measured by measuring the lateral displacement in the image plane of two slits angularly spaced 0.01 degrees at a distance of infinity. The spectra is measured in a way that is in accord with this – Link

The EPD procedure is as follows:

  1. Take the lens to be measured (25/1.2), remove front and rear caps, ensure aperture is fully open, and place it in a v-block or other fixture to hold it in a stable fashion.
  2. With a camera and your macro lens, focus such that the limiting feature of the aperture is in crisp focus and take a picture.
  3. Remove the lens being measured and replace it with a ruler; the camera may move provided the focus ring is not touched.
  4. Capture a picture of the ruler at precisely the same focus; both meaning it is well-focused, and the focus ring was not adjusted.
  5. In photoshop or some other, use the ruler tool to measure some known length on the ruler, e.g. 1cm => 706.24px. This yields the "plate scale," which for the prior values would be 1.4159um/px.
  6. Repeat the prior process with the lens aperture. If you measure, say, 1462.87px, then 1462.87 * 1.4159 / 1e3 = 20.713mm.

This gives the entrance pupil diameter.

Screenshots of my measurements in Photoshop:

The effective focal length and transmittance were measured by AiryDiscus. The EFL is 25.03mm

T_avg[390-670] = 0.789016

Results Summary:

EPD: 19.40mm => f/1.29

EFL: 25.03mm

T#: 1.36

A BIG thanks to AiryDiscus for his guidance and help with this.

Likewise! (I'm glad AiryDiscus suspended his distaste for testing mFT lenses long enough to perform this test.) And a big thanks to you for taking the time to assist AiryDiscus here.

This discussion should probably be continued over at the PST forum, but my reading of these results is that they confirm AiryDiscus' observations in the PST thread regarding the (in)credibility of DXOMark's t-stop measurements generally, let alone with respect to the Oly 25mm f/1.2 in particular. Unless and until more comparative tests are done, it's impossible to say how far off DXOMark is with respect to other lenses and how the Oly 25mm f/1.2 compares. However, it increasingly appears to me that a good deal of the handwringing about the real performance of these fast lenses on cameras with small pixels is - to frame it most positively - premature.

nick779 Regular Member • Posts: 293
Re: Olympus 25mm f/1.2 PRO: Finding T#
2

I wonder how using this method it would compare to the 42.5 nocticron that Dx0 rates at T# 1.7

 nick779's gear list:nick779's gear list
Olympus E-M1 II Olympus 12-40mm F2.8 Pro Olympus 40-150mm F2.8 Pro Olympus M.Zuiko Digital 17mm F1.8 Olympus M.Zuiko Digital 25mm F1.8
MEDISN
OP MEDISN Senior Member • Posts: 1,086
Re: Olympus 25mm f/1.2 PRO: Finding T#
10

knickerhawk wrote:

Likewise! (I'm glad AiryDiscus suspended his distaste for testing mFT lenses long enough to perform this test.) And a big thanks to you for taking the time to assist AiryDiscus here.

Regardless of any "feelings" he has toward mFT lenses, he was gracious, courteous and patient with my elementary questions (and screw-ups) for nearly a week putting this together.  I'm sure he has better things to do.  Folks like him, Roger Cicala, Jim Kasson, Bill Claff and many others provide an invaluable service to this community (sans-compensation no less).

This discussion should probably be continued over at the PST forum,

I posted a link from that thread to this one in case anyone is still interested.  Given this is specifically about the Olympus lens and not DxO testing, I thought it would garner more interest here than the PST forum.

knickerhawk Veteran Member • Posts: 6,574
Re: Olympus 25mm f/1.2 PRO: Finding T#
2

MEDISN wrote:

knickerhawk wrote:

Likewise! (I'm glad AiryDiscus suspended his distaste for testing mFT lenses long enough to perform this test.) And a big thanks to you for taking the time to assist AiryDiscus here.

Regardless of any "feelings" he has toward mFT lenses, he was gracious, courteous and patient with my elementary questions (and screw-ups) for nearly a week putting this together. I'm sure he has better things to do. Folks like him, Roger Cicala, Jim Kasson, Bill Claff and many others provide an invaluable service to this community (sans-compensation no less).

I agree and am also grateful for their efforts, all the more so when they involve extra work (which, as he explained, was the source of AiryDiscus' reluctance to perform tests on mFT systems). I'm also grateful to you for persisting with the issue and taking the time to provide the supporting data to AiryDiscus.

This discussion should probably be continued over at the PST forum,

I posted a link from that thread to this one in case anyone is still interested. Given this is specifically about the Olympus lens and not DxO testing, I thought it would garner more interest here than the PST forum.

I'm guessing it will garner interest in both forums. I'd quibble with you, though, about the relevance of DXOMark's testing and t-stop results. Ever since DXOMark started publishing t-stop measurements and its original analysis of the impact of pixel size, this discussion has generally been framed by DXO's measurements. A good example of that is the original thread on this forum that prompted Jim Sterling's follow-up post on the PST forum. I consider the efforts of you and AiryDiscus to be an interesting and useful contribution to the discussion of, both, Olympus lenses and DXOMark's test results and associated claims.

TomFid Veteran Member • Posts: 3,351
error?
1

MEDISN wrote:

EPD / EFL => F/#

F/# / sqrt(T_avg) => T#

T_avg[390-670] = 0.789016

Results Summary:

EPD: 19.40mm => f/1.29

EFL: 25.03mm

T#: 1.36

Something is amiss here.

F1.29 / sqrt( 0.789 ) = T 1.45, no?

Also, what does this really say about the potential for sensor interactions (microlens effects)?

 TomFid's gear list:TomFid's gear list
Olympus M.Zuiko ED 75-300mm 1:4.8-6.7 II Panasonic 14-140mm F3.5-5.6 II
Ab Latchin Senior Member • Posts: 1,613
WAIT, no WAIT, stop the presses
7

Are you saying DxO has misleading information??!!!??????!!!!!

I will not stand for this, they are the BIBLE, the NEWEST testament for photographers on all things except camera weight and size which we defer to another site which does not make mistakes.

The pro lenses are fantastic. I have 2 now and they are worth every penny, despite what so many of the usual suspects seem to think.

Next on my list is the 45mm f1.2 OR the 24mm LAOWA snorkel lens.

AiryDiscus Senior Member • Posts: 1,928
Re: Olympus 25mm f/1.2 PRO: Finding T#
3

nick779 wrote:

I wonder how using this method it would compare to the 42.5 nocticron that Dx0 rates at T# 1.7

If someone has a ruler, macro lens, and copy of my least favorite lens...

AiryDiscus Senior Member • Posts: 1,928
Re: error?
4

TomFid wrote:

MEDISN wrote:

EPD / EFL => F/#

F/# / sqrt(T_avg) => T#

T_avg[390-670] = 0.789016

Results Summary:

EPD: 19.40mm => f/1.29

EFL: 25.03mm

T#: 1.36

Something is amiss here.

F1.29 / sqrt( 0.789 ) = T 1.45, no?

Clerical error, sorry.  it is 1.45.

Also, what does this really say about the potential for sensor interactions (microlens effects)?

The F/# is ~f/1.3.  If the spectral transmittance is not dependent on radial aperture coordinate (a reasonable assumption, we don't see color gradients in the pupil images, then the F/# of the lens is "overriden" near focus to that of the microlens.

So if this f/1.3 lens is used with an f/1.5 microlens, then the "effective" T# (I do not like such terms, but there is no good name for it) would be 1.5 / sqrt(.789) = 1.689.

AiryDiscus Senior Member • Posts: 1,928
Re: Olympus 25mm f/1.2 PRO: Finding T#
8

knickerhawk wrote:

Likewise! (I'm glad AiryDiscus suspended his distaste for testing mFT lenses long enough to perform this test.) And a big thanks to you for taking the time to assist AiryDiscus here.

I am simply the puppetmaster, Roger or Aaron is responsible for having done the measurements.

This discussion should probably be continued over at the PST forum, but my reading of these results is that they confirm AiryDiscus' observations in the PST thread regarding the (in)credibility of DXOMark's t-stop measurements generally, let alone with respect to the Oly 25mm f/1.2 in particular. Unless and until more comparative tests are done, it's impossible to say how far off DXOMark is with respect to other lenses and how the Oly 25mm f/1.2 compares. However, it increasingly appears to me that a good deal of the handwringing about the real performance of these fast lenses on cameras with small pixels is - to frame it most positively - premature.

Their methodology is reasonable, they are just limited by the microlenses of the cameras used in their test.  When you see a range of T#s for a single lens from them, it tells you about either:

(1) noise / instability in their process (unlikely to span T1.6-T1.8)

(2) the microlense F/#s of the cameras used.

If you look at an ensemble of large aperture (f/1.2 or larger) lenses on a given camera, you can use the group trends to infer more accurately about the microlens F/#.

It's not so much that they are "off" so much as it is that they are unable to look "through" the camera due to their method.

AiryDiscus Senior Member • Posts: 1,928
Re: Olympus 25mm f/1.2 PRO: Finding T#
12

MEDISN wrote:

knickerhawk wrote:

Likewise! (I'm glad AiryDiscus suspended his distaste for testing mFT lenses long enough to perform this test.) And a big thanks to you for taking the time to assist AiryDiscus here.

Regardless of any "feelings" he has toward mFT lenses, he was gracious, courteous and patient with my elementary questions (and screw-ups) for nearly a week putting this together. I'm sure he has better things to do.

My day job is to build space telescopes, and my night job to turn the world of consumer camera lens testing on its head...

bclaff Veteran Member • Posts: 9,199
Re: Olympus 25mm f/1.2 PRO: Finding T#
3

MEDISN wrote:

...

EPD / EFL => F/#
EFL / EPD => F#

F# / sqrt(T_avg) => T#

The relevant standard here is ISO 517
...then 1462.87 * 1.4159 / 1e3 = 20.713mm.

This gives the entrance pupil diameter.

...

The effective focal length and transmittance were measured by AiryDiscus. The EFL is 25.03mm

...

T_avg[390-670] = 0.789016

Results Summary:

EPD: 19.40mm => f/1.29
EPD: 20.713mm => f/1.2

EFL: 25.03mm

T#: 1.36

Nice work. I think the corrections in bold above are right.

Note:

25.03 / 20.713 = 1.208

1.208 / sqrt(0.789016) = 1.3599

If EPD really is 19.90mm then you're going to get T# is 1.45

25.03 / 19.40 = 1.29

1.29 / sqrt(0.78016) = 1.45

-- hide signature --

Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )

TomFid Veteran Member • Posts: 3,351
Re: error?
1

AiryDiscus wrote:

TomFid wrote:

MEDISN wrote:

EPD / EFL => F/#

F/# / sqrt(T_avg) => T#

T_avg[390-670] = 0.789016

Results Summary:

EPD: 19.40mm => f/1.29

EFL: 25.03mm

T#: 1.36

Something is amiss here.

F1.29 / sqrt( 0.789 ) = T 1.45, no?

Clerical error, sorry. it is 1.45.

Also, what does this really say about the potential for sensor interactions (microlens effects)?

The F/# is ~f/1.3. If the spectral transmittance is not dependent on radial aperture coordinate (a reasonable assumption, we don't see color gradients in the pupil images, then the F/# of the lens is "overriden" near focus to that of the microlens.

So if this f/1.3 lens is used with an f/1.5 microlens, then the "effective" T# (I do not like such terms, but there is no good name for it) would be 1.5 / sqrt(.789) = 1.689.

Thanks for the clarity!

 TomFid's gear list:TomFid's gear list
Olympus M.Zuiko ED 75-300mm 1:4.8-6.7 II Panasonic 14-140mm F3.5-5.6 II
AiryDiscus Senior Member • Posts: 1,928
Re: Olympus 25mm f/1.2 PRO: Finding T#

bclaff wrote:

MEDISN wrote:

...

EPD / EFL => F/#
EFL / EPD => F#

F# / sqrt(T_avg) => T#

The relevant standard here is ISO 517
...then 1462.87 * 1.4159 / 1e3 = 20.713mm.

This gives the entrance pupil diameter.

...

The effective focal length and transmittance were measured by AiryDiscus. The EFL is 25.03mm

...

T_avg[390-670] = 0.789016

Results Summary:

EPD: 19.40mm => f/1.29
EPD: 20.713mm => f/1.2

EFL: 25.03mm

T#: 1.36

Nice work. I think the corrections in bold above are right.

Note:

25.03 / 20.713 = 1.208

1.208 / sqrt(0.789016) = 1.3599

If EPD really is 19.90mm then you're going to get T# is 1.45

25.03 / 19.40 = 1.29

1.29 / sqrt(0.78016) = 1.45

The number 20.713 comes from me showing an example worked calculation from the plate scale and is not in any way connected to Olympus' lens.

bclaff Veteran Member • Posts: 9,199
Re: Olympus 25mm f/1.2 PRO: Finding T#

AiryDiscus wrote:

bclaff wrote:

MEDISN wrote:

...

EPD / EFL => F/#
EFL / EPD => F#

F# / sqrt(T_avg) => T#

The relevant standard here is ISO 517
...then 1462.87 * 1.4159 / 1e3 = 20.713mm.

This gives the entrance pupil diameter.

...

The effective focal length and transmittance were measured by AiryDiscus. The EFL is 25.03mm

...

T_avg[390-670] = 0.789016

Results Summary:

EPD: 19.40mm => f/1.29
EPD: 20.713mm => f/1.2

EFL: 25.03mm

T#: 1.36

Nice work. I think the corrections in bold above are right.

Note:

25.03 / 20.713 = 1.208

1.208 / sqrt(0.789016) = 1.3599

If EPD really is 19.90mm then you're going to get T# is 1.45

25.03 / 19.40 = 1.29

1.29 / sqrt(0.78016) = 1.45

The number 20.713 comes from me showing an example worked calculation from the plate scale and is not in any way connected to Olympus' lens.

Thanks. Interesting it works out so nicely !

Regards,

-- hide signature --

Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )

Jack Hogan Veteran Member • Posts: 7,264
Re: error?

AiryDiscus wrote:

TomFid wrote:

MEDISN wrote:

EPD / EFL => F/#

F/# / sqrt(T_avg) => T#

T_avg[390-670] = 0.789016

Results Summary:

EPD: 19.40mm => f/1.29

EFL: 25.03mm

T#: 1.36

Something is amiss here.

F1.29 / sqrt( 0.789 ) = T 1.45, no?

Clerical error, sorry. it is 1.45.

Also, what does this really say about the potential for sensor interactions (microlens effects)?

The F/# is ~f/1.3. If the spectral transmittance is not dependent on radial aperture coordinate (a reasonable assumption, we don't see color gradients in the pupil images, then the F/# of the lens is "overriden" near focus to that of the microlens.

So if this f/1.3 lens is used with an f/1.5 microlens, then the "effective" T# (I do not like such terms, but there is no good name for it) would be 1.5 / sqrt(.789) = 1.689.

Nice job MEDISN and Airy.  A couple of totally naive questions (I am not up to speed on T#s):

1) Does the procedure take vignetting into account?
2) With such small f-numbers, would using the more precise definition of relative aperture [1/(2.sin(theta') ) in air] result in material differences to the obtained figures?

I presume that another potential source of differences to DXO measurements could be the spectra of the illuminants used.

Jack

AiryDiscus Senior Member • Posts: 1,928
Re: error?
2

Jack Hogan wrote:

AiryDiscus wrote:

TomFid wrote:

MEDISN wrote:

EPD / EFL => F/#

F/# / sqrt(T_avg) => T#

T_avg[390-670] = 0.789016

Results Summary:

EPD: 19.40mm => f/1.29

EFL: 25.03mm

T#: 1.36

Something is amiss here.

F1.29 / sqrt( 0.789 ) = T 1.45, no?

Clerical error, sorry. it is 1.45.

Also, what does this really say about the potential for sensor interactions (microlens effects)?

The F/# is ~f/1.3. If the spectral transmittance is not dependent on radial aperture coordinate (a reasonable assumption, we don't see color gradients in the pupil images, then the F/# of the lens is "overriden" near focus to that of the microlens.

So if this f/1.3 lens is used with an f/1.5 microlens, then the "effective" T# (I do not like such terms, but there is no good name for it) would be 1.5 / sqrt(.789) = 1.689.

Nice job MEDISN and Airy. A couple of totally naive questions (I am not up to speed on T#s):

1) Does the procedure take vignetting into account?

On-axis, there is no vignetting.

2) With such small f-numbers, would using the more precise definition of relative aperture [1/(2.sin(theta') ) in air] result in material differences to the obtained figures?

EFL / EPD is exact for an object at infinity.  You only need 1/2sinx for finite object distances.

I presume that another potential source of differences to DXO measurements could be the spectra of the illuminants used.

If they followed a procedure like mine, the illuminant wouldn't matter.

bobn2
bobn2 Forum Pro • Posts: 63,842
Re: WAIT, no WAIT, stop the presses
4

Ab Latchin wrote:

Are you saying DxO has misleading information??!!!??????!!!!!

I will not stand for this, they are the BIBLE, the NEWEST testament for photographers on all things except camera weight and size which we defer to another site which does not make mistakes.

Hold your horses there. These results are completely compatible with DxOmark's:

https://www.dpreview.com/forums/post/61457016

AD explains here:

https://www.dpreview.com/forums/post/61457030

The pro lenses are fantastic. I have 2 now and they are worth every penny, despite what so many of the usual suspects seem to think.

If they are worth every penny to you, then that's what they are worth to you. No need to be defensive about it, it's your money.

Next on my list is the 45mm f1.2 OR the 24mm LAOWA snorkel lens.

-- hide signature --

Ride easy, William.
Bob

bobn2
bobn2 Forum Pro • Posts: 63,842
Re: Olympus 25mm f/1.2 PRO: Finding T#
2

knickerhawk wrote:

MEDISN wrote:

Following Jim’s thread, and ensuing geek discussion, in the PST forum last month, I reached out to AiryDiscus to see if we could provide some measurements of the mZD 25mm f/1.2 PRO. Since the DxO “transmission” values were released (T=1.6, 1.7, 1.8 depending on the camera), there has been ongoing speculation concerning the impact of 19 elements, micro-lens shading, super-secret gain adjustments and so on. Let’s dive in!

EPD / EFL => F/#

F/# / sqrt(T_avg) => T#

The relevant standard here is ISO 517

Entrance pupil diameter (EPD) was measured on my end based on AiryDiscus instruction below. Effective focal length (EFL) is measured by measuring the lateral displacement in the image plane of two slits angularly spaced 0.01 degrees at a distance of infinity. The spectra is measured in a way that is in accord with this – Link

The EPD procedure is as follows:

  1. Take the lens to be measured (25/1.2), remove front and rear caps, ensure aperture is fully open, and place it in a v-block or other fixture to hold it in a stable fashion.
  2. With a camera and your macro lens, focus such that the limiting feature of the aperture is in crisp focus and take a picture.
  3. Remove the lens being measured and replace it with a ruler; the camera may move provided the focus ring is not touched.
  4. Capture a picture of the ruler at precisely the same focus; both meaning it is well-focused, and the focus ring was not adjusted.
  5. In photoshop or some other, use the ruler tool to measure some known length on the ruler, e.g. 1cm => 706.24px. This yields the "plate scale," which for the prior values would be 1.4159um/px.
  6. Repeat the prior process with the lens aperture. If you measure, say, 1462.87px, then 1462.87 * 1.4159 / 1e3 = 20.713mm.

This gives the entrance pupil diameter.

Screenshots of my measurements in Photoshop:

The effective focal length and transmittance were measured by AiryDiscus. The EFL is 25.03mm

T_avg[390-670] = 0.789016

Results Summary:

EPD: 19.40mm => f/1.29

EFL: 25.03mm

T#: 1.36

A BIG thanks to AiryDiscus for his guidance and help with this.

Likewise! (I'm glad AiryDiscus suspended his distaste for testing mFT lenses long enough to perform this test.) And a big thanks to you for taking the time to assist AiryDiscus here.

This discussion should probably be continued over at the PST forum, but my reading of these results is that they confirm AiryDiscus' observations in the PST thread regarding the (in)credibility of DXOMark's t-stop measurements generally, let alone with respect to the Oly 25mm f/1.2 in particular.

I would think, that they confirm the general credibility DXOmark's results. As he points out, the DxOmark tests a not really a T-stop, they are an effective T-stop, taking into account both the actual lens transmission and the microlens shading effect. I think also of interest is that this lens is 1/4 stop slow so far as the f-number is concerned, it would have been more honest to call it an f/1.3.

Unless and until more comparative tests are done, it's impossible to say how far off DXOMark is with respect to other lenses and how the Oly 25mm f/1.2 compares. However, it increasingly appears to me that a good deal of the handwringing about the real performance of these fast lenses on cameras with small pixels is - to frame it most positively - premature.

Not really. It depends whether you want to write off DxOmark's results as incredible, and some kind of plot against mFT, or else use them in combination with these results to arrive at an estimate of microlens speed, camera by camera.

I think more generally, that it does show that in the general solution space for the specific problem of light gathering, the small sensor/small f-number solution is less effective than the larger sensor/larger f-number solution.

-- hide signature --

Ride easy, William.
Bob

Keyboard shortcuts:
FForum MMy threads