Pages: [1]   Go Down

Author Topic: DxO lens measurements  (Read 4463 times)

Hans Kruse

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2106
    • Hans Kruse Photography
DxO lens measurements
« on: December 14, 2015, 08:10:21 am »

DxO has many measurements of lenses and cameras. There is the Mpix which is sort of described but it is not very clear what exactly it means except that a higher number probably means a better resolving lens on a given camera. The accutance profile seems like a god place to judge a lens and camera combination as you can for a zoom lens select the focal length and the aperture. What is not clear to me is how the accutance % relates to resolution. I have read the descriptions http://www.dxomark.com/About/In-depth-measurements/Measurements/Sharpness but still it does not me at least spell out what the curves really mean.

The attached comparison shows Canon 5D III and 5DsR on the Canon 24-70 f/2.8L II lens which is rated at 32 Mpix for the 5DsR and the Nikon 24-70 f/2.8G ED on the D810 rated at 21 Mpix. We can see that the 5DsR and D810 accutance % is the same in the center of the lens, but does that mean that the resolution is the same?

We can also see that the 5D III accutance % goes higher than the D810 at the 40% field position of the lenses. Does this mean that the 5D III has a higher resolution than the D810 between 40% and 100% of the given lenses?

« Last Edit: December 14, 2015, 08:13:26 am by Hans Kruse »
Logged

David Eichler

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 823
    • San Francisco Architectural and Interior Photographer
Re: DxO lens measurements
« Reply #1 on: December 14, 2015, 01:33:48 pm »

DxO has many measurements of lenses and cameras. There is the Mpix which is sort of described but it is not very clear what exactly it means except that a higher number probably means a better resolving lens on a given camera. The accutance profile seems like a god place to judge a lens and camera combination as you can for a zoom lens select the focal length and the aperture. What is not clear to me is how the accutance % relates to resolution. I have read the descriptions http://www.dxomark.com/About/In-depth-measurements/Measurements/Sharpness but still it does not me at least spell out what the curves really mean.

The attached comparison shows Canon 5D III and 5DsR on the Canon 24-70 f/2.8L II lens which is rated at 32 Mpix for the 5DsR and the Nikon 24-70 f/2.8G ED on the D810 rated at 21 Mpix. We can see that the 5DsR and D810 accutance % is the same in the center of the lens, but does that mean that the resolution is the same?

We can also see that the 5D III accutance % goes higher than the D810 at the 40% field position of the lenses. Does this mean that the 5D III has a higher resolution than the D810 between 40% and 100% of the given lenses?

I can't answer your question. However, it appears that you do not understand what acutance is. Acutance refers to edge contrast, which has an effect on apparent sharpness but has no bearing on resolution.
Logged

Hans Kruse

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2106
    • Hans Kruse Photography
Re: DxO lens measurements
« Reply #2 on: December 14, 2015, 01:40:56 pm »

I can't answer your question. However, it appears that you do not understand what acutance is. Acutance refers to edge contrast, which has an effect on apparent sharpness but has no bearing on resolution.

DxO defines as follows:

What is acutance?
An objective measure of sharpness which takes into account the sensitivity of the human visual system to specific spatial frequencies and the viewing distance of an image. Edge acutance refers to the ability of a photographic system to show a sharp edge between contiguous areas of low and high illuminance. Texture acutance refers to the ability of a photographic system to show details without noticeable degradations.

David Eichler

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 823
    • San Francisco Architectural and Interior Photographer
Re: DxO lens measurements
« Reply #3 on: December 14, 2015, 02:22:59 pm »

DxO defines as follows:

What is acutance?
An objective measure of sharpness which takes into account the sensitivity of the human visual system to specific spatial frequencies and the viewing distance of an image. Edge acutance refers to the ability of a photographic system to show a sharp edge between contiguous areas of low and high illuminance. Texture acutance refers to the ability of a photographic system to show details without noticeable degradations.

I think that definition may be misleading. Both resolution and acutance contribute to the effect of apparent sharpness. To me, "edge acutance" seems redundant, since, by definition, acutance refers to edge contrast.
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: DxO lens measurements
« Reply #4 on: December 14, 2015, 02:52:15 pm »

Hi,

I would say that DxO is a bit less than clear in describing the measurements they do.

Anyway, I would guess that the data they show is MTF at a given frequency.

Resolution is by definition the line pair per millimetres were MTF drops to a small value, like 9% or even lower values.

So I would say that resolution is mostly identical to the resolution of the sensor, while sharpness more depends on MTF, and actually more on MTF at lower frequencies than at high frequencies.

So, yes, the interpretation you make is reasonable, although I feel that resolution is a bit of a misnomer in the context.

These two papers by Dr. Hubert Nasse of Zeiss explain things pretty well:

http://www.zeiss.com/content/dam/Photography/new/pdf/en/cln_archiv/cln30_en_web_special_mtf_01.pdf
http://www.zeiss.com/content/dam/Photography/new/pdf/en/cln_archiv/cln31_en_web_special_mtf_02.pdf

One thing to keep in mind is that human perception of sharpness is dominated by low to medium frequency detail. A better lens generally would transfer more contrast at low to medium frequencies than a lens of less image quality. But the real figure to check is MTF of lens * MTF of sensor and there is a lot more to image quality than just MTF.

Best regards
Erik

DxO has many measurements of lenses and cameras. There is the Mpix which is sort of described but it is not very clear what exactly it means except that a higher number probably means a better resolving lens on a given camera. The accutance profile seems like a god place to judge a lens and camera combination as you can for a zoom lens select the focal length and the aperture. What is not clear to me is how the accutance % relates to resolution. I have read the descriptions http://www.dxomark.com/About/In-depth-measurements/Measurements/Sharpness but still it does not me at least spell out what the curves really mean.

The attached comparison shows Canon 5D III and 5DsR on the Canon 24-70 f/2.8L II lens which is rated at 32 Mpix for the 5DsR and the Nikon 24-70 f/2.8G ED on the D810 rated at 21 Mpix. We can see that the 5DsR and D810 accutance % is the same in the center of the lens, but does that mean that the resolution is the same?

We can also see that the 5D III accutance % goes higher than the D810 at the 40% field position of the lenses. Does this mean that the 5D III has a higher resolution than the D810 between 40% and 100% of the given lenses?
Logged
Erik Kaffehr
 

Hans Kruse

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2106
    • Hans Kruse Photography
Re: DxO lens measurements
« Reply #5 on: December 14, 2015, 03:14:10 pm »

Thanks for the responses. So for DxO they have Mpix measurements by focal length and aperture but they only go up to 12 Mpix so if I wanted to see how much better let's say a 5DsR would resolve using the old Canon 24-70 f/2.8L than the same lens on a 6D, then I can't use the Mpix charts since except f/2.8 and above f/11 all is green which means >12 Mpix. The profiles section shows the acutance in % and my real question was if I could conclude that a higher % would likely result in a higher resolution also under certain situations. Like the example above where we have the same lens on two different camera bodies. Screen shot attached. Of course if it is possible to conclude that the resolution is higher the next question would be by how much.

It seems to me that the profiles part is only useful for judging how a single lens and camera body performs from center to edge and in terms of contrast and not resolution. Since I would be much more interested in a measure of resolution by aperture, focal length and center to edge than acutance I was wondering if I had overlooked something. I may be wrong but I only know of DxO that does the testing of lens and camera body combinations. Photozone has now a few tests using the 5DsR. This type of tests would be nice if DxO provided such info for all lens and camera combinations. Given their Mpix numbers I would have thought that they could derive such info given their extensive test database.

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: DxO lens measurements
« Reply #6 on: December 14, 2015, 03:50:58 pm »

Hi,

I have recently spent some on Digital Picture: http://www.the-digital-picture.com

They have interesting things like:

http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=949&Camera=979&Sample=0&FLI=0&API=0&LensComp=486&CameraComp=0&FLIComp=0&APIComp=0

http://www.the-digital-picture.com/Reviews/MTF.aspx?Lens=949&FLI=1&API=2&LensComp=486&FLIComp=0&APIComp=2&CT=AVG

To me it seems like a quite worthwhile resource, albeit with some reservations. One is that test charts may be to small, although I don't know. I feel that we need at least 1:50 size reduction which mean 1.2 x 1.8 m size of charts. The MTF data from OLAF may be in it's infancy.

I cannot support them by buying trough their links, but support with a small monthly donation.

Best regards
Erik
Logged
Erik Kaffehr
 

Dave Ellis

  • Full Member
  • ***
  • Offline Offline
  • Posts: 103
Re: DxO lens measurements
« Reply #7 on: December 14, 2015, 04:21:16 pm »

Hans I share your frustration !

DXOMark obviously does detailed tests on lenses but presents the results without defining adequately the parameters they present. This makes them of limited value IMO. Acutance is quoted as a percentage, but I ask a percentage of what ? Interestingly, in their explanatory article you link to, the acutance figures are shown in lppmm but there is no indication to what mtf the lppmm figures correspond.

Dave
« Last Edit: December 15, 2015, 01:10:07 am by Dave Ellis »
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: DxO lens measurements
« Reply #8 on: December 14, 2015, 04:52:36 pm »

Hi,

I have some issues with both DxO-mark and Megapix rating. The DxO-mark is heavily affected by maximum aperture. That is OK if we are looking for low light performance but not really for on tripod shooting.

The Megapix rating is similar, if we don't know how they are calculated it gives little good info.

Regarding the "acutance curves" I guess they relate to MTF at some lp/mm. It seems that higher resolution systems get better acutance values. So 5DsR > Nikon D810 > 5DIII, this is probably both correct and relevant.

So DxO publishes a lot of data, but quite a bit dumbed down and that dumbing down decreases the value of data.

Best regards
Erik

Ps. You sold me on the 16-35/4 and the 24-70/2.8 LII is on it's way into the shopping list.



Thanks for the responses. So for DxO they have Mpix measurements by focal length and aperture but they only go up to 12 Mpix so if I wanted to see how much better let's say a 5DsR would resolve using the old Canon 24-70 f/2.8L than the same lens on a 6D, then I can't use the Mpix charts since except f/2.8 and above f/11 all is green which means >12 Mpix. The profiles section shows the acutance in % and my real question was if I could conclude that a higher % would likely result in a higher resolution also under certain situations. Like the example above where we have the same lens on two different camera bodies. Screen shot attached. Of course if it is possible to conclude that the resolution is higher the next question would be by how much.

It seems to me that the profiles part is only useful for judging how a single lens and camera body performs from center to edge and in terms of contrast and not resolution. Since I would be much more interested in a measure of resolution by aperture, focal length and center to edge than acutance I was wondering if I had overlooked something. I may be wrong but I only know of DxO that does the testing of lens and camera body combinations. Photozone has now a few tests using the 5DsR. This type of tests would be nice if DxO provided such info for all lens and camera combinations. Given their Mpix numbers I would have thought that they could derive such info given their extensive test database.
Logged
Erik Kaffehr
 

Dave Ellis

  • Full Member
  • ***
  • Offline Offline
  • Posts: 103
Re: DxO lens measurements
« Reply #9 on: December 14, 2015, 10:34:40 pm »

After some more research, I came across this article from Imatest.

This defines Acutance (in a quantitative manner) as an extension of the Subjective Quality Factor concept and apparently comes from the IEEE Camera Phone Image Quality Group. I wonder if DXOMark is using something like this. It would be nice if they told us !

Dave
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: DxO lens measurements (an interesting article)
« Reply #10 on: December 15, 2015, 01:39:27 am »

Hi,

This is some interesting stuff from Roger Cicala: http://www.lensrentals.com/blog/2013/01/a-24-70mm-system-comparison

Best regards
Erik
Logged
Erik Kaffehr
 

Hywel

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 294
    • http://www.restrainedelegance.com
Re: DxO lens measurements
« Reply #11 on: December 16, 2015, 07:13:35 am »

AH, thank for posting that link, Erik, I'd missed that one. Great to see a confirmation of the "higher res sensor with lower res lens" effect from another thread there: the benefit of using the higher res Nikon on the worse-performing Tamron lens is quite clear. Obviously sharpness isn't everything but it is good to see some actual measurements.

I wonder why this is why I'm so fond of the look of vintage lenses on digital? Especially for video. Lenses that might have been too soft when shooting on film or low-res digital are suddenly just the ticket for 4K+ sensors, giving just the right mix of detail and kindness on skin tones.

Cheers, Hywel
Logged
Pages: [1]   Go Up