Ad
Ad
Ad
Pages: « 1 [2] 3 »   Bottom of Page
Print
Author Topic: Dynamic range of human visual system  (Read 15966 times)
bjanes
Sr. Member
****
Offline Offline

Posts: 2756



« Reply #20 on: May 28, 2013, 02:33:29 PM »
ReplyReply

CCD Arrays, Cameras, and Displays by Holst, 2nd edition, defines display contrast as:

Contrast = Lwhite + Lambient * reflectivity / Lblack + Lambient * reflectivity

where L is luminance.

I am not sure how you are defining contrast, but a standard outdoor scene in daylight was always defined by Kodak as about 160:1. A silver print topped out at about 30:1. Contrast should be logarithmic.


Those contrast ratios do not seem realistic to me. Karl Lang, Rendering the print: the art of photography, demonstrates an outdoor scene with luminances of 10-30,000 cd/m^2 for a range of 3,000:1, and states that a typical silver print has a contrast ratio of 250:1. The 160:1 to which you refer may represent the daylight scene as rendered to a silver print (output referred, not the scene referred contrast).

Regards,

Bill
Logged
BJL
Sr. Member
****
Offline Offline

Posts: 5121


« Reply #21 on: May 28, 2013, 02:54:31 PM »
ReplyReply

Karl Lang, Rendering the print: the art of photography ... states that a typical silver print has a contrast ratio of 250:1.
I am skeptical about that second number: numerous sources suggest that the reflectivity of pure black on a glossy silver print is about 1.5% (the darkest natural substances, like fine charcoal, reflect about 4%). Meanwhile pure white anything reflects no more than 90%, and the baryta base of printing paper is not quite the whitest thing around. These two numbers limit the contrast ratio of a silver print to 60:1, or about six stops --- a range often discussed as a target in B&W printing from high contrast negatives.
« Last Edit: May 28, 2013, 02:57:12 PM by BJL » Logged
bjanes
Sr. Member
****
Offline Offline

Posts: 2756



« Reply #22 on: May 28, 2013, 04:02:40 PM »
ReplyReply

I am skeptical about that second number: numerous sources suggest that the reflectivity of pure black on a glossy silver print is about 1.5% (the darkest natural substances, like fine charcoal, reflect about 4%). Meanwhile pure white anything reflects no more than 90%, and the baryta base of printing paper is not quite the whitest thing around. These two numbers limit the contrast ratio of a silver print to 60:1, or about six stops --- a range often discussed as a target in B&W printing from high contrast negatives.

That figure may be a bit optimistic. For comparison, the ICC PRMG uses a high quality virtual print with a 288:1 DR, having maximal and minimum reflectances of 89% and 0.30911% respectively. The corresponding densities are 0.05 and 2.51.

Ilfobrom Galeria fb
chemical print paper has a Dmax of 2.2. Assuming that the maximal reflectivity is 89% or a density of 0.05, the DR would be 2.15 logs or 141:1.

Regards,

Bill

Logged
theguywitha645d
Sr. Member
****
Offline Offline

Posts: 970


« Reply #23 on: May 28, 2013, 05:08:38 PM »
ReplyReply

Those contrast ratios do not seem realistic to me. Karl Lang, Rendering the print: the art of photography, demonstrates an outdoor scene with luminances of 10-30,000 cd/m^2 for a range of 3,000:1, and states that a typical silver print has a contrast ratio of 250:1. The 160:1 to which you refer may represent the daylight scene as rendered to a silver print (output referred, not the scene referred contrast).

Regards,

Bill

You need to use a logarithmic scale. That better represents our perception of light values. I can check my texts again, but I remember the values very well--they were kind of drilled into us at RIT.
Logged
bjanes
Sr. Member
****
Offline Offline

Posts: 2756



« Reply #24 on: May 28, 2013, 05:47:36 PM »
ReplyReply

You need to use a logarithmic scale. That better represents our perception of light values. I can check my texts again, but I remember the values very well--they were kind of drilled into us at RIT.

You do not need a log scale to calculate contrast ratios. Density is log based. If you want perceptual values, you should use L* or at least gamma 2.2, which can easily be done with Bruce Lindbloom's Companding Calculator.

You should recheck your texts, as I think your recalled values are way off as demonstrated in my previous post using Ilford photographic paper. Your outdoor scene contrast ratio is way off. Karl Lang is a serious guy who knows his stuff.

Bill
Logged
BJL
Sr. Member
****
Offline Offline

Posts: 5121


« Reply #25 on: May 28, 2013, 06:01:41 PM »
ReplyReply

That figure may be a bit optimistic. ...
Ilfobrom Galeria fb[/url] chemical print paper has a Dmax of 2.2. Assuming that the maximal reflectivity is 89% or a density of 0.05, the DR would be 2.15 logs or 141:1.
Thanks Bill: it seems my information was not up to date with the latest papers. At http://www.imatest.com/guides/image-quality/print-dmax/ it is suggested that the new record might be a minimum print reflectivity of Dmax=2.3:
"There have been reports that the new Epson Ultrachrome K3 printers have Dmax as high as 2.3 with Premium Luster paper".
That would be a minimum reflectivity of about 1/200, so then allowing for at best 90% maximum reflectivity gives a contrast range of about 180:1. (But that is not a silver print, which is what I was Karl Lang and I were talking about.)

Note: Dmax is log base 10, so in the log base 2 units of "stops" more familiar to photographers, that Dmax=2.2 for Ilfobrom Galeria fb is about 7.3 stops below 100% reflectivity, or 158:1, and so a contrast range a bit less: probably a bit over 7 stops and no more than 150:1. Given that this is the highest Dmax we can find for a silver print paper, I still maintain that Karl Lang is overstating the numbers a bit.
« Last Edit: May 28, 2013, 06:18:05 PM by BJL » Logged
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3444


« Reply #26 on: May 28, 2013, 06:16:05 PM »
ReplyReply

I am skeptical about that second number: numerous sources suggest that the reflectivity of pure black on a glossy silver print is about 1.5% (the darkest natural substances, like fine charcoal, reflect about 4%).

Hi,

But glossy surfaces can reach a much higher D-max, because less light is scattered (also back to the observer) and more is reflected away from the observer. An interesting thing is that an extremely black object can be created by folding a glossy(!) black surface into a deep cone shape ... Part of the light will be absorbed when it hits the surface, and the reflected component will be absorbed a few times more before it can ever return to the observer.

Cheers,
Bart
Logged
bjanes
Sr. Member
****
Offline Offline

Posts: 2756



« Reply #27 on: May 28, 2013, 07:19:24 PM »
ReplyReply

Note: Dmax is log base 10, so in the log base 2 units of "stops" more familiar to photographers, that Dmax=2.2 for Ilfobrom Galeria fb is about 7.3 stops below 100% reflectivity, or 158:1, and so a contrast range a bit less: probably a bit over 7 stops and no more than 150:1. Given that this is the highest Dmax we can find for a silver print paper, I still maintain that Karl Lang is overstating the numbers a bit.

Yes, I don't know what paper Karl is using. For those who don't have Excel or a scientific calculator that does log base 2 calculations, one can calculate log base 10 and multiply by log(10, base 2) or 3.32.

Bill
Logged
bjanes
Sr. Member
****
Offline Offline

Posts: 2756



« Reply #28 on: May 28, 2013, 07:40:19 PM »
ReplyReply

Hi,

But glossy surfaces can reach a much higher D-max, because less light is scattered (also back to the observer) and more is reflected away from the observer. An interesting thing is that an extremely black object can be created by folding a glossy(!) black surface into a deep cone shape ... Part of the light will be absorbed when it hits the surface, and the reflected component will be absorbed a few times more before it can ever return to the observer.

Cheers,
Bart

Bart, thanks for pointing that out. I see that you supplied Norman Koren with a link to a post by Roger Clark on rec.photo.digital describing such a light trap. I understand that the camera looks into the cone from the apex. What should one use for the base of the cone and what is the optimum angle for the cone?

Best regards,

Bill

PS. I see that this news group still exists. Does anyone on LuLa follow them? I abandoned them long ago when my favorite newsreader (Gravity) ceased production at that time.

Logged
Schewe
Sr. Member
****
Offline Offline

Posts: 5419


WWW
« Reply #29 on: May 28, 2013, 10:56:21 PM »
ReplyReply

Yes, I don't know what paper Karl is using. For those who don't have Excel or a scientific calculator that does log base 2 calculations, one can calculate log base 10 and multiply by log(10, base 2) or 3.32.

So, given a D-max and D-min, exactly how does one calculate contrast ratio? Also, I believe that certain Epson papers with UltraChrome HDR inks can hit a D-max of 2.5-2.7 (with EFP being near 2.5 and Ultra Premium Glossy being close to 2.7 (I remember 2.68 as a number but not where I read that and I didn't measure that).
Logged
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3444


« Reply #30 on: May 29, 2013, 02:44:30 AM »
ReplyReply

So, given a D-max and D-min, exactly how does one calculate contrast ratio?

Hi Jeff,

Contrast Ratio =10 ^ (D-max - D-min)
Where ^ denotes the power function

So, e.g. with a D-max of 2.20 and a D-min of 0.05, we'd get 10^2.15 = 141, or as a ratio 141 : 1.

Cheers,
Bart
Logged
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3444


« Reply #31 on: May 29, 2013, 03:32:11 AM »
ReplyReply

Bart, thanks for pointing that out. I see that you supplied Norman Koren with a link to a post by Roger Clark on rec.photo.digital describing such a light trap.

Hi Bill,

That's correct, I got that suggestion from a discussion with Roger.

Quote
I understand that the camera looks into the cone from the apex. What should one use for the base of the cone and what is the optimum angle for the cone?

I'm not sure if there is an optimal angle, because it also depends on how directional the light is that enters the cone, and it's dimensions. However, the angle should be narrower than 45 degrees (less is better), in other words the 'depth' must be at least half of the base diameter, but I'd go for at least as deep as it is wide, if it must remain compact.

It would help if the incident light cannot enter the cone at an angle of more than 45 degrees off-normal, by attaching a glossy black tube that functions as a light shade at the base where the light enters, and that would be at least as deep as its diameter. To keep that light shade more compact and more effective, one could use a short reversed cone with the tip cut off for letting the light in (see attachment).

Cheers,
Bart
« Last Edit: May 29, 2013, 06:08:29 AM by BartvanderWolf » Logged
Schewe
Sr. Member
****
Offline Offline

Posts: 5419


WWW
« Reply #32 on: May 29, 2013, 03:51:01 AM »
ReplyReply

So, e.g. with a D-max of 2.20 and a D-min of 0.05, we'd get 10^2.15 = 141, or as a ratio 141 : 1.


Pardon my lack of understanding (I don't do math all that well) so if one had a d-min of .0318 and a d-max of 2.2639, the difference would be 2.2321...so, in a calculator, how do you calculate 10 ^ (D-max - D-min)...using a calculator?
Logged
francois
Sr. Member
****
Offline Offline

Posts: 6731


« Reply #33 on: May 29, 2013, 04:03:00 AM »
ReplyReply

…how do you calculate 10 ^ (D-max - D-min)...using a calculator?

Since you're using a Mac, simply paste 10 ^ (2.20 - 0.05) in the spotlight field and you'll get the result.
Logged

Francois
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3444


« Reply #34 on: May 29, 2013, 04:28:15 AM »
ReplyReply

Pardon my lack of understanding (I don't do math all that well) so if one had a d-min of .0318 and a d-max of 2.2639, the difference would be 2.2321...so, in a calculator, how do you calculate 10 ^ (D-max - D-min)...using a calculator?

Hi Jeff,

No problem, many hand (or smart phone) calculators have a 10x key (may need to press a shift key).
Just calculate the D-max - D-min difference and once you have the result press that 10x key.

When you have a computer at hand, the suggestion from Francois gives the result as well.

Cheers,
Bart
« Last Edit: May 29, 2013, 04:31:27 AM by BartvanderWolf » Logged
AlfSollund
Full Member
***
Offline Offline

Posts: 128


« Reply #35 on: May 29, 2013, 05:13:32 AM »
ReplyReply

My personal vision is about 14 Ev static and 28 Ev Dynamic.

But my left eye is 10/10 and my right only 6/10 Wink

Thierry

Thanks. How did you measure this?
Logged

-------
- If your're not telling a story with photo you're only adding noise -
http://alfsollund.com/
AlfSollund
Full Member
***
Offline Offline

Posts: 128


« Reply #36 on: May 29, 2013, 05:20:16 AM »
ReplyReply

Take a look at this paper.

It argues the stead-state dynamic range of the visual system is 3.7 log units.

http://www.cs.bris.ac.uk/~reinhard/papers/kunkel_apgv2010.pdf

DAve

Thanks for sharing,

Great paper. But what it really show is that "under specific conditions" for the chosen set-up the steady-state dynamic range of the visual system is 3.7 log units. Please note "This display has a backlight consisting of individually modulated LEDs" and "Our results therefore indicate that current display technologies do not yet adequately match the human visual capabilities".

So I really would have loved to see a very different set-up without back-light monitor to measure the dynamic range of the human vision. But one cannot have all for free  Cool.
Logged

-------
- If your're not telling a story with photo you're only adding noise -
http://alfsollund.com/
bjanes
Sr. Member
****
Offline Offline

Posts: 2756



« Reply #37 on: May 29, 2013, 06:05:52 AM »
ReplyReply

Thanks for sharing,

Great paper. But what it really show is that "under specific conditions" for the chosen set-up the steady-state dynamic range of the visual system is 3.7 log units. Please note "This display has a backlight consisting of individually modulated LEDs" and "Our results therefore indicate that current display technologies do not yet adequately match the human visual capabilities".

So I really would have loved to see a very different set-up without back-light monitor to measure the dynamic range of the human vision. But one cannot have all for free  Cool.

Yes, the paper is of interest. I see that the maximal luminance of the display is 3548 cd/m^2. This would drive the Digitaldog (Andrew Rodney) crazy, since he recommends that one should set the luminance of the monitor to about 140 cd/m^2 (Why are my prints too dark).

For softproofing prints one does not need a monitor with a very high contrast ratio, but the contrast ratio should be greater than that of the target print.

Best wishes,

Bill
Logged
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1666


« Reply #38 on: May 29, 2013, 07:13:00 AM »
ReplyReply

Yes, the paper is of interest. I see that the maximal luminance of the display is 3548 cd/m^2. This would drive the Digitaldog (Andrew Rodney) crazy, since he recommends that one should set the luminance of the monitor to about 140 cd/m^2 (Why are my prints too dark).

For softproofing prints one does not need a monitor with a very high contrast ratio, but the contrast ratio should be greater than that of the target print.

Best wishes,

Bill
Soft-proofing is only one, specific usage of displays.

I see no point in holding back display technology to the limits of prints. Rather, it should strive for "realism", and those that needs to simulate print on screen should use profiles and calibration to accomplish good simulation.

If I look out of my window on a bright summer day, the visual appearance is a very bright image compared to the (probably) relatively darker inside of my room. Should a reproduced image (print, display or whatever) be capable of reproducing that sensation, or should its maximum brightness be normalized to the in-room lighting (such as reflective media)?

-h
Logged
digitaldog
Sr. Member
****
Offline Offline

Posts: 8619



WWW
« Reply #39 on: May 29, 2013, 08:45:59 AM »
ReplyReply

This would drive the Digitaldog (Andrew Rodney) crazy, since he recommends that one should set the luminance of the monitor to about 140 cd/m^2 (Why are my prints too dark).

Actually he doesn't say to set cd/m2 to any other setting than the value which produces a visual match to the print:

Quote
The correct value for luminance is one that produces a visual match. You can start at the so-called “recommended” value, which is often in the neighborhood of 120-140cd/m2. That value may need to be lower or higher. You will need to adjust the display luminance until you get that visual match.
« Last Edit: May 29, 2013, 08:48:42 AM by digitaldog » Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
Pages: « 1 [2] 3 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad