Dynamic range

[IWC Doppel]

Does anyone know how many stops most monitors will show ?

I am aware of colour gamut limitations, but will my iMAC show sufficient dynamic range for my camera (Leica M9-P), I know this is 'only 11.7' stops or so some will say (Erwin putts recorded higher figures from memory), never realy though about this

[ErikKaffehr]

Hi,

9 stops or less, IMHO.

Best regards
Erik

Does anyone know how many stops most monitors will show ?

I am aware of colour gamut limitations, but will my iMAC show sufficient dynamic range for my camera (Leica M9-P), I know this is 'only 11.7' stops or so some will say (Erwin putts recorded higher figures from memory), never realy though about this

[hjulenissen]

Modern computer monitors may have a regular contrast ratio approaching 1000:1 (10 stops). I believe that prints may have in the range of 100:1 or less (7 stops).

http://www.tftcentral.co.uk/reviews/dell_u2713h.htm
http://www.tftcentral.co.uk/reviews/nec_p232w.htm

Dynamic range for cameras is the ratio of clipping point to noise floor. Contrast ratio for monitors is the ratio of brightest to darkest display value. It may not make sense to compare those two directly.

It may be as important to worry about display linearity (gamma), as lcd displays are not inherently linear (as cameras are), and measuring/correcting the shadows seems to be non-trivial.

Even a high-DR display will only be able to show its DR in a room with suitable ambient lighting.

I calibrate my display for paper appearance (low brightness, warm whitepoint compared to defaults), and allow daylight into my room. I also use only 8-bits connection to my computer. So I probably dont have that much real CR.

-k

[Jack Hogan]

Modern computer monitors may have a regular contrast ratio approaching 1000:1 (10 stops). I believe that prints may have in the range of 100:1 or less (7 stops).

To add to what hjulenissen said above, the majority of sRGB monitors out there have contrast ratios of 500:1 or less (i.e. 8 or 9 stops) at normal viewing brightness (90-120 cd/m2).  When you calibrate/profile a monitor some software will report its contrast ratio.  My wide gamut Dell U2410 is just about 500:1, while my other monitor, the run of the mill ST2210, clocks in at about 280:1.

Jack

[Guillermo Luijk]

Manufacturer specs are always optimistic. Some even provide the dynamic DR, i.e. the difference between max and min lightness after doing adjustments in the device's brightness controls.

I measured my monitor's (HP LP2475W) DR through a simple test based on two patches: white and black. The result under regular daylight lighting conditions was 6,7 stops.

This was the test capture:




This was the resulting EV histogram showing the DR:




In the test I also measured print DR resulting only 4,3 stops:






Of course different monitors and adjustments, paper and ink, lighting conditions,... will definitively impact in the achieved DR as perceived by the observer's eye.

Regards

[Jim Kasson]

Modern computer monitors may have a regular contrast ratio approaching 1000:1 (10 stops).

Maybe, but to get that kind of nosebleed range in use, you have to turn off all the lights, wear black clothes, and maybe a black ski mask...

Jim

[hjulenissen]

Maybe, but to get that kind of nosebleed range in use, you have to turn off all the lights, wear black clothes, and maybe a black ski mask...

Jim

Yes, I believe that I mentioned that:

Even a high-DR display will only be able to show its DR in a room with suitable ambient lighting.

The core issue seems to be that "black" display pixels will only be as black as charcoal until some clever nano-tech solves it. So light incident on the display will raise black levels without affecting white levels.

-h

[hjulenissen]

Manufacturer specs are always optimistic. Some even provide the dynamic DR, i.e. the difference between max and min lightness after doing adjustments in the device's brightness controls.

I believe that "dynamic DR" is usually used when you have dynamic backlighting (adjusting the backlight dynamically based on the brightest pixel) and/or zoned backlighting (adjusting the lighting within a spatial window in much the same way). Depending on how the measurement is carried out, this can give CR numbers that are very large but may not correlate with user satisfaction.

That is the reason that I wrote "regular contrast ratio" in my initial post.

I measured my monitor's (HP LP2475W) DR through a simple test based on two patches: white and black. The result under regular daylight lighting conditions was 6,7 stops.

This was the test capture:




This was the resulting EV histogram showing the DR:




In the test I also measured print DR resulting only 4,3 stops:

Of course different monitors and adjustments, paper and ink, lighting conditions,... will definitively impact in the achieved DR as perceived by the observer's eye.

Regards

That is interesting. The same (5 years old) monitor is tested here, and DR is reported to be 357/0.45 [cd/m2] = 793:1
"The LP2475W was tested at default factory settings out of the box using the LaCie Blue Eye Pro and their accompanying software suite."
http://www.tftcentral.co.uk/reviews/hp_lp2475w.htm

So do you use very different display settings (reducing brightness may decrease DR on some monitors), do you pick up more ambient light than tftcentral did, or is there some significant flaw in one of the tests?

-h

[Bart_van_der_Wolf]

So do you use very different display settings (reducing brightness may decrease DR on some monitors), do you pick up more ambient light than tftcentral did, or is there some significant flaw in one of the tests?

Hi,

There is a difference between measuring the light output with an instrument that shields a certain amount of ambient room light (emission only), and a camera shot of the display (emission + ambient). The latter is arguably going to be closer to what we will see ourselves (assuming we don't wear a white shirt which adds light reflection off the screen surface).

Cheers,
Bart

[IWC Doppel]

If it helps I am very aware of how reflection off a matt white screen can effect contrast, but displays are obviously very different, I guess paper is closer to this situation.

Fortunately you do not need to achieve effective black via reducing internal room reflections, you have printed this and the light falling will be even on the paper.

[ErikKaffehr]

Hi,

I measured mine with a spotmeter. Compared white and black in Photoshop and got EV 1.7 - 8.3 -> 6.6 EV or 1:95

Increasing brightness to maximum shifted the to 2.7 -- 9.4 EV, the change in DR was probably less than the error margin.

My monitor is a calibrated Eizo FlexScan SX2461W monitor, it is above a standard monitor but below Color Edge series.

Best regards
Erik

[Jim Kasson]

I measured mine with a spotmeter. Compared white and black in Photoshop and got EV 1.7 - 8.3 -> 6.6 EV or 1:95

Erik,

Did you measure two different screens, one all white and one all black? That will tell you what more or less what the monitor will do with no one in front of it. If you have two equal patches, one black and one white, on the screen at the same time, that's closer to a realistic test. Taking a typical image and embedding a black and a white square is probably the most realistic, especially if you sit with he meter in your hand where you'd be if you were actually editing.

Jim

[ErikKaffehr]

Hi,

A large black patch surrounded by a large white area. Used a Minolta Spotmeter F. Made several measurements which were quite consistent.

I think he Spotmeter F measures peak intensity, but I think the results are quite valid in this case.

I repeated the measurements in a dark room, I got 1.8 -- 8.2 EV : 1:83 or 6.4 stops.

I also measured with a very old Minolta Autometer II and got 2.5 -- 8.5 EV, or 1:64 or 6 stops.

I could also mention that I have analyzed quite a lot of raw images using RawDigger and found a single image having a DR more then 9 stops, and that was a dupe I made from a Velvia slide, after shielding out all light leaks. Most images fit comfortably in a 9 stop range and that is mostly due to lens flare. Yes, I can get a DR of 12 stops on a Stouffer wedge but it is no easy.

Best regards
Erik

Erik,

Did you measure two different screens, one all white and one all black? That will tell you what more or less what the monitor will do with no one in front of it. If you have two equal patches, one black and one white, on the screen at the same time, that's closer to a realistic test. Taking a typical image and embedding a black and a white square is probably the most realistic, especially if you sit with he meter in your hand where you'd be if you were actually editing.

Jim

[Jim Kasson]

A large black patch surrounded by a large white area. Used a Minolta Spotmeter F. Made several measurements which were quite consistent.

Then those are probably good, real-world numbers.  Thanks.

Jim

[Jack Hogan]

I could also mention that I have analyzed quite a lot of raw images using RawDigger and found a single image having a DR more then 9 stops, and that was a dupe I made from a Velvia slide, after shielding out all light leaks. Most images fit comfortably in a 9 stop range and that is mostly due to lens flare.

Very interesting, I am going to have to check mine.  I am curious, did your sample include Exmor sensors and landscapes?

[ErikKaffehr]

Hi,

Yes, the images I checked were mostly Exmors and landscapes and shots of Stouffer wedges (the 41x0.1ND version). I used mostly RawDigger.

I also checked some MF raws (IQ180 and IQ160) I got from Tim Ashley and Doug Peterson.

Most landscapes I have seen generally fit in the 9 EV range. Tim's image was intended to test DR and is in the 12EV range.

I normally use RawDigger and use logarithmic Y-scale and assume that histogram count under 5k are noise.

I hope it is OK to use Doug Peterson's image. He posted the same image before on these forums. I just post the raw histogram, and I hope it is OK.


Best regards
Erik

Very interesting, I am going to have to check mine.  I am curious, did your sample include Exmor sensors and landscapes?

[ErikKaffehr]

Hi,

Here are some more samples. Lighthouse with new Sony Alpha 99 and old Sony Alpha 900.

Also a picture with sun in the image.

The second picture with sun included takes about 12 EV on sensor.

Yes, some of the samples I posted are well beyond 9 EV, which contradicts my initial statement.

Best regards
Erik

[ErikKaffehr]

I looked at some more recent pictures, and found that a landscape image including the sun can have a very excessive DR.

I also found an image I thought would need HDR "The Forsmark Image", but I actually preferred the non HDR image. That one has also wide dynamic range. It actually has decent shadow detail.

So, I have found quite a few images exceeding nine EV, I guess I have proven my self wrong!

Best regards
Erik

[Tim Lookingbill]

Erik, can you show a 100% view crop of shadow detail brightened in one of the sun shots to show how much meaningful detail is present? Those trailing RGB vertical bars on the shadow end graphed in Rawdigger screengrab is difficult for me to relate EV levels to reproducing what the human eyes see.

Thanks,

Tim

P.S. I appreciate the hard work and effort in your demonstrations and input on this subject.

[NikoJorj]

I looked at some more recent pictures, and found that a landscape image including the sun can have a very excessive DR.

A good recipe may also be a sunrise or sunset picture, with the foreground lighted solely by the dawn/dusk sky.
This one has a bit more than 9EV - it's hard to shoehorn that in a goodol'Rebel raw file, and I pityfully underexposed, but can use not having much assistance from the camera as a lame excuse.