High Dynamic Range Image

kikidicksen

Forum Enthusiast
Messages
327
Reaction score
0
Location
plaine-wilhems, MU
Hello folks,as technology is progressing,what will b the dynamic range of a digital camera in the future ?...will it surpass the human eye,in my opinion scientists are already scratching their head to how to do it..by the way if we have to rate the human eye in terms of ASA wat shud it be 20,000 or so ? imagine a camera capable of caturing detail in the extreme highlight and extreme shadow WOW.. so i invite u people to discuss on the matter.

THX
KUNAL

Canon 20D & Nikon D70
 
Hello folks,as technology is progressing,what will b the dynamic
range of a digital camera in the future ?
Presumably more than today.
will it surpass the human eye
Depends on the brightness and contrast of the scene.
by the way if we have to rate the human eye
in terms of ASA wat shud it be 20,000 or so ?
You're way, way off in the wrong direction. This page claims about 1 during daylight, 800 when dark-adapted:

http://clarkvision.com/imagedetail/eye-resolution.html
imagine a camera
capable of caturing detail in the extreme highlight and extreme
shadow WOW.
Not too hard if you combine shots. Now imagine an output technology that shows that full dynamic range. Paper won't do it for you, neither will displays.

--
Seen in a fortune cookie:
Fear is the darkroom where negatives are developed
 
The brain composites images.

You look at a bright scene with the eye stopped down and then a moment later you look at a dark scene with the eye opened up and your brain produces a seemless composite of the image.

On top of being able to stop down, your eye also has variable sensitivity, like changing ISO.

also your eye has great resolution in the centre, but lousy resolution towards the edge of the frame. again the brain composites the detail as it becomes available when the good part of your vision scans that part of the frame.

etc. etc.

It really uses every trick in the book to make such great images.

It's not all about the sensor or the camera. the post processing is a big part too.

--

Graphic Design + Web Design + Photography

http://www.molecule.com.au
 
analysis Mike. We may be chasing a moving target with our search for ever higher dynamic range. I wonder whether digital might already exceed the human dynamic range without the dynamic PP adjustments the human eye constantly makes.

I also wonder what a very high dynamic range picture would look like. Would it, for example, lack definition and contrast. Perhaps the compromises the digital camera makes are the state of the "art", i.e., more pleasing and acceptable than other representations. I see this tendency already where oversaturated images often seem to be "better pictures".



Oly 2040, 4000(deceased), 5050, 8080 http://humphrey.smugmug.com/
 
Today we have (normally) 12-bit AD converters in digicams.

They are fast and cheap and give us 4096 steps. 14 and 16 bits are slower and more expensive. We have them in scanners. They will replace 12 bit

and we will get 16 384 and 65 536 steps. In the PC we use RGB with 8 bits. It will also move up to something like 14 or 16 bits. Converting RAW
to something we can see will be less complex than today.
 
Today we have (normally) 12-bit AD converters in digicams.
They are fast and cheap and give us 4096 steps. 14 and 16 bits are
slower and more expensive. We have them in scanners. They will
replace 12 bit
14 and 16 bit aren't much slower or more expensive, they are simply redundant with most sensors because there's only 12 bits of data to encode.
In the PC we use RGB with 8 bits. It will also move up to something like 14 or 16 bits.
The 8 bits you refer to are usually gamma encoded, the 12 bits on the camera are linear. Both can encode a simillar range. There is no point in moving to 14 bit encodings on the PC, as 16 bits works just fine.

--
Seen in a fortune cookie:
Fear is the darkroom where negatives are developed
 
Hello folks,as technology is progressing,what will b the dynamic
range of a digital camera in the future ?...will it surpass the
human eye,in my opinion scientists are already scratching their
head to how to do it..
Sure. You just have to build a camera that allows the darkest-exposed photosites ample time to collect photons while not blowing out the brightest-exposed photosites.

There are already cameras that are doing this sort of thing. They just haven't made it to consumer still cameras.

But remember that there is no way at the moment to display all that range. We can compress and display it, but....

--
bob

The Blind Pig Guild
A photo/travel club looking for members
http://www.jeber.com/Clubs/Blind-Pig/

Flowers of Asia
A photo club for appreciators of Asian flowers - looking for members
http://www.jeber.com/Clubs/Flowers-of--Asia/

Travel Galleries
http://www.pbase.com/bobtrips
 
at the moment of capture, where we colloct the data needed.

In post process we then need to compress that dynamic range to suit whatever out put device we want, and match those.

That means about 5 stops for monitors and 3,5 stops for prints.

But we still want the data intact so that we can get a pleasent mix of high lights, shadows and mid tones.
--
Anders

Some of my pictures can be seen at;
http://teamexcalibur.se/US/usindex.html

event photography and photo journalism
 
you get more dynamic range by capturing more photons to count without changing the timeframe. The most obvious way of doing that is using larger photosites, which in turn often means using larger sensors.

The bit depth only tells us how fine the difference in luminance is described.
--
Anders

Some of my pictures can be seen at;
http://teamexcalibur.se/US/usindex.html

event photography and photo journalism
 
not bits.
I think he meant bits.
And we're above 12 stops now on digital, so why not...
Most digital cameras can do a maximum of 12 stops. That's the theoretical limit, based on linear encoding 12 bits. Actual number of stops is a bit lower, depends on ISO, and is subjective. Somewhere in the ballpark of 8 or 9 usable stops is pretty good.

--
Seen in a fortune cookie:
Fear is the darkroom where negatives are developed
 
even if I agree that we need sufficient bit depth to encode the data.

Most backs use 16 bit now to describe 12 stops, but I'm confident that low energy, low heat, low interference 20-24 bit ad converters will soon find their way into cameras when we get even bigger photosites.
--
Anders

Some of my pictures can be seen at;
http://teamexcalibur.se/US/usindex.html

event photography and photo journalism
 
at the moment of capture, where we colloct the data needed.
Close, I think.

But large DR is more than interesting. I think its about getting to the processing stage with as much information about brightly lit detail and dimly lit detail as possible.
In post process we then need to compress that dynamic range to suit
whatever out put device we want, and match those.

That means about 5 stops for monitors and 3,5 stops for prints.

But we still want the data intact so that we can get a pleasent mix
of high lights, shadows and mid tones.
--
Anders

Some of my pictures can be seen at;
http://teamexcalibur.se/US/usindex.html

event photography and photo journalism
--
bob

The Blind Pig Guild
A photo/travel club looking for members
http://www.jeber.com/Clubs/Blind-Pig/

Flowers of Asia
A photo club for appreciators of Asian flowers - looking for members
http://www.jeber.com/Clubs/Flowers-of--Asia/

Travel Galleries
http://www.pbase.com/bobtrips
 
even if I agree that we need sufficient bit depth to encode the data.

Most backs use 16 bit now to describe 12 stops, but I'm confident
that low energy, low heat, low interference 20-24 bit ad converters
will soon find their way into cameras when we get even bigger
photosites.
Most digital cameras, including most DSLRs, encode 12 bits. One DSLR (the S3 Pro) encodes 14 bits. Medium format backs typically encode anywhere up to 16 bits.

All of this doesn't much matter, though. You can encode 16 bits, but if you don't have 96dB snr you will not be able to extract 16 bits of useful data.

Quick math -- you need one bit for every 6dB (roughly). Kodak 39MP medium format CCD is 71.4dB. How many bits do you need to encode the output of this sensor?

12.

Kodak data sheet (see page 4):

http://www.kodak.com/ezpres/business/ccd/global/plugins/acrobat/en/datasheet/fullframe/KAF-39000LongSpec.pdf

Converting dB to bits:

http://en.wikipedia.org/wiki/Signal-to-noise_ratio#Fixed_point

--
Seen in a fortune cookie:
Fear is the darkroom where negatives are developed
 
actually agree? ;)

In order to benefit from larger bit depth we need more dynamic range to describe. We COULD chop up the DR we have in finer slices, but unless we really need to create larger files I see no real use for it.

When I was in the music industry I always found it amusing when audiophiles went on about dynamic range and frequency respons.

I still find it amusing;)
--
Anders

Some of my pictures can be seen at;
http://teamexcalibur.se/US/usindex.html

event photography and photo journalism
 
actually agree? ;)
I guess I didn't get it, since you talking about 20 and 24bits. Even with bigger photosites 16 bits should be enough.

But for marketing purposes I can see why digiback and digicam manufacturers would want to advertise "24 bits!" even if the sensor output could be encoded in 12 bits.
In order to benefit from larger bit depth we need more dynamic
range to describe.
Bingo!
When I was in the music industry I always found it amusing when
audiophiles went on about dynamic range and frequency respons.
There are just too many things amusing about audiophiles, or at least what they are suckered into buying:

http://www.ilikejam.dsl.pipex.com/audiophile.htm

My favorite:

http://www.referenceaudiomods.com/Merchant2/merchant.mvc?Screen=PROD&Product_Code=NOB_C37_C&Category_Code=VOLUME&Product_Count=2

--
Seen in a fortune cookie:
Fear is the darkroom where negatives are developed
 
The problem with current DSLRs is that they have so much readout noise at the cameras' lowest ISOs, that this noise limits DR by at lest 3 or 4 stops more than the sensor itself does.

--
John

 
actually agree? ;)
I guess I didn't get it, since you talking about 20 and 24bits.
Even with bigger photosites 16 bits should be enough.
Guess I just didn't want to take a guess on just how large they'll make photosites in the future, but I suspect you are right.
But for marketing purposes I can see why digiback and digicam
manufacturers would want to advertise "24 bits!" even if the sensor
output could be encoded in 12 bits.
In order to benefit from larger bit depth we need more dynamic
range to describe.
Bingo!
When I was in the music industry I always found it amusing when
audiophiles went on about dynamic range and frequency respons.
There are just too many things amusing about audiophiles, or at
least what they are suckered into buying:
Your links reminded me of a discussion I heard some engineers having regarding the audio quality after the (still digital) signals had been sent through spdif vs aes/ebu cables...

Now, where did I put those gold plated optical cables...

--
Anders

Some of my pictures can be seen at;
http://teamexcalibur.se/US/usindex.html

event photography and photo journalism
 

Keyboard shortcuts

Back
Top