![]() |
|
|
|||||||
![]() |
Digital Photography - How Does Auto White Balance Work? |
|
|
Thread Tools | Search this Thread |
|
|
#1 |
|
Posts: n/a
|
I've mad three attempts at Google-searching for a good explanation of
how auto white balance works, and how to tell which method a given camera uses. My feeble understanding gives two ways it works. Method 1 -- The age old method Kodak developed for mass color printing. The entire image is mixed together (like in a blender). The result is assumed to be neutral gray. This result is then used to color balance the entire image.. dpreview's definition seems to say this is *THE* way it works. http://www.dpreview.com/... ..../Glossary/Digital_Imaging/White_Balance_01.htm "Most digital cameras feature automatic white balance, this means that the camera looks at the overall colour of the image and calculates the best-fit white balance, however these systems are often fooled (especially if taking a photograph dominated by one colour, say green)." Method 2 - The algorithm searches the image for the lightest area and assumes this is white. This area is then used to color balance the entire image. I'm under the impression method 2 is more common in prosumer and DSLRs. I don't know how to tell which is being used by a given camera. I assume there may be other methods, but these two are the only ones I'm aware of. Does anybody know where there's a good explanation of auto white balancing they can point me to? jim |
|
|
|
#2 |
|
Posts: n/a
|
jim evans <> wrote in
news:: > I've mad three attempts at Google-searching for a good explanation of > how auto white balance works, and how to tell which method a given > camera uses. Each camera probably has it's own propreitary algorithm -- Lucas Tam () Please delete "REMOVE" from the e-mail address when replying. http://members.ebay.com/aboutme/coolspot18/ |
|
|
|
#3 |
|
Posts: n/a
|
On Thu, 26 Feb 2004 16:54:12 GMT, Lucas Tam <>
wrote: >jim evans <> wrote in >news: : > >> I've mad three attempts at Google-searching for a good explanation of >> how auto white balance works, and how to tell which method a given >> camera uses. > >Each camera probably has it's own propreitary algorithm It seems unlikely there are as many approaches to white balancing as there are cameras. jim |
|
|
|
#4 |
|
Posts: n/a
|
jim evans <> wrote in
news:: >>Each camera probably has it's own propreitary algorithm > > It seems unlikely there are as many approaches to white balancing as > there are cameras. No, but who knows what's in the firmware except for the programmer(s). You maybe able to find some information for Pro grade SLRs, but I doubt for consumer grade cameras. -- Lucas Tam () Please delete "REMOVE" from the e-mail address when replying. http://members.ebay.com/aboutme/coolspot18/ |
|
|
|
#5 |
|
Posts: n/a
|
"jim evans" <> wrote in message news:... > "Most digital cameras feature automatic white balance, this means that > the camera looks at the overall colour of the image and calculates the > best-fit white balance, however these systems are often fooled > (especially if taking a photograph dominated by one colour, say > green)." > > Method 2 - The algorithm searches the image for the lightest area and > assumes this is white. This area is then used to color balance the > entire image. > > I'm under the impression method 2 is more common in prosumer and > DSLRs. > > I don't know how to tell which is being used by a given camera. > > I assume there may be other methods, but these two are the only ones > I'm aware of. > > Does anybody know where there's a good explanation of auto white > balancing they can point me to? It's WAY more complicated than this. In fact, fooling the cameras these days with different kinds of lights and different colors doesn't happen. The camera firmware is significantly smarter about this (at least the Kodak ones) It's also proprietary, so I can't tell you. |
|
|
|
#6 |
|
Posts: n/a
|
"jim evans" <> wrote in message news:... > I've mad three attempts at Google-searching for a good explanation of > how auto white balance works, and how to tell which method a given > camera uses. <snip> I would assume the camera's software computes the average red, green, and blue pixel values, then using the assumption that the entire scene should average out to grey-ish at the recorded luminance level, the software would determine the color bias and subtract the latter from each (R,G,B) triple comprising the photo. If I were a camera, that's the kind of camera I would be, anyway. Funny, I just did a Google search using the arcane and cryptic search phrase "white balance explained" (include quotes) and came up with a ton of hits including http://www.3dvelocity.com/reviews/4800/fujirev-4.htm which contains an explanation agreeing with my very own pre-search hypothesis/conjecture. Damn, but I'm one clever little monkey, ain't I? And I smell good, too! That's why ZZ Top sang "every girl crazy 'bout a sharp-brained man!" Uh-huh. |
|
|
|
#7 |
|
Posts: n/a
|
On Sat, 28 Feb 2004 21:56:36 -0800, "Paul H."
<> wrote: >>I've mad three attempts at Google-searching for a good explanation of >>how auto white balance works, and how to tell which method a given >>camera uses. >> >>My feeble understanding gives two ways it works. >> >>Method 1 -- The age old method Kodak developed for mass color >>printing. The entire image is mixed together (like in a blender). The >>result is assumed to be neutral gray. This result is then used to >>color balance the entire image.. >> >>Method 2 - The algorithm searches the image for the lightest area and >>assumes this is white. This area is then used to color balance the >>entire image. >> >>I'm under the impression method 2 is more common in prosumer and >>DSLRs. >> >>I assume there may be other methods, but these two are the only ones >>I'm aware of. >> >>Does anybody know where there's a good explanation of auto white >>balancing they can point me to? > >I would assume the camera's software computes the average red, green, and >blue pixel values, then using the assumption that the entire scene should >average out to grey-ish at the recorded luminance level, the software would >determine the color bias and subtract the latter from each (R,G,B) triple >comprising the photo. > >If I were a camera, that's the kind of camera I would be, anyway. > >Funny, I just did a Google search using the arcane and cryptic search phrase >"white balance explained" (include quotes) and came up with a ton of hits What a surprise. It's about 678,000 hits to be more exact. The problem is finding a needle in this haystack. The great majority of these hits aren't relevant to auto white balance at all, and most that are relevant explain what auto white balance is, what it does, not how it does it. >including > >http://www.3dvelocity.com/reviews/4800/fujirev-4.htm > >which contains an explanation agreeing with my very own pre-search >hypothesis/conjecture. Damn, but I'm one clever little monkey, ain't I? And >I smell good, too! That's why ZZ Top sang "every girl crazy 'bout a >sharp-brained man!" Uh-huh. Your clever "hypothesis/conjecture" is the method I called Method 1 in my original post -- the post you were responding to. I do not agree that the reference you quote is describing this method. They say: "[T]he AUTO setting looks at the colour of light available, and tries to adjust it so it appears neutral in the final photo." Looking at the "colour of light available" is not the same as blending/mixing/averaging together all the colors/light that is reflected from the scene, which is my Method 1 and your "hypothesis/conjecture." Looking at the "colour of light available" would be a third method: Method 3 -- the camera senses the ambient/available light and uses this measurement to adjust the color of the image. jim |
|
|
|
#8 |
|
Posts: n/a
|
[POSTED TO rec.photo.digital - REPLY ON USENET PLEASE]
In <> on Sun, 29 Feb 2004 13:09:59 -0600, jim evans <> wrote: >On Sat, 28 Feb 2004 21:56:36 -0800, "Paul H." ><> wrote: >>including >> >>http://www.3dvelocity.com/reviews/4800/fujirev-4.htm >> >>which contains an explanation agreeing with my very own pre-search >>hypothesis/conjecture. Damn, but I'm one clever little monkey, ain't I? And >>I smell good, too! That's why ZZ Top sang "every girl crazy 'bout a >>sharp-brained man!" Uh-huh. > >Your clever "hypothesis/conjecture" is the method I called Method 1 in >my original post -- the post you were responding to. > >I do not agree that the reference you quote is describing this method. > >They say: >"[T]he AUTO setting looks at the colour of light available, and tries >to adjust it so it appears neutral in the final photo." > >Looking at the "colour of light available" is not the same as >blending/mixing/averaging together all the colors/light that is >reflected from the scene, which is my Method 1 and your >"hypothesis/conjecture." > >Looking at the "colour of light available" would be a third method: > >Method 3 -- the camera senses the ambient/available light and uses >this measurement to adjust the color of the image. How would a camera (with only the sensor behind the lens) sense "the ambient/available light" without a second sensor, or pointing the camera at the light source? Except for (blue) sky, a camera is normally sensing only *reflected* light. -- Best regards, John Navas [PLEASE NOTE: Ads belong *only* in rec.photo.marketplace.digital, as per <http://bobatkins.photo.net/info/charter.htm> <http://rpdfaq.50megs.com/>] |
|
|
|
#9 |
|
Posts: n/a
|
On Sun, 29 Feb 2004 20:10:50 GMT, John Navas
<> wrote: >>Method 3 -- the camera senses the ambient/available light and uses >>this measurement to adjust the color of the image. > >How would a camera (with only the sensor behind the lens) sense "the >ambient/available light" without a second sensor, or pointing the camera at >the light source? Except for (blue) sky, a camera is normally sensing only >*reflected* light. I've read things that seem to say some cameras have a separate sensor for ambient light which they use either alone or in combination with the light coming through the lens to do auto white balance. I don't understand how all this works, that's one of the reasons for my post. jim |
|
|
|
#10 |
|
Posts: n/a
|
[POSTED TO rec.photo.digital - REPLY ON USENET PLEASE]
In <> on Sun, 29 Feb 2004 23:05:14 -0600, jim evans <> wrote: >On Sun, 29 Feb 2004 20:10:50 GMT, John Navas ><> wrote: >>>Method 3 -- the camera senses the ambient/available light and uses >>>this measurement to adjust the color of the image. >> >>How would a camera (with only the sensor behind the lens) sense "the >>ambient/available light" without a second sensor, or pointing the camera at >>the light source? Except for (blue) sky, a camera is normally sensing only >>*reflected* light. > >I've read things that seem to say some cameras have a separate sensor >for ambient light which they use either alone or in combination with >the light coming through the lens to do auto white balance. ... I think you'll be hard pressed to actually find one. -- Best regards, John Navas [PLEASE NOTE: Ads belong *only* in rec.photo.marketplace.digital, as per <http://bobatkins.photo.net/info/charter.htm> <http://rpdfaq.50megs.com/>] |
|
![]() |
| Thread Tools | Search this Thread |
|
|
Similar Threads
|
||||
| Thread | Thread Starter | Forum | Replies | Last Post |
| Identity Theft at Work @ A True Review | Silverstrand | Front Page News | 0 | 06-27-2006 02:56 PM |
| bush,jr spoke, bush,jr lied | Sid9 | DVD Video | 3 | 06-03-2006 06:37 AM |
| ""Crash" is a white-supremacist movie." by Robert Jensen | JAS | DVD Video | 6 | 05-26-2006 01:12 AM |
| DVD review: GEORGE W. BUSH: FAITH IN THE WHITE HOUSE | RFCSAC627N | DVD Video | 9 | 10-18-2004 03:44 AM |
| auto start, and auto replay possible? | Dik Mus | DVD Video | 3 | 10-28-2003 11:09 AM |