|
|
Posted by Gene E. Bloch on 04/09/07 18:01
Rick Merrill <rick0.merrill@NOSPAM.gmail.com> wrote in
news:XPudnT0hEcqRCorbnZ2dnUVZ_rmdnZ2d@comcast.com:
> Gene E. Bloch wrote:
>> "AnthonyR." <nospam@sorryspammers.com> wrote in
>> news:46152706.3000301@sorryspammers.com:
>>
>>> Richard Amirault wrote:
>>>> "Gene E. Bloch" wrote ...
>>>>> I would be willing to bet (OK, maybe only a nickel) that your
>>>>> camcorder has custom white balance, but I'm too lazy right now
>>>>> to download the manual and find out for myself.
>>>>>
>>>>> Although that's not called manual, it does give a comparable
>>>>> capability.
>>>> I have a Sony Digital 8 that *only* has automatic color
>>>> balance. My Sony Mini-DV has adjustable white balance.
>>>>
>>>> For the original poster ... What kind of circumstances are you
>>>> shootiing? What is the color of the subject? What is the color
>>>> of the background? What kind of room? Daytime? Nighttime?
>>>> Windows? Are the lightbulbs all the same type (mixed
>>>> incandesant / flourescant?)
>>>>
>>> Richard, when you say the camera has automatic white balance,
>>> what do you mean? How does it balance it? How do you tell it
>>> what is white? And if it's automatic, how does it do it?
>>> Usually, every camera I've ever owned even the cheapest, let's
>>> you point to a white object and hit a button so it can adjust to
>>> that being white.
>>>
>>> Does your manual in the Sony Digital 8 explain how it can
>>> possible determine a white subject on it's own (automatically)?
>>> It might have settings for daylight, tungsten, florescent
>>> etc...I can understand that, but auto, how?
>>>
>>> AnthonyR.
>>>
>>
>> How about the simple way? Find out the average color of the
>> scene, assume that that average corresponds more or less to white
>> - or grey, which is a dirty white :-) - and choose a balance that
>> makes that average color look white.
>>
>> This amounts to adjusting the gain of two of the colors so that
>> their average intensity equals that of the other color. Let us
>> say, adjust the levels of red and blue to match that of green.
>>
>
> Your method is too slow. Find the most white and the most dark and
> make them so and interpolate all other intensities between those
> ends.
Perhaps you could restate the above so that it makes sense, not
forgetting that we're in the context of computing white balance...
Since today's typical still camera can compute and display a
complete histogram in real time (some cameras display three
histograms, R, G, & B), and since the white balance has to be done
only once per session (if so desired), I wonder where you got the
idea that my idea is too slow. Also, the computation I described is
just one extra step per pixel (incrementing a count in an array
indexed on the digitized value), so it could indeed be continuous if
desired, with no problem.
It also has the advantage that it is logically correct. I don't
think your proposed method is, but since it is so oddly phrased, I
can't be sure what you're proposing.
--
Gene E. Bloch (Gino) ... letters617blochg3251
(replace the numbers by "at" and "dotcom")
Navigation:
[Reply to this message]
|