Anonymous user
Talk:Grayscale image: Difference between revisions
→Linear color space
(Apples and oranges) |
|||
| (2 intermediate revisions by 2 users not shown) | |||
Line 36:
:::::::That does not make any sense, sorry. --[[User:Dmitry-kazakov|Dmitry-kazakov]] 17:39, 23 January 2009 (UTC)
::::::::These facts make perfect sense to anybody who has even basic knowledge about computers. Obviously you are not one of them. May I suggest that you get one of those "Computers for dummies" books and start learning the basics. --[[User:PauliKL|PauliKL]] 19:45, 25 January 2009 (UTC)
::::::But adding two pixels in non-linear color space does not give correct results. That was my original message. If you are editing an image in an image editor and you want to do it right, you should first convert the image into linear color space. (In case of RGB to grayscale conversion, the image editor probably does the color space conversion automatically.) --[[User:PauliKL|PauliKL]] 11:20, 23 January 2009 (UTC)
:::::::When something does not give correct results, then it is called incorrect. Which in this case is another way to say that it cannot be added, like apples and oranges. See [http://en.wikipedia.org/wiki/Pixel pixel] --[[User:Dmitry-kazakov|Dmitry-kazakov]] 17:39, 23 January 2009 (UTC)
::::::::That is just plain ridiculous. There is no law of nature that would stop people from making mistakes. If there was, accidents would never happen, and for example at school everybody would always get just straight A's. The fact is that it is perfectly possible to perform image editing operations in non-linear color space. And ignorant people like you do that all the time, and therefore get bad results. That is why I gave the advice. --[[User:PauliKL|PauliKL]] 19:45, 25 January 2009 (UTC)
:::::::::I don't know what did you mean by saying this. An ability to implement certain image operations on images represented using certain color model does not imply that you can do it without any model. The rest is just a lack of some basic understanding of the fundamental difference between representation (computer model) and the modeled thing. Did you have a [http://en.wikipedia.org/wiki/Computer_science CS] course? In short, you can imagine it as the memory device having states associated with things being modeled, like number, color point, text, employee, program etc. The program itself is merely description of state transitions, see [http://en.wikipedia.org/wiki/Finite_state_machine FSM]. Then you might try to understand the difference between a hardware instruction (merely controlling transitions) and what this instruction is about to implement (like the [http://en.wikipedia.org/wiki/Ring_(mathematics) ring] operation +). That would help you to learn why there are so many numbers in computing (modular, ranged, saturated, fixed-point, packed decimal, floating-point etc). This might save you many silly discussions in the future. Pixel is not a number in any possible sense. It is not as a model ([http://en.wikipedia.org/wiki/Palette_(computing) for example]), in case you mixed it with what it represents, namely a color point. On the other side a color space is a [http://en.wikipedia.org/wiki/Vector_space vector space], not a [http://en.wikipedia.org/wiki/Ring_(mathematics) ring], [http://en.wikipedia.org/wiki/Field_(mathematics) field] or other mathematical structure associated with vague "number." Furthermore, in the color space the operations on the points can be defined differently depending on semantic meaning of the arguments and the result. For example, when addition of colors means mixing corresponding water-color pigments on white paper as observed by a statistically normal human, or else whatsoever. I hope that helps. --[[User:Dmitry-kazakov|Dmitry-kazakov]] 10:16, 26 January 2009 (UTC)
::::::::::Yes, it helped to confirm that, indeed, you do not know anything about computers, and definitely nothing about computer graphics. You are right in one thing: you do not understand. You do not understand what pixels are. You do not understand that everything in computer memory is numbers and can be manipulated. Heck, you do not even understand that it is possible to make mistakes! Not even after it was explained to you!
::::::::::Obviously, it is hopeless to try and explain things to Dmitry-kazakov. But perhaps there are others who are willing to learn. --[[User:PauliKL|PauliKL]] 20:24, 28 January 2009 (UTC)
: As I said earlier, most '''image processing functions work correctly only in linear [[wp:color space|color space]]''' (gamma 1.0), since that is what they have been designed for. However, most images are in [[wp:sRGB|sRGB]] color space, which uses [[wp:Gamma correction|gamma]] of approximately 2.2. Therefore, it is advisable to convert the image into linear color space for editing, and then convert the finished image back to sRGB.
: Here are some links for more information:
:* [http://www.4p8.com/eric.brasseur/gamma.html Gamma error in picture scaling]
:* [http://www.sjbrown.co.uk/2004/05/14/gamma-correct-rendering/ Gamma-Correct Rendering]
:* [http://softimage.wiki.softimage.com/index.php/Gamma,_Linear_Color_Space_and_HDR Gamma, Linear Color Space and HDR]
:--[[User:PauliKL|PauliKL]] 20:24, 28 January 2009 (UTC)
| |||