[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: re: [cgmo-webcgm] The sRGB color space
Benoit As I understand it, after reading the sRGB description at http://www.w3.org/Graphics/Color/sRGB, for a CGM viewer to correctly render a CGM in which the color model = sRGB, a ICC color profile is required. This color prfile enables Windows to convert sRGB to RGB for the target device. I don't see where the WebCGM profile provides a way to specify a color profile, so I believe we have an incomplete specification as it stands. So on that note, I agree that we should deprecate sRGB. Regards, Don. > Hi cgmo-webcgm, > There were talks about sRGB and RGB during our last FTF meeting. I > asked a friend (no affiliation) of mine a few questions on the > matter: > Question: WebCGM 1.0 allows for RGB and sRGB color spaces, there are > talks of deprecating the sRGB color space. I'm wondering if > implementations are not using sRGB without knowing it? > > Answer: "RGB isn't a color space. It's just raw data that probably > just gets sent to the display hardware directly, which may or may > not have a color space close to sRGB (depends on whether it's a CRT, > what correction curves the video card is using, etc.). So yes, using > raw RGB is sort of like using sRGB, as long as you're displaying the > image on a CRT and the video card's gamma correction is set close to > 1. > It may make sense to deprecate sRGB (or color management in general) > for the types of applications CGM has mostly been used for, but if > CGM docs are ever going to be printed in color, or expected to match > CSS colors on a web page, for example, it would be a good idea to > encourage the use of sRGB. Color calibration may also become more > important as more people use non-CRT displays. > Unless sRGB is the default color space, CGM implementations are > probably NOT using it, just uncalibrated RGB, which isn't the same > thing." > Question: A PNG image is expressed in sRGB right? I know it doesn't > support embedded ICC profiles so it has to be sRGB. Right? If I take > a PNG and display it on screen with the traditional > createDIBSection, BitBlt methods, am I not in fact assuming that the > source in is sRGB values? > Answer: "PNG data isn't sRGB in general (here's the spec: > http://www.w3.org/TR/REC-png-multi.html). By default the colors are > uncalibrated, but there can be optional gamma and color correction > data. I don't know whether that data is sufficient to calibrate it > as sRGB. A lot of apps probably ignore or mishandle the calibration > data if it's present - for example, you might want to read this > article (http://hsivonen.iki.fi/png-gamma/) that I came across a few > days ago." > Question: If you create a bitmap and set the RGBQUAD to be > (0,0,255), aren't those values expected to be in sRGB. > Answer: "I had a quick look at the color management docs for Windows. > Search for the page with title "Using GDI Functions With ICM". Some > functions use CMM, others don't. E.g. > SetDIBitsToDevice > Color management is performed. If the specified BITMAPINFO structure > is not version 4 or version 5, the color profile of the current > device context is used as the source color space profile. If it > doesn't have one, the sRGB color space is used. If the specified > BITMAPINFO structure is version 4 or version 5, the color space > profile associated with the bitmap is used as the source color > space. > > The DC has to be enabled for CMM (I don't know how exactly - I've > never read any of these docs). Most apps probably don't use it, in > which case all bitmaps RGB values are device-dependent (i.e. > uncalibrated). > Note that in the description of SetDIBitsToDevice it says the > bitmap's color space is considered to be the same as the DC's for > version <= 3, which means no color correction will be done, so it's > equivalent to having CMM turned off. In other words the RGB values > are equivalent to uncalibrated, NOT sRGB. > In summary, it looks like colors are not assumed to be sRGB when > unspecified - they're treated as uncalibrated." > Regards, > > -- > Benoit mailto:benoit@itedo.com
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]