On 18/01/13 17:58, Richard Hughes wrote: > I'm wondering about the basics of what gamma is and also how to > measure it. Some of the questions I'm asking myself: > > * on real life hardware, can we assume gamma_red == gamma_blue == gamma_green?
I'd say 'mostly', but it depends on the device as to how well 'gamma' models the device. There are variations of the basic equations that do or do not assume an offset (sometimes two), gain, gamma and their various orderings in the equations. Most of this comes from the CRT days when the physical process involved was modelled reasonably well by this. (Monitors with poor power supplies could have channel/power level effects which break it badly). With LCDs and other technologies, it very much depends on how well the manufacturer chooses to emulate this behaviour. On higher end LCD e.g. EIZO's CG series, or HP Dreamcolor, (other manufacturers are free to send me a sample monitor to eval :-), the internal systems do a good job of controlling the response curve as well as maintaining channel independence. In these cases I've found that if you can measure (or cleverly estimate) what the zero emission level is for the display you can subtract that from your measurements and get a good handle on the response curve and that they then fit well to a GOG/GOGO model. > * how do we measure gamma given there's an offset at zero for anything > other than LED displays - pretending the backlight is zero and > offsetting everything to that seems a giant hack given our perception > of light isn't linear. This assumes your measuring in the dark and thus there is no externally generated 'flare' contribution to black. > * roughly how many points does it take to calculate the gamma assuming > the hardware is well behaved (e.g. monotonic) -- three seems the > obvious answer, but the backlight at 0,0,0 and measurement accuracy > makes that tricky. there are some approaches where you can model the channel outputs as a combination of signal + fixed bias (noise). If you do this you can get a handle on the level of emission at zero for whatever you want e.g. tristimulus. One of the methods I've used is to assume channel independence and that each channel is constant colour and then model the device output as: traistimulus measurement = black + 3 RGB components where black is unknown I then minimised the deviations generated by the distribution of chromaticity/whatever measure of colour you want for the channels. With most devices if you plot the chromaticity of a red ramp you will see that at peak intensity you pretty much get the primary of the device, as you lower the intensity the true colour is contaminated by more and more of the black emission, repeat this for a number of ramps and you'll notice the lines appear to intersect... the rest is just careful statistics and with the numerical precision regarding the number of points to measure, well the distribution of points along the scale can also be important - should you bias your measurements so you have more nearer black for instance? > * how does the 2.4 v.s. 2.2 gamma adaption for viewing conditions > work? Is that a function of the luminance of the room, in which case > we should probably measure ambient first and do something more > technical than += 0.2. 2.4 vs 2.2 vs 2.6 is an approximation of course, it doesn't account for perception of colour due to intensity of the device, flare, etc. But is a good start. I'd further recommend Charles' books/website as a good starting point as well as those from Mark Fairchild. Disclaimer - I know Charles, but did buy his original book long before meeting him. Kevin -- | Kevin Wheatley, Cinesite Ltd. | Nobody thinks this | | Head of Software Engineering | My employer for certain | | (and Colour Management, ... ) | Not even myself | ---------- This email and any attachments to it are confidential and are intended solely for the use of the individual to whom it is addressed. Any views or opinions expressed are solely those of the author and do not necessarily represent those of Cinesite Ltd. If you are not the intended recipient of this email, you must neither take any action based upon its contents, nor copy or show it to anyone. Please contact the sender if you believe you have received this email in error. Warning: The recipient should check this email and any attachments for the presence of viruses. Cinesite Ltd accepts no liability for any damage caused by any virus transmitted via this email. ---------- Cinesite Ltd. Registered Office: Medius House, 2 Sheraton Street, London, W1F 8BH. Company registration # 7913144 ------------------------------------------------------------------------------ Master Visual Studio, SharePoint, SQL, ASP.NET, C# 2012, HTML5, CSS, MVC, Windows 8 Apps, JavaScript and much more. Keep your skills current with LearnDevNow - 3,200 step-by-step video tutorials by Microsoft MVPs and experts. ON SALE this month only -- learn more at: http://p.sf.net/sfu/learnnow-d2d _______________________________________________ Lcms-user mailing list Lcms-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/lcms-user
