SD Q Colour

Started 3 months ago | Discussions thread
DMillier Forum Pro • Posts: 21,521
Re: SD Q Colour

I don't understand your claim that colour is subjective and that we all see it differently. There are umpteen colour reference charts and tools. It's trivial to test whether colour is accurate from a camera: you photograph one of these tools and then measure the deviation in different patches. There is nothing subjective about it.

Perhaps you mean that colour preference is subjective?

There is certainly little evidence that viewers of photographs prefer accurate colour. Quite the opposite. Most people prefer some kind of romanticised colour - whether it be higher saturation, warmer hues or whatever. This preference for deviations from accurate colour is clearly individual and subjective, hence the different deviations produced by different brands.

I have no training in colour science but I learnt a bit recently in correspondence with the chief developer at Datacolor. In particular, I learned a bit how to make colour correction profiles better. It turns out the profiles are generated by a process of measurement and interpolation: a colorimeter measures a fixed number of colour patches across the spectrum and assumes a device/inkset is well behaved in between the measured patches. But not everything is - for example the particular inkset I was having trouble with turned out to have odd non-linearities and strange crossover points between channels. This made standard profiling not particularly effective.

The advice was to measure more patches: instead of 200 patches, measure 1000. Colorimeters ship with simplified colour patch charts and more complex ones. Often the simple charts are sufficient. However, the larger charts' extra data points give the profiling algorithm more to work with. Obvious really. The point being that there is nothing subjective about colour - it's simply a matching/correction process.

With respect to Foveon colour, I have never used a Quattro model. My sense from nothing more sophisticated than looking at published Q images is that the colour is improved over previous models. It seems more even, more natural, more Bayer-like.

All my Sigma cameras produce inconsistent colour: sometime it looks nice, sometimes it looks damn odd. Roland (Kalpanika) mentioned a number of times that Foveon colour can be troublesome. Because of the translation matrix from the captured data to RGB, sometimes certain colours go out of bounds and weird things happen. Foveon may or may not have greater colour discrimination but it certainly has the ability to go wrong on occasions. Sigma are forever tweaking and playing around with colour from model to model in order to arrive at an acceptable compromise. The Q models (to my eyes) look more natural but that is not to say they are necessarily accurate.

Bayer colours are subject to tweaks by each of the manufacturers and anyone can make their own custom raw profile (I've made loads - even for my Sigma cameras) so colour can be almost anything but the basic capture does seem to be more predictable and easier to work with.

-- hide signature --
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow