[12 October 2011]
What is a linear workflow? Many people find linear workflows confusing but the concept is rather basic. Most displays use a gamma (contrast compression) of 2.2. The way we compensate for this is by applying a gamma value of 0.45 (sRGB standard gamma) to offset the images contrast. This is referred to as gamma correction.
When for example you take a photo with a digital camera this gamma correction is applied automatically. It gets a bit more complex when working with realistic computer generated imagery. Within an artificial 3D environment light is computed in a linear manner just as it douse in the real world. This means the images need to be compensated post render. One of the problems is that any textures used within the scene have most likely already been adjusted and end up with an affective gamma of 0.20. So all textures must be manually linearised before hand with a 2.2 gamma. Many systems now have the option to do this automatically when an image is imported into the scene itself. This only applies to 8/16bit images as 32bit imagery tends to already be in linear space.
When the image has been rendered it can then be gamma corrected. Within a linear workflow this is done after the compositing so that when inputting adjustments the values produce a predictable and consistent result in linear space. I find compositing incredibly useful and time efficient when used correctly.
This is an example of a title animation I did using this linear workflow.
I first created a 32bit multi-pass .exr file in linear space with a universal base colour. I did this because attempting to approximate the output colour after the scene lighting takes effect never turns out perfect. Also if anyone wants to change the colour later the whole scene would need to be re-rendered. I then used an ID pass to selectively give elements colour.
(this image has had gamma correction for display purposes)
Here is the finished result.
This is another title animation I did using the same workflow.