On 25-7-2017 18:07, clipka wrote:
> Problem #1:
> 
> Linear greyscale gradients (which is what you get by default with
> `assumed_gamma 1.0`) are typically perceived as non-linear.
> 
> Solution:
> 
> Introduce a mechanism to interpolate colour gradients in a non-linear
> fashion. Enter `blend_mode 2`. (`blend_mode 1` was defined as a
> mechanism to interpolate gradients in a linear fashion even when
> `assumed_gamma 2.2` or similar is used.)
> 
> 
> Problem #2:
> 
> Non-linear interpolation of colour gradients, if done on RGB values,
> causes colour gradients to exhibit a midway dip in brightness and a poor
> midway hue (note that this effect can be seen not only with `blend_mode
> 2`, but also with default blend mode if `assumed_gamma 2.2` or similar
> is used).
> 
> Solution:
> 
> Introduce a mechanism to interpolate brightness in a non-linear fashion
> while interpolating chromaticity (the relative ratio of R:G:B) in a
> linear fashion. Enter `blend_mode 3`.
> 

Thank you indeed Christoph, much appreciated. My thoughts were not a 
criticism towards the methods but more a failure from my side to 
/really/ understand what was going on, and I did not go too deep into 
experimentations. Your explanation is (as always) welcome and clear. 
Next time I need some blending I shall certainly explore deeper.

-- 
Thomas

Post a reply to this message