PDA

View Full Version : Choose Colour Space for Gradient?



raw-m
03-03-2013, 03:53 AM
We all know how you can't get very far without using gradients! I've been using them a lot for displacements. The colour channels are esp useful for getting positions (see attached - I'm basically dynamically moving the one null to another) - but only when working in Linear Color Space. I understand why but need to work in sRGB, which obviously sends my colour channel numbers out of whack.

Is there a system that will swop the Gradient for something that uses absolute numbers, or is there another approach?

RebelHill
03-03-2013, 05:25 AM
Yeah... there was a bug in gradients (and a couple other nodes) with them not respecting/treating colourspace the right way... this could be due to that. Simple fix... shove a colour correct node onto the output, then u can gamma 2.2/0.454 (whichever) it as needed.

raw-m
03-04-2013, 03:06 AM
Hi RH. Thanks for the tip, I thought this might be a solution but I can't get the positions to line-up in sRGB space and some very strange results occur when Positions go negative.

Out of the Gradient, I'm using db&w Simple Colour Corrector, Color into Color and changing the Gamma setting accordingly. Is this what you meant? I guess I could always split out the x,y,z into scalars and do it the long way.

When you say its a bug, should the Gradient node stay in either one colour space or the other, regardless of CS you're in?

XswampyX
03-07-2013, 12:11 PM
No wonder sometimes things just don't seem to work properly. If you were doing maths with any of the vectors you could get any result. Look at the difference!!!


http://youtu.be/ZMTV5MbtG6g

Thanks for the info raw-m. :thumbsup:

raw-m
03-11-2013, 05:35 AM
Hey XswampyX, just seen this (the email notification on these forums is a bit whack sometimes!). Thanks for the clarification, glad it isn't just me scratching my head!

Is this a bug or how it should work? Surprised it hasn't been noticed and tackled earlier if it is a bug.