<p><br>
On May 20, 2012 5:24 AM, "Brecht Van Lommel" <<a href="mailto:brechtvanlommel@pandora.be">brechtvanlommel@pandora.be</a>> wrote:<br>
><br>
> On Sun, May 20, 2012 at 11:35 AM, François T. <<a href="mailto:francoistarlier@gmail.com">francoistarlier@gmail.com</a>> wrote:<br>
> ><br>
> > True, Yet Brecht is right any look/creative LUT should be done down the pipe</p>
<p>Well known and understood, as already pointed out by the LMT quote above, however slightly tangential to the a discussion of the statement:</p>
<p>>If we apply it before they wouldn't be both in the same linear space.</p>
<p>> I guess the world "linear" is a bit loaded. But after such an LMT, the<br>
> colors are definitely not linear with scene radiometry anymore. I<br>
> would call that a different space but it's just a matter of semantics.</p>
<p>The scene referred intent may be skewed.</p>
<p>The sole point I was making was that the end transform LUTs do the heavy lifting of manipulating workspace value foo into expected result bar.</p>
<p>In the ACES case, the RRT / ODT always interpret the working space as ACES, no matter how mangled / contaminated / broken the scene intent may be, creative of otherwise.</p>
<p>The good news however, is that it would be identical across all touch points in a pipeline and even across vendors / services, assuming the output devices are calibrated[1][2].</p>
<p>With respect,<br>
TJS</p>
<p>[1] A discussion of output device profile LUT in the final position would probably be expected.<br>
[2] The same could be said of an sRGB LUT assuming the chain was accountable and delivered to the standards for primaries, white points, and curvature.</p>