<div class="gmail_quote">On 21 May 2012 22:01, Olivier Galibert <span dir="ltr"><<a href="mailto:galibert@pobox.com" target="_blank">galibert@pobox.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div class="im">On Mon, May 21, 2012 at 07:16:35PM -0700, Paul Berry wrote:<br>
> - Add the front-end API to allow the client to request per-sample shading<br>
> (I assume it's glEnable(GL_HAM_SANDWICH) or something like that, in which<br>
> case it should be easy)<br>
<br>
</div>There is a "MIN_SAMPLE_SHADING" value in the [0,1] interval which is<br>
set by glMinSampleShading(clampf value). It tells the minimum ratio<br>
of samples that must be individually shaded. For instance, if you're<br>
doing 4xMSAA and mss=0.75, then you must shade at least 3 samples.<br></blockquote><div><br></div><div>Thanks, Olivier, that's exactly the information I was looking for.</div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<br>
Always shading all of them as soon as mss is non-zero is compliant to<br>
the spec.</blockquote><div><br></div><div>Ok, cool. I think we can improve on this slightly by doing persample dispatch when mss >= 1/N (where N=number of samples per pixel). Of course in practice it's hard to see why anyone would set 0 < mss <= 1/N on purpose :)</div>
<div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> Always shading all of them all the time is actually<br>
compliant to the spec, as far as I can say, but would be a bad idea.<br>
<span class="HOEnZb"><font color="#888888"><br>
OG.<br>
</font></span></blockquote></div><br><div><br></div>