Aha! Ian Grant's workaround does the trick! in fact, in my case it can be simplified to just adding a Clear in the RII patch (which I had already), with Clear Color and Clear Depth unchecked (so in theory, it shouldn't actually do anything).
Very obscure... a|x --- On Sun, 23/11/08, Alex Drinkwater <[EMAIL PROTECTED]> wrote: > From: Alex Drinkwater <[EMAIL PROTECTED]> > Subject: Re: Render In Image and GLSL noise() > To: [email protected] > Date: Sunday, 23 November, 2008, 11:08 PM > Thanks very much vade and cwright for getting back to me. I > as under the impression noise() was supposed to run on the > GPU, but maybe that's not always the case. There > certainly must have been a time when it didn't, which > presumably is why it didn't make it into the CIKernel > subset of the GLSL language. Doesn't seem to run too > slowly on my MBP/X1600, though. > > I will give Ian Grant's workaround a go. Definitely > seems like a RII bug to me, which doesn't surprise me > too much. Thanks for the link d > CW. > > Cheers, > > a|x > > > --- On Sun, 23/11/08, vade <[EMAIL PROTECTED]> wrote: > > > From: vade <[EMAIL PROTECTED]> > > Subject: Re: Render In Image and GLSL noise() > > To: "Christopher Wright" > <[EMAIL PROTECTED]> > > Cc: "Alex Drinkwater" > <[EMAIL PROTECTED]>, > [email protected] > > Date: Sunday, 23 November, 2008, 10:38 PM > > right, because it *always* renders in software because > > noise() punts to sw, since noise() doesnt run on > normal > > "non workstation" class cards. At least, > this is > > what I understood to be the case when I attempted some > > similar vertex displacement code using noise3() and > ran into > > massive slowdowns. A kind "GL wizard" > explained > > that to me and told me I should not rely on noise() on > the > > GPU, since it will almost always be slow. THis may > have > > changed, but I figured it would be the same. > > > > Now, that doesnt explain why there are artifacts, but, > it > > may mean there is an issue in the SW renderer? No > idea. :) > > :P > > > > > > On Nov 23, 2008, at 5:10 PM, Christopher Wright wrote: > > > > >> Well, im not sure about the rendering > artifact, > > but noise3() and all noise() functions are typically > not GPU > > accelerated, from what I understand, so i think part > of the > > code punts to software. Maybe that is a clue? > > > > > > > > > Not sure if I support that hypothesis -- > rendering in > > all-software mode still produces an artifact (a > transparent > > triangle, no less, see attached picture). No > > 'puntage' in this case, since there's > nothing > > hardware accelerated. > > > > > > <Image-1.png> > > > > > > Probably some kind of CoreImage/GL bug in QC. > > It's been reported before on the list, perhaps > there was > > a more in-depth explanation at that point, or there > were > > other conditions where it happened, to refine > what's > > happening. > > > > > > (Here's your link: > > > http://lists.apple.com/archives/Quartzcomposer-dev/2008/Aug/msg00129.html > > credits to Ian Grant for starting the thread) > > > > > > -- > > > [ christopher wright ] > > > [EMAIL PROTECTED] > > > http://kineme.net/ > > > > > > > _______________________________________________ > Do not post admin requests to the list. They will be > ignored. > Quartzcomposer-dev mailing list > ([email protected]) > Help/Unsubscribe/Update your Subscription: > http://lists.apple.com/mailman/options/quartzcomposer-dev/the_voder%40yahoo.co.uk > > This email sent to [EMAIL PROTECTED] _______________________________________________ Do not post admin requests to the list. They will be ignored. Quartzcomposer-dev mailing list ([email protected]) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/quartzcomposer-dev/archive%40mail-archive.com This email sent to [EMAIL PROTECTED]
