btw Rob, (after a long break away) back to that earlier problem of the art gallery: I know this is contrary to most of the advise - but you know i still think it worthwhile in principle to render it using brute force - ie without illums on the coffering (i think that was Martin's original suggestion). The coffering detail doesn't look complicated enough to have really caused Radiance probs - and there are enough apertures in the space that it seems like they should have been picked up by a reasonable setting for -ad & -as and provide a well behaved indirect lighting distribution.
I am sure the folks here in Berkeley with me right now are all grinning in bemused enlightenment. The -ad parameter has taken on a very clear meaning to us now, following some excellent discussion and presentations by John Mardaljavec. I'll have to look at those settings again when I get home; maybe the -ad was pretty low. However, I must say, mkillum is a great tool and I believe that it should be used when appropriate, and this is certainly a good use for it (I'm actually using far less light sources now, because of this technique; I'm getting excellent results in far less time!) I will certainly have a look at the -ad in my earlier explorations of this scene regardless, I'm very curious to see if I can affect some positive change on the non-illum version of the scene.
And i'd agree with Martin again in that it's good to move away from relying on RAD - far better to get to grips directly with specifying rpict parameters. I find that once settled on generally appropriate params for buildings (accounting for model size in -ar and -aa) - usually switching between high/low quality, slow/fast rendering can be controlled enough just via -ab and image size and then -ad -as.
Let's put it this way: I use rad to get started, but I tweak from there. The more I learn about the -d* and -a* settings of rpict, the more I will rely on manual tweaking. To be clear, I am not simply setting rad and hoping for the best (well, actually that's true, and then when the best does not happen, I start adding render= lines!). rad, like mkillum, are both excellent tools that work too well to be ignored, at least for this novice user. I agree, the "under the hood" aspects of rtrace need to be understood, but I still like to use rad to at least steer me to a ballpark setting. This art gallery exploration has been useful for me to learn both mkillum and the -a* parameters. The more of these that happen, I'm sure the more I will understand, and thus, the less I will rely on rad. But I honestly don't think I will ever totally abandon it, at least for general stuff.
The fact that you have taken the time to follow-up with me is indicative of the extremely generous nature of the majority of this list membership, and all week here at the Workshop that same spirit -- that same bubbling enthusiasm -- has been flowing through buildings 90 and 70A at LBNL. I wish you could be here!
Rob Guglielmetti
[email protected]
www.rumblestrip.org
···
On Wednesday, September 24, 2003, at 09:55 AM, Alex Summerfield wrote: