|
|
As many of you know, I modified Samuel's grass macro to use trace(). This
morning, at around 2AM, in the middle of my desperation trying to improve
the macro, an idea came to me, and I tried it out. The blades of grass in
Samuel's macro are made of one triangle and three boxes. I turned this into
a small mesh of 7 triangles. This reduces memory consumption (in my tests,
7500 blades went from 64MB to 46MB), and speeds up rendering a bit. The only
trade-off is that due to the use of trace() and a special condition I added,
the parsing is slower. I think it's worth it, though. I'll be posting the
source soon. I'm thinking of adding control over the pattern that gets
evaluated. Should I do that before releasing the code, or should I just post
what I have right away?
PS: I have posted an image in .images, which consists of 55,000 blades. More
details there...
Post a reply to this message
|
|
|
|
"Tony[B]" <ben### [at] panamac-comnet> wrote in message
news:3959166d@news.povray.org...
| The blades of grass in
| Samuel's macro are made of one triangle and three boxes. I turned this into
| a small mesh of 7 triangles. This reduces memory consumption (in my tests,
| 7500 blades went from 64MB to 46MB), and speeds up rendering a bit. The only
| trade-off is that due to the use of trace() and a special condition I added,
| the parsing is slower.
| I'm thinking of adding control over the pattern that gets
| evaluated. Should I do that before releasing the code, or should I just post
| what I have right away?
|
| PS: I have posted an image in .images, which consists of 55,000 blades. More
| details there...
I hadn't understood how Samuel's was done. It never looked like a single
triangle before. That's a pretty good improvement your way but 5+ hours and
329MB memory still seems an awful lot for a small hillside of grass. But then
maybe I'm just pessimistic.
Bob
Post a reply to this message
|
|