 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Trying to whip up a spiroform compact fluorescent light bulb, all seemed
well until I flicked on the high ambient and radiosity. It took freakin'
4 hours for a little 320x320 swatch!
______________________________________________________________________
Scene Statistics
Finite objects: 562
Infinite objects: 1
Light sources: 6
Total: 569
Render Statistics
Image Resolution 320 x 320
Pixels: 117663 Samples: 260097 Smpls/Pxl: 2.21
Rays: 25682366 Saved: 2994586 Max Level: 15/15
Ray->Shape Intersection Tests Succeeded Percentage
Box 12167475 6235321 51.25
Cone/Cylinder 52542338 26269426 50.00
CSG Intersection 19066704 10143355 53.20
CSG Union 15876905 15818031 99.63
Plane 42394448 12473732 29.42
Sphere 34522020 32168741 93.18
Sphere Sweep 62860863 30115671 47.91
Torus 31929828 8356481 26.17
Torus Bound 31929828 9443106 29.57
Clipping Object 25660773 7068664 27.55
Bounding Box 921869187 436347103 47.33
Light Buffer 11780860 6728420 57.11
Vista Buffer 2208478 2144762 97.11
Function VM calls: 88
Roots tested: 217585910 eliminated: 43232
Calls to Noise: 3544 Calls to DNoise: 36760
Shadow Ray Tests: 34071387 Succeeded: 8334916
Reflected Rays: 9021048 Total Internal: 17
Refracted Rays: 7998251
Transmitted Rays: 970
Radiosity samples calculated: 42010 (9.07 %)
Radiosity samples reused: 421288
Smallest Alloc: 18 bytes
Largest Alloc: 92896 bytes
Peak memory used: 8283571 bytes
Total Scene Processing Times
Parse Time: 0 hours 0 minutes 1 seconds (1 seconds)
Photon Time: 0 hours 0 minutes 0 seconds (0 seconds)
Render Time: 4 hours 7 minutes 38 seconds (14858 seconds)
Total Time: 4 hours 7 minutes 39 seconds (14859 seconds)
CPU time used: kernel 946.48 seconds, user 11311.77 seconds, total
12258.25 seconds
Render averaged 8.35 PPS over 102400 pixels
______________________________________________________________________
#include "rad.inc" version 1.1-2006nov11. Radiosity is ON.
radiosity
{ brightness 1.000
count 200
error_bound 0.450
normal on
pretrace_end 0.01000
pretrace_start 0.080
recursion_limit 1
}
______________________________________________________________________
Incredibly, POV-Ray seems to have been granted only 82% of the CPU. I
suspect that the remainder was dominated by my Web browser being kicked
around by ad requests and YouTube pulls. However, my Internet connection
is so capricious that I really hesitate to close any windows, especially
YouTube. (When it takes an hour to download a 4 minute video--and this
after several attempts spanning hours--you don't want to let it go.)
Normal is on because it defaults that way in my include file. I could
have sworn I'd changed it. In any case, I don't think it make much, if
any, difference for this scene. Radiosity notwithstanding, I suspect that
the real slowdown was in the sphere_sweep:
Spiral form with radiosity 12258 seconds (3:24:18)
Spiral form without radiosity 736 seconds (0:12:16)
Dummy shape with radiosity 82 seconds
Dummy shape without radiosity 6.25 seconds
Ratios of spiral to dummy were 150:1 and 118:1.
Ratios of radiosity to non were 17:1 and 13:1.
That it took 12 minutes *without* radiosity should have been a tip-off.
Another slowdown may have been in my attempt at limb-darkening, which
involved refraction and a near-duplication of the tubing. While the
real-life effect far exceeds the dynamic range of the typical CG image, it
is noticeable enough in real life that a saturated CG render looks flat
and unnatural. (I didn't do a controlled timing test of that feature
because of the trouble it would take to dismantle it; I may do it later.)
The second attachment is welder's-eye view mock-up (the lighting is all
staged) of the effect that I'm trying to achieve.
--
<Insert witty .sig here>
Post a reply to this message
Attachments:
Download 'compact_fluo-swept.jpg' (13 KB)
Download 'compact_fluo-look.jpg' (7 KB)
Preview of image 'compact_fluo-swept.jpg'
Preview of image 'compact_fluo-look.jpg'
|
|
| |
| |
|
|
|
|
| |
| |
|
|
That is a pretty cool bulb, and the way the glass and phosphors look viewed
from nearly parallel to the bulb surface is quite accurate (though the glass
could be a touch thicker).
I feel your pain with those render times...I've got a scene thats been going
for over two days now and is 12% complete at 1024x768... :-(
Granted, I'm using media in the water, media clouds, isosurfaces and
radiosity (media on)...
ian
"Cousin Ricky" <ric### [at] yahoo com> wrote in message
news:op.### [at] your-727a0a4e7cvipowernetnet...
> Trying to whip up a spiroform compact fluorescent light bulb, all seemed
> well until I flicked on the high ambient and radiosity. It took freakin'
> 4 hours for a little 320x320 swatch!
>
> ______________________________________________________________________
>
> Scene Statistics
> Finite objects: 562
> Infinite objects: 1
> Light sources: 6
> Total: 569
>
> Render Statistics
> Image Resolution 320 x 320
>
> Pixels: 117663 Samples: 260097 Smpls/Pxl: 2.21
> Rays: 25682366 Saved: 2994586 Max Level: 15/15
>
> Ray->Shape Intersection Tests Succeeded Percentage
>
> Box 12167475 6235321 51.25
> Cone/Cylinder 52542338 26269426 50.00
> CSG Intersection 19066704 10143355 53.20
> CSG Union 15876905 15818031 99.63
> Plane 42394448 12473732 29.42
> Sphere 34522020 32168741 93.18
> Sphere Sweep 62860863 30115671 47.91
> Torus 31929828 8356481 26.17
> Torus Bound 31929828 9443106 29.57
> Clipping Object 25660773 7068664 27.55
> Bounding Box 921869187 436347103 47.33
> Light Buffer 11780860 6728420 57.11
> Vista Buffer 2208478 2144762 97.11
>
> Function VM calls: 88
>
> Roots tested: 217585910 eliminated: 43232
> Calls to Noise: 3544 Calls to DNoise: 36760
>
> Shadow Ray Tests: 34071387 Succeeded: 8334916
> Reflected Rays: 9021048 Total Internal: 17
> Refracted Rays: 7998251
> Transmitted Rays: 970
>
> Radiosity samples calculated: 42010 (9.07 %)
> Radiosity samples reused: 421288
>
> Smallest Alloc: 18 bytes
> Largest Alloc: 92896 bytes
> Peak memory used: 8283571 bytes
> Total Scene Processing Times
> Parse Time: 0 hours 0 minutes 1 seconds (1 seconds)
> Photon Time: 0 hours 0 minutes 0 seconds (0 seconds)
> Render Time: 4 hours 7 minutes 38 seconds (14858 seconds)
> Total Time: 4 hours 7 minutes 39 seconds (14859 seconds)
> CPU time used: kernel 946.48 seconds, user 11311.77 seconds, total
> 12258.25 seconds
> Render averaged 8.35 PPS over 102400 pixels
>
> ______________________________________________________________________
>
> #include "rad.inc" version 1.1-2006nov11. Radiosity is ON.
> radiosity
> { brightness 1.000
> count 200
> error_bound 0.450
> normal on
> pretrace_end 0.01000
> pretrace_start 0.080
> recursion_limit 1
> }
> ______________________________________________________________________
>
>
> Incredibly, POV-Ray seems to have been granted only 82% of the CPU. I
> suspect that the remainder was dominated by my Web browser being kicked
> around by ad requests and YouTube pulls. However, my Internet connection
> is so capricious that I really hesitate to close any windows, especially
> YouTube. (When it takes an hour to download a 4 minute video--and this
> after several attempts spanning hours--you don't want to let it go.)
>
> Normal is on because it defaults that way in my include file. I could
> have sworn I'd changed it. In any case, I don't think it make much, if
> any, difference for this scene. Radiosity notwithstanding, I suspect that
> the real slowdown was in the sphere_sweep:
>
> Spiral form with radiosity 12258 seconds (3:24:18)
> Spiral form without radiosity 736 seconds (0:12:16)
> Dummy shape with radiosity 82 seconds
> Dummy shape without radiosity 6.25 seconds
>
> Ratios of spiral to dummy were 150:1 and 118:1.
> Ratios of radiosity to non were 17:1 and 13:1.
>
> That it took 12 minutes *without* radiosity should have been a tip-off.
>
> Another slowdown may have been in my attempt at limb-darkening, which
> involved refraction and a near-duplication of the tubing. While the
> real-life effect far exceeds the dynamic range of the typical CG image, it
> is noticeable enough in real life that a saturated CG render looks flat
> and unnatural. (I didn't do a controlled timing test of that feature
> because of the trouble it would take to dismantle it; I may do it later.)
> The second attachment is welder's-eye view mock-up (the lighting is all
> staged) of the effect that I'm trying to achieve.
>
> --
> <Insert witty .sig here>
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
The second one is very nice!
;-)
Paolo
>Cousin Ricky on date 19/02/2009 05:24 wrote:
> Trying to whip up a spiroform compact fluorescent light bulb, all seemed
> well until I flicked on the high ambient and radiosity. It took
> freakin' 4 hours for a little 320x320 swatch!
>
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"Cousin Ricky" <ric### [at] yahoocom> wrote:
> Trying to whip up a spiroform compact fluorescent light bulb, all seemed
> well until I flicked on the high ambient and radiosity. It took freakin'
> 4 hours for a little 320x320 swatch!
Here's my quick attempt with mcpov. Yours looks much better, but it's possible
mine may render more quickly.
The bulb is a mesh just generated by extruding a cross-section along a spline,
all in pov, of course. There's not much control for orientation or twisting,
but it's fine for a circle. I used an aoi pattern for the bulb, although it is
almost completely blown out. It'd be nice to have HDR output for mcpov, but oh
well. I will post the source to p.t.s-f shortly.
- Ricky
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"triple_r" <nomail@nomail> wrote:
> Here's my quick attempt with mcpov. Yours looks much better, but it's possible
> mine may render more quickly.
Oops. Oh, and that wasn't a challenge, just a helpful suggestion.
- Ricky
Post a reply to this message
Attachments:
Download 'cfl.jpg' (41 KB)
Preview of image 'cfl.jpg'
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"triple_r" <nomail@nomail> wrote:
>
> Here's my quick attempt with mcpov. Yours looks much better, but it's possible
> mine may render more quickly.
I noticed that you used one of those hard-to-find daylight bulbs. I actually
prefer those (I use one IRL), but I used a soft warm light here because it
seemed to fit the mood of the recent Tiffany thread. Makes for an ugly-looking
green through the welding glass, though. :^P
> The bulb is a mesh just generated by extruding a cross-section along a spline,
> all in pov, of course. There's not much control for orientation or twisting,
> but it's fine for a circle.
That sounds like a good solution. I'll take a look.
> I used an aoi pattern for the bulb, although it is
> almost completely blown out. It'd be nice to have HDR output for mcpov, but oh
> well. I will post the source to p.t.s-f shortly.
Almost blown out actually sounds like decent goal. It gives the desired
hypersaturated impression while just avoiding the flatness I mentioned earlier.
I didn't realize that POV had angle-of-incidence. Is that an MCPOV thing?
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"Cousin Ricky" <ric### [at] yahoocom> wrote:
> I noticed that you used one of those hard-to-find daylight bulbs. I actually
> prefer those (I use one IRL), but I used a soft warm light here because it
> seemed to fit the mood of the recent Tiffany thread. Makes for an ugly-looking
> green through the welding glass, though. :^P
I don't remember them being hard to find, but that's the beauty of compact
fluorescent bulbs--you can hardly remember the last time you purchased one!
> That sounds like a good solution. I'll take a look.
I've been meaning to do some benchmarks for a while to compare the rendering
time of spheres or more complex primitives to their mesh equivalents. When
mcpov was originally released, the mesh-based scenes seemed to work much better
than scenes with many primitives. This should be no surprise, but indicates
that perhaps if a mesh alternative is available, it could be worthwhile.
> I didn't realize that POV had angle-of-incidence. Is that an MCPOV thing?
It's a megapov thing, and since mcpov is based on megapov, it has the feature as
well. You did a fine job of accomplishing the effect though.
- Ricky
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
|
|
