 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Hi,
i'm rendering large images, the last ones with pov-ray 3.62.
now i switched to v3.7 rc5 and then rc6, and i ran into a big problem:
for many images it takes several hours to just pre-trace the radiosity,
and since i just can use my computer for about 12 hours, i got used to
continue_trace in v3.62, which worked just fine.
when the pre-trace was done and the final rendering was in progress, it could be
interrupted and continued the next day where i left.
but in v3.7 rc5 and rc6 povray is redoing the pre-trace everytime restarted,
even with radiosity_from_file=on and ..._to_file=on.
since i wanted to work with sslt, it has to be v3.7, but the pre_trace of my
actual image takes 11 hours, and i really don't want to do that again and again.
is there no way pov-ray v3.7 will re-use the already calculated values and just
continue_trace ?
or am i missing something ?
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Am 30.09.2012 21:41, schrieb Asfar:
> Hi,
> i'm rendering large images, the last ones with pov-ray 3.62.
>
> now i switched to v3.7 rc5 and then rc6, and i ran into a big problem:
>
> for many images it takes several hours to just pre-trace the radiosity,
> and since i just can use my computer for about 12 hours, i got used to
> continue_trace in v3.62, which worked just fine.
> when the pre-trace was done and the final rendering was in progress, it could be
> interrupted and continued the next day where i left.
>
> but in v3.7 rc5 and rc6 povray is redoing the pre-trace everytime restarted,
> even with radiosity_from_file=on and ..._to_file=on.
>
> since i wanted to work with sslt, it has to be v3.7, but the pre_trace of my
> actual image takes 11 hours, and i really don't want to do that again and again.
> is there no way pov-ray v3.7 will re-use the already calculated values and just
> continue_trace ?
> or am i missing something ?
POV-Ray will always try to satisfy the various radiosity quality
requirements, even if it loads radiosity samples from a file. That
means, if your first run did not produce anywhere near enough samples
according to the quality criteria (because pretrace_start and
pretrace_end would result in too few pretrace passes), POV-Ray will do
another pretrace to compute more samples.
One solution should be to set both pretrace_start and pretrace_end to 1
in the subsequent renders, giving you virtually no pretrace at all.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
thanx for quick response.
i will try pretrace_start and pretrace_end = 1
but i don't quite get it-
usually i render with pretrace_start = 0.08 and _end = 0.005 and i thought, when
pretracing is done and rendering begins, povray will fetch from pretraced
values.
i don't quite understand why v3.62 works like this i.e.
10 hours pretracing
2 hours rendering- then i stop
and next day continue_trace with
x hours of rendering until i stop (following days wiht x hours of rendering)or
image done
while v3.7 gives me
10 hours pretracing
2 hours rendering- until i stop
and next day its again
10 hours pretracing
x hours rendering
where the second (and third and fourth...) pretracing gives me exactly the same
"pixels" as the first pretrace... starting with value 0.08 again even if last
days pretrace 0.005 was finished...
maybe i'm too stupid to see, what's behind this...
thx for enlightment...
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On 30-9-2012 23:11, Asfar wrote:
> while v3.7 gives me
> 10 hours pretracing
> 2 hours rendering- until i stop
> and next day its again
> 10 hours pretracing
> x hours rendering
> where the second (and third and fourth...) pretracing gives me exactly the same
> "pixels" as the first pretrace... starting with value 0.08 again even if last
> days pretrace 0.005 was finished...
[I got second thoughts about this, so I cancelled my earlier post on the
news reader, but maybe it is still there on the web site. Disregard what
I wrote. Sorry for any confusion]
Using +C works correctly on a render without saved radiosity data, as
radiosity calculations restart again for the not-rendered part of the
image and only for that part. So, what you may be seeing is not a
complete re-calculation but only for the not-rendered region? That would
be correct.
From your post I assume you do not save/read radiosity data? For very
long renders using sslt in particular I think that would be necessary.
with the pretrace value at 1 for the reading sessions as Clipka wrote.
Thomas
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Thomas de Groot <tho### [at] degroot org> wrote:
> Using +C works correctly on a render without saved radiosity data, as
> radiosity calculations restart again for the not-rendered part of the
> image and only for that part. So, what you may be seeing is not a
> complete re-calculation but only for the not-rendered region? That would
> be correct.
>
yes, that's the way it is if i don't save rad-data (either v3.62 or v3.7 rc5/6-
sometimes i deleted the v3.62.rca file to see the unrendered part again in
pretrace)
> From your post I assume you do not save/read radiosity data? For very
> long renders using sslt in particular I think that would be necessary.
> with the pretrace value at 1 for the reading sessions as Clipka wrote.
>
but i did use v3.7 rc5/6 with radiosity_file_name="xyz", radiosity_to_file = on
and radiosity_from_file = on
and i think, this is the way to save rad-data, is it ?)
and even with that the pretrace (for the unrendered part) is starting all over
again, from pretrace_start-size to pretrace_end-size.
yesterday i did start a clean new render with v3.7 rc6:
radiosity pretrace completely finished after 11 hours. then 3 more hours gave me
18% of render. then i stopped.
today i continued with pretrace_start and _end set to 1 as suggested and as i
started the render, the image was as i left it, but povray was at 83%.
rendering for 11 hours gave me nothing... still 83%, still the same image.
well, maybe it is a point that really will take long to render, longer than 11
hours, but why the 83% ?
sorry, but i'm confused.
all other things work great, maybe i should just shorten the rendertime :)
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
after several sped up tries, i found out, there is no real problem, just in my
head :)
v3.62 does pretrace only once, no matter if you continue the render some other
time, while v3.7rc5/6 does pretrace everytime the render is started/continued
(unless one specifies pretrace_start = pretrace_end = 1)
right ?
thanx anyway !!!
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
|
|
