clipka <ano### [at] anonymousorg> wrote:

> That's a temporary file that is deleted after a successful render, but
> left on disk if you abort, to be loaded automatically when you continue
> the aborted render, so that the radiosity cache data doesn't get lost.

Of course, now I have a few questions. (So what else is new?!)

In light of this info, I did a radiosity scene test (nothing really complex), to
try and understand what difference(s) the temporary .rca file *stored* on disk
makes to a final *non*-aborted radiosity render. (The only way I know of to
judge any difference is either visually, or through the total parse/render time
of the scene.) In other words, here's what I did--three different images of the
same scene, all run with radiosity (in v.3.6.1--I know things have changed since
then!)

1) Run the scene, but abort it before the radiosity pass is finished. This
produces the temp .rca file and stores it. Then re-run the scene again, letting
it go until finished. (Which *uses* the stored .rca file in some way, as
mentioned, but wipes it off the disk afterward.) This produces image #1.
2) Repeat the same thing as in 1), but DELETE the temp .rca file that's
produced, before running the scene again until it finishes.  This produces image
#2
3) Run the scene all the way through, *without* aborting it. This produces image
#3.

What I see is that there is NO difference at all between the renders, either
visually (image quality) or in render time. (I compared them in Photoshop, using
the 'difference' blend mode there.)

My little test may not be definitive; but it seems to me that the temp .rca file
isn't doing much! *If* that's also the case in the *betas* (with its newer
radiosity code), it leads me to ask: Is the temp .rca file stored on disk really
necessary? (That is, for an aborted render?) The only thing I see it doing is
taking up disk space--and it can be a whoppingly large file. (There are
instances in the past where I've aborted a rad run, then put the scene away for
awhile--which leaves that temp .rca file intact, necessitating hand removal.) If
any of my reasoning or results are valid, it seems to me that the file should be
erased even after an aborted run. Comments?

Ken

Post a reply to this message

Am 29.04.2010 14:59, schrieb Kenneth:

> My little test may not be definitive; but it seems to me that the temp .rca file
> isn't doing much! *If* that's also the case in the *betas* (with its newer
> radiosity code), it leads me to ask: Is the temp .rca file stored on disk really
> necessary? (That is, for an aborted render?) The only thing I see it doing is
> taking up disk space--and it can be a whoppingly large file. (There are
> instances in the past where I've aborted a rad run, then put the scene away for
> awhile--which leaves that temp .rca file intact, necessitating hand removal.) If
> any of my reasoning or results are valid, it seems to me that the file should be
> erased even after an aborted run. Comments?

I don't know for sure whether the temp .rca file really ever worked as 
intended in 3.6.

Note that in order for the temp .rca data to be re-used, you'll have to 
start the second render in abort-continue mode.

Furthermore, 3.6 only stored "top-level" radiosity samples, i.e. those 
produced at recursion depth 1. If you choose a higher recursion depth 
and POV-Ray decides that it needs more samples in the final render than 
are available in the cache, it will have to re-compute quite a lot of 
"deeper" samples.

Post a reply to this message