"Lance Birch" <-> wrote...
> Hiya, well, this is my first use of POV-Ray in a few years now (probably 3
> years, wow, it's been a while), so I've been catching up on a lot of
> features that were never in the POV-Ray I knew.  This is just a test of
> using Photons and Radiosity together... it took about 20 minutes to render
> which doesn't seem too bad, but I'm guessing the "stripe" effect in the
> photons is just because I haven't set the count high enough?

Increasing the photon jitter (up to 1.0) should get rid of that, but it
might introduce some noise into the caustics... overall, high jitter should
work well with this scene, though.

-Nathan

Post a reply to this message

Nathan Kopp wrote:
> Increasing the photon jitter (up to 1.0) should get rid of that, but it
> might introduce some noise into the caustics... overall, high jitter
should
> work well with this scene, though.

Thanks, I did and it made a big difference :)

Lance.

thezone.firewave.com.au
www.firewave.com.au

Post a reply to this message

After a little experimenting I found that it was simply that Windows 2000
was sharing the pvengine process across both processors, so that it could
load balance other processes better.  The affinity by default is set on CPU0
+ CPU1, even if the process only contains a single thread.  As more
processes are set to a specific processor, it throttles the other processes
more towards the opposite processor to ensure they're being balanced in
terms of cycles.  I tried forcing pvengine to affinity CPU0, and it does
take up 100-99% of that processor, so the thread is getting the equivilant
of 100% of a processor's time.  As soon as it's switched back to CPU0 + CPU1
it balances again.

The process shows as taking up 50% of the total CPU usage, which is correct.
(graph of the cycles from switch affinty from CPU0 to CPU0 + CPU1 included
just for fun ;)  Even with distributed.net running it still gets 50-49% of
the total usage, while distributed.net is throttled to the other processor,
and gets the remaining 49-50%.  Works well :)

Anyway, that's getting really off-topic (kinda), so back to the image.

I tried raising the jitter from 0.3 to 1 as was suggested by Nathan and it
made the image look a lot better, but it was getting dark and light patches,
so I figured it still needed more photons.  I bumped the count up to 3
million from 2 million, but it was still patchy, so I went to 4 million,
then 5 and then 6 million (heh) and it seems to have smoothed it out a fair
bit.  I probably should read the documentation a little more carefully...
maybe there's an option I've missed to help a little bit, but then again
maybe there isn't (I guess that's what jitter is for really), so I'm happy
with pushing the count up.  It's still a little patchy in some of the more
sparse areas of the shadows, but no where near as bad as it was :)

I've also replaced the standard light with an area light and it looks a lot
better (I'm a big fan of soft shadows, heh) but there is an atefact I've
having difficulties getting rid of (the harsh edge to the shadows that
appears on the shadows of the first and second torii on the left side of the
image, just where it goes really soft).  Any ideas how to get rid of it?  I
tried using 32x32 lights in the area light too (instead of the 18x18 in that
image) but it was still there :/

Anyway, I can see I'm going to have a LOT of fun with photons and radiosity!

Lance.

thezone.firewave.com.au
www.firewave.com.au

Post a reply to this message

Attachments:
Download 'photon_tracing_2.jpg' (71 KB) Download 'affinity_graph.gif' (5 KB)

Preview of image 'photon_tracing_2.jpg'

Preview of image 'affinity_graph.gif'

The image looks very good.
  However, there's something strange there. I think that you clearly wanted
to make glass torii. However, for some reason they look like made of plastic
instead.
  What is their ior? Also perhaps the highlights should be sharper.

-- 
#macro N(D,I)#if(I<6)cylinder{M()#local D[I]=div(D[I],104);M().5,2pigment{
rgb M()}}N(D,(D[I]>99?I:I+1))#end#end#macro M()<mod(D[I],13)-6,mod(div(D[I
],13),8)-3,10>#end blob{N(array[6]{11117333955,
7382340,3358,3900569407,970,4254934330},0)}//                     - Warp -

Post a reply to this message

Warp wrote:
>   The image looks very good.
>   However, there's something strange there. I think that you clearly
wanted
> to make glass torii. However, for some reason they look like made of
plastic
> instead.
>   What is their ior? Also perhaps the highlights should be sharper.

Yes I haven't played with the highlights yet, so I left them fairly dull
(more like some kind of plastic).  The IOR is 1.15, so the highlights should
be much tighter.

Lance.

thezone.firewave.com.au
www.firewave.com.au

Post a reply to this message

"Lance Birch" <-> wrote in message news:3bdcc5f5@news.povray.org...
> I'm not sure, with RC5 running I can get both processors doing 100% (if not
> with a little bit or kernel action here and there).  I'll have to
> investigate it further.  AFAIK POV-Ray is single-threaded as well, but even
> still, setting it to the highest thread priority should get it using 100% of
> one processor.  I'll have to try manually setting it in the task manager on
> pvengine.

Incidentally, if you download beta.7 (available in the next 24 hours), you will
find that on NT it prints a summary of actual kernel+user time after each render.

I added this to allow for more accurate benchmarking, regardless of what other
processes are running on the machine (in the process of which I discovered that,
for some reason, NT doesn't seem to include time spent updating graphics or the
UI in either time count, which to an extent suits me, though I wasn't expecting
it).

-- Chris

Post a reply to this message

Chris Cason wrote:
>
> "Lance Birch" <-> wrote in message news:3bdcc5f5@news.povray.org...
> > I'm not sure, with RC5 running I can get both processors doing 100% (if
not
> > with a little bit or kernel action here and there).  I'll have to
> > investigate it further.  AFAIK POV-Ray is single-threaded as well, but
even
> > still, setting it to the highest thread priority should get it using
100% of
> > one processor.  I'll have to try manually setting it in the task manager
on
> > pvengine.
>
> Incidentally, if you download beta.7 (available in the next 24 hours), you
will
> find that on NT it prints a summary of actual kernel+user time after each
render.
>
> I added this to allow for more accurate benchmarking, regardless of what
other
> processes are running on the machine (in the process of which I discovered
that,
> for some reason, NT doesn't seem to include time spent updating graphics
or the
> UI in either time count, which to an extent suits me, though I wasn't
expecting
> it).

Thanks Chris, this'll be an interesting thing to watch.  I've been
monitoring the way it's prioritised across both processors and it seems to
get the equivalent of 100% of one processor (regardless of the assigned
affinity), which is good :)

Lance.

thezone.firewave.com.au
www.firewave.com.au

Post a reply to this message

Nice image. Can you share the photons and radiosity parameters with us?

Ruy

Post a reply to this message

Lance Birch <-> wrote:
: The IOR is 1.15

  That explains a lot. They just didn't have that glass-refraction look.

  The IOR for glass is 1.5, and for water it's 1.33. So what you have modelled
are torii made of some gas which is a bit denser than air. In fact, those
torii couldn't exist in the real world (a material of that density would just
vaporize as any gas). No wonder why they looked odd... :P

-- 
#macro N(D,I)#if(I<6)cylinder{M()#local D[I]=div(D[I],104);M().5,2pigment{
rgb M()}}N(D,(D[I]>99?I:I+1))#end#end#macro M()<mod(D[I],13)-6,mod(div(D[I
],13),8)-3,10>#end blob{N(array[6]{11117333955,
7382340,3358,3900569407,970,4254934330},0)}//                     - Warp -

Post a reply to this message

Warp wrote:
> Lance Birch <-> wrote:
> : The IOR is 1.15
>
>   That explains a lot. They just didn't have that glass-refraction look.
>
>   The IOR for glass is 1.5, and for water it's 1.33. So what you have
modelled
> are torii made of some gas which is a bit denser than air. In fact, those
> torii couldn't exist in the real world (a material of that density would
just
> vaporize as any gas). No wonder why they looked odd... :P

Yes I know all the values for different materials... I was more just going
for artistic mertic ;)

Lance.

thezone.firewave.com.au
www.firewave.com.au

Post a reply to this message