A quick note about something I just discovered...

   When using two-pass radiosity, if you use a little focal blur during the
first pass (just a few blur samples is enough), the you can scale up the
render size on the second pass without suffering the typical artefacts (and 
you don't need to use focal blur on the final pass).

   I've used this trick successfully with the first pass at 320x240 and the 
second at 1024x768. So, for the first pass I'm now using that in the camera:

   aperture .01 blur_samples 7 focal_point <,,>

   Regards,

--
Jaime

Post a reply to this message

Would jitter do the same?  It seems focal blur is:
1. relatively expensive
2. introduces an artifact, as it's depth-dependent

Jaime Vives Piqueres wrote:
>   A quick note about something I just discovered...
> 
>   When using two-pass radiosity, if you use a little focal blur during the
> first pass (just a few blur samples is enough), the you can scale up the
> render size on the second pass without suffering the typical artefacts 
> (and you don't need to use focal blur on the final pass).
> 
>   I've used this trick successfully with the first pass at 320x240 and 
> the second at 1024x768. So, for the first pass I'm now using that in the 
> camera:
> 
>   aperture .01 blur_samples 7 focal_point <,,>
> 
>   Regards,
> 
> -- 
> Jaime

Post a reply to this message

Dan Connelly nous illumina en ce 2009-01-09 13:45 -->
> Would jitter do the same?  It seems focal blur is:
> 1. relatively expensive
> 2. introduces an artifact, as it's depth-dependent
> 
> Jaime Vives Piqueres wrote:
>>   A quick note about something I just discovered...
>>
>>   When using two-pass radiosity, if you use a little focal blur during 
>> the
>> first pass (just a few blur samples is enough), the you can scale up the
>> render size on the second pass without suffering the typical artefacts 
>> (and you don't need to use focal blur on the final pass).
>>
>>   I've used this trick successfully with the first pass at 320x240 and 
>> the second at 1024x768. So, for the first pass I'm now using that in 
>> the camera:
>>
>>   aperture .01 blur_samples 7 focal_point <,,>
>>
>>   Regards,
>>
>> -- 
>> Jaime
With this little samples, it's not realy expensive. Also, the sampling is 
normaly jittered.
What it does here, is to force the taking of some more radiosity samples during 
the first pass.

-- 
Alain
-------------------------------------------------
You know you've been raytracing too long when you've gained twenty pounds 
sitting at the computer, but can't tell because your beard covers your stomach.
Taps a.k.a. Tapio Vocadlo

Post a reply to this message

Alain wrote:
> With this little samples, it's not realy expensive. Also, the sampling 
> is normaly jittered.
> What it does here, is to force the taking of some more radiosity samples 
> during the first pass.
> 

Is the depth dependence a concern?   At the focal depth, there should be no effect.

Post a reply to this message

Dan Connelly escribió:
> Alain wrote:
>> With this little samples, it's not realy expensive. Also, the sampling 
>> is normaly jittered.
>> What it does here, is to force the taking of some more radiosity 
>> samples during the first pass.
>>
> 
> Is the depth dependence a concern?   At the focal depth, there should be 
> no effect.

   Now that you mention it, I was using a focal point very close to the 
camera location, so the focal blur effect must have been more or less 
uniform, I suppose.

--
Jaime

Post a reply to this message

Jaime Vives Piqueres <jai### [at] ignoranciaorg> wrote:
> A quick note about something I just discovered...
>
>    When using two-pass radiosity, if you use a little focal blur during the
> first pass (just a few blur samples is enough), the you can scale up the
> render size on the second pass without suffering the typical artefacts (and
> you don't need to use focal blur on the final pass).
>
>    I've used this trick successfully with the first pass at 320x240 and the
> second at 1024x768. So, for the first pass I'm now using that in the camera:
>
>    aperture .01 blur_samples 7 focal_point <,,>
>
>    Regards,
>
> --
> Jaime

Always a man of many tricks! :)

good to see you're back at povving!

Post a reply to this message

Jaime Vives Piqueres <jai### [at] ignoranciaorg> wrote:
>    When using two-pass radiosity, if you use a little focal blur during the
> first pass (just a few blur samples is enough), the you can scale up the
> render size on the second pass without suffering the typical artefacts (and
> you don't need to use focal blur on the final pass).

- Did you see any speed (or quality) advantage over using a higher-resolution
image?

- How does Anti-Aliasing with jitter perform in comparison?

Post a reply to this message

> - Did you see any speed (or quality) advantage over using a 
> higher-resolution image?

   The advantage is only speed, and the quality is almost the same. I just
did 3 quick tests with the same rad settings:

   1) 1024x768 for both passes as reference for quality and render time

   2) first pass of only 256x192, second 1024x768 -> 38% render time, but
there are many artefacts not present on case 1.

   3) same as 2, but with focal blur (7 samples) -> 65% render time, no
artefacts visible, and the image feels almost identical to case 1.

> - How does Anti-Aliasing with jitter perform in comparison?

   It works also to obtain better results for "scaled up radiosity", but
doesn't eliminates all the artefacts as the focal blur trick does. And
increasing the aa depth, the results were not much better, even with render
times bigger than with focal blur.

--
Jaime

Post a reply to this message

Jaime Vives Piqueres <jai### [at] ignoranciaorg> wrote:
>    1) 1024x768 for both passes as reference for quality and render time
>
>    2) first pass of only 256x192, second 1024x768 -> 38% render time, but
> there are many artefacts not present on case 1.
>
>    3) same as 2, but with focal blur (7 samples) -> 65% render time, no
> artefacts visible, and the image feels almost identical to case 1.
>
> > - How does Anti-Aliasing with jitter perform in comparison?
>
>    It works also to obtain better results for "scaled up radiosity", but
> doesn't eliminates all the artefacts as the focal blur trick does. And
> increasing the aa depth, the results were not much better, even with render
> times bigger than with focal blur.

What type of artifacts are we talking about? Black splotches? Generic "dirty"
look? Or the horizontally-smeared artifacts typical for samples gathered during
final trace?

If we're talking about the latter, then those are very strange results; maybe
there's a secret hidden in this approach that might help design a faster
pretrace algorithm.

Does changing the focal blur aperture have an effect on this?

Post a reply to this message

> What type of artifacts are we talking about? Black splotches? Generic
> "dirty" look? Or the horizontally-smeared artifacts typical for samples
> gathered during final trace?

   Look at the images I just posted on p.b.i. (just forget about the ugly
compression) : the number on the file names reflect the cases I listed. Look
at the second image and you will see artefacts on the edges between
surfaces. These can be worse in "real" scenes with finer radiosity settings,
appearing as horizontal lines on the shadows.

> If we're talking about the latter, then those are very strange results;
> maybe there's a secret hidden in this approach that might help design a
> faster pretrace algorithm.
> 
> Does changing the focal blur aperture have an effect on this?

   Hmmm... didn't try this, I just assumed that aperture didn't matter. Will
try it later...

   Regards,

--
Jaime

Post a reply to this message