|
good read
On 6/5/06, Francois Lord <francoislord(at)gmail.com> wrote:
And to illustrate that:
http://www.lamrug.org/resources/samplestips.html
middle of the page.
Halfdan Ingvarsson wrote:
> The filter size *does* have a say in how many samples are taken. If
> the filter needs samples that lie outside of the tile boundary, more
> samples will be taken for that tile even if they overlap what has
> already been sampled in surrounding tiles. Sample data is not shared
> between tiles.
>
> With a scene where min/max sample is zero, if your tile size is set to
> 32, and the filter size is set to 1x1, the number of samples is 1089,
> which corresponds to 33 x 33 pixel corner samples. If you whack the
> filter size to 3x3, the number of samples goes to 1225, which
> correspomds to 35 x 35 corner samples (1 extra pixel on each side,
> since that's what the filter now covers from the pixel center).
>
> You see, the scene is sampled at pixel corners but filtered at pixel
> centers, with x and y filter size controlling the width and height of
> the filter (not radius!) This is the main reason as to why it is not
> possible to get a non-filtered output from mental ray (is that a
> double negative?).
>
> - ½
---
Unsubscribe? Mail Majordomo(at)Softimage.COM
with the following text in body:
unsubscribe xsi
|
