 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>I also tried making a variety of modifications to media, including evenly
>spaced samples and supersampling, but I couldn't get supersampling to work
>properly either. I now know what I was doing wrong. Make sure that when
>you super-sample, you give the super-samples less weight in the average.
>I.E. if you add a new sample between two existing samples, the new weights
>of the existing samples plus the weight of the new sample should still add
>up to the sum of the original weights of the two original samples. Ok...
>that sounds a lot more confusing than I intended.
Makes perfect sense to me.
A1 = original pixel 1
B1 = original pixel 2
A2 = new pixel 1
B2 = new pixel 2
C2 = supersampled pixel between A2 and B2
A1+B1 = A2+C2+B2
Right?
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
TonyB wrote:
>
> >I also tried making a variety of modifications to media, including evenly
> >spaced samples and supersampling, but I couldn't get supersampling to work
> >properly either. I now know what I was doing wrong. Make sure that when
> >you super-sample, you give the super-samples less weight in the average.
> >I.E. if you add a new sample between two existing samples, the new weights
> >of the existing samples plus the weight of the new sample should still add
> >up to the sum of the original weights of the two original samples. Ok...
> >that sounds a lot more confusing than I intended.
>
> Makes perfect sense to me.
>
> A1 = original pixel 1
> B1 = original pixel 2
> A2 = new pixel 1
> B2 = new pixel 2
> C2 = supersampled pixel between A2 and B2
>
> A1+B1 = A2+C2+B2
>
> Right?
Well, not exactly or you wouldn't need the supersampled pixel at all :)
but ideally yes (and as far as weight is concerned, yes)
Jerome
--
*******************************
* they'll tell you what can't * mailto:ber### [at] iname com
* be done and why... * http://www.enst.fr/~jberger
* Then do it. *
*******************************
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> I also tried making a variety of modifications to media, including evenly
> spaced samples and supersampling, but I couldn't get supersampling to work
> properly either. I now know what I was doing wrong. Make sure that when
> you super-sample, you give the super-samples less weight in the average.
> I.E. if you add a new sample between two existing samples, the new weights
> of the existing samples plus the weight of the new sample should still add
> up to the sum of the original weights of the two original samples. Ok...
> that sounds a lot more confusing than I intended.
I think I know what you are saying, but the way I'm doing it fits in with the
new sample_media function, so each time a supersample is taken it's added into
the total emission. Then later it's divided by the total number of samples for
that interval.
I think I found the problem though. I was looking through one of the functions
for where the values are added and noticed the variable d0 was still being used
even though I took d1 out of the function. I can only assume it's value is 0
because I forgot to pass it into the function. I'm going to try to fix that
and I hope, oh how I hope it works! I've spent so much time on this and really
haven't got anywhere.
-Mike
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Mike <pov### [at] aolcom> wrote in message news:37F03EC1.A53D7734@aol.com...
> I think I know what you are saying, but the way I'm doing it fits in with
the
> new sample_media function, so each time a supersample is taken it's added
into
> the total emission. Then later it's divided by the total number of
samples for
> that interval.
>
That will cause a problem, though. Uneven sampling (which is the result of
supersampling) needs uneven weights for the averaging process. A simple
sum-and-divide will not give desired results. I know, since I tried it.
;-)
-Nathan
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> That will cause a problem, though. Uneven sampling (which is the result of
> supersampling) needs uneven weights for the averaging process. A simple
> sum-and-divide will not give desired results. I know, since I tried it.
> ;-)
I can only assume you are correct since I can't get it to work. I'm still
trying to get a handle on how everything goes together in media.c.
-Mike
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
