 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On 25 Feb 2002 08:30:30 -0500, Warp <war### [at] tag povrayorg> wrote:
> Could you tell me what is the difference between a smooth and an
> interpolated image?
With term "smooth" I refered to previous post of Thorsten. It was (probably
wrong) shortcut to describe AA.
ABX
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Skiba <abx### [at] babilonorg> wrote:
> But when calculations of this functions are very complicated and this pattern
> is intersected many times (for example via reflections) then image function is
> afaik only alternative to speed things up.
The function will still have to be called a lot. Further, it is extremely
simple to just write a function like this and get any supersampling effect you
want with much more flexibility and about the same speed:
(foo(u+fdelta,v+fdelta) + foo(u-fdelta,v+fdelta) + foo(u,v)*4 +
foo(u+fdelta,v-fdelta) + foo(u-fdelta,v-fdelta)) / 8}
Thorsten
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Mon, 25 Feb 2002 14:38:31 +0100, "Thorsten Froehlich" <tho### [at] trfde>
wrote:
> The function will still have to be called a lot. Further, it is extremely
> simple to just write a function like this and get any supersampling effect you
> want with much more flexibility and about the same speed:
>
> (foo(u+fdelta,v+fdelta) + foo(u-fdelta,v+fdelta) + foo(u,v)*4 +
> foo(u+fdelta,v-fdelta) + foo(u-fdelta,v-fdelta)) / 8}
I think you took my posts wrong way.
I didn't requested it is necessary to implement AA to image function so no
need to show me workarounds (note I wrote proximity/blur macro i
text.scene-files as example how such thing could look). I only pointed that
subsampling has sense and showing above code you proved this imo.
ABX
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Skiba <abx### [at] babilonorg> wrote:
> I think you took my posts wrong way.
> I didn't requested it is necessary to implement AA to image function so no
> need to show me workarounds (note I wrote proximity/blur macro i
> text.scene-files as example how such thing could look). I only pointed that
> subsampling has sense and showing above code you proved this imo.
No, you asked for this effect, so there it is. It still does not make sense,
but as you insisted that it would be useful ... now you can try it and see for
yourself that you gain nothing :-)
Thorsten
____________________________________________________
Thorsten Froehlich, Duisburg, Germany
e-mail: tho### [at] trfde
Visit POV-Ray on the web: http://mac.povray.org
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Mon, 25 Feb 2002 18:22:19 +0100, "Thorsten Froehlich" <tho### [at] trfde>
wrote:
> No, you asked for this effect, so there it is.
I asked note in documentation becouse I think documentation is not only
specification but user-friendly story about how it works and what is (or
isn't) performed during calculations.
> It still does not make sense,
> but as you insisted that it would be useful ... now you can try it and see for
> yourself that you gain nothing :-)
You can check what the difference is in
http://news.povray.org/oh4n7u47ee1bmgh4mipmf04msaatvcuoj6%404ax.com
Both images have the same resolution (600x600), both have interpolation 2
applied, but right one has your own code applied before interpolation.
(sorry Thorsten for accidental mail)
ABX
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Skiba <abx### [at] babilonorg> wrote:
>> It still does not make sense,
>> but as you insisted that it would be useful ... now you can try it and see
>> for yourself that you gain nothing :-)
>
> You can check what the difference is in
> http://news.povray.org/oh4n7u47ee1bmgh4mipmf04msaatvcuoj6%404ax.com
>
> Both images have the same resolution (600x600), both have interpolation 2
> applied, but right one has your own code applied before interpolation.
Sure it looks different as expected, but that doesn't say anything about the
usefulness. You can just use the function directly you want to get rid of the
"pixel" effect. Image functions are after all there specifically to get an
image of the function. If you use them for something else, well, what do you
expect?
I really see no point in explaining that an image is composed of pixels while
a function is evaluated for every point with its precise coordinates (or
whatever way to explain it), because that is the only way they can and do work
everywhere...
> (sorry Thorsten for accidental mail)
No problem.
Thorsten
____________________________________________________
Thorsten Froehlich, Duisburg, Germany
e-mail: tho### [at] trfde
Visit POV-Ray on the web: http://mac.povray.org
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Tue, 26 Feb 2002 15:30:41 +0100, "Thorsten Froehlich" <tho### [at] trfde>
wrote:
> You can just use the function directly you want to get rid of the
> "pixel" effect.
That's obvious. The problem is to use image to _avoid_ direct function
calculation for the same coordinates. Imagine: You make poster 3000x3000.
There is flat object with pattern based on complicated function. This object
is projected onto image into area 300x300. This object has also many (20?)
reflections on other objects. Each reflection is smaller or the same size.
When I use AA of rendering and direct function in pattern then effect is
smooth but rendering is veeery slow becouse some calculations are performed
for similiar coordinates of pattern. So I want to speed things up making image
function. I don't want waste additional memory for high resolution of this
image function becouse memory is important in such large poster rendering. So
I use additional amount of time during parsing to prepare blured image
function. Looking at my experiments I can save 50% of memory used by not
blurred image with the same effect.
If above is still useles for you then EOT for me. I don't know how to convince
you of it.
ABX
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Skiba <abx### [at] babilonorg> wrote:
> If above is still useles for you then EOT for me. I don't know how to convince
> you of it.
At some point ray-tracing involves either memory or time or usually even both.
So indeed your argument does not convince me.
Thorsten
____________________________________________________
Thorsten Froehlich, Duisburg, Germany
e-mail: tho### [at] trfde
Visit POV-Ray on the web: http://mac.povray.org
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
