abx### [at] abxartpl news:7glic0h9hssmch7sq5rh4ne9jj2tofmesv@4ax.com

> povray +SR76 +ER76 +SC64 +EC158
> povray +SR77 +ER77 +SC60 +EC140

And parse each 10,000 lines of .pov, load 10 .df3 and 30 textures for each 
line? No, thanks ;)

-- 
http://www.raf256.com/3d/
Rafal Maj 'Raf256', home page - http://www.raf256.com/me/
Computer Graphics

Post a reply to this message

On 11 Jun 2004 06:51:00 -0400, "Rafal 'Raf256' Maj" <spa### [at] raf256com> wrote:
> > povray +SR76 +ER76 +SC64 +EC158
> > povray +SR77 +ER77 +SC60 +EC140
>
> And parse each 10,000 lines of .pov, load 10 .df3 and 30 textures for each 
> line?

You already have to do it after changing finish in isosurface as in example
you gave. Or do you have at hands ready algorithm where it is easy to find
pixels in image where changed finish is visible (including reflections and
transparency of other objects)?

If you already know you can recreate those few pixels because there is no
mirror or transparent I do not believe that all 10.000 lines, 10.df3 files and
30 textures are *all* visible in those two short rows of pixels. Conditional
parsing is popular solution.

ABX

Post a reply to this message

abx### [at] abxartpl news:3n3jc09gk9e7bvt26knb90sk50b62ug0i6@4ax.com

>  Or do you have at hands ready algorithm where it is easy to find
> pixels in image where changed finish is visible (including reflections
> and transparency of other objects)?

Ofcourse we have. The raytrace procedure does *exacly* it, only need to add 
extra value returned by it (nor only R,G,B but also information like 
bool was_object_XXX_hit)
 
> If you already know you can recreate those few pixels because there is
> no mirror or transparent I do not believe that all 10.000 lines,
> 10.df3 files and 30 textures are *all* visible in those two short rows
> of pixels. Conditional parsing is popular solution.

Its also a good idea.


-- 
http://www.raf256.com/3d/
Rafal Maj 'Raf256', home page - http://www.raf256.com/me/
Computer Graphics

Post a reply to this message

On 11 Jun 2004 07:23:17 -0400, "Rafal 'Raf256' Maj" <spa### [at] raf256com> wrote:
> abx### [at] abxartpl news:3n3jc09gk9e7bvt26knb90sk50b62ug0i6@4ax.com
> >  Or do you have at hands ready algorithm where it is easy to find
> > pixels in image where changed finish is visible (including reflections
> > and transparency of other objects)?
>
> Ofcourse we have.

Of course we haven't. We could have if somebody would implement it. IIRC you
were talking about it already in the past and AFAIR you promised to deliver
patch with something like this. But you didn't IIRC. Did you?

> The raytrace procedure does *exacly* it, only need to add 
> extra value returned by it (nor only R,G,B but also information like 
> bool was_object_XXX_hit)

With the risk of repeating it again, are you aware of how many objects are hit
in radiosity scenes for each pixel with every intersection? Do you know memory
behind it? Do you know that antialiasing increases it? Do not theoretize
please, Rafal. Take a compiler which you already have and finaly prove us
something! I would be happy to see it.

ABX

Post a reply to this message

abx### [at] abxartpl news:ek6jc0pqg1313gv43dqo485du7mag1h7re@4ax.com

> With the risk of repeating it again, are you aware of how many objects
> are hit in radiosity scenes for each pixel with every intersection? Do
> you know memory behind it? 

Who said that each ray should have full history of with objects is he 
depending on (with object do this ray or its children etc. hit ;) ?

Its another way around, YOu have one object i.e. 27-th isosurface in scene, 
YOu store its pointer globaly, and each ray only have a boolean information 
did he hit it - YES or NO

> Do you know that antialiasing increases it?

it don't in case a above

> Do not theoretize please, Rafal. Take a compiler which you already
> have and finaly prove us something! I would be happy to see it.

I would also heapy if I just had some free time...


-- 
http://www.raf256.com/3d/
Rafal Maj 'Raf256', home page - http://www.raf256.com/me/
Computer Graphics

Post a reply to this message

On 11 Jun 2004 11:07:16 -0400, "Rafal 'Raf256' Maj" <spa### [at] raf256com> wrote:
> > With the risk of repeating it again, are you aware of how many objects
> > are hit in radiosity scenes for each pixel with every intersection? Do
> > you know memory behind it? 
>
> Who said that each ray should have full history of with objects is he 
> depending on (...) YOu have one object i.e. 27-th isosurface in scene, 
> YOu store its pointer globaly, and each ray only have a boolean information 
> did he hit it - YES or NO

So your "proposition" is supposed to work only as long as you are modyfing one
object?

(note object pattern does not use Trace routine)

ABX

Post a reply to this message

"Rafal 'Raf256' Maj" <spa### [at] raf256com> wrote:
> abx### [at] abxartpl news:ek6jc0pqg1313gv43dqo485du7mag1h7re@4ax.com
>
> > With the risk of repeating it again, are you aware of how many objects
> > are hit in radiosity scenes for each pixel with every intersection? Do
> > you know memory behind it?
>
> Who said that each ray should have full history of with objects is he
> depending on (with object do this ray or its children etc. hit ;) ?

You would need it, in order to know which pixels to re-render.

> Its another way around, YOu have one object i.e. 27-th isosurface in scene,
> YOu store its pointer globaly, and each ray only have a boolean information
> did he hit it - YES or NO

Oo... Global Variables = Bad (sorry, couldn't resist) :)

Anyway, this way would actually use more memory - you would have an array of
(image_width * image_height) of booleans *PER* *OBJECT*.  A list of rays
which hit would be better, but would probably be equivalent to each ray
having a list of object which affect it.

In fact, something like this could probably be implemented for tiles,
instead of single rays, but it would only be useful for interactive
modeling.  The way POV-Ray currently works would not support it.

....Chambers

Post a reply to this message

> The way it works is the master opens a new session, delivering
> "command-line" arguments over RPC. Then, the MegapovXRS slave parses the
> scene, shoots any photons and then waits for tile requests. When tiles
> come in they are rendered *without* having to parse or shoot photons
> again. This should provide a substantial boost (compared to a pure
> script based approach) when it comes to network rendering scenes with
> long parse times.

Wow! That's good.

Is it Linux- or Windows-based ?


> I have a highly functional beta here and I'm working as we speak on
> preparing it for release to my beloved POV-Ray community for review.

Can't wait to see that!

Best,

    S.

>
> Best Regards,
> George Pantazopoulos
> http://www.gammaburst.net
>
>

Post a reply to this message

Steven Pigeon wrote:
>>The way it works is the master opens a new session, delivering
>>"command-line" arguments over RPC. Then, the MegapovXRS slave parses the
>>scene, shoots any photons and then waits for tile requests. When tiles
>>come in they are rendered *without* having to parse or shoot photons
>>again. This should provide a substantial boost (compared to a pure
>>script based approach) when it comes to network rendering scenes with
>>long parse times.
> 
> 
> Wow! That's good.
> 
> Is it Linux- or Windows-based ?
> 
> 

Hi Steven,

MegapovXRS is linux-based. What platform do you use?

> 
>>I have a highly functional beta here and I'm working as we speak on
>>preparing it for release to my beloved POV-Ray community for review.
> 
> 
> Can't wait to see that!
>

Thanks, same here. I'm trying to work out how to best handle the many 
library dependencies and also getting some bugs out (eg. transferring 
the rendered images works over a LAN but not always over the net). It 
can be difficult. Due to the use of libraries there's a substantial 
amount of code that I didn't write, yet still have to debug along with 
my own code when something goes wrong :)

Regards,
George Pantazopoulos
http://www.gammaburst.net

Post a reply to this message

> Hi Steven,
>
> MegapovXRS is linux-based. What platform do you use?

Actually, both. XP to get the scene right, and Linux to do the
very lengthy renders.

Best,

    S.

--
Steven Pigeon, Ph. D.
ste### [at] stevenpigeoncom
ste### [at] videotronca

Post a reply to this message