Hi all,

I've worked on povray in fits and starts, and have correspondingly lurked
around here.  This isn't my first post, but it might be my second.  I know
it doesn't fit the rules exactly, but there is a checker pattern involved
in the floor texture,  and spheres were involved in creating some of the
other stuff, and.... oh, never mind.  :)

Some explanation:  I started out thinking to create three-dimensional
versions of parts of one of Frank Stella's large multi-canvas paintings.
The idea was to create, ex post facto, the subjects of paintings that were
meant to be purely abstract.  This explains the shapes of the three
foreground objects, and the texture on the one.  Where the three squares
are on the wall, I would show the corresponding 2D versions.

Now, having come to this point, I have choice to make.  Do I procede with
the above idea, and introduce the remaining concentric ring textures, or do
I leave Stella behind and rework it with my own shapes and perhaps somewhat
less busy textures?

Fortunately, this image rendered in about four and a half hours, so I can
run pretty high quality tests overnight.  I do wish the focal blur gave
better antialiasing along high-contrast edges (such as on the ring
texture).  This image used 100 samples with variance 0.001 and confidence
0.95.

Anyway, I'd be happy to have your comments and suggestions.

  -Ed

Post a reply to this message

Attachments:
Download 'min_wip.jpg' (25 KB)

Preview of image 'min_wip.jpg'

Ed Jackson wrote:
> Hi all,
> 
> I've worked on povray in fits and starts, and have correspondingly lurked
> around here.  This isn't my first post, but it might be my second.  I know
> it doesn't fit the rules exactly, but there is a checker pattern involved
> in the floor texture,  and spheres were involved in creating some of the
> other stuff, and.... oh, never mind.  :)
> 
> Some explanation:  I started out thinking to create three-dimensional
> versions of parts of one of Frank Stella's large multi-canvas paintings.
> The idea was to create, ex post facto, the subjects of paintings that were
> meant to be purely abstract.  This explains the shapes of the three
> foreground objects, and the texture on the one.  Where the three squares
> are on the wall, I would show the corresponding 2D versions.
> 
> Now, having come to this point, I have choice to make.  Do I procede with
> the above idea, and introduce the remaining concentric ring textures, or do
> I leave Stella behind and rework it with my own shapes and perhaps somewhat
> less busy textures?
> 
> Fortunately, this image rendered in about four and a half hours, so I can
> run pretty high quality tests overnight.  I do wish the focal blur gave
> better antialiasing along high-contrast edges (such as on the ring
> texture).  This image used 100 samples with variance 0.001 and confidence
> 0.95.
> 
> Anyway, I'd be happy to have your comments and suggestions.
> 
>   -Ed
> 
> 
> ------------------------------------------------------------------------
> 

1)  I really love the idea!  Almost a kind of whimsical revisionist 
history.

2)  All the radiosity and focal blur doesn't have to be such a factor 
initially as you develop the idea, or other variations, does it?

Post a reply to this message

Jim Charter <jrc### [at] msncom> wrote:
>
> 2)  All the radiosity and focal blur doesn't have to be such a factor
> initially as you develop the idea, or other variations, does it?

No, of course not.  I can render reasonable previews without all that, for
purposes of texturing and composition, in a minute or less.  On the other
hand, running one overnight with all the goodies turned on gives me a
chance to tweak those settings too, without waiting until the last minute.
Naturally, it's also a bit of self-indulgence.

I really am a little frustrated with the antialiasing provided by the focal
blur.  I'd love to hear others' experience with where the diminishing
returns set in with variance and confidence settings.

As for "whimsical revisionist history", that's just what I was going for.
I'm glad it at least worked for you.

  -Ed

Post a reply to this message

"Ed Jackson" <nomail@nomail> wrote in message 
news:web.42d7397bdb87661e437d093a0@news.povray.org...
>
> I really am a little frustrated with the antialiasing provided by the 
> focal
> blur.  I'd love to hear others' experience with where the diminishing
> returns set in with variance and confidence settings.

I think you have a good idea going here, and it seems to be rendering out 
fine enough-- aside from the AA trouble you mentioned.

If you were to use the beta of version 3.7 then you could combine regular AA 
with the focal blur. I know that might be problematic to do but if you were 
to try it I can only suggest that AA method 2 is the thing to work with. 
There's currently some trouble with method 1, I believe. In fact, using the 
command-line +a0.2 +r2 +am2 could be adequate.

Since version 3.6 cannot combine the antialiasing and focal blur you can 
only supply more 'blur_samples' and use the most accuracy confidence 
possible. From what I've seen of v3.7 beta it might have better DOF when AA 
(method 2) is added.

Bob Hughes

Post a reply to this message

"Bob Hughes" <bob### [at] charternet> wrote:
>
> If you were to use the beta of version 3.7 then you could combine regular AA
> with the focal blur. I know that might be problematic to do but if you were
> to try it I can only suggest that AA method 2 is the thing to work with.
> There's currently some trouble with method 1, I believe. In fact, using the
> command-line +a0.2 +r2 +am2 could be adequate.
>

Thanks for this pointer!  I had been wishing for +am2 along with focal blur,
but I didn't realize they were doing it for 3.7.  Unfortunately for me,
though, the beta is only available as a windows executable.  Since I'm a
card-carrying Linux hippie, I guess I have to wait for the release.

I should be able to get decent enough results with 3.6 though.  Thanks again
for helping me stay up-to-date.  ;)

  -Ed

Post a reply to this message