Nope, no df3s here, it's a granite pattern with a lot of turbulence. The 
bands are parallel to the bounding shape, because in the distance each ray 
passes through a much longer range of clouds but uses the same number of 
samples (why doesn't pov adjust the min samples based on the distance it's 
sampling over?!), so you tend to get artefacts relating to large changes in 
depth of the bounding shape, which in this case means horizontal lines 
(effectively contours of depth of the bounding shape). And, before you say 
it, the bounding shape has to be like that because otherwise the clouds 
occupy such a small region of space that they mostly vanish when I use other 
bounding shapes, because no rays intersect with them.

Annoyingly media method 3's "anti-aliasing" seems blind to this artefact, so 
the only thing that fixes it is more samples => proportionally more render 
time :(

This one has 4 times as many samples and the artefacts are better, but still 
visible. Based on that I'd expect a 1500x1000 render without artefacts to 
take more than 24hrs! aaargh! So I'm not going to bother :)

-- 
Tek
http://evilsuperbrain.com

"David El Tom" <dav### [at] t-onlinede> wrote in message 
news:456bf263$1@news.povray.org...
> Tek schrieb:
>> I put the camera somewhere more interesting, and tweaked the lighting and
>> atmosphere a bit. Though this much higher-res image shows up some errors 
>> in
>> the clouds. Also this view shows some obvious levitating rocks, and I'm 
>> not
>> so fond of the boring grey rock texture...
>>
>> Anyway, still looks pretty dramatic so I thought I'd show it :)
>>
>> I'm probably going to use it as a backdrop for some 2D stuff in 
>> photoshop,
>> with some figures on the overhang on the left.
>>
> nice surreal appeal ...
>
> it looks like your using media clouds based on df3 data (banding/cell
> strukture). Instead of increasing the resolution of the df3 data you may
> add some turbulence *after* importing the density file. Will slow down
> the render process slightly but you get rid of the cell strukture, which
> is an inherent problem of density files.
>
> ... dave

Post a reply to this message