Jaime Vives Piqueres <jai### [at] ignoranciaorg> wrote:
> Bruno Cabasson wrote:
> > For cumuli, the problem is different than clouscapes and quite tricky. I think
> > I'll have to do with single clouds. But I am afraid that the turbulence feature
> > is not completely appropritate for the 'roundness' of cumuli. I tried with the
> > current system, but it is not suitable for cumuli and I could never get
> > acceptable results (mainly because the density is based on standard patterns).
> > Perhaps a particle system will be necessary, but I absolutely do not master
> > that. Advice/ideas welcome.
>
>    For my cumuli I used the object pattern, using as object a conglomerate
> of spheres, distributed in a loop to resemble the shape of a cloud, then
> used this with media and turbulence. But as you said, I never got realistic
> results with turbulence... it should be my fault, because Gilles used it
> very well on his DF3 clouds.
>
> > It can define the terrain either as
> >     *) pigment-based function (user-defined), allowing overhangs (isosurface)
> >     *) Height field out of pigment function
> >     *) Height field out of image_map
> >     *) Image_map generation from pigment function
>
>    I would add "isosurface based on a external heightfield". I have some WIP
> on this subject, and it is very handy, as you can control the general shape
> of the terrain with a heightfield "guide" and then add overhangs with an
> extra function. I even developed a method to add the overhangs only on the
> vertical parts, although it is very trick (requires calculating a "slope
> map" of the heightfield first).
>
> > It can (will) define the sky as
> >     *) Media, like the skies I already posted. Allows cloudscape to cast shadows
> >     -) Sky-dome image map (probably a sphere)
> >     -) DF3 generation for single clouds.
>
>    For the skydome imagemap (which can be very useful indeed), take into
> account the possible output as HDR. This means getting a good balance
> between the intensity of the sky and the sun.
>
> >     *) Water
>
>    Plane, heightfield or isosurface? For close-ups to the shoreline, an hf
> or isosurface is very convenient.
>
>   > One big piece will also be to incorporate vegetation. It can use
> POV-Tree, and
> > stuff from PlantStudio or so. I would like to provide for pattern-based
> > placement macros (I already have something for that).
>
>    I think it's better to let the user supply their own plants/trees, you
> only need to provide the placement macros.
>
> > I want all be declarative, so that a MMI can be easily plugged. For example, for
> > the last version of my sky system, and for the sky that is currently rendering:
>
>    Looks like the right direction to me...
>
> > PS: By essence, skies require intensively the media feature. What about media
> > with 3.7?
>
>    IIRC, it will get ride of the container artifacts you are going to meet
> with individual clouds... ;)
>
> --
> Jaime

Thanks for advise and kindness Jaime. You are one of our masters and I am
flattered that my modest work gets your attention.

I obtained 'correct' individual clouds that ressemble cumuli using turbulence,
somewhat like Gille's. But sometimes I see real huge round and nearly solid
cumuli occupying a large protion of the sky. I could never approach this
result.

OK for the HDR feature and format.

My next step now will be the Sun in order to complete the sky system.

At the moment water is flat with fade_power/fade_distance/fade_color,
reflection, normal and media in some occasions. I'll see later for HF's and
isosurface.

Concerning vegetation, I had in mind that TerraPOV could provide a basic yet
suffficient vegetation system, quite easily extracted from PlantStudio.
However, I'll see this when all the rest is OK.

I was many many many times confronted with the container artifact for media.
Glad 3.7 will fix it.

    Bruno

Post a reply to this message

> Nicolas Alvarez <nic### [at] gmailisthebestcom> wrote:

>>> PS: By essence, skies require intensively the media feature. What about media
>>> with 3.7?
>> What's your problem, too slow? Send me a zip with the scene and all
>> needed files (even the *standard* include files) and settings you want.
>> As long as you don't use radiosity, I can use tiled rendering on it to
>> spread it on a dozen machines.
> 
> Thanks for the proposal! I'll think of it if needed!
> 
> Concerning 3.7, I made a few tests and the results were different in brightness
> and color. Maybe I'll try again... I also noticed that it was faster on the
> contrary.

Note that my renderfarm runs *3.6*, since I needed to modify the source 
code. I should merge the 3.6.1 changes... but it's not exactly a 
priority :) I'd rather wait for 3.7 code and rewrite all my changes (the 
new .pov-state resume file will be really useful!).

Post a reply to this message