I've been trying and trying to get media to produce realistic flames
like halo used to, but to no avail.

My first problem stems from the fact that an emitting media doesn't seem
to allow the user specify a transmit or filter value, like halo used to
do, so flames disappear against certain background colours.  Adding an
absorption media in the same object helps some, but (a) it doesn't seem
to produce the same results, and (b) it slows things down even more.

My second problem is that when I introduce turbulence, the media looks
grainy no matter what I do to try and fix it.  I've tried increasing the
samples, using finer anti-aliasing, and so on, but I can't seem to get
smooth blends of colour like I did with a relatively low-sample (and
even un-antialiased) halo.

The usual advice I've seen is to go back to 3.0 for those scenes, but
that means that other advances -- such as Bezier-spline lathes and
prisms -- then cannot be used.

So if anyone can help me understand how to get media to look like the
fire and flame which the old halos could produce, I'd greatly appreciate
it.


-- 
Jeff Lee         shi### [at] gatenet         http://www.gate.net/~shipbrk/

Post a reply to this message

Jeff Lee wrote:
> 
> I've been trying and trying to get media to produce realistic flames
> like halo used to, but to no avail.
> 
> My first problem stems from the fact that an emitting media doesn't seem
> to allow the user specify a transmit or filter value, like halo used to
> do, so flames disappear against certain background colours.  Adding an
> absorption media in the same object helps some, but (a) it doesn't seem
> to produce the same results, and (b) it slows things down even more.
> 
> My second problem is that when I introduce turbulence, the media looks
> grainy no matter what I do to try and fix it.  I've tried increasing the
> samples, using finer anti-aliasing, and so on, but I can't seem to get
> smooth blends of colour like I did with a relatively low-sample (and
> even un-antialiased) halo.
> 
> The usual advice I've seen is to go back to 3.0 for those scenes, but
> that means that other advances -- such as Bezier-spline lathes and
> prisms -- then cannot be used.
> 
> So if anyone can help me understand how to get media to look like the
> fire and flame which the old halos could produce, I'd greatly appreciate
> it.
> 
> --
> Jeff Lee         shi### [at] gatenet         http://www.gate.net/~shipbrk/

The best advice I can offer is to search the scene files groups and the images
groups with "media" in the subject line. There have been several flames posted
with source - even a couple by me. The parameters change from scene to scene
and I wouldn't know where to begin to advise you without seeing your source
plus having the time to work on it (not much of that lately I'm afraid).

-- 
Ken Tyler

mailto://tylereng@pacbell.net

Post a reply to this message

My memory fails me... But if I'm not mistaken, halo used the same types -
emitting, absorbing and scattering. The basic principle behind them has
remained the same, i.e. emission colour is added to the backround,
absorption colour is subtracted etc. The only problem here is getting the
right density, which is now dependent on the length of the rgb vector. I
can't help you much here, except that a) you should definitely use
absorption along with emission, lest the media should disappear against a
lighter background, and b) absorption and emission should have roughly the
opposite colours (in terms of hue). One very important thing to remember is
that although emission <1,0,0> is red, absorption <1,0,0> is _not_ red, but
cyan - red is blocked out.

As for the second problem, in my experience it's pretty much a matter of
luck. Sometimes media remains grainy no matter what. But scattering media is
by far the worst in this respect. Emittin and absorbing should give more or
less smooth results with a bit of tweaking. You might try scaling it a bit,
up min samples and confidence/variance... But as I said, it's a matter of
luck.
I personally (and it seems many others too) miss the old sampling method.
Perhaps it becomes an option in the next release.

I have seen many good media flames in these ng-s and IRTC. Although none
come to mind that I would call 100 per cent realistic. A flame is a complex
phenomenon and not easy to model with the current density types. Perhaps
someone would implement a special "flame" pattern with options like origin,
whorl, rise, type (candle, campfire etc). Any takers? Didn't think so...

Margus

Jeff Lee wrote in message <376a75e2@news.povray.org>...
>I've been trying and trying to get media to produce realistic flames
>like halo used to, but to no avail.
>
>My first problem stems from the fact that an emitting media doesn't seem
>to allow the user specify a transmit or filter value, like halo used to
>do, so flames disappear against certain background colours.  Adding an
>absorption media in the same object helps some, but (a) it doesn't seem
>to produce the same results, and (b) it slows things down even more.
>
>My second problem is that when I introduce turbulence, the media looks
>grainy no matter what I do to try and fix it.  I've tried increasing the
>samples, using finer anti-aliasing, and so on, but I can't seem to get
>smooth blends of colour like I did with a relatively low-sample (and
>even un-antialiased) halo.
>
>The usual advice I've seen is to go back to 3.0 for those scenes, but
>that means that other advances -- such as Bezier-spline lathes and
>prisms -- then cannot be used.
>
>So if anyone can help me understand how to get media to look like the
>fire and flame which the old halos could produce, I'd greatly appreciate
>it.
>
>
>--
>Jeff Lee         shi### [at] gatenet         http://www.gate.net/~shipbrk/
>

Post a reply to this message