I'm making a POV model of NASA's Cassini spacecraft. Needless to say 
it's built largely using metals, meaning I've been using stuff from 
metals.inc extensively. The problem is that these metals do not look 
realistic in the blackness of space because the shadowed parts of the 
spacecraft become 'self illuminating' and therefore have a very flat and 
unrealistic appearance. The reason is the high ambient values used in 
metals.inc. I've tried making my own metals by using a far lower value 
for ambient but I'm not happy although this resulted in some 
improvement. Most of the spacecraft now looks rather dark, even after I 
added several secondary lightsources. Any suggestions ?


-- 
Bjorn Jonsson / bjj### [at] ZZZmmediais
Address changed to avoid spam. Remove YYY and ZZZ to reply.
http://www.mmedia.is/~bjj

Post a reply to this message

> Any suggestions ?

Well, first of all you need some reference images. Maybe you have already
studied some photos taken by NASA. This is step one, when you're trying to
get that particular look. What kind of metal was used? Not all metals act as
complete mirrors; some diffuses the light. Some have a distinct texture (due
to irregularities) that helps us identify the material as metal. You don't
need to know the *name* of the metal, but to analyse how much light is
diffused (sometimes that's a lot) and how much texture there is.

Always set ambient to zero - always, in materials that reflects light rather
than emit light. The 'ambient' setting is a feature from old days when
computers were much slower. Have you tried global illumination instead? Even
a fast-rendering radiosity can do wonders to a scene.

Often, a metallic surface gives a blurred reflection. It may also help to
imitate this.

Regards,
Hugo

Post a reply to this message

news:MPG.1a56abe37b4346cb98968d@news.povray.org...
> Most of the spacecraft now looks rather dark, even after I
> added several secondary lightsources. Any suggestions ?

In addition to what Hugo said:

You need to define your lighting set-up: is it solar light only or a
combination or solar light and planetary radiance? Don't forget that lights
can be > 1 and in fact it's rather recommended in this sort of scene if you
want to have sharp contrasts. Radiosity can help a lot there. I did once a
rendering of the ISS that way
(http://www.oyonale.com/iss/english/museum_08.htm), The model isn't mine and
the pic could be much better (just to a few hours to set up, it was mostly a

for altitude and earth size. I based it on reference images for the
materials but didn't try too hard (I should have coded the model myself for
that).

And do you want it to be pretty (sci-fi like, with lots of secondary lights,
even if they do not make sense in deep space) or just realistic (all parts
not lit by the sun or a planet should be black) ? Artistic renderings of
spacecraft are rarely realistic in terms of lighting: see
http://icb.nasa.gov/cassini.gif : most of the light comes right from the
artist's mind...

G.

-- 
**********************
http://www.oyonale.com
**********************
- Graphic experiments
- POV-Ray and Poser computer images
- Posters

Post a reply to this message

On Fri, 26 Dec 2003 22:39:03 +0100, "Gilles Tran" <gitran_nospam_@wanadoo.fr>
wrote:

>I did once a
>rendering of the ISS that way

LOL What is that car doing there? 
Regards
        Stephen

Post a reply to this message

In article <3fecaa85$1@news.povray.org>, gitran_nospam_@wanadoo.fr 
says...

> news:MPG.1a56abe37b4346cb98968d@news.povray.org...
> > Most of the spacecraft now looks rather dark, even after I
> > added several secondary lightsources. Any suggestions ?
> 
> In addition to what Hugo said:
> 
> You need to define your lighting set-up: is it solar light only or a
> combination or solar light and planetary radiance? Don't forget that lights
> can be > 1 and in fact it's rather recommended in this sort of scene if you
> want to have sharp contrasts. Radiosity can help a lot there. I did once a
> rendering of the ISS that way
> (http://www.oyonale.com/iss/english/museum_08.htm), The model isn't mine and
> the pic could be much better (just to a few hours to set up, it was mostly a

> for altitude and earth size. I based it on reference images for the
> materials but didn't try too hard (I should have coded the model myself for
> that).
> 
> And do you want it to be pretty (sci-fi like, with lots of secondary lights,
> even if they do not make sense in deep space) or just realistic (all parts
> not lit by the sun or a planet should be black) ? Artistic renderings of
> spacecraft are rarely realistic in terms of lighting: see
> http://icb.nasa.gov/cassini.gif : most of the light comes right from the
> artist's mind...

To reply to what Hugo and Gilles said:

I have been using lots of images of the spacecraft from 
http://saturn.jpl.nasa.gov/multimedia/images/spacecraft/index.cfm as a 
reference, plus Wavefront obj files (polygon meshes) from 
http://samadhi.jpl.nasa.gov/models/ which includes no useful texture 
info but highly useful 'geometric' info. I suspect many of the metallic 
parts are aluminum.

Here are two renderings made using the metals.inc file supplied with 
POV:

http://www.mmedia.is/~bjj/povstuff/spacecraft/cassini_69_bright.png
http://www.mmedia.is/~bjj/povstuff/spacecraft/cassini_70_bright.png

These are much more recent versions than the ones on the webpage at my 
site in case anyone has looked at them (I plan to replace them Real Soon 
Now). Only one lightsource is used. In particular, the flat area below 
the big antenna disc look unrealistic, it should be largely in shadow. 
Closeup renderings can become especially unrealistic using these metals. 
The model is extremely (and probably ludicrously !) detailed so nice 
closeups should be possible.

Here are the same renderings done using a modified metals.inc where the 
ambient component is almost 0. In addition I'm using two secondary 
lightsources, both of intensity <1,1,1>/2 and I'm ware that the lights 
can have an intensity >1.

http://www.mmedia.is/~bjj/povstuff/spacecraft/cassini_69.png
http://www.mmedia.is/~bjj/povstuff/spacecraft/cassini_70.png

The area below the anteanna disc is much more realistic. However, the 
magnetometer boom (the big, golden boom) is almost invisible. Also the 
Huygens probe (the big, 'circular', golden object attached to Cassini's 
side) looks very dark. This can be compared to this photo showing 
Huygens being attached to Cassini:

http://www-pao.ksc.nasa.gov/kscpao/images/large/97pc1098.jpg

Overall the renderings using the modified metals.inc look better to me 
but I'm not completely happy. One additional problem is that the photos 
I have of the spacecraft are all taken inside a well lit room where the 
bright walls get reflected from the spacecraft.

I plan to make mainly sci-fi like renderings with several secondary 
lights.

-- 
Bjorn Jonsson / bjj### [at] ZZZmmediais
Address changed to avoid spam. Remove YYY and ZZZ to reply.
http://www.mmedia.is/~bjj

Post a reply to this message

Indeed, there are several differences between the photo and your renderings.
As you guessed, photos taken inside a room cannot be compared to the
darkness of open space. The most "realistic" look is usually achieved when
we see something we're familiar with, and that is (for most of us anyway)
being inside rooms, or outside where light bounces off many surrounding
objects. My own obervation is that even photos from space aren't completely
"realistic" looking.

However, photos are still a tad better than your rendering. There are a few
things you could do:

1)  Your rendering is hardly realistic because everything is "perfect". By
this, I'm NOT referring to dirt (though I see some dirt on the photo). Dirt
is not a top priority here. But the reflections are too perfect, and the
shadows have too sharp edges. By blurring those 2 aspects of your rendering,
the photographic quality will improve.

2)  The light-intensity seems off, even for a space rendering. I see at
least 2 lightsources: Above and below the craft. Both have almost the same
intensity. Both seems to cause highlights that are too dim, and diffuse that
is too bright. As you know, sunlight is much stronger than anything your
computer screen can produce. It's also much stronger than any photo can
hold. A camera that takes a photo of a landscape, for example, and faces the
sun, will either produce a completely white sky with the ground "intact", or
a completely black ground, with the blue sky "intact"... When some cameras
manages to do both things, it's because they have a very efficient light
responce curve... Now, this is what you need to 'copy' in order to get a
photo-realisitc rendering.

Talking about these things, and doing them, are 2 different things of
course. I haven't yet tried a project like yours (metal in space) so I don't
know how easy the solution will be. But a good 'start' that will get rid of
these very artificial materials, are to use only 1 light source for the sun,
and have this illuminate the entire craft. First of all: Directly where the
craft faces the sun. Secondly, where the craft reflects light upon itself
(indirect illumination) and third: A planet nearby - if that's your plan
anyway.. Adding another lightsource to fake the presense of planet, is not a
good idea. Using a big blue sphere is a good idea. Just be sure to use
radiosity for all this to work.

When you've done this (look at "rad_def.inc" for using radiosity) it's time
to blur the reflections, which is also very easy. Just add a normal modifier
to the surfaces that say:  normal { bumps .2 scale .0000001 }  and then
render the scene with this anti-alias:  +AM2 +A0.0 +R2 -J

I would be very interested in seeing the result after this. Then you can do
2 more things, actually: Add some blur caused by the lens of the virtual
camera - this will cause the areas that are very bright to "glow" which is
very realistic since the human eye does the same thing (an example of this
can be seen if you look at the ocean during a sunny day .. note the
beautiful flares)... Last - but this is a cheap trick - you could add a tad
of noise over the entire image.... or some dirty materials as they can be
seen on the Cassini photo.

Good luck! And thanks for reading this very long post...!!  ;o)  Hope it
wasn't too long.

Regards,
Hugo

Post a reply to this message