PDA

View Full Version : Lightwave Light bugs



ingo
07-26-2004, 08:02 AM
Hi all, maybe somebody can help me out here, i found two nasty bugs in Lightwave regarding lighting. In the first picture you can see my scene. It has a spot in every window to give an ambient light and a sunlight.

If you look at the blue wall you can see some annoying vertical stripes, anybody has an idea where they come from ?

The second bug is more annoying. In the second picture you can see the circles of the intensity falloff, as you can see they are only 2 or 3 m, and the office is 15 m wide. But in the rendering you can see that the inside walls close to the windows get light from the lights of the other side, although light intensity has to be zero at this point. That makes the intensity falloff nearly useless. Maybe anyone with LW8 can tell if this bug is still there ??

TIA

ingo
07-26-2004, 08:02 AM
Okay, here is the rendering :

caesar
07-26-2004, 08:31 AM
Bug 1 = ?
Bug 2 = ambient light is set to 0%?

The floor is carpet texture? Looks great

Mylenium
07-26-2004, 09:19 AM
I'm not quite sure what you are getting at. These are quite obviously no bugs, but normal behavior.

a) If you use anything other than linear falloff, your light will never fade away entirely. The range in this case only acts as kind of a threshold where your intensity reaches a certain amount.

b) Your wall seems to have smoothing errors. Either there are some points that aren't merged, overlapping polygons in the same place or your smoothing angle is simply to high.

BTW, you should re-think your light placement. While the overall look of the room is quite satisfactory, the incorrect specular shading on the window panes gives away everything.

Mylenium

ingo
07-26-2004, 10:30 AM
Thanks for your comments. I've tried now another test and the first bug has gone away.
Instead of shadow maps i used raytrace shadows, that makes the vertical stripes, although the shadow maps are at the minimum of 2000, thats weird. Oh, and the modell is okay, so no merging points or smoothing problem.
Youre right with using inverse distance or inverse distance/2, but it has to be near zero after 15 meters, depending on the rounding. But at the left wall i still have nearly 50 % of the light intensity, so the calculation is obviously wrong. And ambient is at 0 %.

ingo
07-26-2004, 10:48 AM
BTW, what do you mean with the incorrect specular shading ??

Mylenium
07-26-2004, 11:31 PM
Ah, you used Shadow maps... that explains it then - your light cone angle is too great and the wall only showed stepped artifacts of your map. 2000 pixels isn't really that big, especially at your large angle. This could make one pixel in the map as large as 30 by 30 cm even in such a small room.

Have you checked your ceiling lights? They have no fallof and may illuminate your left wall.

Your window panes are showing bright spots where there should be none. Especially on the right side, where it should only be dark, there are some highlights on the side that does not receive any sun light. Anyone with even the slightest insight into CG would suspect a spotlight there. The same goes for the left side. It just feels unrealistic .

Mylenium

ingo
07-27-2004, 02:53 AM
Thanks again for your help, seems logical now even for me with the shadowmap size. Otherwise the ceiling lights are off, so the bright light on the left walls comes only from the spots in the windows on the right, with a falloff of 2 meters, and no specular on the walls, just on the windows. But i try now to get a good lighting with the inverse distance^2 setting since the falloff is much faster.

If youre talking about the spots to the left and the right of the windows, they are of course realistic, the only problem is they still look to much like wideangle spotlights. Most of the light coming inside a room is indirect light, coming from every angle. As one of my client said to me last time : "Just move your eyes away from your monitor, there are amazing things going on outside".

ingo
07-27-2004, 04:34 AM
Here's a little update. While watching the rendering going on i found out that most of the light on the right side comes from the sun shining through the top. So i decided to get back to FormZ and subdivide the outer shell at every 3,40 m (window axes) and now it gets darker inside.

Otherwise i realized that LW has problems with polygones with more than four sides, that explains FPrime's problems with them too. Reminds me a bit of good ol' Lightscape....

ingo
07-27-2004, 06:38 AM
Okay, i found the big problem, but please tell me, is this logical and does it makes sense ?????

The main reason was that i used a shadow color other than 100 % black for the sun. When i switch to black everything works as expected, when i use a different color it always looks like the sun is going through the building. But here comes the strange thing, it affects only polygones that can be seen by the sun, the opposite ones are totally black, which is a bit confusing....and i think its wrong.

toby
07-31-2004, 05:36 PM
Lights that cast colored or gray shadows will not cast any color outside of it's influence. Ray-traced and Shadow-mapped shadows are not 'blocked' light as in real life, they add darkness (or color) to the pixels where they're supposed to - otherwise they couldn't add color, could they?

In other words, a lights that cast a color through solid objects is totally unrealistic, so don't expect them to behave in a realistic manner.

ingo
08-02-2004, 02:11 AM
I have added a picture to explain the weird behaviour. The scene has the basic distant light (although the icon shows the wrong spotlight icon) with standard settings, and the windowless room has the standard materials and the standard render flags.

The left rendering shows the correct rendering with 100 % black shadows, the left rendering shows the wrong rendering with the shadows colored red. The weirdness starts even with a dark grey (99% black) shadow.

Lightwolf
08-02-2004, 02:39 AM
ingo:
This is normal behaviour.
The polys facing away from the light are actually not even submitted for a light/shadow test with the light, so they can't react to the shadow colour.
The polys facing the light are actually submitted to this test, are found to be in the shadow, and are thus shadowed accordingly.
Coloured shadows are extremely restrictive, and I wouldn't call this a bug, but unexpected behaviour (...even though it is logical).
I think what you're really after is more like the effect of ambient light.

Cheers,
Mike - and greetings to Hamburg from sunny Stuttgart :D

ingo
08-02-2004, 03:57 AM
Hmmm, at school they told me without light no shadows; but that was looong ago. So in a dark room without windows i expect total darkness, like in the first picture, even with white shadows. Maybe i should ask Schroedingers cat ? ;)

greetings from a much more sunny Hamburg to Stuttgart

Lightwolf
08-02-2004, 04:02 AM
Hi ingo,
But a shadow is defined as an area without light, after all, if it was lit, it wouldn't be in the shadow now, would it?

Anyhow, how would you know if the box is actually black inside without looking in? Schroedingers cat indeed ;)

Cheers,
Mike - above 30 and sunny as hell, I love it :D

ingo
08-02-2004, 04:51 AM
I somewhat feel like Snoopy in this comicstrip.......:D

http://www.snoopy.com/comics/peanuts/archive/peanuts-20040720.html

caesar
08-02-2004, 06:30 AM
Originally posted by ingo
I somewhat feel like Snoopy in this comicstrip.......:D

http://www.snoopy.com/comics/peanuts/archive/peanuts-20040720.html

Peanuts is amazing goog!