PDA

View Full Version : Confusing air poly behavior?



Johnny
05-18-2004, 11:35 AM
I'm trying to get my glass bottles and vases to render more accurately by creating air polys for them. my method is to copy the vase or bottle, to paste it in place, flip it, name it "air", give it the same surface as the glass, only with refraction of 1.0.

as you can see in this rest render, I'm getting what looks like black inside the glass. I haven't modeled any liquid in either of these containers. in this render, raytraced refraction is turned on. before the air polys were added, each container passed and refracted the background you see, and appeared like solid blobs of glass, as they lacked "air" inside. So, that's what I'd expect before adding the air polys.

Is there something else you need to do to the air poly geometry or surface to make it behave like clear air?

Thanks

J


http://home.earthlink.net/~zoomin/blackglas.jpg

cagey5
05-18-2004, 12:25 PM
I recently posted a thread after I first attempted glass using air/glass surfaces.
This is the link
http://vbulletin.newtek.com/showthread.php?s=&threadid=22800
Don't worry if you don't have fPrime it's not needed, it just helped me get the set-up quicker. Within the thread I posted the scene and objetcs. Have a look at them and see if they help at all.

cagey5
05-18-2004, 12:35 PM
Additional.
I also merge points after I've pasted and flipped the air surface and in addition to making refraction 1.0 I make transparency 100%

Edit btw what have you set the refraction of the glass surface to. It looks pretty extreme from your pic, could that be part of the problem. I use about 1.5.

toby
05-18-2004, 04:20 PM
also make sure your ray recursion is high enough, 3 or 4 can give you darkness. 8 is in the ballpark.

Johnny
05-18-2004, 04:23 PM
Originally posted by toby
also make sure your ray recursion is high enough, 3 or 4 can give you darkness. 8 is in the ballpark.

that was the lucky charm! in fact, as soon as I entered 8 and hit return, the positive change showed up in the Fprime window.

big help..thank you and everyone else who added suggestions!

J

cagey5
05-18-2004, 04:32 PM
That just shows I don't optimise my scenes enough! I don't think I've ever dropped it below the default 16 so it's not a problem I've seen before. Nice one.

Johnny
05-18-2004, 04:35 PM
Originally posted by cagey5
That just shows I don't optimise my scenes enough! I don't think I've ever dropped it below the default 16 so it's not a problem I've seen before. Nice one.

I have been sticking around 4 just to keep render times lower, but my experience level is growing...

cagey; I noticed that in your transparency node, you used a couple of grads to handle thickness and I believe incidence angle (not at that LW computer)

How did you arrive at the values to those? I can't quite visualize it.

tx

J

toby
05-18-2004, 04:38 PM
cagey! you are going to save hours, uh, DAYS rendering reflections!

cagey5
05-18-2004, 04:43 PM
I read through a glass tute in 3D World I think and the next day just booted up Lightwave and fPrime and had a go at what I could recall from memory so the settings were what looked right as I tweaked. If you haven't got fPrime I can't recommend it highly enough for tweaking textures especially in Radiosity.

I'll see if I can track down a link to that tute.

Edit: This is it http://www.3dworldmag.com/stoppress/realisticglass_tutorial.pdf

btw. The surface thickness settings I have always found difficult to set up, but a bit of experimentation should get you where you need to be. It all depends on the scale of the object you are working with.

Johnny
05-18-2004, 04:49 PM
thanks for that link..that tute looks pretty hardcore.

J

cagey5
05-18-2004, 04:57 PM
I can't say I worked through it. But it gives you all the basic info by reading through and of course it's Lightwave so it's ideal as a reference.

Playing around with the luminous rectangles is useful too. Post another pic now you've sorted your problem btw. Let's see what it should look like.

Johnny
05-18-2004, 05:05 PM
This is a render of one bottle from the "real" scene I'm developing; the other image was just a quick and dirty model while I was tangling with the glass issues.

this bottle is supposed to have a hand-blown look, which accounts for the green irregular areas around the bottom perimeter of the bottle.

previously, this same model with the same surface was giving me that dark/black interior nonsense before I cranked the recursions up to 8 or more.

Johnny

http://home.earthlink.net/~zoomin/good.jpg

Johnny
05-19-2004, 03:00 PM
can anyone say from experience whether more recursions is better when it comes to glass?

If 8 is the minumum to avoid that darkness problem, as mentioned above, is 16 better? 32? Even higher?

Let's say you have time to burn and can let a render go all night long, will you see even better results with more recursions?

J

Wade
05-19-2004, 03:09 PM
And what kind of time hit do you take when rendering from going from 2 to 8 to 16.... ? If any knows off hand.

cagey5
05-19-2004, 03:19 PM
I don't thinkk it works like that Johnny, though I'm no expert by any means. As far as I know you just need as many recursions to get the ray to where it's going. In your case going through a number of surfaces with refraction and reflection will require more than if it was a solid matt surface.
If you had lots of self reflecting objects in the scene for instance with a low recursion limit. You may only see one object reflected in another. Whereas with a higher count you would see the reflections in the reflected object. Does that make sense? So in that case upping the number is only of benefit in reflection transparency heavy scenes. If you are looking for continuous improvement. Look at fPrime.

Johnny
05-19-2004, 05:38 PM
Here's an early render of the scene I'm developing..the scene that prompted my original question.

compared to the renders I was getting without air polys, this is a major improvement toward realism and the kind of lightplay I'm going for in my glass.

thanks again for the pointers!

Johnny


http://home.earthlink.net/~zoomin/nuglas.jpg

toby
05-19-2004, 10:57 PM
still seeing some blackness - more rays!

More rays than what you need will give you a more accurate render, but that doesn't mean 'better', just different. Do some tests. When rendering animations, it's usually not worth the extra render time. It won't go higher than 24 btw

Johnny
05-20-2004, 08:46 AM
Originally posted by toby
still seeing some blackness - more rays!

More rays than what you need will give you a more accurate render, but that doesn't mean 'better', just different. Do some tests. When rendering animations, it's usually not worth the extra render time. It won't go higher than 24 btw


I'll give that a try, but I *think* that some of the blackness in this still might be due to scene lighting, which is low. Other renders with brighter lights show the glass to be fully "glassy" and transmissive.

Also, I'm not using RT transparency, bcs render times were getting ridiculous with it on.

When to use RT transparency and when not to?

J

Johnny
05-20-2004, 10:13 AM
A larger number of recursions DOES make a difference! I went from 8 to 16 and those darker areas of the glass (at about 85 degrees from dead-on) are now clear, where they used to be very dark, even in low light levels.

thanks again!

J

MiniFireDragon
05-20-2004, 10:47 AM
About your hand blown glass, are u using this in an old time setting?? It will have to be really really old. Most hand blown glasses and bottles look as smooth and nice as a machine drawn bottle. The irregularites occur in the glass itself. Cheap glass has a green tingle, where the more expensive glass is cleaner (that is in regards to pyrex, aka hard glass).

Softglass is the same way, just takes less heat to alter.

This comes out of experience as my father is a hand sculptured glass artist and I been around glass all my life.

www.sculpturedglass.com

Johnny
05-20-2004, 11:09 AM
Originally posted by MiniFireDragon
About your hand blown glass, are u using this in an old time setting??

Interesting points...as my renders improve, I'm seeing some modeling aspects I'd like to fix.

Time isn't an element in this scene; it's all about the glass, light, form.

Would you say that some glass artists strive to achieve pieces with varying thicknesses, waves and bubbles?

I'm partly thinking of this Mexican glass you can get at various stores...bubbles throughout, waves, imperfect, yet perfect at the same time.

J

MiniFireDragon
05-20-2004, 12:55 PM
Cheap glass can/will contain air bubbles. As for the wavyness, I suppose it can be art or from an inconsistant heat source, or once again, cheap glass. The bottles are made with glass tubing, the end is drawn down small enough so they can put their mouth over it and blow into the glass, causing it to balloon at the hot spots.

If you ever have the chance, goto the Pennsylvania Renisance Fair outside of Lancaster(maybe the other fairs have them too) and watch the glass blower just up the street from The Crystal Palace (opposite direction of Cloth's By Rebecca).

http://www.sculpturedglass.com/gallery/ms-ppig.html

the above link is a sample of hollow hard glass.

Johnny
05-21-2004, 10:59 AM
I'm puzzled by RT transparency and RT reflection in this scene I'm working on.

with RT refraction and transparency, my scene renders in about 14 minutes, with caustics, 24 recursions, enhanced high, and 640x480

if I add RT reflection to the settings, the same computer cooks away on the first pass for about 1/2 hour!

do you have to choose between great glass and great reflections? I can guess that with recursions at 24, in a scene dominated by reflective surfaces, it's going to take a lot of juice to do all the calculations.

how do people deal with this aspect of glass & reflection?

j

toby
05-21-2004, 12:44 PM
RT reflection multiplies the render time, not just add to it, especially with recursions at 24 - think of it this way: 2 bottles next to each other will have 24 bottle reflections rendered onto them, a reflection of a bottle inside:

a reflection of a bottle in a reflection of a bottle in a reflection of a bottle in a reflection of a bottle in a reflection of a bottle in a reflection of a bottle in a reflection of a bottle in a reflection of a bottle in a reflection of a bottle in a reflection of a bottle in a reflection of a bottle in a reflection of a bottle in a reflection of a bottle in a reflection of a bottle in a reflection of a bottle in a reflection of a bottle in a reflection of a bottle in a reflection of a bottle in a reflection of a bottle in a reflection of a bottle in a reflection of a bottle in a reflection of a bottle in a reflection of a bottle.

And it has to do caustics and refraction at the same time. You probably don't need RT transparency for this image. It's for things that don't show up behind reg. transparency, like HV.

If you have to have high recursions for the refraction, render the reflections in a separate scene with lower recursions -

JulianW
05-22-2004, 11:10 AM
If you're finding the black annoying go to the surface editor and you'll see that you can select different settings for Reflection/Refraction Options.

The standard Ray Tracing + Backdrop should be fine in many cases *provided* you set the backdrop colour to a dominant colour in your scene.

The way it works is that when a ray exceeds the number of recursions LW cheats and takes the current Backdrop colour as the colour for that ray. Since the default Backdrop colour is black you get obvious black patches in the renders.

For more complicated scenes Ray Tracing + Spherical Map is worth looking into.


JW.