Interior Render (Spiral Staircase)

[Nicolas Jordan]

This is my first interior render. After much tweeking and experimenting I am calling it finished. Any feedback is welcome.

[Karmacop]

I think it looks great :)

[Dave Jerrard]

The only thing that jumped out at me was the glass, which doesn't seem to be refracting or have a fresnel effect on it. Other than that, beautiful render.


He Who Hasn't Done Any Architectural Stuff In A Long Time Now And Misses It.

[Matt]

Like the design, one thing though, is that chair on the left the correct size? Looks huge!

[Nicolas Jordan]

All the glass has a fresnel effect applied. I think the chair on the left just appears large because of my camera settings and a wider lense. I will be making more renders at different angles.

I am going to see if I can get the glass to look better.

[Nicolas Jordan]

Here is a newer render. I set the refraction up a bit for the glass. Does that look convincing?

[ANS]

Glass looks better.. maybe more reflection? Matt has a good eye- chairs looks too big. especially this on left..

Cheers.

[Nicolas Jordan]

I guess I shouldn't have been so quick to put this in the finished gallery. I will know for next time to leave in work in progress gallery a bit longer for final critics to polish it more. Here is a newer render with the chairs resized a bit.

[Karmacop]

The smaller chairs and the refraction looks better :)

[ShawnStovall]

Very Nice!

[Dave Jerrard]

Another glass question. Is it using an inner air surface? For thin glass (I'm guessing it's supposed to be about 1/4 inch thick), the refraction of the windows behind it seems a bit high. This is just technical nit-picking mind you. It's a beautiful image.

He Who Knows How Easy It Is To Lose Track Of Details After So Much Time.

[Nicolas Jordan]

Another glass question. Is it using an inner air surface? For thin glass (I'm guessing it's supposed to be about 1/4 inch thick), the refraction of the windows behind it seems a bit high. This is just technical nit-picking mind you. It's a beautiful image.

He Who Knows How Easy It Is To Lose Track Of Details After So Much Time.

I dont currently have air polys implemented, but I think I will do that and see how it looks. It will probably turn out much better!

[righteous]

mmmmmmmmmmmm, reflection bluring. How long are render times?

[Nicolas Jordan]

mmmmmmmmmmmm, reflection bluring. How long are render times?

I am using F-Prime to render. I just let it render long enough to get most of the noise out. 3-4 hours or so for 800x600.

[Dave Jerrard]

That's what's causing the exaggerated effect then. When teh rays from the camera hit the glass, they're being refracted, and they continue along their new path until they hit something else, 'thinking' they're in whatever material that caused the refraction for the entire distance. They're not away that they only passed through a thin layer of glass because they never saw the other side. Nothing told them they were leaving glass and entering air again. This is where an air surface is needed, which tell the rays that they're leaving glass and entering air, which has a refraction index of 1. Then the ray will come from the camera, hit glass and be refracted, then hit air, and 'unrefract.' Without this air surface, glass (or any refractive transparent surface) will tend to look much thicker than it really is.

The easiest way to make an air surface is to follow this procedure in Modeler:
Select the glass polygons
Copy them (don't deselect anything)
Flip them (again, don't deselect anything)
Change surface to air.
Paste the original polygons you copied
Merge points (optional).

When you copy the glass polygons, you're storing a copy of these in memory, which are unaffected by the rest of the operations. All the changes done later are done only to the polygons you still have selected in Modeler.
These changed flip the polygon so they face inward, and change the surface name for them. Air polygons need to face inward so a ray cay see when it's exiting the object. They can't normally see a surface that's facing away from the ray.
When you paste, you'r epasting the original glass polygons back into the object, completely unaffected by those changes. You now have the original glass and a set of inward facing air polygons that share the same shape.

Now, just make the air surface 100% transparent, 0% diffuse, and give it a refraction index of 1.0.

Making a surface like this double sided will not work. Refraction only occurs between materials that have different refraction indexes. Double-siding a surface will result in a ray entering glass, being refracted, and then entering glass again. Since it's already in glass, there's no change in refraction. This isn't a rendering error, it's actually how light behaves.

Here's a few examples. The first teapot is only comprised of a glass surface. There's no air surface used. From this render, the teapot looks more like a solid glass sculpture of a teapot rather than being an actual working one. It doesn't appear to have anywhere inside to hold tea. There appears to be a specular highlight coming off the back side, but the real source will be evident later.

The second render is the same teapot with the lid removed and placed behind it. With the lid removed, you can now see there is actually an interior to the model, but you couldn't tell from the way it was refracting. Now, you can see into the pot, and through the back side. The front of the pot still has the same refraction of the checkerboard, and still looks solid. Similarly, the back side looks like it's pretty thick too. The IOR of this glass is only 1.5, so it's not that dense, but the refractions are a bit exaggerated because they rays are acting like the glass fills the entire region from the glass surface to the ground plane behind it. That specular highlight in the first image was from the interior.

Remember what I mentioned about refraction only happening between materials of different IOR? In this second image, you can see the lid through the back of the teapot (and the handle), but it doesn't look solid. The checkerboard floor is visible through the lid, but is not disorted by it - it's only darkened slightly by the glass not being 100% transparent (no physical surface is - the most transparent materials known are only about 98%). It's the same refraction as the rest of the teapot, and since the rays have not been told they've left the glass, they're entering a material that's the same as the on they're leaving, so there's no refraction occuring.

The third image is the same as the first, but with the addition of an inner air surface. Nothing else is changed. The difference is immediately apparent. Even with the lid on, you can see the teapot has an interior. The spout is not a solid glass extrusion, but an actual functional spout. The handle on the lid is not solid, but is actually hollow. The true thickness of the object is apparent. You can also see that specular highlight on the inside surface of the pot clearly too.

The last image is the same as the second, but again, the only difference is the air surface. The lid seen through the pot is now refracting the checkerboard in addition to what the pot is doing.

The importance of air surfaces becomes even more apparent when you start using thickness attributes on a surface, like thickness gradients or subsurface scattering since the model can suddenly start to appear very dark, or opaque, as I show in this thread (http://www.newtek.com/forums/showthread.php?p=416254#post416254), and in Lightwiki (http://tinyurl.com/ndn8z).


He Who Has Been Getting A Lot Of Use Out Of This Teapot.

[Nicolas Jordan]

Here is my the render after implementing air polys. Thanks alot for the detailed info of refraction and air polys Dave! It helped me out alot! I have been using lightwave for a while and havent dabbled with refraction and galss much until now. Here are some screen grabs of my setting for the surfaces and fresnel. Thanks again! :)

[Dave Jerrard]

That got it!

I completely forgot all about Fast Fresnel. Ever since 6.0, I've been using gradients for all my incidence angle stuff. They can be affected by textures too. :)


He Who has Been Messing Around With The Thickness Node lately.

[Silgrin4D]

I`m jealious that you`ve done your 1st architectural render so well. I`m finishing my 2nd one and I`m still far behind:devil: Maybe `cause I don`t use LW yet:)

The only crit from me is that there`s hardly any environment behind the win. There`s just a blue anything here. But the modeling, texturing and especially lighting is great as for me.

[Matt]

Last render is looking much better! Well done!

[Nicolas Jordan]

Some guys over at Spinquad wanted to see my light setup and a close up render of the staircase so I thought I would post it here to. I am thinking of still making a few minor changes to the scene regarding color on the stair carpet and walls.

I will also be puting a background scene is for the window. Im heading off for downtown right away to take some photos.:)

[Pavlov]

Very good work, expecially if it's your first interior shot.
My suggest: raise the overall light, raising the arealight out of the doorand Fprime intensity % (for interiors you can go easily over 400). Try to control overall color with a low intensity, strongly colored ambient light.
Try also to give outside light a blue/turquoise color and all other spots a clear yellow/orange color. You'll get a less constant lighting and more interesting hues, which should appear in this situation.

bye
Paolo

[lightscape]

Is it my correct understanding that airpolys are in the same exact location of the glass poly except that they are flipped and have different surface setting? Wont this have rendering error?

[Dave Jerrard]

Is it my correct understanding that airpolys are in the same exact location of the glass poly except that they are flipped and have different surface setting? Wont this have rendering error?
Yes, and possibly. You'll run into the posibility of render errors if the polygons are non-planar. Generally, air polygons are completely invisible though, so even if the polygons are non-planar (to a certain extent), they shouldn't cause any problems on the camera-facing sides of the object. Ray traced non-planars, on the other hand, are less forgiving. Whether they're double sided or not, ray traced non-planars can frequently look like they have gaps between them, resulting in ugly lines around the polygons and cracks in the shadows.

He Who Has Seen Some Pretty Bad Double-Sided Errors In LightWave 3.0.