View Full Version : micropoly displacement?

05-30-2006, 06:40 AM
Anyone know if there will be something like micropoly displacement in the 9.x development?

05-30-2006, 07:05 AM
dear Werner
actual APS is very near to micropoly displacement like a result, is fast and flexible to use. it's optimized to pixel resolution and you can have same result of micropoly displacement with very few second/minutes of render (i used an old p4 2,4 to test it...

05-30-2006, 08:25 AM
I find that in SOME ways APS is easier to use and more reliable than mr sub-pixel displacment. It's predictable which says a lot in my book.

05-31-2006, 12:47 AM
Some quick "space-cheeze" using CC, SSS, and displacement at 5 polygons per pixel.

The first one is radiosity, the second standard raytracing. The last is the shape without displacement or the nodal surface.

As you can see it works just like sub-pixel displacement for all intents and purposes. The RAM consumption was only 141MB.

05-31-2006, 01:22 AM
5 polygons per pixel.
Did you perhaps mean 5 pixels per polygon? :stumped:

Cool-looking renders, but I don't think APS can go quite that extreme, not in 141MB of RAM at least. :D

05-31-2006, 02:49 AM
Cheeeeeze Grommit :p

05-31-2006, 03:26 AM
Weepul: He might indeed have meant 5 polys per pixel - if you set your pixels per poly to being 0.2 that's what you'd get. Mind you, at a memory usage of only 141MB I doubt it too... :D


05-31-2006, 10:09 AM
actual APS is very near to micropoly displacement like a result, is fast and flexible to use.

OK, can anyone tell me what APS stands for? Keep seeing it but don't know what it is.

05-31-2006, 10:16 AM
Adaptive polygon subdivision. :) Or is it pixel? One of the two :)

05-31-2006, 10:50 AM
Adaptive polygon subdivision. :) Or is it pixel? One of the two :)
It's pixel. :D

05-31-2006, 03:57 PM
Doah- Your right- five pixels per polygon! But yes we could do five polygons per pixel. The lowest I've tried is .5 on some very simple small objects. It worked, but you'd be hard pressed to have a reason for using it instead of some AA.

05-31-2006, 05:05 PM

i would like to know if it's possible to keep the same amount of poly in a scene during a zoom animation ?

05-31-2006, 05:37 PM

i would like to know if it's possible to keep the same amount of poly in a scene during a zoom animation ?

Can I as why you'd want to do it with APS? I think you can if you drive things correctly but I haven't tested it.

05-31-2006, 06:23 PM
There are three settings now for render subdivision level. For any of the three settings, you can now apply a texture to determine the value of subdivision level.

1. per object - the same as what we had before. The entire mesh will subdivide evenly.

2. per poly - allows you to determine subdivision level on a per-poly basis. Distance to camera via a gradient is one example, but you could also set the subdivision level based on a weight map, or slope, or a procedural texture or image map. You can additionally set the subdivision to be a layered combination of any of these parameters. So to answer your question, yes, you could keep the same amount of polys during a zoom animation. You simply don't include a distance to camera setting in the texture value.

3. pixels per poly - More like true micro poly displacement, the subdivision is based on rendered polys. So if you have it set to 16, it would subdivide a poly until it is 4 pixels x 4 pixels (for 16 pixels total). This one is a bit weird, because the higher the number, the lower the subdivision. A setting of 1 would mean that any given poly is only 1 pixel in size, and a setting of .5 would mean any given poly is only .5 pixels in size.

I could be mistaken on this, but my understanding of true micro poly displacement is that the division happens during rendertime on a per pixel basis, so you can render polys at a subpixel level of subdivision without having to store billions of polys. The one downside of APS is that it subdivides the entire mesh at the beginning of rendering, so levels too high can bring your computer to a grinding halt.

At any rate, APS does give some really nice options for adaptive subdivision. In addition to distance to camera, For example, you can set the subdivision higher based on incidence angle, so the edge on view has more subdivision than the face view, where bump mapping looks good enough on it's own. You can also set the subdivision level lower for polys behind the camera, so you aren't creating a bunch of polys that won't be seen.

The one thing that does bum me out is that it only works on subdivision surfaces. The funny thing is, back at Siggraph, when they were just demoing it as a plugin, it worked on any mesh. When they integrated it, this ability was lost. According to someone I talked to about it, the plugin would convert the mesh to a CC surface anyways, but it was still neat that you could apply it to any mesh, not just a subdivision surface.


05-31-2006, 10:39 PM
APS is brand new to LightWave. I would be very surprised if we didn't see some pretty amazing enhancements to it in either the 9.x cycle or in LW 10. Things such as iterative subdivision or perhaps some cool way to handle the memory of the mesh without requiring loads of RAM. NewTek's got some really bright guys working on LightWave now. :D

05-31-2006, 11:09 PM
What is "iterative subdivision"?


06-01-2006, 12:49 AM
What is "iterative subdivision"?

- subdivide a polygon
- check if it has reached a (user supplied) limit
- repeat if it hasn't reached the limit

That's it...


06-01-2006, 09:30 AM
What is "iterative subdivision"?

Currently if you have a single large polygon and you want to use it for a landscape that is subdivided more in the foreground and less in the background you cannot accomplish this with APS - unless the mesh that makes up the landscape is already subdivided to an extent. A single polygon will always subdivide evenly with the current APS system. Iterative subdivision would allow the renderer to subdivide the polygon once, then check to see if your specified division settings are met and if not subdivide only the areas that still need more division.

And as Mike said, it would iterate as necessary.

06-01-2006, 11:08 AM
Well even in the beta the amount of memeory use by APS has greatly improved. I hope and have faith it will get even better.