Maxwell Render

Here's my first try with simulating volume lights. I let this run until SL 20 (to tell the truth, I forgot it was running... ). I forgot to take note on the final render time and benchmark, but I know initially the BM is around 9 (pretty darn low) but this could be a result of all the teapots in the scene as well......

As you can see, my first attempt, despite the high SL, still has quite a bit of noise. However, I'm not sure if this was a result of the renderer, or corresponding/overlapping pixels from my material. I used the multi-plane technique on this one, similar to my rain effect, but with a very fine transparency/layer mask bitmap. What I neglected to do with this test was to offset the individual planes in order to introduce a bit of randomness to the layers of material.

The good news is that this noise was very 'clean' and I was able to use the PS, dust and scratches plugin to remove them fairly easily. Unfortunately, using this feature does kill the sharpness of the image........

My next step is to first try a render with some randomization in the positioning of the numerous planes. I'm then going to play around with some different transparency maps to see if I can speed things up a bit and get a clearer image.

Stay tuned!

Raw image:



Postwork image:

_________________
"Skill without imagination is craftsmanship and gives us many useful objects such as wickerwork picnic baskets. Imagination without skill gives us modern art." - Tom Stoppard

My modern art gallery at: http://nws.carbonmade.com/

well, the dust & scratches filter sucks bit time...

try with the reduce noise filter, probably with 2 or 3 passes, and you'll definitely get a much better result

_________________
www.andrearettori.com

looks spooky! would look great with skulls/bones as opposed to teapots

_________________
=LES=

Efficient exploration of Maxwell's boundaries, as always !

Do you plan to develop micropoly displacement next?

I prefer the raw image... it looks like dust in the air.

Great tests!

_________________
http://www.ftella.com

ya, nice
so basically your using planes. .not cubes textured with a trans mat.. (1sided planes?) in one of your previous fog tsts you were using a cube/box in front of the camera.

I like these results.. as mentioned above i too like the 1st image withthe dust look to it.

could you post a quick wire of your scene?

Thanks guys! Here's another quick test - using three colored lights - with the planes randomly offset. Rendered for 20 hours to around SL 17.4. I also used half the number of planes, and made the fog material denser by decreasing the UV size.

Still a bit too noisy for my liking.... (got your skull and bones LesliePere! )



Here's a wireframe of the scene. Very simple - just a series of planes facing the camera.

_________________
"Skill without imagination is craftsmanship and gives us many useful objects such as wickerwork picnic baskets. Imagination without skill gives us modern art." - Tom Stoppard

My modern art gallery at: http://nws.carbonmade.com/

what do you gain by using planes instead of the box you used for the fog tests (which btw are fantastic and so are these)??

Have you thought of layering variations of fog material? Or are the planes less memory intensive?

One of the big problems I have encountered with my original fog material is that I simply cannot get it to work properly for volume lighting. That method basically uses a very 'thin' SSS material applied to a simple box encompassing the scene. This works fine for scenes lit from all directions by physical sky, sun, or sky dome. However, I have been unable to get it to react properly to regular individual emitters. I think that this is related to the big problems we have all been having with the current SSS feature: it simply does not work correctly. A focused 'beam' of light couldn't penetrate the SSS material correctly in order to illuminate any standard objects in it's path. No matter what settings you use, the current implementation of SSS seems to over-scatter light passing through it. Hopefully, SSS will be one of the things fixed with 1.5 and could be used for this sort of effect.

So, I had to go back and think about what fog really is - which is very fine dust and/or water particles suspended in the air. Since SSS wasn't working properly, that left me with only two options: I could either use a particle system and create a gazillion particles (truly emulating fog) which would be hugely impractical, or I could fake it by using planes with a particle transparency map.

This is the main reason I've gone to a planar system for the volume lights. I played with several different versions of maps, but finally settled with this one:



Each dot represents a single dust/water particle and each is only a pixel wide. My original thought was to use this simply as a clip map. However, using the dual BSDF ghost layer technique gives me more flexibility in adjusting the transparency and roughness of the particles. By playing with the size of the UV map, transparency and roughness, I can increase or decrease the density of the 'fog' - depending on what I need for the scene).

Now, the downside of using this method is that you cannot achieve true 3D shadows. With the sample below, it works fine, since the objects casting the shadows are intersecting the fog planes. However, this would not work if they were rotated 90 degrees (parallel to the fog planes). For most cases, this isn't really an issue as you would normally not see these sorts of shadows within the fog itself (since they would be obscured by the portions of the fog in front of them)



My next test will be to see if I can get this to work in a room environment....

_________________
"Skill without imagination is craftsmanship and gives us many useful objects such as wickerwork picnic baskets. Imagination without skill gives us modern art." - Tom Stoppard

My modern art gallery at: http://nws.carbonmade.com/

WOW... this is the best fake I've seen so far!



Doing a tridimensional grid, so having planes on xy, xz and yz?

Very good,
Ricky

thanks for the explanation

Nice.

_________________
Product Design
Linkedin

wow

Why do you doubt it could not work with an interior scene? rendertime?
Great research et results!

Edit: i'd be curious to see what we would see if you'd place a big mirror facing the camera, in the last plan behind the spots

I wonder if anymore works has been done with this? Above michaelplogue mentions that SSS materials were part of the problem. Do the new SSS in1.7 help?

Is there an updated technique for Volume Light?

Thanls!

T

I was also hoping for a good Way to do an underwater scene without faking it in postprod

_________________
Home Setup /// Mac Pro 2 x 2.8 GHz Quad-Core Xeon 6GB OS 10.5.6. // Macbook Pro 15" 2.4 Ghz 6GB OS 10.4 ///
Work Setup /// Mac Pro 2 x 2.4 GHz Dual-Core Xeon 16GB OS 10.6.4 ///