What a juggling act. You can’t take a nicely lit scene and simply add fog to some of the lights - you have to re-set all the light intensities in the entire scene all over again, and by different amounts depending on what type and what brightness they are. Then if you’re disatisfied with the fog density and change it ( or a supe tells you to change it), you have to adjust every single light intensity all over again.
I appreciate the realism, but that’s not flexible enough. It’s like forcing us to use both diffuse and reflection for every light all the time, for the sake of realism.
For the record, mental ray is no better, with or without volumetric lighting. It’s about as predictable as an asshole on your elbow. Why? Because the fundamental light equations are wrong, as I’ve been saying for years. Maxwell was close, but even he couldn’t separate light in his equations.
the light / rendering / precalc part of all of this can be like * juggling a bunch of saws while standing on a net of rope. one adjustment and the whole things fails but it is improving. After spending years grinding towards a perfect beauty right in the viewport without any post- its safe to say post almost always wins. There should however be a way to render these types of physical phenomenon with current available computer strengths. Arnold is great although can get a little hungry at times with my machine resources.
What about precalc GI ? the moment you move your lights its the same story anyway- so in those cases the workflow is very linear. I don’t know of any people go in and merge their IRmap/LCmap passes together manually, they just say screw it and recalculate the thing.
Assuming you’re a 2D effects artist, sure. Have you ever done an animated volumetric light, that interacts with other 3d objects, in post? I know a little bit of comping, but I have no idea how that’s done - I’m sure they’d need to bring 3d geo into nuke. All that extra knowledge and time required for something that a supe would ask for just on a whim -
What about precalc GI ? the moment you move your lights its the same story anyway- so in those cases the workflow is very linear. I don’t know of any people go in and merge their IRmap/LCmap passes together manually, they just say screw it and recalculate the thing
Baking is a whole different subject - if you bake something, then change it, of course you have to bake again! Having to re-light your scene just to add one lighting effect though, that can be a no-go, and I’m not going to use it at all now.
Seems to me all that’s needed is voxels contained within the light’s frustrum - but, no, I’m not going to claim to be even half as smart as vlado or other render programmers - and who knows, maybe it’s just not a high priority.
Yes, I actually have- and yes it became a nightmare after a while. With opacity mapped tree leaves no less and the camera was crossing line of sight for the sun source. It was actually part of the reason for my posting at the time. Would I risk that with what I didn’t know about rendering this effect in a production environment- probably not likely. These days though if I had the time for a personal project I’d say exploring deep could be a possibility.
Also correct. I was trying to be relative- but I did leave of the focus of the post. On another note if the light effect is purely additive like “gods rays”, it is possible to do that with post and not having to relight the scene.
Maybe point cloud geo can do what we can’t figure out in this way? Hey I’m just glad the post is coming alive and well- rendering always had its lions share of problems.
