OTOY Forums

Hello world,

I have read in a few threads that displacements will not be coming any time soon because cuda doesn’t have support for this in unbais ray tracing.

Then how about a scan line rendering option so we could comp this in?
I don’t know if that would be possible but it would be very nice to have.

Also spot light support would be great.


Thanks,

the lights wont happen im almoste positive, but you will get material emmitters, which if you think abit about it is what happens in real life, there is no such thing as a cg lamp construct in the world around us.
What generates evey single light in the world is a surface or a group of surfaces that emmits the light.

The displacement is very difficult to treat efficiently on a gpu. I dont think they want to compromize and do a scanline hack or whatever it is you mention.
They have been talking about micro displacement though. You got to wait and follow the announcements etc.

-mads

Spotlight helps you focus or if you like directs attention to areas in and image (because this is and image) that in the “real world” you wouldn’t normally see. Done well it can add drama to a your “real” but monotonous image.

So if it acts like a "real world" stage spotlight then I am all for it.

As far as I thought micro polygon displacement is the one and only displacement mapping, which is what I was asking for but as a scan line option not a “Hacked” version.

Spotlights in the real world are spots because they have barndoors and so forth, provide your material emitter a proper environment as the one mentioned and you get a spot.
CG spots are fake they have an unrealistic aproach, they are as far away from the realworld as anything can get.
A material emitting is the closest you can get, it is the users job to build the lamp however, its not something that comes along.

To get a spot light in an unbiased environment, you actually have to model a simple spot light and put an element in it that you would give an emitting material and crank up the power.

This brings up some questions regarding the whole 'physically based/no basis in reality' argument.

On one hand things like single/double sided polys are dismissed, but other things like bump mapping are accepted. Bump mapping is a huge fake, where is it in reality? It is a cheat, as are spotlights.

I am not trying to wind up anyone here, these are my observations since I started looking at Octane and it's peers.

PhilBo wrote:To get a spot light in an unbiased environment, you actually have to model a simple spot light and put an element in it that you would give an emitting material and crank up the power.

ay to that, what i said.
Can it be any nicer? ( dont think so )

Gristle : well a valid argument is that lights are not bump maps

They cant be compared as it would be to compare a needle to the smell of icecream.

Cant solve bump mapping other than with meshdisplacements, and that is very expencive to produce.
A material casting rays is alot more realistic than your average near far falloff spot light.

Its all a compromis ofcourse, but some stuff gets closer alot easier than others. Just because one thing is hard to do doesnt mean you have to put everything under the same "light"

That said, we still have to remember what the developers want to achieve with octane. And that is not some kind of hybrid "if it works it works (brazil,mental ray etc among other hybrids "where it benefits the looks" ( doesnt mean they are bad)) aproach"

-mads

True, a bump map is not a light. I was not trying to say they were similar, apart from the fact both are "cheats"

I am just watching from a distance regarding quoted realistic solutions versus others. It is entertaining to some extent; where does one draw the line, seeing as it is all fake! As you said, it is about what the developer wants and where they are heading. They draw the line...

gristle wrote:True, a bump map is not a light. I was not trying to say they were similar, apart from the fact both are "cheats"

I am just watching from a distance regarding quoted realistic solutions versus others. It is entertaining to some extent; where does one draw the line, seeing as it is all fake! As you said, it is about what the developer wants and where they are heading. They draw the line...

I guess it all depends on the clientele...
products like maxwell, fryrender and indigo follow the physically corrrect philosphy.
they won't ever include any non mesh-emitter lights, no double/single sided surfaces, etc...

Octane is meant to be the same, eg a GPU based equivalent.
however I'm not %100 strict like them and a will add features based on demand, and if they are necessary, eg they can also be done with another approach within the existing features.

One thing is that, i'd really like to focus on getting the engine complete feature wise, MLT, area lights, portals etc, before i start spending time on those...

Radiance

Fair play, sounds like enough to do first.

I guess if the developers you mentioned take such a high brow path, it makes it really hard for them ego wise if for what ever reason they need to compromise.