PROJECT PEACH DEV: Approximate AO, faster, banishes noise, NOW IN SVN

[Ace Dragon]

http://peach.blender.org/index.php/a...ent-occlusion/

A new AO method that's faster, and even less noise then 10 samples with Broken's QMC AO.

The indirect lighting test looks pretty neat and I hope that the prototype code Brecht used for the picture gets into 2.50 despite it not being used for Peach.

We could call it AAO due to approximate being in the name of the method.

[Mmph!]

Cool man, thanks for pointing this out to us.

[IamInnocent]

The last sentence of Brecht post:
"Currently this new ambient occlusion is not in SVN yet, so you will have to wait a bit to test it, but it will be there soon."

Jean

[blenditall]

2.5 is going to be so awesome!

The traditional methods will be kept though, I hope. I kind of like the noise sometimes

[BlendRoid]

Yeah, the noise can make renders look really cool. Not to say that I'm not amazed and excited about this awesome new method (I read the article ealier too)!

Ooh, I didn't notice the indirect lighting before. That would be really cool to have quick rendering radiosity to some degree! Blender just keeps getting better and better.

[gat]

Thats good stuff, I am really impressed by the technicality of this.

[Kai Kostack]

hats off, another important achievement for blender and its community. i already thought a fast GI algorithm couldn't made it into blender during the peach project, but now it's almost there. indirect lighting and easy animateable AO, this sounds to good to be true.

now i hope he holds it not back for too long as i'm having a few animated scenes with lots of AO in the pipeline ready to get rendered.

[Sago]

I've NEVER cared much about blender development, new features and so on (I rather depend on my own skills and ideas than on the tool), but this is really making me cream my pants. Amazing stuff.

It's so fast and good, it might mean a new era in the blender animation. So instead of some crappy animations every now and then, there will be TONS of crappy animations ALL THE TIME.

Originally Posted by blenditall The traditional methods will be kept though, I hope. I kind of like the noise sometimes

Use nodes.
You don't wanna go back once you tried the new AO with that speed.

[ropsta]

Originally Posted by Sago It's so fast and good, it might mean a new era in the blender animation. So instead of some crappy animations every now and then, there will be TONS of crappy animations ALL THE TIME.

AHAHHA

That's pure unfiltered GENSUI

[Felix_Kütt]

Originally Posted by ropsta AHAHHA That's pure unfiltered GENSUI

yeah, not genius, but even better!

[Ace Dragon]

Originally Posted by Sago It's so fast and good, it might mean a new era in the blender animation. So instead of some crappy animations every now and then, there will be TONS of crappy animations ALL THE TIME.

I can see that, crappy animations using AO, but who am I to speak for animators, I have never made and rendered an animation in Blender

[3pointEdit]

Now ALL those array tests can be remade, everyone can be an expert with just a pinch of A AO. Cool for dummys (like me).

[Coalth]

I hate to ask such a stupid question, but is there a (windows) build availiable somewhere? I needs to play with feature and make crappy animations.

[BeBraw]

The feature is not available in the SVN version yet. Please wait a while.

[Cessen]

Originally Posted by blenditall The traditional methods will be kept though, I hope. I kind of like the noise sometimes

If you like noise, it's easy enough to add it in compositing.

However, the "traditional" method will be kept, I suspect, because it's accurate. This approximate solution has artifacts of its own due to approximating surfaces with sets of disks. For instance, a room with a window and light streaming in (using AO) would be a case where this new method would not be well suited.

[JoOngle]

Originally Posted by Cessen This approximate solution has artifacts of its own due to approximating surfaces with sets of disks. For instance, a room with a window and light streaming in (using AO) would be a case where this new method would not be well suited.

Maybe that could be solved using a´edge detection system where you´d render
normal AO near objects (distance specified in a "x.xxx" value setting) and anything
over the "x.xxx" value would render with the blurring discs?

[Ace Dragon]

A AO NOW IN SVN
http://lists.blender.org/pipermail/b...ry/012500.html

Link to release log page in this link. Raytraced AO methods still there for stills.

[homebody]

can't wait to render my crappy animation of bouncing ball with AO and QMC shadow and all the fancy stuff!

[IamInnocent]

My surfaces develop spotting, even pin holes when I mix AAO and OSA.

Jean

[bigbad]

Originally Posted by Cessen If you like noise, it's easy enough to add it in compositing. However, the "traditional" method will be kept, I suspect, because it's accurate. This approximate solution has artifacts of its own due to approximating surfaces with sets of disks. For instance, a room with a window and light streaming in (using AO) would be a case where this new method would not be well suited.

Isn't this method almost like a dirtmap shader. Like the once we see in XSI?