Lux-a-thon?

General Project and community related discussion.

Moderator: coordinators

Re: Lux-a-thon?

Postby paco » Sat Oct 29, 2011 12:47 am

SATtva wrote:I don't think we actually need FLMs in the case of animation


I was thinking of perhaps rendering each frame to say 400 S/p - and then if we've got time just repeat the process - so at worse we'll get a grainy video, and if we get heaps of support then we'll get a nice squeaky clean one. I thought it would be worse to only get 30% or so of the frames done to the desired quality. That way I thought we could combine the .flms to allow this?

aldozang wrote:I like the idea of render a Gigapixel image. I have resources for render in my lab (a cluster with aproximately 40 core i7 machines). If you modelling the giga scene I can render it. ;)


Nice offer! Again the problem is the requirements for such an image.... If anyone can produce something that will look decent at 30,000 x 40,000 and run in 1-2Gb let me know! If geometric smoothing becomes a reality and someone can produce nice looking procedural textures then we might have a chance.... alternatively we could implement an iso-surface function like the mandelbulb and render that - but it would be a fair bit of work.
paco
Developer
 
Posts: 456
Joined: Sun Feb 07, 2010 1:37 am

Re: Lux-a-thon?

Postby SATtva » Sat Oct 29, 2011 4:36 am

paco wrote:I was thinking of perhaps rendering each frame to say 400 S/p - and then if we've got time just repeat the process - so at worse we'll get a grainy video, and if we get heaps of support then we'll get a nice squeaky clean one. I thought it would be worse to only get 30% or so of the frames done to the desired quality. That way I thought we could combine the .flms to allow this?

Good idea.
Linux builds packager
聞くのは一時の恥、聞かぬのは一生の恥
User avatar
SATtva
Developer
 
Posts: 5487
Joined: Tue Apr 07, 2009 12:19 pm
Location: from Siberia with love

Re: Lux-a-thon?

Postby paco » Wed Nov 23, 2011 7:43 pm

Bit of a bump....

Was thinking about the bandwidth involved regarding assembling the images into a single frame. I'm currently thinking that because it's an animation we could get away with people uploading tonemapped jpgs and then combining them into a single frame (weighted by S/p). I was thinking about the FLM or EXR formats but they are quite large and it could get quite difficult.

If we go with jpgs, then probably rendering each frame to say 100 S/p or something might be the way to go. We could then just use a straight average on the result - or some fancy outlier rejection (?median / mode) to simplify things.

Thoughts?
paco
Developer
 
Posts: 456
Joined: Sun Feb 07, 2010 1:37 am

Previous

Return to General Discussion

Who is online

Users browsing this forum: No registered users and 1 guest