Very interesting indeed and doesn't have the drawbacks of most other methods like storage of large amounts of data per pixel. Using color and depth which are already available in the film, this algorithm should give pretty good results.
It also says source code will be released soon, LordCrc seemed to think the algorithm was pretty straight forward. I think it would be a great shiny feature for 1.0 final
Have any of the devs taken a look at this yet? I know Everyone has been busy with v.1.0 so I have no expectations yet but it seems like a very interesting technique...
I have looked at the technique and it looks feasible to integrate it in LuxRender, but I am busy on oher stuff right now. I think Lord Crc had a look a t it too, maybe others.
That sounds good. Am I correct if I assume it is a post effect that could be enabled via the lux gui or is it something that needs being set by the exporter? I might ask Lord Crc what he thinks of this.
It seems very powerful for giving more complete previews of the final render of luxrender. Say for example if the render API of blender allows luxrender to run interactively in the viewport it would be great for adjusting the scene. Of course I do not know if the luxrender architecture would allow tweaks to a scene after the it has started to render the scene, but that is a whole other question.
I wonder this would probably blur textures if they were not sampled enough (as with other noise reduction)? Perhaps a shadeless render could be done (eg in blender before export) and used as an extra "dimension" of information in addition to the depth (forgive me if this is stupid, I don't understand the techniks )