stratified wrote:I've replied
here but let's discuss in this thread.
Hi,
Octane is a ray tracer and traces multiple rays per pixel, as a consequence, all edges will be anti-aliased. You can make the edges sharper by decreasing the filter size in the kernel node but you will still have smooth edges. If you take less samples per pixel, you will only get jagged edges and it will look weird.
I don't understand what you are trying to do. The picture on the right doesn't look like an alpha mask to me. It has the outlines of each objects but how would you use this? You can render proper alpha masks of each individual object via the render layers. Or you can use the render layers to render each isolated object and you don't need an alpha mask.
cheers,
Thomas
Hi Guys, I had some issues for videos basically as said before we use Nuke for composting the problem of the
ID we have now we can separate the alphas in a perfect way (I tried many different always photoshop does but not perfect)
We use a old techniques used for many year at the industry like
Coverage pass (many people stop using to create so many passes I don't know why?)
here a example of this method
http://www.sigillarium.com/blog/lang/en/379/So we use Mental ray to generate this two passes one with flat color to be easier to select and have perfect selection and then we multiply with the
coverage to add the aliazing information. We still use Mentalray because we have automatic scripts on Nuke were we clic at the object and nuke generate the alpha for it quite faster and easier.
My point is the
ID provide with Octane is focus to use magic wand with tolerance settings each is not excellent alpha.