Fusion : HoS_MayaToxic2DMV.plugin

eyeonfusion120A Fusion plug-in that converts a Maya/MentalRay 2D Motion Vector (2dMVToxik) render pass to a normalized data set that Re:VisionFX’s ReelSmart Motion Blur can us.

All you as user need to do, is render out your 2dMVToxik pass as you normally would (either embedded as extra EXR pass or as a single RGB(A) file) and plug the following channels in the Background input:

red -> 2dMVToxic X-motion (the red channel by default.)
green -> 2dMXToxik Y-motion (the green channel by default.)

and plug the following channel into the Foreground input:

alpha -> the Alpha channel of the image that needs to be motion blurred.

When you have done that, the plug-in normalizes the channels according to the Displacement value so that when placed into the FG input of a ReelSmart Motion Blur Vectors node with the Max Displace set to the same value as the Displacement value of this plug-in, it should work exactly like the data that is produced by a shading plug-in like,  say lm2dmv.

This plug-in has been used extensively in our pipeline without any troubles so far, but should you run into something please let me know.
At the moment i have only compiled a x86 version and some small updates are planned, so keep an eye out for those.

download : hos_mayatoxic2dmv


9 thoughts on “Fusion : HoS_MayaToxic2DMV.plugin

  1. Just curious…
    Is there a reason this method was chosen over using the 2dMVToxic pass directly with the Motion Vector tool in Fusion?
    Cool tool BTW 🙂

  2. You mean the Vector Motion blur that comes with Fusion by default? Or the RSMB tool?

    The standard Vector Motion blur doesn’t handle (overlapping) edges between different blur values well, RSMB does that a bit nicer, though it would be nice to have the blurring like the default one at times combined with the RSMB one.

    Would be nice if RSMB could use a depth map to define what overlaps what for an even nicer effect.

    So, to answer you initial question, we use RSMB and thus the motion vector data need to be normalized as described here : http://www.revisionfx.com/support/faqs/motion_vector_FAQs/motion_vector/

    It can be done quite easely with a Custom Tool, a compiled plugin written just for this task tends to be faster though

  3. Hello.

    >Would be nice if RSMB could use a depth map to define what overlaps what for an even nicer effect.

    I am looking forward to this feature.

    The artifact appears when RSMB velocity channel has complex object shape.
    Isn’t there method of evading the artifact of the each object?

    Sorry bad English.
    Best regards.

  4. Not that we figured out, at least, not without rendering each object separately (or splitting out a render using object id’s), we tend to check which tool gives the best result per shot and use that.

  5. Thanks for the reply.

    Because the artifact appears in RSMB, I often use Fusion’s vectorblur.
    However, the result of RSMB is good when RSMB velocity channel has simple object shape.

    Best regards.

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.