Motionblur Mismatch

[brasco]

Hi,

I'm getting an issue when rendering to picture viewer where the motionblur differs from the picture viewer:




I can work around this by baking to Alembic and then rendering the Motionblur in Transform/Vertex Mode, but it would be nice to not have to keep baking out to test the looks.

cheers
brasc

[aoktar]

please read motion blur info in manual. I wrote that's not guaranteed in LV exactly same results. There is several limitations, you'll find it with some practice.
And pls don't forget, there is no other renderer can do it this stuff in realtime except our's.

[brasco]

Hi Aoktar,

I've read the manual on this, and I know how the LV is different to the PV but in this case, the PV is the wrong result, when it's usually the other way around, or is the whole Motionblur system not ready for production yet?

cheers
brasc

[aoktar]

I don't like this kind of generalisation with a few words. Mostly should be well. Motion blur has also several problems on c4d's own renderer. is this dynamic calculated stuff? Pls Try with baking.

[brasco]

Yes, it's a baked X-particle simulation, it's fine when I bring it back in as an Alembic cache and use Vertex mode, so that's what I'm doing (as I mentioned in the first post).

I'm really just trying to understand what cases Motionblur works as it should and where it doesn't, it's great to have but of course we want all of our toys to be used in production

cheers,
brasc

[aoktar]

only some problems are appearing on uncached objects and some a few situations. Therocally our method is relaible mostly, similar to c4d's. Biggest difference is c4d produces with forward keyframes. But my integration uses previous(with previous option) or previous/next keyframes. So some objects may not give required values in iterations. We will see what's missing with real projects.