Maxwell Render | Official public forum

Just noticed the MB cuts out around the time the cloth tears, C4D can render the MB correctly. Is there a workaround? or the future possibility to code for this?

My guess would be that the number of triangles in the mesh change when it tears (assuming you are using PLA). If that's the case, I don't know of any workaround, besides ensuring that the triangle count remains constant across the blur time.

Are NL planning to fix this?, should I post this in the wish list? (I'm infering from your comment that none of the plugins can render this situation at the moment)

I'll just change the sim so that it doesnt tear - more rubbery..

All I can really tell you is that a mesh has x-number of points, each possibly having been assigned two locations in time. If the mesh has a different set of points at those two times, though, then there's not much I can do -- I can't even write it at all, since there's no place (or too many places, depending) to put the data.

eric nixon wrote: C4D can render the MB correctly.

Is that so? I tested in physical render and it broke the frame the tearing started..

Interesting..., I tested with scanline and substeb MB and it looked ok, but I only tested one frame, the same frame where I first noticed a problem in maxwell. I didnt render an animation out, maybe try googling this? I wonder which 3d softwares can cope with this.. maybe none?

Before I thought that after the tear the MB on that object was failing for the rest of the frames, but now I found that it recovers a few frames later, so the problem becomes minor.

The problem is that a mesh doesn't contain enough data to deal with something like this; not that it is impossible -- one can imagine approximating each mesh state using a topology-independent intermediate representation, morphing between the two, and then deriving new meshes from that -- but this is far outside the scope of what Maxwell currently aims to do. Such a capability is not even present in a modeling application like Rhino, which specifically deals with NURBS and meshes; it's something along the lines of what nPower provides with their SubD <-> CAD translators.

The thing is, the substep MB is fake anyways, so not really valid to compare. I believe Maxwell takes the vertices into account, and as these change (in numbers, ids etc) during the tearing process of the cloth, there isnt proper info for the real motion blur to take place.

Same is happening with other renderers I believe, as not even the c4d native physical render can deal with that. I dont have octane or vray, so cant tell though.

Yet I wonder how Maxwell deals with the consistent change of RealFlow Mesh sequence and motionblur?

My desktop today, very organic workflow with the new plugin, some glitches due to me doing crazy things, esp with mapping, but easily workaroundable, and could be c4d glitches. One annoyance seems to be decimal values.. Only whole numbers are working for many things such as f-stop, intensity values etc.





I dont know where this is going, but i hope it will 'gel' soon, mostly low-poly photomapped background stuff, and some 2d trees too.

Photomapping..., quite addictive and frustrating.. prob wont use this one, but its the ideal kind of photo..