Cloth Cache appears in wrong position #32944

New Issue

Daniel Salazar · 2012-10-21T20:19:14+02:00

Daniel Salazar commented

2012-10-21 20:19:14 +02:00

%%%Open environment/Envionment.blend

Cloth cache should work fine in viewport

Now render

Cloth is far away from camera%%%

%%%Open environment/Envionment.blend Cloth cache should work fine in viewport Now render Cloth is far away from camera%%%

Daniel Salazar commented

2012-10-21 20:19:14 +02:00

Changed status to: 'Open'

Lukas Tönne commented

2012-10-21 22:30:46 +02:00

%%%Problem here is basically the same as in #32846:
https://projects.blender.org/tracker/index.php?func=detail&aid=32846&group_id=9&atid=498

When dupli-objects are generated in the dupli group, the object->obmat matrix is temporarily set to the total dupli instance matrix. Each time a dupli instance is generated it also calls the cloth modifier, which reads the cached state and applies it to the DM vertices. However, the cloth cache is stored in world space, so the cloth mod applies the inverted obmat transform to supposedly get vertices in object space. But since the obmat is actually the instance matrix, this ends up looking like the instance is completely untransformed ...

In order to work correctly the cloth mod would have to know the actual unchanged base object transform, but this is not available. A serious design flaw with dupli objects IMHO ...

I'll try to find a fix, assigning to myself.%%%

%%%Problem here is basically the same as in #32846: https://projects.blender.org/tracker/index.php?func=detail&aid=32846&group_id=9&atid=498 When dupli-objects are generated in the dupli group, the object->obmat matrix is temporarily set to the total dupli instance matrix. Each time a dupli instance is generated it also calls the cloth modifier, which reads the cached state and applies it to the DM vertices. However, the cloth cache is stored in world space, so the cloth mod applies the inverted obmat transform to supposedly get vertices in object space. But since the obmat is actually the *instance* matrix, this ends up looking like the instance is completely untransformed ... In order to work correctly the cloth mod would have to know the *actual* unchanged base object transform, but this is not available. A serious design flaw with dupli objects IMHO ... I'll try to find a fix, assigning to myself.%%%

Ton Roosendaal commented

2012-10-22 10:33:32 +02:00

%%%Wouldn't the flaw be in the point cache then? World space isn't useful for instanced groups.%%%

Lukas Tönne commented

2012-10-24 15:16:27 +02:00

%%%Ok, it's not a nice situation but apparently there is no easy fix to make this way of linking point cached data possible. Several people have suggested complicated workarounds, but i didn't understand a single one in detail. Closing the report.%%%

Lukas Tönne commented

2012-10-24 15:32:49 +02:00

%%%Reopened and assigned to Campbell. Ton said you made some sort of workaround for this, maybe want to comment on this?%%%

Daniel Salazar commented

2012-10-25 10:24:27 +02:00

%%%I think I figured out what it is. It's because the linked dupligroup was moved away from the scene origin before creating the proxy. This creates an offset between the group's origin and the scene origin that is ignored by the cloth cache%%%

Ton Roosendaal commented

2012-10-25 12:24:10 +02:00

%%%This will get added to the standard checklist for recode work on proxy, groups, depsgraph, point cache, name it.. ;)%%%

Ton Roosendaal commented

2012-10-25 12:24:10 +02:00

Changed status from 'Open' to: 'Archived'

Ton Roosendaal closed this issue

2012-10-25 12:24:10 +02:00

Sign in to join this conversation.

No Label

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

Cloth Cache appears in wrong position #32944