Packed Primitives Sim - Swap Hi Res and Lo Res

[bandini]

I couldn't find much information on this, but after about a half day I managed to cobbled enough information together to set up an example file.  Just putting it up here so the community has access to it.  The file shows how to simulate with low resolution packed primitives and then swap it out for a higher resolution version later.  Right now, this is dependent on all pieces having the same numbering, but I am going to try setting up another example where the numbering/fragmentation does not 100% match.

Packed_LoHi_Swap_02.hipnc

[cwhite]

There's also the "Transform Input Geometry" option on the DOP Import SOP, which transforms the input (high-res) geometry based on the transforms of the objects imported from the DOP network. It uses the 'name' primitive attribute on the high-res geometry to figure out which object in the sim each primitive corresponds to.

 

I've attached an example of that for your file, and there's also a similar example file (although it doesn't use packed primitives) on the DOP Import SOP help page.

Packed_LoHi_Swap_02.hipnc

[bandini]

Thanks, Cameron.  I'll check it out.

 

Also, here's an example of using the transforms of low res geometry to drive clusters of high res geometry.  I think this is pretty useful.

 

 

Packed_LoHi_Swap_03.hipnc

[bandini]

There's also the "Transform Input Geometry" option on the DOP Import SOP, which transforms the input (high-res) geometry based on the transforms of the objects imported from the DOP network. It uses the 'name' primitive attribute on the high-res geometry to figure out which object in the sim each primitive corresponds to.

 

I've attached an example of that for your file, and there's also a similar example file (although it doesn't use packed primitives) on the DOP Import SOP help page.

 

Definitely works for unpacked geo, but fails for packed, which is why I was exploring the method I posted.  Not sure if there are any speed advantages to using packed vs. standard geo in this circumstance (swapping).  Anyone have any info there? I am thinking also about the ifd-writing process... Not sure if packed prims offer speed advantage in ifd generation or not.

[bandini]

Last file on this topic.  This has already been covered in many threads (no new info here), just posting it here to keep some related concepts together in a single thread.  In this file, I show a technique to get rid of the voronoi look. (Interior detail subnet was copy pasted from the voronoi fracture sop).

Packed_LoHi_Swap_05.hipnc

Edited May 11, 2014 by bandini

[cwhite]

Definitely works for unpacked geo, but fails for packed, which is why I was exploring the method I posted.  Not sure if there are any speed advantages to using packed vs. standard geo in this circumstance (swapping).  Anyone have any info there? I am thinking also about the ifd-writing process... Not sure if packed prims offer speed advantage in ifd generation or not.

 

It does work for packed, but you need to make sure that the high-res packed primitives have a suitable 'name' primitive attribute so there's a little extra work involved (see the 'transform_hires_packed' node in my previous file).

 

You probably won't see much of a performance difference in transforming packed vs. unpacked, and you can always run the high-res fragments through a Pack SOP later if needed. 

[bandini]

Ah, I see now.  I was missing that "name" had to be a primitive attribute.  Looks like transforming the nearest cluster of pieces is as simple as attribute transferring the name attribute from the packed geo and promoting it to a primitive attribute.

 

Thanks for the help and clarification, Cameron.

[SaleVonGeist]

Thanks for the very interesting technique guys! I'm wondering how would you go about releasing some of those new clusters/fragments from their driven motion to simulation (in a new DOP network). Thanks in advance for any tips and pointers regarding that!

 

Cheers,

Sasa