Loading...
> >

LOD's in Carrara

ProPoseProPose Posts: 527
in Carrara Discussion

One of the main reasons I still use Generation 4 figures (V4 and M4) is because of LOD I can load onto the figure.  Genesis 1,2, or 3 don't use this feature as far as I can tell.  Is there a way to create LOD's in Carrara for other things such as trees and other landscape features.  Carrara is obviously capable of displaying LOD's based on camera distance using V4M4 figures,  so I can't see why this feature can't be implimented for other things,  I just don't know how.  So if someone does, please share the wealth

Comments

  • evilproducerevilproducer Posts: 9,050
    Interesting question. I have no idea, but I would love to hear if anyone has any ideas.
  • FenricFenric Posts: 351

    You can load Levels of Detail for anything that is Poser rigged (CR2), as it is a feature of the CR2Importer module. Making the detail objects is relatively straightforward: you just need software that will do a decimation while retaining existing vertices.

     

    That doesn't really help for props, but it's really not hard to slap one bone onto an object in Poser or Studio and save it out to a CR2 - then you can use the LOD feature.

  • ProPoseProPose Posts: 527
    edited October 2015
    Fenric said:

    You can load Levels of Detail for anything that is Poser rigged (CR2), as it is a feature of the CR2Importer module. Making the detail objects is relatively straightforward: you just need software that will do a decimation while retaining existing vertices.

     

    That doesn't really help for props, but it's really not hard to slap one bone onto an object in Poser or Studio and save it out to a CR2 - then you can use the LOD feature.

    That's kind of what  I thought.  I was actually thinking of you as I posed the question, So thanks for your input.  Would the decimation function that exists in the vertex modeller serve that purpose?  Or even the decimator in Daz Studio?

     

    I was hoping I could use the ERC plugin to toggle the visibilty of an object that has multiple LOD's depending on the distance it is from the camera. And would that work on replicated objects?

    Post edited by ProPose on
  • FenricFenric Posts: 351

    As far as I know, Studio and Carrara both change the vertices. (They're in good company: as far as I can tell, Maya's "Reduce" won't preserve the original vertices either.) I believe the DAZ ones were made with Lightwave, and I have made them with 3DS Max.  Blender changes all the time, so it might be worth checking - I don't know.

     

    And yes, ERC can do that for individual objects - it's quite easy. But ERC cannot operate on the instances within a replicator, no - that information isn't exposed in a way ERC can handle it: ERC only works on keyframe-able values, and the instances within a replicator are not exposed to the sequencer.

     

    I also would caution that I would be surprised if you get as much of a benefit from LOD as you hope, especially given the amount of work it ends up being. I have quit using the LOD feature, because it actually made memory pressure worse to have the multiple geometries loaded, and it didn't really help the render time particularly. Texture map size hits much worse than geometry density: reduce texture map sizes overall and use a mip-map shader like Sparrowhawke's, and it will make more of a difference.

  • ProPoseProPose Posts: 527

    Thanks for your input,  very much appreciated.  I guess it all boils down to efficient scene /asset management. 

  • evilproducerevilproducer Posts: 9,050
    For landscapes and such, replicators can be your friend, and they would be more efficient than LOD in my opinion.
  • YofielYofiel Posts: 204
    edited November 2015

    While I agree LOD is alot of work, it can significantly improve renders of distant objects, and the memory impact should never be more than double. This is because LOD uses a continual sequence of progressively smaller textures which are each half the width and length of the previous one. Being half the size on each axis, it uses 1/4 as much memory. So  as per Zeno, it should never acftually reach double the memory, no matter how many LODs there are in the set...and even though Zeno did not take into account any compression overhead, in practice, LOD file sets are never more than double the size of the original. 

    What typically happens is people divide in two camps. Some people want as many 4K textures as possible for high-quality closeups, but you can do LOD for half the memory cost of doubling a texture size, so in games people generally prefer 2K textures max with LOD, for better render performance amnd betterdistant views. That is, the texture is half the upper Kbyte size, with LOD, as a compromise. 

    Post edited by Yofiel on
  • FenricFenric Posts: 351

    Carrara's LOD feature does nothing to the texture size at all: that's mip-maping which is an entirely separate feature. LOD is only for geometry, and you have to specify the geometry files to use at each distance, it is not automated.

  • YofielYofiel Posts: 204

    Oho thats why I couldnt figure out how it works, lol, thanks for telling me

Sign In or Register to comment.