Question for the Poser Engineers and Geeks

  • I'm not an engineer, but I think I recognize the reasoning behind the idea of cutting up an figure's geometry into the various groups/actors. It basically allowed each body part to move as needed like a joint. And at the time there were not other ways of accomplishing the same thing, I guess. So my question is, is it not possible to accomplish movement and joint flexibility today without the need to physically cut the figure up? Isn't weightmapping itself one of those ways? So, even allowing for backward compatibility issues, wouldn't it make more sense for Poser to phase out the mesh cutting for newer figures going forward?

  • At the time there were other ways to go but Poser's methods were easier in a way - and some aspects of it still works better than you'd get in vanilla Maya. That said, however, with the inclusion of weight mapping in Poser it's now up to the creator to decide if they want to cut up their models or not. I wrote a brief tutorial about this very thing as well as created a video on how you can rig without the need to separate the model into groups within Poser. The video may not be around anymore but I believe the written explanation should be on the forum somewhere.

    The Poser team could take out support for breaking up the models but that's going to also likely kill support for a lot of older content (which I've always been OK with - people need to let old toys stay buried). Since doing that will just cause people to complain and in turn, request support for the old content be reinstated, why not just suggest to new content creators that they stop doing things the old way? Then you get the best of both worlds. You can use the old content with it's weirdness and the new content with all its weirdness.

    So I'd say put the question to the content creators.

  • I would love to find that tutorial. One day I will look for it. This is something I'd like to try.

  • @rokketman said in Question for the Poser Engineers and Geeks:

    I would love to find that tutorial. One day I will look for it. This is something I'd like to try.

    I think this maybe the tutorial he's referring to.

  • Poser Ambassadors

    For those that have access and know where to look ? Search for 31615.
    This is my area where I have been pushing the old and new Poser team. (fingers crossed for P12)

    As Teyon already stated. => You do NOT need multiple vertex groups in weight-bulge mapped figures, clothing or props.
    We can build and rig complete figures/clothing using only ghostbones and a single vertex group.
    And Poser simply "can" not break it up as there is only one. LOL.

    But : It is god dammed long and hard work in current Poser versions. => You have to "rig" all joints completely manually in all directions and scales. => For a standard figure that means hundreds of joints to take case of, one by one, and in the correct order.

    One of these days, pffft, one of these days I am going to make a video, because using only a few vertex groups and move the rest with ghostbones, and letting Poser do "most" of the work, does speed up the proces.

    But even when using only 2 vertex groups?
    Poser will do the split at save, and you have to replace the obj file. => No problem but you HAVE to do it.

    @eclark1849 earl ?
    Be careful what you wish for, because a "definitive fix for this issue"?
    => WILL break 99% of exiting content if not done properly with a checkbox for backwards compatibility.

    Allow me to end with a single example to show how deep this goes: Growing dynamic hair on a figure/clothing.
    To grow hair directly on a figure/clothing, the hair room creates a new vertex group to grow the hair from.
    => Paaf, kaboem, bingo => Double vertex, and here we go again. LOL.

    Ok, the above was not the end. LOL.

    Remember the "old style conventional rigging"?
    A "smart prop" with inherit bends checked on? => Followed all bends from its parent group, one group UP and one group DOWN.

    The same thing soes NOT work on weight-bulge mapped figures.

    But? On weight and bulge mapped figures it works in the setup-and fitting room where you need one less vertex group then bones. => The rig will follow one bone UP and one bone DOWN from the last vertex group.

    Man I need 36 hrs days.

  • @vilters Tony, can strand hair be affected by magnets ?

  • Poser Ambassadors

    one of the ways to style dynamic hair is by using magnets.

    You know that you can also weightmap the magnet zones to control the fall off.
    You style a hair.
    Then with the head twisted or bend or side side? Human hair only moves partially. => You can "style" the hair using magnets, and control the fall off during head turning or bending by painting the fall off zone.

  • @vilters You do a video on that and you got my attention.

  • @eclark1849 That's not it but some of the ideas are explained in that video. The video I spoke of was one I had made a year before that webinar and was on my personal youtube page. It's not anymore. I took it down. It did feature the same character model oddly enough.

  • I think this is where @adp ' s method to make weight maps via Python comes in.
    Basically if you make all your shirt geometry the 'chest' actor, you want all vertices of forearm to respond with weight 1.0 to joint movement of shoulder and collar joints, and then with its own weights to the movement of the forearm joint.
    The script can do the tedious work.

  • @vilters from my recent investigations into dynamic strand hair, the strands are grown from a new prop derived from the selected facets on the underlying actor(s). What the hair group on the original actor is required for, is actually just to map all deformations of the underlying actor's vertices due to bending or morphing, onto the hair growth prop's vertices (without that, the roots of the hair strands don't follow the actor's surface below). Since the hair group is not used as an actor definition in the figure's rigging, it does not cause any vertex duplication in the figure at all.

    I have had to perform tedious workarounds with a strand-hair furred figure, as I've described in other threads (replacing the actor's geometry reference with an objFileGeom instead of a geomFileGeom, so that the figureResFile reference can remain unchanged). Thus, I think that the assertion that dynamic hair would break is certainly relevant, but not actually valid, given the hair facet groups are not referred to in the figure's rigging. Only rigged groups get the vertex duplication treatment, having performed direct comparisons of saved figure geometries which have had hair grown on them.

  • [expletive deleted internet lag and stupidly short edit interval prevented adding this QED image to previous post]
    0_1533140407036_Wild Thing?.jpg

  • Poser Ambassadors

    What you write is correct. The issue is deeper.

    One can work in Poser, morph, re-texture, create a new character, whatever, and when you save to library? ? ?

    • Poser will ONLY re-save a fresh cr2 and all is good. (It continues to use your properly vertex groped and welded obj file and does NOT re-save a fresh obj file if you did not do obj file work) => (Morph data is stored in the cr2)

    But? ? ? When you grow a hair group, all that changes.
    When grow hair on the obj file and save the character to library, Poser sees this as an obj file change (extra vertex groups where added) and ? ? ? Guess what?

    • Poser saves a fresh cr2(good) AND it saves a fresh obj file (not good) because Poser unwelds the vertex groups in its newly saved obj file.

    Bad result : All vertex groups unwelded and NO way to turn back without loosing your dynamic hair. => And you have to repair or replace the original obj file.

    I could write a whole boo about this unwelding. LOL.

  • @vilters OK, now we're on the same page. Dynamic hair breaks the obj because it adds groups which prompts a rewrite which bakes in the actor seam duplicated vertices to the new obj, not because it directly causes vertex duplication itself. :-) So my workarounds for the newly added hair groups saved as individual actor geometry obj files is obviously still lumbered with the duplicated vertices (necessarily, since the actor geometry has to include all of that actor's vertices, even if they're shared with another actor at the seams), but manually restoring the original figureResFile reference allows the pretense that nothing has changed on a whole figure basis.

    None of which is helped by Poser's somewhat misleading references to welding as the act of forcing the normals of duplicated seam vertices to be identical, giving the illusion that the vertices have been consolidated into one, in complete contrast to the Grouping Tool's Weld function, which will actually consolidate vertices within a tolerance distance of each other (though not across actor seam boundaries, it would appear).