Loading...
> >

G3 bones

jmucchiellojmucchiello Posts: 888
in Technical Help (nuts n bolts)

I'm used to G9. I've dipped into G3 for middle distance figures. (Anywhere a lowpi figure is too low poly and where a G9 figure is too much.)

What is there Shoulder Bend and Shoulder Twist? They have the same three rotations paramters just in a different order: Front-Back, Bend, Twist versus Twist, Bend, Front-Back. The values map to one another: The twist is the same regardless of which one is moved. Same thing happens with Forearm Bend and Forearm Twist and Thigh Bend and Thigh Twist.

And there's a set of X/Y/Z rotates spread out between the pairs that are seperate from the FB/Bend/Twist group. Perhaps these rotations are like the G9 "twist 1 and twist 2".

Just curious.

Comments

  • DrunkMonkeyProductionsDrunkMonkeyProductions Posts: 1,204

    Short version, they all exist to solve the same problem, mesh warping when posing/animating.

    The reason for the difference in order, afaik, is due to studio orgazing transforms/rotation based on order of operations, using the Name, not Label, of the parameter.

    g3 uses YZX and G9 uses XYZ for Shoulder bend and Upper arm, respectively. The other nodes use different orders, such as ZXY, or YXZ.

    If you're seeing any parameter labeled X, Y or Z rotation, in the arm nodes, it means you've got "show hidden parameters" enabled, and should probably turn that off if you don't know what you're doing. Also, by default the hidden nodes should be locked, so they don't really do anything normally.

     

    For G3, while you see the same parameters in the Should/forearm Bend and Shoulder/forearm Twist, they are not all "real" morphs, but Alias morphs, which are just pseudo copies/control/convenience morphs that don't contain any geometry data themselves.

    Twist in the Bend node is an alias, while front-back and bend are aliases in the Twist node. The actual rotation parameters in each node are locked by default.

     

    G9 works a bit differently, with more complex rigging, and additional weight maps associated with that rigging. The Twist 1 and 2 nodes.

    No alias dials are used, due to this difference, and you shouldn't be seeing any parameters other than Twist and scale in the Twist nodes.

    In general, you shouldn't mess with these, unless you're trying to make a particular pose look better due to mesh warping in thos particular areas.

     

    One last thing, You don't really save all that much vram/ram using g3 over g9.

    In general, it's only going to be around 15-20MB per character, with render sub-d set to 1 on both characters.

    This will vary depending on character preset used, clothing, hair and other accessories, which may make the difference even smaller.

     

     

     

     

  • jmucchiellojmucchiello Posts: 888

    DrunkMonkeyProductions said:

    If you're seeing any parameter labeled X, Y or Z rotation, in the arm nodes, it means you've got "show hidden parameters" enabled, and should probably turn that off if you don't know what you're doing. Also, by default the hidden nodes should be locked, so they don't really do anything normally.

    I said I was new to G3. Not new to Daz. You need to see hidden parameters for lots of different things involving morphs.

    In general, you shouldn't mess with these, unless you're trying to make a particular pose look better due to mesh warping in thos particular areas.

    So, all the time.

    One last thing, You don't really save all that much vram/ram using g3 over g9.

    This goes against all my anecdotal experience. I can squeeze 6 G9 figures into my 4090 usually. I've had 30 G3 figures in a scene with room to spare.

    In general, it's only going to be around 15-20MB per character, with render sub-d set to 1 on both characters.

    Ah. Well, I suppose. But I'm more likely to put "heavier" hair on a G9 figure.

  • crosswindcrosswind Posts: 9,713
    edited 9:27AM

    DrunkMonkeyProductions said:

    Short version, they all exist to solve the same problem, mesh warping when posing/animating.

    The reason for the difference in order, afaik, is due to studio orgazing transforms/rotation based on order of operations, using the Name, not Label, of the parameter.

    g3 uses YZX and G9 uses XYZ for Shoulder bend and Upper arm, respectively. The other nodes use different orders, such as ZXY, or YXZ.

    Actually both of the bones use XYZ as Rotation Order in G3 and G9. Only the difference is that G3 loads the Rotation properties in the order of YZ + the alias of X of its child twist bone... while G9 loads the Rotation properties in XYZ order without using Aliases anymore ~~ You can find all these definitions in the DSF files of G3 and G9 ~~

    Besides, Shoulders on G3 equal to Upper Arms on G9... while Shoulders on G9 equal to Collars on G3. This is another difference from their skeletal perspective.

    Post edited by crosswind at
  • jmucchiellojmucchiello Posts: 888

    crosswind said:

    DrunkMonkeyProductions said:

    Short version, they all exist to solve the same problem, mesh warping when posing/animating.

    The reason for the difference in order, afaik, is due to studio orgazing transforms/rotation based on order of operations, using the Name, not Label, of the parameter.

    g3 uses YZX and G9 uses XYZ for Shoulder bend and Upper arm, respectively. The other nodes use different orders, such as ZXY, or YXZ.

    Actually both of the bones use XYZ as Rotation Order in G3 and G9. Only the difference is that G3 loads the Rotation properties in the order of YZ + the alias of X of its child twist bone... while G9 loads the Rotation properties in XYZ order without using Aliases anymore ~~ You can find all these definitions in the DSF files of G3 and G9 ~~

    Besides, Shoulders on G3 equal to Upper Arms on G9... while Shoulders on G9 equal to Collars on G3. This is another difference from their skeletal perspective.

    Yeah, I was confused initially by Collar. Both naming conventions are have pluses and minuses: Spine 1-4 in G9 is better than upper/lower ab/chest in G3. But Carpal 1-4 in G3 is better than index, middle, pinkie, ring in G9 since they're in "order".

     

  • Richard HaseltineRichard Haseltine Posts: 109,321

    jmucchiello said:

    DrunkMonkeyProductions said:

    One last thing, You don't really save all that much vram/ram using g3 over g9.

    This goes against all my anecdotal experience. I can squeeze 6 G9 figures into my 4090 usually. I've had 30 G3 figures in a scene with room to spare.

    Using similar materials and texture resolutions?

  • jmucchiellojmucchiello Posts: 888

    Richard Haseltine said:

    jmucchiello said:

    DrunkMonkeyProductions said:

    One last thing, You don't really save all that much vram/ram using g3 over g9.

    This goes against all my anecdotal experience. I can squeeze 6 G9 figures into my 4090 usually. I've had 30 G3 figures in a scene with room to spare.

    Using similar materials and texture resolutions?

    No. But, as I said: without resorting to something like scene optimization. Anything that changes texture resolutions takes time, even with Scene Optimization to do the work for you. SubD = 1 at least is fast. But messy to undo.

    In my workflow, I might do a dozen renders with different foreground depths all using the same characters. Needing to reload so I can get back to the original hirez model for a closeup is time. (And sure, I could be more organized and pose scenes in depth order. But that's error bound for consistency of poses between shots.) Consider a big family dinner. The main characters are sitting at either end of the table and one in the middle. But the cousins and such are lower rez assets. Swapping in and out when there's a close up at one end and then the other end of the table is time consuming.

    In any case, I was asking about why the bones are duplicated in G3, not about scene optimization.

Sign In or Register to comment.