Be ready to have someone write some help if you ever do, without a decent system to reduce them to sparse data and to work do the work GPU side 800 shapes move at the speed of a brick chained to a column, especially in Maya :) XSI 5 however was managing it respectably well already on Pentium III and 4s back then :)
On Sun, Jan 18, 2015 at 2:44 AM, Greg Punchatz <g...@janimation.com> wrote: > Thanks Raff, > > I have used both techniques but never heard some of those terms.... ICE > made doing this work much easier for me. > > I have never ended up with 800 shapes.... but give me the time and the > budget and that sounds like a blast :) > > > On Sat, Jan 17, 2015 at 5:06 PM, Raffaele Fragapane < > raffsxsil...@googlemail.com> wrote: > >> You can blame Bay Raitt for some of the names being thrown around, and >> the LISP community he grew up in :) >> >> Combination sculpting comes in two flavors, FACS based, with expressions >> tabled out and combinations being largely corrective and flattened out, and >> twitch based, with shapes representing individual muscles as roots, >> combinations of nearby muscles in couples or triplets as first branch, and >> so on to full face compensation, usually you stop at tier three or four, >> which can easily get you hundreds of shapes (Charlotte in Charlotte's web >> was twitch combinations and amounted to 802 shapes, Gollum in return of the >> kind was FACS and I think Bay ended up in the 820 or so range in the end). >> >> You can use something like stretch mesh (or ideally better) equalisation >> process after that to reduce drift if you're in a hurry with the broad >> strokes. >> >> Combinatorics are shapes that bridge two other shapes by correcting their >> conflict (additive) rather than by replacing them (you can combine with C = >> abs(A-B) in the former, or suplant with C = abs(A-B) and then subtract C's >> intensity from A and B). >> >> On Sat, Jan 17, 2015 at 10:59 PM, Greg Punchatz <g...@janimation.com> >> wrote: >> >>> We (Brad did all the ICE magic) worked up some pretty niffy tricks for >>> our head tech demo. >>> >>> We could pose our head which was a slightly enhanced FR rig export a >>> reference head into ZB... bring it back into soft the subtract the the >>> deforms of the mesh and reapply only the differences from the corrective >>> shape. >>> >>> Point drift is caused most of the time by subdividing the model in >>> Zbrush. If you do a subdivision in Z all your base point will shift. In >>> our case the mesh was dense enough that was not an issue, we could still >>> clearly see the forms without subdividing while in Zbrush. Brad wired up a >>> ICE tree for the imported corrective shapes to be triggered by pulling >>> different distances from the rig. Of course drift can happen from someone >>> moving points they have no business of moving, or even worse they move >>> points in the wrong direction for the correction or shape. I always work in >>> a stepped process to avoid this for shapes, whether I sent to Zbrush or >>> not. I am at first only focused on how the point mass moves first. I try to >>> get this done with as few proportional moves as possible. Then I test the >>> motion in Soft and on the rig., take a look at what it looks like with the >>> jaw open etc. Then I slowly massage the shapes into place checking the >>> sculpt in action >>> >>> I don't remember if the zbrush link busts your rig, in our case the >>> workflow was to use separate reference geo. >>> >>> It is better if it when done all under one roof but if my point count >>> goes high enough I will jump through a few hoops to get to a better point >>> manipulator. >>> >>> Raf I have never heard the term combinatorics before, and when I looked >>> it up I could not find any references that clearly showed me how it applied >>> to shape animation or rigging. Can you point me to a reference that might >>> help fill in my knowledge gap : ) >>> >>> Also Eric, I had heard of folks having a different neutral vs skinning >>> pose but I have not really seen a good explanation of the idea. I have >>> modified a sculpt to be better for rigging, but that shape then becomes my >>> base shape. What is the difference? >>> >>> >>> >>> On Sat, Jan 17, 2015 at 2:31 PM, Raffaele Fragapane < >>> raffsxsil...@googlemail.com> wrote: >>> >>>> If you're doing combinatorics you don't model the shapes in isolation, >>>> you tweak a base and need to see the result on the combination, which might >>>> be one to four tiers of combinations away. >>>> You don't do combination sculpting without the rig because you don't do >>>> combination sculpting on the final shape half the time if you're sensible >>>> and can't waste a lot of time in kickbacks. >>>> >>>> Doing shapes in ZBrush is doable, but they all need a lot of work after >>>> coming back in because by the nature of ZBrush you will have shit drifting >>>> all over the place. When they will add more than a single morph and a few >>>> simple vector operations to wire the morphs it will then be the ultimate >>>> tool for it, right now it's like trying to drive a truck out of a parking >>>> lot with a small gate. Blindfolded. On iced out ground. With a monkey >>>> hitting you on the head with a baseball bat every five seconds. Technically >>>> doable, but not worth the bother unless you get to show the mental >>>> breakdowns on TV and cash them in :) >>>> >>>> If you're doing cartoony or largely procedurally shaded stuff you can >>>> take a fair amount of drift. if you're doing something that has hundreds of >>>> rigid scales or precisely styled hair bound to the UV space it's an >>>> unmitigated disaster when you don't have something like Soft (or a shitton >>>> of stuff piled on top of Maya) around to do the work. >>>> >>>> On Sat, Jan 17, 2015 at 6:16 PM, Greg Punchatz <g...@janimation.com> >>>> wrote: >>>> >>>>> Raff while what you say is true about needing to check the results of >>>>> your sculpts in combination with with other shapes and deformers. There is >>>>> no reason those edits should not be done in the tool-set best suited to >>>>> sculpt. >>>>> >>>>> Using something like Zaplink or a few scripts can make the back and >>>>> forth seamless. ICE made it so much easier to to pose based deformations >>>>> and corrective shapes using Zbrush to edit. >>>>> >>>>> That being said I still do a great bit of my shape work in soft, >>>>> unless its a very dense mesh, then I whip out the Z >>>>> >>>>> On Fri, Jan 16, 2015 at 6:16 PM, Raffaele Fragapane < >>>>> raffsxsil...@googlemail.com> wrote: >>>>> >>>>>> The problem with ZBrush, or any modelling app that doesn't have your >>>>>> full rig in it, is that for things like combination sculpting they are >>>>>> useless, because you need to see multiple timelines of the shapes >>>>>> converging as you refine them for the result to be any good. It's also a >>>>>> ton easier to get combinatorics started in Soft since you can start any >>>>>> shape from any number of others with ICE. I so miss that in any other app >>>>>> (that last bit is literally the only one where Houdini could compete or >>>>>> even surpass Soft, actually, though it's somewhat painful to wrangle the >>>>>> shit together when you hit a certain degree of complexity and you end up >>>>>> spending more time making an uber rig than you do working the shapes' >>>>>> alignment). >>>>>> >>>>>> On Fri, Jan 16, 2015 at 11:27 PM, Jordi Bares Dominguez < >>>>>> jordiba...@gmail.com> wrote: >>>>>> >>>>>>> Thanks for the notes, there has been quite a lot of changes but it >>>>>>> is true there are a few of your comments still pending, the most >>>>>>> pressing >>>>>>> to me is speed and the viewport needs still lots of love. >>>>>>> >>>>>>> BTW, I was not advocating to use Houdini for modelling though, >>>>>>> rather use Zbrush to be honest and now that Zbrush is getting closer to >>>>>>> a >>>>>>> full set of traditional modelling tools it is pretty obvious it is the >>>>>>> route to go. >>>>>>> >>>>>>> My feeling is that character work is certainly more painful but at >>>>>>> least you get some serious gains and unfortunately there are no options >>>>>>> so >>>>>>> we are in a transition moment. >>>>>>> >>>>>>> So far they are listening and moving forward so I will stick to >>>>>>> Houdini for the time being and keep an eye on others. >>>>>>> >>>>>>> :-) >>>>>>> >>>>>>> jb >>>>>>> >>>>>>> >>>>>>> On 16 Jan 2015, at 21:28, Raffaele Fragapane < >>>>>>> raffsxsil...@googlemail.com> wrote: >>>>>>> >>>>>>> A lot of quality rigging, despite piles of papers trying to sell the >>>>>>> public on the contrary, is still manually tweaked. Taking things out of >>>>>>> the >>>>>>> app where you have the full rig makes authoring a major pain. The most >>>>>>> basic example is shapes, doing shapes work in XSI for something like a >>>>>>> combination sculpting setup was as easy as it got, especially after ICE. >>>>>>> The way data is presented and accessible, the speed on large meshes, >>>>>>> the modelling toolkit, it all lent itself to that kind of work in a >>>>>>> perfect >>>>>>> storm scenario. >>>>>>> Doing the same in Maya, comparatively, is beyond painful and >>>>>>> requires a pretty big staging effort to separate work and write >>>>>>> accessory >>>>>>> tools, in Houdini you don't even have a particularly intuitive modelling >>>>>>> toolkit, and the handling of large meshes was pretty meh with it (at >>>>>>> least >>>>>>> up to 12, it seems to be getting better and promising to be getting >>>>>>> better >>>>>>> again). >>>>>>> >>>>>>> The toolkit in general is pretty hard to impossible to give to a >>>>>>> modeller with little inclination to learn something like Houdini, while >>>>>>> with both Maya and Soft that's not a big challenge. >>>>>>> >>>>>>> I haven't tried the muscle system in a while, so my comment might be >>>>>>> dated to the point of not being valid, but the last time I did it was a >>>>>>> bit >>>>>>> of a joke. No arbitrary topology for the deformers unless you cloth >>>>>>> collided (and the cloth solver was anything but acceptable), only some >>>>>>> weak >>>>>>> superset of metaballs, rather slow, but at least it was relatively >>>>>>> stable, >>>>>>> and overall clunky and requiring the lot a lot of micromanagement and a >>>>>>> lot >>>>>>> of SOPs that often refused to play nicely with the rest of the app. >>>>>>> Mind, I haven't found a single commercial muscle system I would use >>>>>>> if they paid me for it, which is pretty embarrassing given when we >>>>>>> needed >>>>>>> one for WWD we got a rather intuitive one done in just a few weeks that >>>>>>> worked for over 99% of the show meshes without manual intervention of >>>>>>> any >>>>>>> sort on the sim, and literally only a dozen mesh fixes across over 800 >>>>>>> shots. >>>>>>> >>>>>>> On top of all that, and again this is pre-14, most pre-13, it's >>>>>>> slow. Mind boggingly slow to articulate a decent animation rig. I >>>>>>> suspect >>>>>>> this last point has been, or is about to be, superseded though since the >>>>>>> viewport has been getting some love. >>>>>>> >>>>>>> The main issue though remains that preparing an asset in Houdini >>>>>>> remains a long and involved process which very few people from other >>>>>>> departments, some times nobody, can be recruited into, it's born, lives >>>>>>> and >>>>>>> dies in the hands of TDs. >>>>>>> >>>>>>> I've always had a soft spot for Houdini, and I'd give my money to >>>>>>> SideFX rather than many other companies any day of the year, but as a >>>>>>> company their commitment to character work of artistic or hybrid nature >>>>>>> has >>>>>>> always been patchy (and I don't necessarily blame them for it) and >>>>>>> subpar. >>>>>>> They have a lot of work to make up for it, but they seem to be >>>>>>> slowly doing it while making sure they don't lose their core business >>>>>>> with >>>>>>> FX and end-to-end clients. >>>>>>> >>>>>>> I will certainly be looking at H14 as soon as some space for it in >>>>>>> the stash of stuff I need and want to do before clears up :) >>>>>>> >>>>>>> >>>>>>> On Fri, Jan 16, 2015 at 6:45 PM, Jordi Bares Dominguez < >>>>>>> jordiba...@gmail.com> wrote: >>>>>>> >>>>>>>> May I ask you to elaborate the “complex character rigging and tuned >>>>>>>> deformation”, I may be missing something. >>>>>>>> >>>>>>>> To start with you have muscles in Houdini which you don’t, let >>>>>>>> alone FEM simulations and a universal physics engine to cope with >>>>>>>> pretty >>>>>>>> sophisticated things… >>>>>>>> >>>>>>>> Certainly it is easier in Softimage and more artist friendly to >>>>>>>> setup but I see the rigging side as one very strong point. >>>>>>>> >>>>>>>> If you are talking about screen space corrections, blend shapes and >>>>>>>> advanced contact collision its certainly doable with the toolset. >>>>>>>> >>>>>>>> :-| >>>>>>>> >>>>>>>> thx >>>>>>>> jb >>>>>>>> >>>>>>>> >>>>>>>> On 16 Jan 2015, at 16:59, Raffaele Fragapane < >>>>>>>> raffsxsil...@googlemail.com> wrote: >>>>>>>> >>>>>>>> It's only true for some definitions of rigging. >>>>>>>> If you need proceduralism of course it does spectacularly well, and >>>>>>>> assets are simply best of breed in the industry and have been for >>>>>>>> years, >>>>>>>> end of story. >>>>>>>> For the hand-crafted complex character rig and tuned deformation >>>>>>>> kind of job though, no, it's not nicer than Soft, and I'd be hard >>>>>>>> pressed >>>>>>>> to make an argument for it over Maya (which I consider pretty bottom >>>>>>>> barreling already without a ton of custom work). >>>>>>>> >>>>>>>> Some of the upgrades in H14 and some of the future roadmap do bode >>>>>>>> well for that though. >>>>>>>> >>>>>>>> On Fri, Jan 16, 2015 at 1:28 PM, Gerbrand Nel <nagv...@gmail.com> >>>>>>>> wrote: >>>>>>>> >>>>>>>>> Well I say nicer, because there are allot of toys to play with. >>>>>>>>> I think rigging is the part where you need a non destructive >>>>>>>>> procedural work flow the most. >>>>>>>>> In Maya it feels like you have to make damn sure you are done with >>>>>>>>> step A before moving onto step B. >>>>>>>>> Houdini is flexible to the point where you become reckless with >>>>>>>>> your work flow :) >>>>>>>>> Bit more complex when you get started, but worth it. >>>>>>>>> The auto rig at the very least doesn't break like the soft one >>>>>>>>> used to in 2011 :) >>>>>>>>> G >>>>>>>>> >>>>>>>>> >>>>>>>>> On 16/01/2015 14:08, Mirko Jankovic wrote: >>>>>>>>> >>>>>>>>> Riggin nicer then Soft? >>>>>>>>> Will have to check it out then.. In maya rigging and enveloping is >>>>>>>>> huge crap and biggest reason that I don't wanna ago back int othat >>>>>>>>> hell at >>>>>>>>> first place. >>>>>>>>> >>>>>>>>> On Fri, Jan 16, 2015 at 12:11 PM, Gerbrand Nel <nagv...@gmail.com> >>>>>>>>> wrote: >>>>>>>>> >>>>>>>>>> After trying to learn maya for about 6 months, learning houdini >>>>>>>>>> is a breath of fresh air!! >>>>>>>>>> It is not softimage, but I think its the only thing that will >>>>>>>>>> come close to the flexibility and power of soft for small studios and >>>>>>>>>> freelancers. >>>>>>>>>> Once you get into it, It is even more power. >>>>>>>>>> I tried learning it about 2 years ago, and gave up because I >>>>>>>>>> thought my time would be better spent getting better in soft (the >>>>>>>>>> future >>>>>>>>>> was still bright back then) >>>>>>>>>> Back then it seemed complicated, but after dealing with maya, it >>>>>>>>>> feels sooo much friendlier. >>>>>>>>>> The way I see it, you get the operator stack, and ice tree, all >>>>>>>>>> in one place, the network view >>>>>>>>>> So its one thing to learn. >>>>>>>>>> In Maya I feel like I have to learn new software every time I do >>>>>>>>>> something else. >>>>>>>>>> Rigging I found nicer than soft, and the animation editor in >>>>>>>>>> houdini feels like a polished version of the soft one. >>>>>>>>>> Houdini engine is still blowing my mind.. like it doesn't stop!! >>>>>>>>>> At $300 you cannot ignore this as a piece of your pipeline! >>>>>>>>>> I'll probably do allot of work in maya because I need to fit into >>>>>>>>>> teams of Mayans, but with the houdini engine, I can do the work in >>>>>>>>>> the >>>>>>>>>> software best suited for it, without forcing the rest of the team to >>>>>>>>>> conform. >>>>>>>>>> G >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> On 16/01/2015 12:08, Mirko Jankovic wrote: >>>>>>>>>> >>>>>>>>>>> modeling and character riga nd animation wise it is I assume >>>>>>>>>>> sitill nt as suser friendly as SI right? >>>>>>>>>>> how us ievrall generalist and smalls tudio experience? >>>>>>>>>>> SI is more or less out of the box great steramlined solution.. >>>>>>>>>>> >>>>>>>>>> >>>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> -- >>>>>>>> Our users will know fear and cower before our software! Ship it! >>>>>>>> Ship it and let them flee like the dogs they are! >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>> >>>>>>> >>>>>>> -- >>>>>>> Our users will know fear and cower before our software! Ship it! >>>>>>> Ship it and let them flee like the dogs they are! >>>>>>> >>>>>>> >>>>>>> >>>>>> >>>>>> >>>>>> -- >>>>>> Our users will know fear and cower before our software! Ship it! Ship >>>>>> it and let them flee like the dogs they are! >>>>>> >>>>> >>>>> >>>> >>>> >>>> -- >>>> Our users will know fear and cower before our software! Ship it! Ship >>>> it and let them flee like the dogs they are! >>>> >>> >>> >> >> >> -- >> Our users will know fear and cower before our software! Ship it! Ship it >> and let them flee like the dogs they are! >> > > -- Our users will know fear and cower before our software! Ship it! Ship it and let them flee like the dogs they are!
