Even with no C++ knowledge you should be able to take different routes around it, just to see if the bottleneck is specifically in one of the wrappers or far enough upstream. Give it a shot.
On Tue, May 21, 2013 at 1:40 PM, Enrique Caballero < [email protected]> wrote: > Thanks raf and everyone for the advice, its really helpful. > > > I will now reduce their ability to key scaling on the majority of the rig. > I will also try to code my way around it, problem is I don't know c++ so > I'm stuck with python which I doubt will be able to save me when it comes > to slow keying. > > > I did test in 2014 btw, and it is a bit faster, but its still massively > slow. > > I've sent the scene to Autodesk, we still have a year on Subscription so > hopefully they can send us a QFE > > > > > On Tue, May 21, 2013 at 11:36 AM, Raffaele Fragapane < > [email protected]> wrote: > >> Sounds like a regression. >> I've had rigs with controls in the hundreds of objects with quite a few >> added properties and custom parameters, adding up to packs of thousands of >> keyframes at a pop. It's never been blazing fast, even with rig-centric >> dedicated commands, but I would have been skinned alive if it took more >> than a second or two to key out a couple thousand curves. >> >> Consider trying to write your way around it to see how pervasive it is, >> and if not too far reaching possibly stop them from staring daggers at you. >> >> Keying a ref model will always be slower, there's no two ways about it >> when you have to deal with delta tracking rather than simply adding data >> somewhere, but shouldn't be in that measure, it should be just a few >> percentage points at the most. >> >> >> On Tue, May 21, 2013 at 1:32 PM, Enrique Caballero < >> [email protected]> wrote: >> >>> the keys are being set just by pressing the K key and the keying mode >>> set to "Key all Keyable" >>> >>> this is the command that gets spit out >>> >>> Application.SaveKeyOnKeyable() >>> >>> >>> >>> On Tue, May 21, 2013 at 11:30 AM, Enrique Caballero < >>> [email protected]> wrote: >>> >>>> thanks guys, >>>> yep I've already sent the scene to Softimage. Its definitely a >>>> Softimage 2013 Sp1 issue though and not scene related. >>>> >>>> The parameters that the animators can key is already fairly limited as >>>> I'm pretty careful with keyable parameters. but I will strip down what i >>>> can for now. >>>> >>>> It is slow whether or not the Fcurve editor is open. Basically setting >>>> a key on a gear rig referenced model with 180 objects takes 6 seconds, a >>>> local model is instant. >>>> >>>> I've stripped down our workgroup to nothing but gear, same problem. >>>> >>>> Really quite distressing! The animators are giving me dirty looks! >>>> >>>> >>>> >>>> >>>> >>>> >>>> On Tue, May 21, 2013 at 11:21 AM, Raffaele Fragapane < >>>> [email protected]> wrote: >>>> >>>>> Have you tried changing how the keys are set? >>>>> 150 objects with the entire local transform set isn't that many >>>>> curves, we have had issues but that's with thousands piled up on more >>>>> thousands. >>>>> Lastly, is that with the FCurve editor open or not? >>>>> >>>>> I suggest you send the scene to Soft if it can be packaged up. >>>>> >>>>> >>>>> On Tue, May 21, 2013 at 1:16 PM, Enrique Caballero < >>>>> [email protected]> wrote: >>>>> >>>>>> well its a gear rig so there are a fair but of custom parameters. but >>>>>> not an obscene amount. And as far as I know, no parameters driven by ICE. >>>>>> although I am using the dual quaternion skinning compound for the >>>>>> envelope. >>>>>> I've already tested with this removed and it wasnt the issue. >>>>>> >>>>>> I have tested in 2014 and it is better, by like 30 % but its still >>>>>> massively slower than it would be if the model was in Local Mode. >>>>>> >>>>>> Seeing as how well Softimage's referenced model and Deltas work >>>>>> together, it is a major reason we use Softimage here at this studio, I >>>>>> would expect the speed of keyframing to be very fast on referenced >>>>>> models, >>>>>> it should be just as fast I dare say, otherwise its a massive hit on >>>>>> their >>>>>> usability. >>>>>> >>>>>> I have also tested with other rigs as well, Gear rigs, the malcolm >>>>>> rig, and rigs of my own making, its a pretty major speed difference when >>>>>> setting and manipulating keys on referenced models vs local models. >>>>>> >>>>>> >>>>>> As far as how many fcurves. well... A lot, its an entire rig, its >>>>>> like 150 objects, and we are keying their entire srt. as the animators >>>>>> are >>>>>> currently blocking animation, so they key the entire character at once. >>>>>> >>>>>> >>>>>> 2014 is a bit faster, but not massively so. And we don't have the >>>>>> luxury of moving our pipeline to 2014 right now, our tools are written >>>>>> for >>>>>> 2013 at the moment and we are going through a major tools development >>>>>> phase >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> On Tue, May 21, 2013 at 8:24 AM, Raffaele Fragapane < >>>>>> [email protected]> wrote: >>>>>> >>>>>>> There are two issues, a regression, which Matt does a good job of >>>>>>> pointing out and that should be fixed in 2014 (to my knowledge, but >>>>>>> haven't >>>>>>> tested), and other things we found out when a mix of ICE and custom >>>>>>> parameters are involved (which is not related to ICE slow at setting >>>>>>> them, >>>>>>> which was addressed in 2013 already, I believe). >>>>>>> >>>>>>> >>>>>>> On Tue, May 21, 2013 at 9:28 AM, Jeremie Passerin < >>>>>>> [email protected]> wrote: >>>>>>> >>>>>>>> I heard the same thing here, and also heard it's much faster in >>>>>>>> 2014. Have you tested that ? >>>>>>>> >>>>>>>> >>>>>>>> On 20 May 2013 16:20, Raffaele Fragapane < >>>>>>>> [email protected]> wrote: >>>>>>>> >>>>>>>>> How many custom attributes and specifically some feeding into ICE >>>>>>>>> do you have? >>>>>>>>> And how many FCurves at a time are we talking about? >>>>>>>>> >>>>>>>>> We encountered several related issues (and occasionally solved or >>>>>>>>> had confirmation of them, and some QFEs that helped a lot) >>>>>>>>> >>>>>>>>> >>>>>>>>> On Mon, May 20, 2013 at 8:57 PM, Ivan <[email protected]>wrote: >>>>>>>>> >>>>>>>>>> Thanks! >>>>>>>>>> >>>>>>>>>> Sent from my iPhone, please excuse for typos. >>>>>>>>>> >>>>>>>>>> On 20 May, 2013, at 6:02 PM, Enrique Caballero < >>>>>>>>>> [email protected]> wrote: >>>>>>>>>> >>>>>>>>>> Hey Ivan, >>>>>>>>>> Thank you, Yep I do, I will send it to you in a few minutes, >>>>>>>>>> just packaging up the referenced models >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> On Mon, May 20, 2013 at 5:54 PM, ivan tay < >>>>>>>>>> [email protected]> wrote: >>>>>>>>>> >>>>>>>>>>> Hi Enrique, >>>>>>>>>>> >>>>>>>>>>> Do you have a scene file for this ? >>>>>>>>>>> >>>>>>>>>>> Thanks >>>>>>>>>>> Ivan >>>>>>>>>>> Email : [email protected] (please remove nospam >>>>>>>>>>> from email) >>>>>>>>>>> >>>>>>>>>>> >>>>>>>>>>> On Mon, May 20, 2013 at 5:33 PM, Enrique Caballero < >>>>>>>>>>> [email protected]> wrote: >>>>>>>>>>> >>>>>>>>>>>> Sorry this is in softimage 2013 sp1 >>>>>>>>>>>> >>>>>>>>>>>> >>>>>>>>>>>> On Mon, May 20, 2013 at 5:15 PM, Enrique Caballero < >>>>>>>>>>>> [email protected]> wrote: >>>>>>>>>>>> >>>>>>>>>>>>> Hey everyone, >>>>>>>>>>>>> I am running into a distressing problem. >>>>>>>>>>>>> >>>>>>>>>>>>> An animator just came up to me and complained about it being >>>>>>>>>>>>> very slow to key their animation. >>>>>>>>>>>>> >>>>>>>>>>>>> When they select all of the controls on the rig and drag the >>>>>>>>>>>>> keys around or simply set a key there is a fairly major delay. >>>>>>>>>>>>> >>>>>>>>>>>>> I just did some testing and verified that it is quite slow. >>>>>>>>>>>>> >>>>>>>>>>>>> >>>>>>>>>>>>> *I made the referenced model local, and the speed of setting >>>>>>>>>>>>> or manipulating keys increased quite dramatically.* >>>>>>>>>>>>> >>>>>>>>>>>>> We are using GEAR rigs on this project, which I've never had >>>>>>>>>>>>> an issue with before, and I don't really think its gears fault. >>>>>>>>>>>>> Maybe its >>>>>>>>>>>>> something I layered ontop. >>>>>>>>>>>>> >>>>>>>>>>>>> Has anyone run into this problem before? Our pipeline is >>>>>>>>>>>>> heavily dependent on referenced models and I have no plans to use >>>>>>>>>>>>> local >>>>>>>>>>>>> models in our pipeline at all. I'm hoping that its just one >>>>>>>>>>>>> little >>>>>>>>>>>>> oversight that is causing this. >>>>>>>>>>>>> >>>>>>>>>>>>> Any insight would be appreciated. >>>>>>>>>>>>> >>>>>>>>>>>>> -Enrique >>>>>>>>>>>>> >>>>>>>>>>>> >>>>>>>>>>>> >>>>>>>>>>> >>>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>> -- >>>>>>>>> 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!
