Matt I'm guessing what triggered your response was the word "slow". Notice that I was mentioning me as an instruction creator being slower with code to get to what I wanted not execution performance wise as nodes give me that super power: Do more mistakes faster which is helpful to path-find an algorithm if you're extremely smart and educated :D. It also didn't help the briefing was not crystal clear to me from the start so I invested time on several fronts, that was quite rich experience even though time was wasted. But next time I need them, I'll rename that to "investment" :)
But at the moment the goals are clear and I nailed them with all the inputs I received here too, so I'm thankful. https://gfycat.com/FearfulRemorsefulAnaconda Like I said, I got that C++ course and will dive into it, but at the moment I would not have produced anything meaningful for the task. Thanks again! Pedro On Fri, Apr 22, 2016 at 11:29 PM, Matt Lind <[email protected]> wrote: > Um....I don't think your assessment is correct. > > The SDK is single threaded, but in this context ICE is too. Past tests > have > shown that ICE performs ~30% slower than the SDK in single threaded tasks > because of the additional overhead with nodes/tree evaluation. If you can > parallelize the task, then the game changes, but with all the issues you've > run into, that's not likely to happen. > > You don't have to write the operator in C++. Scripted operator will also > work and be less hassle. In fact, you could've been done with it already. > If you have to deliver, the amount of time it takes to find the ICE > workaround (and a reliable one) will exceed the amount of time it would've > taken to write the operator using the standard SDK. > > do what you want. > > > Matt > > > > Date: Fri, 22 Apr 2016 13:48:13 +0100 > From: pedro santos <[email protected]> > Subject: Re: Distribute between knots? Know the percentage position of > a knot? > To: Softimage Mailing List <[email protected]> > > Will do Matt, one day. Got that C++ plus course I referenced. Seems great. > But now I have to deliver something x) I apologize if this is cringy for > you!! Yes the bits you talk about are the most annoying for me, building > the algorithms in straight up code is already a bit slower but the SDK part > is going beyong solving the problem at hand many times, and these are > usually handled for you with node. Hail nodes! :D I'll meet that C++ > bipolar madame soon :) > > Was able to re-purpose "Curve Distance to Curve U" from Dual Curve Deform > by Antton Tapani, to make my dual curve conform along curve length. > > Thanks everyone (!) for the inputs, they were helpful as what was requested > of me kept updating! Here's the result so far: > https://app.box.com/s/tcvzwph4b0j2jf48yqcuscg08lt17gtq > > Cheers > Pedro > > ------ > Softimage Mailing List. > To unsubscribe, send a mail to [email protected] > with "unsubscribe" in the subject, and reply to confirm. > -- *------------------------------[image: http://i153.photobucket.com/albums/s202/animatics/probiner-sig.gif]Pedro Alpiarça dos Santos >> http://probiner.xyz/ <http://probiner.xyz/> <http://probiner.x10.mx/>*
------ Softimage Mailing List. To unsubscribe, send a mail to [email protected] with "unsubscribe" in the subject, and reply to confirm.
![http://i153.photobucket.com/albums/s202/animatics/probiner-sig.gif]Pedro](http://i153.photobucket.com/albums/s202/animatics/probiner-sig.gif]Pedro){kind=link}