Fork me on GitHub
#datavis
<
2015-12-17
>
meow17:12:34

@kephale: how's the mesh stuff going?

meow18:12:45

I'm getting the impression that toxi isn't terribly interested in having collaborators on his library.

meow18:12:00

Which is a shame, but I can't let that keep me from developing the tools that I need to meet my own needs.

kephale18:12:56

well, it certainly has been a one-man army effort up to this point, so I suspect he has something in mind for his baby, probably even partially driven by the demands from the http://thi.ng workshops he has been giving

kephale18:12:30

i’m under a journal revision deadline at the moment, so i’m still stuck in C++ land with my existing mesh simulations

kephale18:12:48

that should change more around new years when the revisions get sent in

kephale18:12:17

we’ve actually been playing a bit with @eggsyntax cljs-mathbox implementation as well

kephale18:12:44

just to get a faster cycle of iterations during development

kephale18:12:45

a thought that is floating around here is to take a pass at having the cljs-mathbox implement some of the protocols from http://thi.ng, and see how that experience goes

kephale18:12:04

although that is most likely to be for graphing/plotting utilities first

kephale18:12:18

so targeting the geom.viz namespaces first

meow18:12:20

My current focus is on the general area of "polygon mesh subdivision modeling" with an eye towards learning all the features of Wings3D and MeshLab and implementing similar functionality in Clojure. I'll keep building on the protocols provided by the http://thi.ng geom mesh as I have been. But at some point it might make sense to look at the underlying implementation to see if it can be improved. To the extent that that overlaps with your need for real-time mesh simulation and deformation I'd love to collaborate.

eggsyntax18:12:03

@kephale: ooh, interesting thought!

meow18:12:21

No hurry on any of this. Just trying to keep everyone that's interested in the loop.

eggsyntax18:12:34

Everything still going smoothly with cljs-mathbox, no hidden pain points as yet?

kephale18:12:22

@eggsyntax: no pains at all, but i admit i’ve been nearly 24/7 on the simulations i’m dealing with

kephale18:12:35

2 of us here are poking at it now though

eggsyntax18:12:56

Cool. Keep me posted simple_smile

eggsyntax18:12:12

& if you find any bad choices, I’m likely down to fix ‘em.

kephale18:12:58

yes, i’ll be vocal if i run into anything, but i’m optimistic based on my read throughs

meow18:12:03

@kephale: haven't seen that before, and the web page won't load for me.

kephale18:12:16

@meow: it might be a webgl verison thing, it works in chrome

kephale18:12:30

its, if not all, most of meshlab in javascript

meow18:12:43

I'm on chrome and do a lot with webgl. Bummer.

kephale18:12:01

i’ve had some slowdowns for meshes with millions of verts

kephale18:12:14

but for more concise meshes its quite comfy

kephale18:12:57

i’m a clojurescript neophyte at the moment, but i can’t help but to wonder if there is a way of using meshlabjs as a dependency

kephale18:12:15

to avoid rewriting the meshops from MeshLab

meow18:12:15

I can see the code in their github repo.

kephale18:12:17

i definitely need a lot of operations from meshlab, but need to get a better idea about which ones are useful by manually poking around, which is terribly slow

kephale18:12:39

it would be a little annoying that it would restrict to clojurescript usage only

kephale18:12:56

but that wouldn’t be the worst tradeoff

kephale18:12:20

clearly pure clojure meshops are most ideal

meow18:12:22

I'm learning a lot about mesh ops by writing them myself, plus I usually end up not liking how others have implemented this stuff so I don't mind building it all up from scratch.

meow18:12:25

I'll have to look at how they structure their meshes and how it relates to the structure and protocols that toxi has defined.

kephale18:12:08

mmm, what i really want though is to screen a handful of meshlab functions sampling a bunch of parameters, but it would be preferable to not have to write the functions that won’t be useful in the long run, etc..

kephale18:12:10

that kind of thing

meow18:12:51

not really sure what you mean

kephale18:12:51

i already have a bunch of ops from my c++ code to port already : X

kephale18:12:00

so i want to avoid porting functions that i wont use

kephale18:12:08

like vertex sampling for example

kephale18:12:29

only some of those are useful for me, but i dont reallllly know which until i test with a range of sampling parameters

meow18:12:45

oh, I see

kephale18:12:04

and sure one could write scripts to run meshlab with them

kephale18:12:20

but if we can bind to an existing meshlab implementation and just call from clojure

kephale18:12:23

then phew, all the better

meow18:12:30

I don't like the meshlab scripting stuff at all.

meow18:12:14

I'll look at the js version and see what it looks like.

kephale18:12:16

cool, i’d be interested to hear your take on linking to it

meow21:12:19

Yep, the new Slack image viewer is way bigger. 😉

meow22:12:55

@kephale: So the majority of meshlabjs is C++ transcompiled to js

meow22:12:12

I don't have much interest in working with it.