Jump to content
FabricPaul

Fabric for Houdini is in testing

Recommended Posts

Hi guys - it's taken a bit longer than expected, but we're now testing Fabric for Houdini. Guillaume Laforge has been working on this and has now opened up the repo - we're going to be maintaining and supporting it from here on out so this is very much an 'official' integration.

 

A couple of videos of current progress: 

https://vimeo.com/126371420

https://vimeo.com/127362431 (most recent)

 

If you want to give it a spin and help us with testing, sign up is here: http://fabricengine.com/canvas-testing-program/ 

 

One thing we're trying to work out is where it's helpful for you guys, so we're really keen to get your help and feedback.

 

Looking forward to hearing from you soon.

 

Cheers,

 

Paul

  • Like 3

Share this post


Link to post
Share on other sites

It looks interesting Paul, I am curious what the advantages are with using Splice over using Vops?

 

I look forward to testing it.

Share this post


Link to post
Share on other sites

I wouldn't say there are necessarily advantages within Houdini - more that the portability of Fabric tools between multiple DCCs has value. One area that consistently comes up is rigging, which is something we are pushing very hard. There'll be some demos online soon showing what we covered at FMX that should give you an idea of what we're up to.

  • Like 1

Share this post


Link to post
Share on other sites

I wouldn't say there are necessarily advantages within Houdini - more that the portability of Fabric tools between multiple DCCs has value. One area that consistently comes up is rigging, which is something we are pushing very hard. There'll be some demos online soon showing what we covered at FMX that should give you an idea of what we're up to.

Hi Paul,

 

Do you think is easy for houdini users to integrate FE in the pipeline? Is it much different from vops?

Share this post


Link to post
Share on other sites

I think I will play safe and say 'we shall see' - Fabric is certainly well within the technical comfort zone of most Houdini users and offers similar programmatic depth. In terms of similarities to vops, I think you guys are better placed to comment than us.

Share this post


Link to post
Share on other sites
Guest tar

Hey Paul - out of interest is Fabric fully optimised as it's all modern code?  i.e. do you have any unimplemented tricks that can make it go faster later on.

 

Thanks!

Share this post


Link to post
Share on other sites

Hi Marty - there's always more optimizations to be had, both from our side and also from LLVM. We also have our ongoing work on GPU compute: http://on-demand.gputechconf.com/gtc/2015/video/S5553.html

 

Down the line there's also things like distributed compute and server-side compute, but that's something for next year and beyond. We've designed with those things in mind though, so it'll be cool when we get to it :)

  • Like 1

Share this post


Link to post
Share on other sites

The videos look cool. Are there any benchmarks that show how fast it is to create, delete, modify geometry, etc in Fabric? And how strong the capabilities of the Fabric library to mesh particle fluids for example? Do these have to be written by the users?

Share this post


Link to post
Share on other sites

The Fabric for Houdini project is still in development, the best way for you to answer these questions is to join the testing program: http://fabricengine.com/canvas-testing-program/

 

Remember that Fabric is a platform for building tools, not a complete application. So we provide lots of components for building a pretty wide range of things, but you do actually have to put them together yourself. 

Edited by FabricPaul
  • Like 1

Share this post


Link to post
Share on other sites

People are comparing Splice to vops, which is comprehensible but not very accurate.

Vops are living (dont want to write restricted) to their context: points, prims or shaders. 

As in in Splice youre able to create your own context. Say, a Character or an Asset could be one. You decide what goes in, is processed and goes out.. Its hard to explain without any examples, but I am sure that these will be coming.

Everything is portable to different applications, plattforms aaaaand hardware. It will be always as fast as possible because of the on runtime compilation. Same goes for vex of course.

 

It wont replace any functionality of Houdini, I dont think there is a need for that. Hou is dope as it is, (minus the occasional bug ;) ).

Share this post


Link to post
Share on other sites
Guest tar

Out of interest is anyone testing Fabric against Houdini's performance?  

 

I'm seeing great performance in the Fabric demo video's but then I can get the same speed in Houdini by tweaking some of the default tools. i.e. Cloth Proxy can be brought up to speed by using a Point Deform etc.

 

Thanks!

Share this post


Link to post
Share on other sites

I am quite interested in Fabric Engine as a platform as a whole, but it is a significant investment for a studio to embrace that. Dneg and MPC seem to be heading that way.
We will test this at Method in various departments over the next year and evaluate if it is something we want to pursue.

Coming from a houdini toolbuilding mindset is quite powerful to do certain things with it. Personally I am quite interested in the procedural rigging aspect of fabric engine, which can then lead into crowds and locomotion type behavior.

You can do (limited) fx things with it as well, but Houdini is already really strong at that. It is mainly the rest of the pipeline that needs to catch up to some more node based parallel processing, this is where I think Fabric could really shine. Especially procedural modeling/layout, procedural rigging, crowds and ai systems, rendering and data management for rendering, signal processing and real-time interaction, previz,... could benefit a lot of Fabric engine.
 

Also how you test Fabric against Houdini will be interesting. Simply because if you are using Houdini 'the right way', you should be using data acceleration structures like openVdb and pointclouds. Those kind of data structures are not really currently implemented. And sure, you can write your own code to implement them, but there is not always enough time or expertise in production to do that. I fully expect in a few years when more of these accelleration structures are implemented that people will be doing amazing things with it.

 

I also think that by lowering the cost of entry for an individual or a small studio (up to 50 licenses to test it out) they are trying to foster adoption. Global site wide licenses will be a significant cost to a studio. It is exciting tech.

Share this post


Link to post
Share on other sites

For me there is only one area where FE is interesting: realtime performance.

As a data processor, for what we have to do in VFX, I think Houdini is the best platform.

Share this post


Link to post
Share on other sites

Same here. I'd be really very interested in Fabric Engine for making stand-alone tools. In Houdini - not so much. I feel H has all I need and even more I haven't yet mastered.

However, the idea of such high-performance rigging is quite appealing!

Share this post


Link to post
Share on other sites

Hi Alexey,


 


I saw your youTube videos and the Houdini modeling tool like the zModeler.


Is there is any way I can use it and give you some feed back :)


please let me know.


 


I posted this in another of yours. I am copying here in case you miss that :)


Thx


 


Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×