Jump to content

Rule of Thumb on When to Favour Mantra over Arnold


up4

Recommended Posts

Hi,

 

Now that the 1.0 version of the HtoA plugin is available, I would like to know if there are production-proven rule of thumb for choosing Mantra over Arnold for rendering Houdini scenes (or Houdini Alembic output files). Ideally without starting a flame war. I'm curious to hear from those who had the chance to use Renderman's RIS framework as well.

 

Regards,

 

Vincent

Link to comment
Share on other sites

I saw this interview when it came out. And I remain convinced that the very goal of Solid Angle through their notion of "artist time" economy is the elimination of the necessity of secret rendering tricks. And that less bias means less saucy tricks. And that unbiased means no sauce at all, only cinematography be it in-camera or not.

 

Anyways, until we get realtime PBRT (and we are getting closer with NVIDIA's newly released Maxwell GPUs), I know there is a place for bias and rendering tricks. And that would be a reformulation of my question: with Arnold around, (exactly) when is the use of Mantra and its rendering tricks still required (and for how long) in the Houdini ecosystem for photorealistic cg in movies?

Link to comment
Share on other sites

I'm not using Arnold yet but AFAIK they are equal rendering technologies. 

 

Can you explain the 'tricks' bit?  ie. Arnold has no caustics, but Mantra does by using a 'trick'

 

Arnold has simplified the UI but Mantra shaders are a simple UI too.

Link to comment
Share on other sites

I haven't used Arnold Houdini since it's very early conception, but know people who use it in production.

The general consensus seems to be that using "ass" files is useful for setting up pipeline between houdini artists and maya lighters.

Arnold handles A LOT of points with instancing and hard surfaces/points are a no brainer.

Mantra volumes are simply better.

 

I don't like not having complete control of the shaders and love tricks... ;)

C

  • Like 1
Link to comment
Share on other sites

To get some reasonable rendertimes I'm using some tricks in Arnold (for XSI) too (separating environment reflection from env. lights), turning off some ray types for some objects based on materials/details etc. to speed up rendering.

 

Ass files are good for compatible pipeline and speeding up scene compilation/conversion time.

Arnold needs simple shader networks to benefit from high speed.

Link to comment
Share on other sites

So all of this kind of confirms the impression that one currently gets: the balance between lookdev/lighting effort vs. render time vs. accuracy is something that Arnold is generally nailing best at the moment. And one should look for Mantra's output at lookdev time for specific situations like things requiring bidirectional path tracing (like caustics) or for which Mantra was designed to be better at being Houdini's internal renderer (mostly volumes, but also maybe also anything FXish). And deep-composite everything later in Nuke with Peregrine Labs' Bokeh. Or something like that.

Link to comment
Share on other sites

Mantra isn't bidirectional afaik. I would dig deeper and look at sss features too. Indirect sss in particular.

It'd be worth doing a step by step feature parity test to really nail this. Thus far in testing Arnold for houdini is a v1 for sure, at least on OS X. Ipr makes houdini jerky and changing lights or textures seems to require you to stop start Ipr iirc.

Link to comment
Share on other sites

So, how does Mantra do caustics if it isn't bidi. Like Arnold?  ;) https://support.solidangle.com/display/mayatut/Refractive+Caustics

 

As Marcos says in the interview, caustics look good and all, but are rarely required in production (the moonlit interior pool with white walls scenario).

 

I agree that ipr integration is broken (I also tested on Mac only for now). But I have access to Linux and Windows machines too.

 

Can we agree on benchmark hip files (more evolved than the Cornell box) for which there would be both Mantra and Arnold versions and we test for speed and post screenshots?

Link to comment
Share on other sites

One big reason for using Mantra is that it's a hell of a lot cheaper than arnold.

 

You can also get the same result with Mantra as Arnold if the artist knows what he is doing. 

 

SESI just needs to spend some more time on usability and documentation so that people can easier get started. They also need to stop making up their own names and just call stuff what artist are used to it being called. Reflection and refraction angle makes sense if you have studied a lot of math but for your average vray/arnold ligher they have no idea. They are looking for glossiness/roughness.

Link to comment
Share on other sites

Im going to start with informal testing to get familiar, then more structured.

Re naming. No way ;) I hate the idea of conforming to other so called standards. Angle is as complex as other terms upon analysis.

Link to comment
Share on other sites

In my opinion the angle is clearly more obvious than the "usual" glossiness/roughness/smoothness values. The ranges tend to differ between applications, and it's often quite hard to find out what the values actually correspond to in the real world. And when you do find out they are often defined as something like variance of microfacet normal vectors (square of standard deviation), which is far more abstract than the angle.

 

Then again, being the odd one out does incur a price - but with Houdini we're the odd one out in any case... :)

Link to comment
Share on other sites

So, with all being said. I tried some examples from the Arnold support site. And the least I can says is that the integration into Houdini is not seamless right now. For the time being, I will stick to Matra and wait a little longer for the Arnold plugin to mature a little more and compare the two.

Link to comment
Share on other sites

So, with all being said. I tried some examples from the Arnold support site. And the least I can says is that the integration into Houdini is not seamless right now. For the time being, I will stick to Matra and wait a little longer for the Arnold plugin to mature a little more and compare the two.

 

Hi up4, I'm the main developer of HtoA.

 

I won't deny the plugin has some rough edges, and it will take some time for the integration to be as good as Mantra, obviously. For this reason, I'd be most interested in getting feedback, good or bad, to assess where we should focus our efforts. 

 

Could you please detail the issues you encountered during evaluation? Of course, I extend this request to anybody who wants to share any feedback on HtoA and Arnold.

 

I don't want to hijack this very interesting topic, so you can PM if you prefer.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...