Anybody here have a view on Houdini vs Katana for lookdev/lighting workflows.

[Jonathan Moore]

Hi all,

I have a client (an advertising network with their own production facilities) that currently have a pipeline involving Maya and Houdini with RenderMan and Redshift as rendering options. There's a smattering of Max and Modo for asset creation but that's beyond the scope of my enquiry.

We're currently going through the process of deciding whether Katana would be an effective tool to add to their pipeline as their are moving into longer form branded content as well as their existing advertising output.

I have a major cognitive bias going into this assessment that Houdini can be used for Katana style deferred rendering workflows as well as it's FX bread and butter. Introducing Katana will come at a considerable cost so I'm wondering what others think and feel about Katana, particularly if they've already gone through a similar thought process. It doesn't matter whether you use Katana in you pipeline (or have used it in the past) I'm just looking for any considered views ref Katana benefits.

And Jordi, if you're reading this, I would love your take on Houdini as a lookdev/lighting toolset as I understand that's exactly how you use it at Framestore.

Funnily enough, the more deeply I research this, the more I'm reminded how ahead of the game the Softimage team were. The whole models workflow (and underlying philosophy) was incredibly flexible as well as powerful. Sure it had some gnarly aspects much like any referencing system (from what I hear, Katana it littered with these referencing cul-de-sac's too). 

My internal bias towards Houdini is that is has so many strengths with regard to deferred procedural loading, packed disc primitives etc etc, and to be frank, shading networks in Katana suck right now. Plus Houdini pretty much invented the nodal shading game with VOPs.

As a positive for Katana, I'm really impressed with the 3delight integration, and it's promise of seamless a seamless pipe with Maya (a necessary evil not a preferred choice). I've always had a soft spot for 3delight and the new OSL driven, artist centric presentation layer/UX is something that connects with my own thoughts about delivering flexible rendering power without the need to have all the wiring on show.

Apologies for the lengthy post. I'm hopping that one or two of you have gone through similar considerations as you've gradually planned your move away from Soft.

As ever, thanks in advance.

jm

[Tim Crowson]

I used Katana at MPC, and lately have been using Houdini at Method. Bear in mind that my use of these products has been for feature films, with medium to heavy shot content. I have not used them in any other context. Bear in mind also that both platforms (because that’s really what they are) require some degree of custom development to achieve efficiency in lighting (as I define efficiency, at least).

I far prefer Katana.

In my view Katana makes it FAR easier to manage scene data without losing your mind. K is also much more elegant in how it handles per-pass overrides. Houdini’s options for per-ROP overrides (on things that are not the ROP itself, which is vital to be able to do) are problematic for me, personally.

Katana also makes it much easier to read the state of things, simply by looking at the graph. Houdini’s paradigm presents you with a bunch of disconnected nodes that don’t seem to be related at all, forcing you to inspect parameters to see what is going on. You adapt to that, but it does create extra mental steps that have to be taken while working. One of my pet peeves is the single-line string field used in the Objects tab on ROPs. It’s a good deal of work to properly read that kind of field, even on mild shots. It’s just a space-delineated list of paths. Translating that into meaningful information takes more time than it should. 

Houdini’s takes are interesting, although the pros where I am never use them because of awful past experiences. And the few times I have tried to use them they bugged out and simply didn’t work reliably. Besides, at the conceptual level, I don’t agree with storing scene states (or overrides) abstracted from a ROP, *unless* you can combine them later. You wind up making one take per ROP, which then makes me wonder why they aren’t just stored on the ROP in the first place.

Katana make sure it incredibly easy, in my view, to not only visualize the data flow, but also to assetize the overrides themselves, for use elsewhere or in other Katana files, combined in any way you like.

On the lookdev and lighting fronts alike, Katana’s CEL statements absolutely demolish the equivalent syntax available in Houdini. CEL statements are simply more advanced and “smarter” in what they let you target within a scene graph.

For me, lighting especially comes down to efficient data management. In film it’s far more technical of a discipline than people think. The artistic part can be done pretty quickly. Managing how a shot is broken down into layers, in a way that makes responsible use of available resources, is the bigger challenge. And in my view Katana is the king here (though Image Engine’s Gaffer is very similar, from what I understand).

I have been using Houdini lately on Aquaman and I guess it’s the stress of production building up, but it’s really just getting on my nerves. Seems like there are far too many possible points of failure and bugs, unless you design a strong custom UX front end, and that’s a lot of work. Getting Katana up to production-ready status requires less development effort, in my view.

But there is that insane Foundry price tag...

I am curious to hear from others, because my exposure to Houdini is admittedly limited.

------
Softimage Mailing List.
To unsubscribe, send a mail to softimag...@listproc.autodesk.com with "unsubscribe" in the subject, and reply to confirm.

[Ivan Vasiljevic]

I know this is totally out of your focus but have you tried Octane Standalone?

You also do have it as Houdini plugin and it can easily transfer complete scene data to a standalone app for further look dev/rendering, with a single click of a button...

Just my 2ct.

Cheers.

Ivan

--

Ivan Vasiljevic
-

Lighting TD

Founder, Digital Asset Tailors
-
web:    http://digitalassettailors.com/

email:  iv...@digitalassettailors.com

[Jonathan Moore]

Thanks Tim, I was really hoping that you'd have some thoughts on the matter.

Gaffer is a real temptation especially if the client decides to give on the power of RenderMan for the artist centric workflow of 3Delight (which is really seems to have come of age recently). Unfortunately it's not RenderMan capable, AppleSeed, Arnold and 3Delight only. But then there's the equally insane price tag... of $0! Gotta love the Linux open source scene. ;)

Good to hear your take on Katana vs Houdini. My client isn't short of TD talent to customise either and currently do use a customised Houdini pipeline but at a far smaller (advertising) scale. 

I'm equally taken by the elegence of CEL statements and the manner in which flexible templates can be built for different aspects of the pipeline. My core question though is whether it's overkill for the scale of projects an advertising agency puts though it's internal production resource. If the pricing was closer to ADSK or Houdini Core (FX licenses are limited at their facility), it would make the decision a little easier.

[Tim Crowson]

Katana might be overkill for that. But it’s not entirely clear specifically what kind of work they need to do, specifically.

Also remember that Katana doesn’t *create* 3D content, it just renders it. Houdini would provide them with a full CG toolset if they need it. It would also allow them to develop elements and pass them off to lighting without having to worry about caching things out in a way that Katana likes.

[Tim Crowson]

Oh and Gaffer is very very beta.

[Angus Davidson]

Have you had a look at

https://www.isotropix.com/products/clarisse/look-development

Havent used it myself buy have heard very good things about it for lookdev from friends.

Kind regards

Angus

--
ICT Project Manager
Digital Arts
Wits School of the Arts
angus.d...@wits.ac.za
011 717 4683

---

From: softimag...@listproc.autodesk.com [softimag...@listproc.autodesk.com] on behalf of Tim Crowson [tcro...@gmail.com]
Sent: 05 July 2018 04:59 PM
To: Official Softimage Users Mailing List. https://groups.google.com/forum/#!forum/xsi_list
Subject: Re: Anybody here have a view on Houdini vs Katana for lookdev/lighting workflows.

This communication is intended for the addressee only. It is confidential. If you have received this communication in error, please notify us immediately and destroy the original message. You may not copy or disseminate this communication without the permission of the University. Only authorised signatories are competent to enter into agreements on behalf of the University and recipients are thus advised that the content of this message may not be legally binding on the University and may contain the personal views and opinions of the author, which are not necessarily the views and opinions of The University of the Witwatersrand, Johannesburg. All agreements between the University and outsiders are subject to South African Law unless the University agrees in writing to the contrary.

[Angus Davidson]

Octane stand alone is an amazing piece of software. We use it at Wits for our student renderings. Incredibly fast and powerful. Its what Shake would have been as a 3d renderer ;)

--
ICT Project Manager
Digital Arts
Wits School of the Arts
angus.d...@wits.ac.za
011 717 4683

---

From: softimag...@listproc.autodesk.com [softimag...@listproc.autodesk.com] on behalf of Ivan Vasiljevic [kleb...@gmail.com]
Sent: 05 July 2018 05:30 PM

To: Official Softimage Users Mailing List. https://groups.google.com/forum/#!forum/xsi_list
Subject: Re: Anybody here have a view on Houdini vs Katana for lookdev/lighting workflows.

[Jonathan Moore]

Katana might be overkill for that. But it’s not entirely clear specifically what kind of work they need to do, specifically.

The reason that Katana has come into the equation is that the client is looking for greater control on 'look' over and extended period of time. The nature of advertising has changed drastically over the last 10-15 years. Advertising budgets first saw a shift to online spend (which frankly wasn't very interesting from a production perspective) but the last 5-10 years have seen a shift into a range of 'experiential' content with long form filmic content distributed through various digital channels and experiential situations (attempting to avoid the words AR/VR here and oops just failed!). And within all this flux the lines have blurred between the creative idea space of advertising agencies and what multi-form content production facilities have been doing outside of their core 'VFX' output. Elements of the advertising industry are attempting to muscle back a bit of client relationship control they lost to the smarter production facilities.

So where does Katana/Houdini fit into this? One of the problems with all these emerging new channels for clients to spend their money is that consistency of the way brand assets are presented over time has been less than ideal. And this is a key area where advertising/branding agencies are strong, so anything that enables them muscle back as the key client creative partner is something seen as being a worthwhile investment. They're not creating feature length content, but over the same time period of a typical feature production, the client can often be creating more content for distribution through different channels. The challenges are not so very different to managing a major show from a project/production management perspective.

[Jonathan Moore]

Ref Clarisse Angus, I'm a massive fan. In many ways it's design philosophy is very similar to Katana. The downside is that it only comes with it's own renderer.

For environmental shots and matte painting it seems to be the go to solution and further than that, I think it's DNEG that have brought on board for all their broadcast output. But they also use Katana for the features pipeline.

Katana being Redshift friendly is a big plus point over Clarisse, and over the next 12-18 months, RenderMan, 3Delight and Arnold will have their own GPU solutions too, so that's another black cross Clarisse. And outside of indie licenses Clarisse is that much cheaper than Katana.

Great product none the less, I just wish they allowed 3rd party renderers, even though they may break many of Clarisse's biggest selling points such as the fluidly of the viewport with multi-million poly assets.

[Jonathan Moore]

Just to clarify Tim. I gave you a reason for the business buy-in for greater control of 'look'  through the pipeline (and over time) but the technical bias I have is that I'm far more familiar with the Houdini Python API than I am with the Katana API (and with Vex vs LUA). From my admittedly spewed perspective, I also believe the documentation of the Houdini Python API is better than the Katana equivalent.

My original questions arose from the acknowledgement that my own views are skewed by a bias based on my existing knowledge of Houdini, which by far outweighs my knowledge of Katata.