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. On Thu, Jul 5, 2018 at 8:45 AM Jonathan Moore <[email protected]> wrote: > 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 > ------ > Softimage Mailing List. > To unsubscribe, send a mail to [email protected] > with "unsubscribe" in the subject, and reply to confirm.
------ Softimage Mailing List. To unsubscribe, send a mail to [email protected] with "unsubscribe" in the subject, and reply to confirm.
