“Solomon Kane” is a film directed by Michael J. Bassett based on a character from pulp author Robert E. Howard, who is widly known for “Conan the Barbarian”. The fantasy movie was produced by a consortium of French, Czech and British Companies and was mostly filmed in the Czech Republic. The opening title was designed by Benjamin Cuénod (scripted reality) and done in Houdini by Christian Kelch (Rebels of Design). He spoke to us about the production pipeline, advantages of Houdini, and future plans.
by Lauren Kozak and J.-E. Burkert

Please tell us something about yourself and when you started to use Houdini for your work.
Hi! My forum name is calix, I’m 33 years old and living in Munich/Germany. For more than ten years I have been working as a freelancing 3D Artist for different Companies, mostly in the field of Product Visualization and Commercials. Unfortunately for my daily work, I still have to use Softimage, but two years ago I started to spend more and more time with Houdini!

Why did you decide to work with Houdini on this job?
Simply because I didn’t know how to handle this job in another software on one workstation. Houdini is extremely scalable, so it was easy to write every part of the simulation reliable to disk. It was very straight forward to work on small steps, the modeling of the broken letters, the RBD, and Smoke simulation. Another positive side effect of this choice was learning a lot about the procedural work flow in Houdini.

Can you tell us a little bit more about the production and the pipeline?
A company I work for told me that Benjamin Cuénod is searching for a 3D Animator to work on the 3D title animation for a feature film. I never heard about “Solomon Kane” before, but we got the first ten minutes of the film and started to work on the breakup sequence. I did many versions of the effect with RBD simulations until I asked Martin Matzeder (sanostol) if he could help me to do a more controllable setup. In the end, the breakup which the director liked most was a simple RBD simulation to the left with dissolving pieces. We worked for about four weeks on the title animation.

I started out modeling the individual letters in Rhino because I wanted more control on the bevels and insets. I exported it with a high poly count to Houdini to have a nice topology for generating the chunks with the Cookie SOP. The designer wanted the letters to have a sword slice through some of the letters so I had to work on it one by one, make groups for the pieces and write every character to disk. Now I was able to read parts of the logo into Houdini and start to work on the simulation. In the end, I added a dissolving of the pieces, which was a simple scaling with the help of the Foreach SOP. Finally, I had to simulate smoke layers for the pieces, but the fluids simulation was way too slow, so I replaced the chunks with simple spheres. The lighting was done with the environment light and a HDR image, and I used a slightly modified old metal shader from the material palette for the letters.

There really was no pipeline. I worked and rendered on a Mac Pro 8-Core workstation and delivered the final 2,5K passes to Benjamin, who comped everything in After Effects. The final grading was done by LipSync Post in London.

The rendering was done completely in Houdini? Please tell us some details about this.
I worked with the old metal shader preset, which already had image planes like diffuse and reflection defined, so I only had to render an additional Ambient Occlusion pass. Rendering was very fast with the Micropolygon Renderer, although the shader had displacement and I had to use high pixel sampling values because of the environment light. The smoke was rendered with the billowy smoke shader and a simple spot light.

Can you tell about upcoming productions and Houdini?
Sadly, very few Companies are working with Houdini here in Munich . A few users do monthly meetings with interested artists to build up a little Houdini community. It takes some time, but more and more people are getting a grasp about the possibilities Houdini has to offer, and the Companies are starting to get interested.

I already did a few little jobs now with Houdini, and I’m trying to get more and more projects where I can use the benefits of a procedural work flow. I’m also working on ideas for an animated short, which will of course be done with Houdini.

Information:

Solomon Kane: Official Site
IMDB: Solomon Kane
Wikipedia: Solomon Kane

Fxguide has a nice interview with Aliens in the Attic Visual effects supervisor Doug Smith and co-visual effects supervisor Mike O’Neal.
Read the Interview Here

od[force]: You’re currently head of 3D/commercials at Framestore CFC in London; what responsibilities are involved with that?
Dan Seddon: A range of responsibilities really: it varies from bidding on jobs, to supervising specific projects, deciding on the general running of the department, to staff issues, and more I’m sure…

od[force]: Do you work on specific projects or do you supervise a whole bunch of projects running at the same point in time?
Dan Seddon: Well both. Currently I’m not on a specific project and I’m watching a number projects and will be likely to do so to a greater extent as more come in. However over the last year (since becoming head) I’ve worked on 7 projects and directly supervised 5. Which was a tad tiring…

od[force]: How many people are under your supervision?
Dan Seddon: Good question – somewhere between 20 and 30.

od[force]: And what range of tools are used in your department?
Dan Seddon: Our main packages are Houdini and Maya – with the TD-ing being about 50/50 and the animation being solely Maya.
We use Maya’s renderer and Mental Ray in Maya and PRMan and Mantra in Houdini (mainly PRMan).
We will also use Z-brush and Boujou, 3D Equaliser and Photoshop + other smaller packages. Also Shake (we are currently trying to move over to Nuke).
Most of our jobs go through Flame in the end.

od[force]: Now it would be interesting to learn a little bit about you: How did you arrive at Famestore? Where did you work before?
Dan Seddon: I’ve been in Framestore for slightly less than 7 years – I worked in Film when I first arrived on a number of Houdini FX based projects and then moved over to Commercials about 3 years ago.
Although film can be very satisfying in terms of the high profile work you can get, I find commercials more interesting due to the length of the projects and the greater personal involvement.
Prior to Framestore I worked in a small commercials company called “Clear” – where I mostly used Houdini on a number of promos and commercials; before that I worked from within Clear in a smaller company which I ran along with Jerry Corda-Stanley.
Before that… I worked at the BBC on the Human Body series of documentaries – which the Planets was a follow up to and I also worked on with Jerry.

od[force]: So you’ve been a Houdini user for a long time?
Dan Seddon: I started with version 1.0…

od[force]: Great! Did you use PRISMS before?
Dan Seddon: Nope – I like to think I’m too young to have

od[force]: So basically all along your career you have used Houdini.
Dan Seddon: Pretty much – I started out using Softimage but quickly picked up Houdini when I got frustrated with being able to do so little with it’s “modelling relation” and could see the benefits of proceduralism.
I also use Maya a little now of course, although it would be fair to say I’m primarily Houdini.

od[force]: You’re a supervisor but speaking from a personal point of view, which area of 3D you feel more attracted to? Development, effects, character?
Dan Seddon: Ermm… all of it, if that’s not a weak answer…
I guess I like to build systems – which is why I’m attracted to Houdini. I don’t tend to lean towards FX as much – which is what many people expect from Houdini people (fire water etc). I guess I’ve done a lot of character stuff in the Commercials department – in terms of fur and feathers.
I enjoy that quite a lot – but I like anything with visual complexity, which Houdini is very good for – it’s being able to control almost all aspects of what you are doing. Houdini is for control freaks .

od[force]: Speaking of that, I understand that Houdini/Maya ratio is around 50/50. What is Houdini used for? You said TD-ing, could you elaborate a little?
Dan Seddon: Well… To answer that I’ll talk through some of the work we’ve done.
When I started 2 years ago (when Andy Boyd was running the department) we did the “Guinness Noitulove” commercial – we did a range of stuff on that project. We did trees growing (using lsystems) rocks eroding, ice forming and also some grass growing (time lapse style) across a field, along with other foliage and things. That was very much an all round use of the package on an FX level – we didn’t know every effect we ended up doing in advance.
od[force]: That’s a great commercial! Did you use only the built-in stuff or also some proprietary extensions?
Dan Seddon: We didn’t write any “code” for Guinness – but we developed lots of “methods”. In my case for the rock erosions – I use displacement feedback loops to gradually erode geometry away. For the grass I used ribbons of grass growing sequences and copied them to particles.

The next project after that was our first fur project (using Houdini). This was the Rexona project for which we developed our own fur pipeline – from the grooming inside of Houdini to the fur generation using a renderman DSO. This is where I personally got in to witting DSO’s and this has given me a slight slant towards Renderman since. But, also a good basis for the department doing a lot of projects since then. We did two Rexona projects in total.

After Rexona we did a project for British Airways where a child saw dolphin shapes swimming through the clouds as he looked out of the plane window. We used some of Framestore Film’s Houdini tools (developed for Superman) to generate the dolphin shapes in the Clouds for that project – it was very successful. For that we used mantra and i3d. It was a system of copying “i3d sprites” to points to build larger cloud shapes.

After that we did a couple of smaller projects another Rexona and then I worked on Famous Grouse. Which was my first “feathers” job. It was a short project (surprise surprise) of 6 weeks and that time the feathers were done with textures on grids – the textures being scanned from a stuffed grouse.
We have the grouse in the department still – it makes most people sneeze as it’s fairly old.

od[force]: You say character jobs: this means also the animation of the characters? or was that done in Maya?
Dan Seddon: We had two jobs where the animation was done in Houdini. The wolf was animated in Houdini for Rexona and the Grouse was animated in Houdini on Famous Grouse. Generally that doesn’t happen however – usually the animation gets done in Maya. I’d say, on the whole, there aren’t so many benefits as you might be thinking to keeping the animation in the same package as long as you have a good system for getting stuff between them.

od[force]: Do you find the FBX tools useful? Or do you have your own format/system for scene exchange?
Dan Seddon: Yes we’ve used it more recently instead of some in house scripts for exporting cameras and geometry. It’s really quite good – though not everything works yet.

od[force]: What do you like about Maya over Houdini and about Houdini over Maya (and if you have experience with XSI, include it too)
Dan Seddon: Houdini you can do almost anything in and it takes little time to solve almost any problem. Maya is good if you want to do something relatively simple quickly. Maya has a very good selection of default shaders via Mental Ray – using final gather and or refraction/reflection is very straight forward and subsurface is very easy too. I think this should be fairly easy to do in Houdini – there should be a better range of default shaders, but we don’t seem to have them.

od[force]: Is there something you really dislike in Houdini?
Dan Seddon: I don’t like the management of VOP assets. If you make a VOP SOP or VOP Shader there is no easy way to save that out as an asset and bring it back in it’s VOPs form at a later date. You could argue that other packages don’t have VOPs – but other packages do have those nice “default” tools and the thing stopping Houdini having them to me is this one thing. Currently any VOP asset you develop becomes written in stone by turning it in to *.vex or gets lost in an old scene.
od[force]: How true.

od[force]: I understand you use Renderman/PRMan mainly. What’s your opinion on Mantra? some Mantra vs Renderman/PRMan vs Mental Ray opinion?
Dan Seddon: PRMan is very-very stable and has a large development team. Mantra has many of the features that PRMan has. But I usually find PRMan gives better results on the whole – at the time I last used Mantra, it was much more difficult to do HDR type lighting. Of course, since then they’ve introduced their PBR renderer – which gives amazing results, but is very slow…… PRMan has some great features when it comes to it’s point cloud rendering – the occlusion, subsurface and colour bleeding is incredibly quick. I am also more familiar with Renderman than other renderers – I can write DSOs in Renderman whereas I wouldn’t know where to start with mantra as the HDK is pretty hard.
Mental Ray gives results very quickly and is easier to use for the novice – but is more of a black box and is difficult to develop for.

od[force]: what OS are you using at Framestore commercials?
Dan Seddon: We are on Linux.

od[force]: What’s your relationship with Side Effects? Did you need to involve them in some projects?
Dan Seddon: We talk quite a lot – I know a few of them fairly well. I often given opinion to them on where I think Houdini should be going. I like them quite a lot in terms of their responsiveness. I don’t always get my RFEs though .
They need to clone themselves.
od[force]: twice

od[force]: Since you mentioned it: where do you think Houdini should be going?
Dan Seddon: I think they have the interface now so they need to start building in more push button stuff. I think the toolbar stuff is brilliant – such as the FX stuff – and as it builds over time will be the answer to much of the Maya type tools.
I think the biggest thing is shaders though. I think they should get a copy of Maya and copy the whole darn lot – I don’t think it would take that long for a good shader writer. I think if they employed a good one for a year they’d have it done – I think they should look at Renderman Studio and rip that off too whilst they’re at it
Also, they should get their Mental Ray interface working and move some of that methodology over to Mantra PRMan – currently it’s very easy to develop shaders in Houdini but quite hard to use them. They need a higher level interface (perhaps built around materials and I’m sure this is what they are thinking) that is more similar to materials in Houdini 1.0 like Maya’s Hypershade – currently if you want to drive a parameter by a texture you need to edit the shader (something that will scare many people) or build a massive shader with options for everything. To take this one further – what if you want to put a ramp in to one of the inputs instead of a texture. I think this may mean VOPs and or Mantra need a little bit of development, but it should be possible and once it’s been developed then this type of interface would be more appropriate to Mental Ray as well of course, given the similarity in what I mention above to Maya/Xsi.
So. That was the shading and “tooling” up. I think the next thing should be “speeding” up – currently many of the SOPs seems a little slow and in the future should probably take advantage of threading. I’d like to see SOPs like the ray sop (nearest point) get a lot quicker – I think if people want to drive rigs with Houdini this is going to be necessary. Threading is something I think they know about.
Oh – one other thing. We’ve been quite slow to pick up 9 compared to other departments because we are so PRMan centric. The support has gone backwards a bit in 9 (no auto archives for instance) . I’m pretty sure they’re working on that too though.
It’s funny. I keep mentioning things I want them to do in Houdini. But I’m pretty much hooked on Houdini! It’s mainly because I’m so keen on using it that I feel so strongly about them doing more stuff.
od[force]: I think many users feel the same. they like Houdini so much that they’d prefer not using other package for anything so they pile SESI with requests.

od[force]: What else? I mean all those are (important) developments of existing things. There are people who say they’d like completely new stuff like 3D tracking for example (I personally strongly disagree) or Massive-like tools, 3D paint. Is there any special functionality you’d like to see?
Dan Seddon: I don’t want them to waste time on stuff like tracking either (they have too much to do). I think they should hold off on stuff like Massive. I think there’s a lot of development to get something with the same level of AI – I think the most they should do is write some more interaction stuff for POPs, DOPs, that would be interesting as an on going development. I think 3D paint would be very useful – I think they need to create another attribute type “point, primitive, vertex, detail, texture”. This would enable you to paint/rasterise attributes to a higher level of refinement than the topology – it would clearly benefit texturing as you could just paint “spec” and this would be promoted at render time in the same way as primitive/detail attributes (by reading it as a texture instead). But, clearly this could also be used for many other things such as fur and fluid sims.
To go back to Massive. I think the power of Houdini is what a TD can do with it. I think Massive is too “high level”. Maybe massive like functionality could be added later when they’ve solved the other issues I’ve mentioned.
I think at the moment Houdini needs more market penetration – I don’t think having Massive like functionality will help that. But I think better default shaders will.

od[force]: Tell us about Houdini people. How hard is for you to find people to hire? What do you look for in someone who would use Houdini at Framestore commercials?
Dan Seddon: It can be very tricky finding good Houdini guys (although Bournemouth was a bumper year last year). The main thing I would look for is keenness and the ability to creatively solve problems – it’s also good to look at someones reel, as it’s very easy in this industry to spend a lot of time playing but not actually make any pictures
I think a lot of the younger guys are learning Houdini as opposed to other packages because they like the look of the power it gives them – it will be interesting to see if this changes things at all. I think young people can be very good.

od[force]: What would be your advice to someone who thinks of learning Houdini? And why would you advise someone to learn Houdini in the first place?
Dan Seddon: I think Houdini is the best learning tool as it teaches you the basics of how many packages work. This is going to sound patronising – but there are far fewer Maya guys who truly understand what is going on under the bonnet than there are Houdini guys. Maybe that’s a good thing, that Maya is able to let you work without you knowing those things. But, I think if you want to learn 3D, Houdini is a great place to start. Even if you don’t intend to be a Houdini TD in the long term it’s worth having a look at Houdini in just the same way it’s worth reading a good book.
I would start in SOPs – as this to me is the engine room of Houdini and where you can learn it’s true power. Most things can be done here and it’s the one part of Houdini you can surely say no other package has.

od[force]: An area getting lots of flak here on od[force] is Houdini’s Cloth. The general consensus is that cloth is not exactly “ready to race” right now. What do you think about Houdini cloth? Where do you place it in regard to Maya’s nCloth which gets lot of praise generally?
Andy Boyd: I have not tried it, but heard it’s not the best cloth solution.  I have had a play with nCloth in Maya and it’s super easy to use, very fast and when I have a job that needs it I will go the nCloth route.

od[force]: What improvements would you like to see most in Houdini?
Andy Boyd: I want to see Mantra keep improving, no matter how good the software is, it’s the rendered images that count the most. I want to see Mantra beat Mental Ray, I think it’s almost there.

od[force]: What does it lack compared to Mental Ray?
Andy Boyd: Speed when it come to multi proc rendering. In fact on my Windows machine it’s broken! I can only ever preview with one proc while the other 3 do nothing.

od[force]: I’m not happy with it on my Windows install either, but on my Linux install it screams.
Andy Boyd: Yes, I have heard Linux is better.

od[force]: What is the place of Houdini at Method? How much of the overall 3D production is done in Houdini?
Andy Boyd: They only started using it when I started 7 months ago. There are 12 3D artists and 3 of us use Houdini – but all the others are very interested.

od[force]: That’s interesting to hear. So there is potential for growth
Andy Boyd: We have budgeted to have 2 Masters and 2 Escapes for April onward. Then when big jobs come in we will just rent, works out cheaper.

od[force]: Based on your experience, what do you feel the generic “Houdini user” is lacking? Are they too technical as it is generally considered? What do you recommend to someone who chooses Houdini as his primary package and is looking to find a job at Method?
Andy Boyd: I find advanced users are very good at all things – but character setup seems to be low on Houdini users “to learn” list! However as our team grows and I interview for junior 3D positions I found that most newbies who start using Houdini just know how to do particle effects, or some cool DOPs setups as they think it is an effects tool. Please learn to light, render, build shaders with VOPs and read Craig’s book (Houdini on the Spot). This will make you more useful in production.

od[force]: That’s good advice
od[force]: Tell us a little about the squirrel in the Bridgestone commercial; how did you approach the fur? Was it the built in toolset or some proprietary stuff?
Andy Boyd: I used the new fur procedural in H9 and CVex for shaping the fur – worked very well. Busy doing it again on a kangaroo!

od[force]: Any simulations? Wire dops?
Andy Boyd: Yes the workflow is paint attributes onto geo > comb normals for groom > copy curves (guides) that are shaped by attribute and groom > guides run though a DOP wire solver > new simulated guides are pasted to fur procedural
There is a video with the Mandrills from a commercial I did at Framestore
The fur procedural renders great in mantra, was handling a million plus hairs with motion blur and depth of field.

od[force]: Great! Thanks for that.

od[force]: What was your role in the “End of innocence” short movie?
Andy Boyd: CG supervisor/TD. I created all the environments 3D – everything but a few shots on the grass was filmed blue screen. Biggest challenge was creating the 3D forests. I rendered them in one go, not in layers, using the new delayed load in H9. The individual trees were created with L-systems then written our to disk to be read at render time.

od[force]: Fantastic, thanks so much for the interview.

Interview by Dragos Stefan

From his roots in South Africa, to the head of 3D Commericals at Framestore CFC in London, to his new job at Method Studios in Los Angeles; Andy has plenty of high end projects on his portfolio. Here he talks to us about the journey and how he came to use Houdini for many of the later projects he’s worked on.

od[force]: Tell us a little about yourself: where are you from, how did you started in 3D
Andy Boyd: I am South African. I studied printing and animation was just a hobby.
When I was 21 I decided to make my hobby my job. I applied for a job at a small (one of two) 3D company in Johannesburg. I made my show reel by taking a personal loan and buying a 3d studio R3 and upgrading the RAM in my computer to 32MB!!! After offering to work for free (guess my reel was not good enough!), I got the job. Only not as a 3D animator but as a multimedia programmer. However I was allowed access to the SGI 3D machines.  I would come into work at 4am and use Jason Iversen’s machine (his was the fastest!) and at 9am go upstairs and wait the day out and return at night for another session! After doing this for about 4 months I knew enough to get on the machine full time!

od[force]: You worked in this South African studio for how long?
Andy Boyd: I worked at Digital Directions (now the Refinery) for 3 years before leaving for London

od[force]: Did you use Houdini or PRISMS there?
Andy Boyd: No, Jason and Marc Horsfield did. At the time I was really into lighting and animating which was much easier in Softimage, and then in Maya.
I had a look but it seemed too hard to do simple things, but I always liked the nodal approach, though.

od[force]: …and then you went to London, directly to Framestore?
Andy Boyd: No, when I moved to London and went door-to-door looking for work and finally did at Glassworks.
Funny though when I started there they were still using Softimage – it was like traveling back in time! I started to head up the Maya front slowly converting people, but it was only after about 3 years at Glassworks that I started to get into Houdini
It all started when I was talking to a friend about people learning new software. I said that software companies should have a free learning version available for download. He said Houdini already has this! So I downloaded apprentice version 5.5 and so began my Houdini life.

od[force]: I see… so you studied it for yourself or was Glassworks interested in Houdini at some point?
Andy Boyd: At first I did all my jobs twice, first in Houdini – then I would try to get Glassworks to buy it, they would say no – then I would have to redo it in XSI.  One of my reasons for moving to Framestore was that they had Houdini.

od[force]: So you liked it that much?
As I understand you are (or were) a user of all the big 3: Maya, XSI, and Houdini.
Andy Boyd: I was bored with Maya – nothing had changed since version 3! XSI had rubbish scripting abilities and bad handling large scene. Houdini on the other hand was coming into its prime, new version every 6 months. Loads of development happing.

od[force]: and do you use all 3 now?
Andy Boyd: No, now I use Houdini 90% of the time and Maya for the rest. Depends on what I’m doing!

od[force]: Excellent! So, after 3 years at Glassworks you went to Framestore?
Andy Boyd: Well, I actually went to Mill Film for 6 months to work on the spiders for Harry Potter, when I realised that I prefer commercials so I moved to Framestore.
So, Digital Directions(3 years) > Glassworks (4years) > Mill film (6 months) > Framestore (4 years) > now in LA baby!!! Working at Method in Santa Monica

od[force]: That is great! It’s nice that you mentioned the commercials vs. features because I was just going to ask that: you were almost always involved in commercials, and I see this was intentional. Many people think of features as being the nirvana of CGI, why do you like commercials more?
Andy Boyd: Commercials are more diverse, every 3 months you have new set of problem to solve. Also there is much more involvement in creating the job – being on shoots, filming techniques, direct contact with directors and agencies to help realise their visions.

od[force]: I agree with that. How involved is someone from Framestore or Method when a commercial is done? At which stage are you called?
Andy Boyd: From the very start, the directors and agencies make sure that it’s all doable in time and budget before selling the idea to their client.

od[force]: And then you are consulted in the development of the script / concept?
Andy Boyd: No, it’s more of what should be real, what’s better done in 3D and how it should be filmed to work in post…

od[force]: I see. What I’d like to ask you now is some generic stuff about the technical environment, that is, what applications and operating systems are you using in your daily work?
Andy Boyd: Our render farm is Linux, our workstations are Windows (soon to be Linux). Software is Houdini, Maya, Shake, After Effects, Photoshop, but most 2D comps are done in Flame.

od[force]: How would you describe Houdini’s place (as a tool), among Maya and XSI?
Andy Boyd: That is a hard one as every one is different in their way of working. I would say that if you are tired of writing MEL scrips to do tricky stuff or can’t write your own C++ plugins, use Houdini. If you want to have the procedural flexibility in your workflow, use Houdini. If you want to be on the beach on the weekends and not fighting with Maya to get what you want, use Houdini!

od[force]: You seem quite enthusiastic about Houdini and it is great to hear that. Now, let’s suppose at some moment you want to go freelancing or to open your own small shop (you and 2 friends). Would you think of going with Houdini as the primary package for the shop?
Andy Boyd: Absolutely, but I think I would have still a few Maya’s as there are so many freelancers that know it.

od[force]: So the reason would be availability of people, not technical?
Andy Boyd: Try to find a good animator/rigger for Houdini and you can chose between 3 people! For Maya you have 10% of the world population!!

od[force]: That is true. Speaking of that, did you use Houdini on character jobs at Framestore or Method?
Andy Boyd: Only for the rendering of the fur. Calin Casian rigged the wolf for me, on the Rexona project when I was at Framestore – that was the only one.

od[force]: Why? because the character pipeline was already based on Maya?
Andy Boyd: Yes, all the character people knew Maya.

od[force]: So do you feel the only reason for Houdini not being used more for characters is inertia and general industry acceptance? Do you think there are also technical reasons?
Andy Boyd: Yes, I think it might also be easier in Maya – if you are a student who wants to animate and you prime concern is getting a job you most likely will go for Maya or XSI.

od[force]: Since we talk about “technical reasons”, what do you feel as being a major problem when it comes to Houdini? In what kind of situation would you rather NOT use Houdini? I’m speaking about technical stuff, not user availability.
Andy Boyd: Hmm, I think I would always rather use Houdini.

od[force]:
od[force]: Do you use H9 in production? What’s your opinion about it? What do you like, what do you dislike?
Andy Boyd: Yes I have been using H9 for 6 months now. I love the new Mantra 9, can’t think of anything I don’t like – I know that most things I don’t like are currently bring improved, like better example files

od[force]: Have you used fluids in production?
Andy Boyd: Yes, we are doing some 3D chocolate. Yummy! I have used it on other jobs too, one for Hummer.

od[force]: what do you think about them, compared with already established solutions like Real Flow?
Andy Boyd: Well there are 2 solutions for fluids. There is the particle fluid solver (like Real Flow) and there is the container/gas type solvers. I don’t think the particle solver is better than Real Flow, it’s just much nicer to have it in the package – keeping it all procedural. The container fluids is much better than Maya’s. However unless you are a DOPs king you will be using the new shelves in H9 get you going. In version H8 it would take me 2 hours to set simple DOP networks now it takes 1 click!

od[force]: What difficulties have you encountered when using fluids for the Hummer spot?
Andy Boyd: It was hard was to create a feeling of scale. When you don’t need a glass of water splashing and you need a swimming pool of water – thats when things get tricky! I used a combo of particle fluids and normal particles all surfaced with the new particle fluid surfacer SOP.

Please stay tuned for part II

Part II of our interview with Jordi. If you haven’t read it yet, part I can be found here.

od[force]: Was the project done in Houdini 8?
JBD: The project was done with 8 because we started it before the release of 9, yes.

od[force]: In the end how many people worked with Houdini for the shot? You, Cezar Niculescu and were there any others?
JBD: Just Cezar and myself. It was a small team

od[force]: And how do you feel about the “output” of a Houdini expert? I’m talking about the perception from the Houdini community that Houdini would be more productive in the hands of an expert than a competitive package (also in the hands of an expert). Do you think this is a correct assumption?
JBD: I believe we could say so. The more complex the project is, the more exploration needs to be done.

od[force]: What did you use for rendering?
JBD: Mantra 8. It was superb and although v9 is even better I would happily recommend Mantra for such a complex job any time.

od[force]: In what sense it was superb? speed ? memory? features?
JBD: Well, considering we had to render 496 books with 300 pages each, full motion blur with shadowmaps (and some raytracing too on some shots), and something like 4000 textures for the pages assigned procedurally I am pretty happy with both speed and memory.
We only split the scene on two blocks for the flame to have more control in the comp, not for technical reasons.

od[force]: Was it a deliberate decision from the start to use Mantra or you did consider mental ray too?
JBD: We did not really consider mental ray for this job, we have plenty to do for the integration of mental ray in Houdini to be clear for us, although we intend to, there is too much luggage The Mill has and wants to use. Tools, plugins, shaders are key for our production output so as you can imagine, we are working on that.

od[force]: How is your relationship with Side Effects? Were they involved (support or otherwise) in the project?
JBD: We had some early talks to them so they were aware we were jumping on a huge job although we managed to not bother them too much. Some questions here and there and some posts on the forums did the trick, although the level and quality of support they provided us has proved that kind of thing and assures us we can build on top and not be abandoned on the wild. They have been extremely helpful all the time and I look forward to keep on working with them.

od[force]: You said you checked Houdini 9. What are your thoughts about it? OK, Mantra is awesome, but what about the UI? Do you find it better than the previous one? Also, what about the fluids? Have you tried them in production conditions?
JBD: The UI is far better, much more approachable and above all the tiny pieces of code that do some initial work for you are very helpful for someone like me which still does not have the, let’s say, vocabulary to do it from scratch. In my opinion at some point I will turn off those shelves but for now they are very very very useful.
Also the new UI, which I don’t mean only from the graphical point of view, is much more friendly to me in the sense of manipulating and handling of the scene. I imagine many more people trying Houdini in this condition than with version 8.1 which although great, was far more obscure.
I have not tried fluids seriously, although from our little research we are planning to build our fluid tools on top of Houdini so I can already tell you Houdini is here to stay.

od[force]: That is nice to hear. Now, a purely subjective question: what do you like most about Houdini?
JBD: I see Houdini as an operating system rather than an application, the approach is far more abstract than a classical tool but is far more powerful. The way we used it for this job more closely resembles the compilation of a program than a classical 3D workflow.

od[force]: If you were to decide on only *one* big improvement /feature (which you’d think would improve Houdini significantly) what would that be?
JBD: Better integration with Mental Ray would be the one, also I would ask them to add a 2D tracker to the compositor and improve the shader library with a better supershader.

od[force]: What do you dislike about Houdini?
JBD: I guess I don’t know enough to pinpoint things that I dislike, there are a few I would ask them to do in a slight different way but not much more.

od[force]: Would you consider Houdini for character jobs? It has a terrific toolset now, but I wonder if the studios are aware of that?
JBD: I am really willing to explore that so let’s see…

od[force]: Thank you for your time
JBD: My pleasure

Jordi Bares is the joint head of 3D and a VFX Supervisor at the Mill in London. Recently the Mill decided to use Houdini on a new Guinness commercial, and Jordi was kind enough to talk to us about the decision and what he thinks of Houdini and it’s future at the Mill.

Take a look at the commercial here (click on the Guinness link) and then come back to read the interview and see some making of clips.

od[force]: Tell us a little about your background and how you arrived at The Mill.
Jordi Bares Dominguez: I am Spanish, born in Barcelona and 38 years old (gosh). My background, like many people that started a long time ago, is very weird. But I did study Architecture and on the way did work as a C programmer for Siemens Nixdorf, did some traditional animation, and specially freelanced a lot when my architectural visualization work at the university started to call the attention of some teachers. From there via a project I was doing at the Image lab in the university I ended up doing advertising for TV and from then I started to work doing this as it was really good fun.

From my last position in Spain as Animation Director in Filmtel I moved to London to do a freelance job for Jim Henson’s in London and after another few projects I landed at The Mill, which is the place I really had in mind as I really do love the really dynamic and challenging world of advertising. I basically came from Architecture and moved to London for the fun of the challenge.

od[force]: Please tell us about your position at The Mill.
JBD: I am currently Joint-Head of 3D and act also on production as VFX Supervisor. I basically run certain kind of jobs and assist the director to make sure we can get the most out of the production, supervise the shoot and once back lead the team of artists to make the whole thing happen.

od[force]: Excellent
JBD: Yeah, I love it

od[force]: So being the head of 3D, you are responsible for coordinating the 3D department and choosing the approaches and the tools?
JBD: Together with the R&D team and production we are drawing a plan and steering the ship to accommodate the huge changes that happen from year to year. I am one of the people involved in the recommendation of the tools we are going to need.
This environment is so dynamic that it is really complicated to draw a master plan for the next 5 years as things can really change and we need to be able to respond quickly to the competition and client’s desires.

od[force]: Which are the main tools for the 3D department?
JBD: The Mill has always been a Softimage 3D facility but times have changed and it is now a blend of tools. Although we still keep the same render engine so we could say we are a Mental Ray facility. Our main tools are XSI and Maya, around 50%. On top of that we use all sorts of tools from Massive, Realflow, Houdini, Endorphin to anything that can really do the job.

od[force]: How did you became interested in Houdini, and why?

JBD: Houdini has been on the agenda for me for such a long time I can barely remember. But the fact is that the actual production needs are pushing the boundaries so much that neither Maya nor XSI can cope with the needs and you end up spending a lot of time writing code to do things rather than just doing them. So right now we are facing the fact that Houdini flexibility takes us there without the need of programming. Also exploring things is far easier and lets us carry on with multiple lines of thought rather than committing to a far too early decision from the very beginning due to time constraints or natural changes from production. I would say Houdini is extremely well positioned to be the backbone of any production which is something I am more than willing to test.

od[force]: Is the Guinness commercial the first production to use Houdini?
JBD: We have been using it in many occasions for auxiliary work to support, for example, the import limitations from other packages. So yes, this is officially the very first project to output frames for real shots.

od[force]: What were your responsibilities for the Guinness commercial?
JBD: When the job arrived we talked to the director in detail and one of our matte painters (Jimmy Kiddell) did some concept art to root the talks onto something so we could discuss all the possibilities and understand the nature of the job. After those first conversations it became clear to me this was a very special beast as we were talking about substituting the real pint of Guinness that appears on every single commercial with a CG version of it, clearly we needed to approach it in a way that all parties would be able to change anything as the case may be.
We did propose the mechanism and design the whole structure, although later on set, the art department did a working prototype that basically locked the original concept and they were keen that we could work based on something physical.
My responsibility was to supervise the shoot and lead the team to put the director’s vision on the screen, so I decided to get the help from a Houdini expert to build the pint based on my specs and he did improve my original design due to his understanding of Houdini. From then it was a huge learning curve for me as the concept kept moving forward and we started to add extra things and layers of controls. The animation changed dramatically and we even added dust and bookmarks falling from the book so I had to learn to use Houdini better and basically jump onto it.

od[force]: …and was that process harder than you expected, or easier?
JBD: The process was much easier than I though in terms of workflow but setting it up would have been an impossible task without the help from Cezar Niculescu who did almost all the TD work.

Stay tuned for part II.

Calin Casian is involved with R&D at Side Effects Software with a particular focus on character animation features and development. He is responsible for the wonderful AutoRig, and is currently busy improving an already excellent toolset for the next releases of our favourite software.

od[force]: Tell us about your role at SESI.

Calin Casian: Hmm, my role at SESI. I think there isn’t one role that I can pin down. Probably that’s the fun part of being in R&D, you have to deal with all sorts of problems that are pretty much all over the map. If I have to pick one in particular, with the highest percentage, that would be character development. You may say that character development in itself is a huge area and you’ll want me to be more exact so here I am: I don’t do C++ coding, I’m leaning towards TDs so scripting is a must but no more than that. Right now for example I’m focusing on procedural rigging and autorigs.

od[force]: So what exactly in R&D? Do you stay at SESI coming up with new concepts for Houdini? Do you work with customers for project-based solutions? Are you involved in production?

Calin Casian: Well, let’s say R&D; I have an idea and I need to prototype it and test it, see if it works. Houdini is probably the best 3d package that helps you prototype ideas, concepts that the developers of the software didn’t think about but they provide you with such a complex set of building blocks that it’s impossible to find something that a TD can’t do. After I built it in Houdini, with Houdini’s nodes we analyze it and see if somebody from R&D, a developer, can code the functionality in a single node because it would run much faster. Sometimes though, it’s better to keep it as a digital asset, or as a tool, so other people can use/dissect it and extended it.

Yes I do production consulting as part of SESI’s services package but that’s usually the part about I can’t talk much because of the NDAs.

od[force]: So there are improvements coming in the autorig included in Houdini?

Calin Casian: Well as you probably know H9 is around the corner and we have a brand new UI and with a brand new UI it looks good to have a brand new set of rigging tools and autorigs.

I can’t say much about the system but it will be a very flexible system that will empower the user in the same way Houdini does it in FX. It’s based on production character pipelines that were tested for years.

od[force]: What other character areas are you focusing on? Right now there’s much talk about facial animation and automation. Does SESI intend to develop something along these lines?

Calin Casian: Another concept that we introduced with H8.1 is metacapture and metadeform as a side effect of developing ways of doing muscles in Houdini. Metacapture and metadeform are a brand new way in capturing/skinning and deform a character in Houdini. It’s using metaballs instead of bones. Why metaballs? because of some very nice features that metaballs offers out of the box. One being the fact that they fuse together and can build a nice smooth surface, it’s perfect when you want to build the muscle structure underneath the skin. Other concepts are using polygonal or NURBS based fusiform muscle but they have a very sharp edge when two muscles intersect each other. If you want to keep the skin tight to the muscles you’ll see the gap. Of course there are tools to deal with such problem but why use 2-3, sometimes 4 tools to achieve the same result the metaballs offer right away. Another plus is the fact that metaballs with their nature gives you a cheap way of volume preservation which most of the time is what you want. Production surfaced the fact the direct-able solutions are far better that physically correct simulation so with our muscle solution you get almost the same result that a simulation but interactively.

I have nothing to say at this moment about a facial auto-rig but one can use all the traditional methods plus the new metacapture and metadeform.

od[force]: Tell us a little about what do you like about SESI and working for them and what do you think sets SESI apart from other 3D software development companies?

Calin Casian: Probably the first thing that struck me when I joined SESI were the people; I know it sounds cliche but that’s the truth. The team is relatively small but the brains, oh my God, the brains. Each developer is like a mini genius.

od[force]: What do you think should be improved in Houdini after version 9? Or what would you like improved?

Calin Casian: I don’t think that some parts of Houdini needs to be improved because they don’t quite work or they are old, it all makes sense in Houdini’s world. I can’t talk for everybody but I’ll try to create wrappers and expose the concepts in the industry’s common language.

od[force]: So do you think Houdini should become more like the mainstream packages, or least at the first contact?

Calin Casian: Well, let me return your question with another question: Define mainstream? If you think as the new user should find it’s way moving from other packages because the main concepts are the same, tool, action, result, then yes Houdini should be that package. If you think mainstream as certain button should be in the left or right corner of the screen then I would say no.

od[force]: We know that when somebody thinks about character animation, Houdini is probably the last package that comes to mind. Why do you think somebody should consider Houdini for character animation? What can one find in Houdini and not in the established character packages?

Calin Casian: Well Houdini is quite young in terms of character animation but that doesn’t mean we are behind in tools, features and capabilities, I think we proved with C.O.R.E. and The wild that Houdini can do a feature animation from head to tail. I think Houdini was always perceived as a developer’s package and it’s hard to get rid of this stigma, funny enough in my experience Houdini is the least developer’s package as long as you are familiar with a node based application. To tie the “young” concept more, the truth is that there aren’t too many Houdini animators and I think it’s up to the artist if he/she is willing to learn a new package or use the one from school and be done with it. Most of the time, with some exceptions, the artist picks his/her tool.

Schools are another thing as they teach Houdini as a FX package, so I think the problem is spread out and I don’t know all the details so I would encourage you to go and ask the artists as well as the teachers.

od[force]: Can you expand this to the whole pipeline instead of only characters? So why would I choose Houdini for my CGI studio and base my pipeline around it?

Calin Casian: Well, that’s a very good question. Houdini is used today and it’s well known for it’s robustness and flexibility as the back end of the pipeline or the glue between departments with minimum amount of custom code. So I would say that Houdini seems a logical choice for a small studio that doesn’t have the resources to maintain a R&D department. Also we are unique in the way Houdini deals with artists or anybody that’s not so technical, so basically they want to use the package without breaking the rigs or the pipeline. We have the concept of Digital Assets. A rig is one of the best examples. Many other packages have different methods of protecting the artists from “breaking” the rig. With digital assets the TD builds the rig and then creates a node, asset, character and he makes only the controls that should be animatable available for the animators and nothing else.

All the guts are still inside the character but all the parameters for all the nodes are locked so from outside you can’t touch them. The same way goes for the objects inside your character. The artist can’t parent/un-parent, delete, add new nodes in the character’s hierarchy. If you ask me the Digital Asset alone is worth using Houdini as your primarily tool.

od[force]: And are those advantages working for people? I mean, in the real world, are people switching? are there success stories, studios switching from competing packages to Houdini and gaining from this, or studios built from the ground up on Houdini and being more efficient?

Calin Casian: I may not be the right person to be asked about this but if you read our customer stories from our website you’ll get your question answered.

od[force]: By reading the names of the developers on the Houdini forum, it occurred to me that SESI’s team seems quite multi-cultural. Is that true? in that context, tell us a little about where are you from and how you got to work at SESI

Calin Casian: Well Canada it’s a multicultural country due to it’s immigration policy. I’m from Romania, Eastern Europe, and I moved to Canada 3 years ago and I joined SESI’s team 2 weeks after my landing. In Europe I was working in production and I thought it would be so cool to work in a company that builds the software I was using and somehow shape it’s form.

od[force]: Thank you for your time; keep up the good the good work for Houdini 9!

Calin Casian: If any of the readers are interesting to find out more about characters in Houdini or Houdini in general check out http://www.sidefx.com or drop by to our booth at Siggraph 2007 in San Diego, I know you’ll come to our party.

Dragos Stefan (digitallysane)

Photo by Leyla Tirgari