Learning Houdini: You're Doing It Wrong and How to do it Right | Matt Estela | SIGGRAPH 2019

[Music] [Music] this is Houdini you're learning it wrong I'm hoping this will be the most click Beatty title use you'll see at suger this week I met Estella I'm the cydia wiki guy I'll just give you warning now I'm not Matt s Lia the YouTube capitalization if you want that talk there is a wood graph down the road you want to go to that anyway City wiki people know me for this kind of stuff lots of different many examples penny resource as a quick cheat sheet for Houdini but now also a teacher running a master's degree in VFX and real time out at the University of Technology Sydney it's super fun so just as a quick show of hands who's the students here in law students whose industry should be the exact opposite okay and who here reviews show reels okay good he was here to pick a fight and heckle yeah okay so learning Houdini it's easier than ever before if you typed learn Houdini into Google you get 8.1 million hits social media is also super busy so I think this is fairly popular Facebook group has 29,000 members and they're posting video is pretty much every day so you'll see lots lots of this kind of stuff the stuff is being posted here is pretty representative of most show reels if you look at these long enough you see certain tropes repeat certain memes having watched a lot of these kind of reels at Animal Logic when I was a effects Lee down there and lots of student reels now that I'm teaching you can kind of do this which is play show reel bingo so if you pick your favorite Cheryl cliche they're all there notice the the the triple score row for nipping riddens five now the this isn't a dig at these tutorials they're all fantastic there's a reason that they're really popular and you should watch more but people who do review show reels play this the next time you're actually doing that stuff hopefully by this time next year after this video goes out you might be able to score any bingo cards so Who am I calling out here it's the students so let me explain here's something I've seen far too often so this is a carbon copy of the end results of the nipping Bridget's 5 so what is it showing me this is I mean that Steve nipping is a great teacher it's showing me that this student has spent money on a tutorial that they can follow instructions but frankly little else about this person this is essentially a Cheryl by proxy for Steve nipping as far as I know he's doing fine it doesn't need help but compare this to this the same basic technique but what's different well now it's been applied to a vertical surface the feel and the timing have changed it's gone through a few few iterations to get to the state sorry I've got cigarettes throws now this would be much more useful shots in a show real and this kind of reveals a truth about reels that most beginners forget which is that what you're showing isn't what Studios want to see you want to show that you can button click along with an online tutorial that might impress your mum that might impress your non-industry friends but not Studios series need to know more about you so you show them that so while this title is pretty click baby really it should be this so what should you be showing that studios want to see well it's broadly this so it's tech eye and social skills so let's break this down so they take is obvious and it's what you get from your online tutorials and from the school at a minimum you know all the buzzwords some the workflow and the eyes a bit trickier that's largely down to just basically feedback in practice most people can look at an experienced effects reel and look at a beginner effects reel and know that they're different but not be able to quantify why that takes time in practice and an ideally regular feedback which leads to the social skills so this is my longest rant is slide and this to me is probably the most important thing in a real and covers it for huge range of things like Kenny responds to notes in dailies I have a good team player do you know the difference between an artisan and an artisan so an artists will think of all of the tasks and effects its high art and they get angry if a suit says its final then they don't think it is versus an artisan where you recognize that essentially we're all fancy carpenters and we're making pretty things but it's still a budget and to a deadlock a deadline can you deliver things on time or will you take two months to do a shot can you take a generic rig to do moats and you say yep I'll do it gonna run it on please 40 shots because someone has to do it can you be given a task and solve it and so on now two questions can you get this from and on my tutorial and how can I show this in my real I don't think you can get that from an online tutorial CG is a team sport now you might be able to learn a few soccer skills from YouTube but you don't really understand the game until you get out on the field and play with other people I learned most of this in the first few years that I was working in the industry made countless mistakes and I put my foot in my mouth no but terrifying amount of times but I'm a fast learner and the industry was less crowded 19 years ago so I managed to stay in it and keep a job these days it's such a crowded market with juniors on tap and there's less patience with from studios to deal with slightly difficult genius worse a lot of studios just don't have time to mentor and guide juniors so they might have their contract finish they never get us back and the juniors have no idea what they did wrong so this way the right kind of school can be a help where if they create the right environment they set up a place where people can go and make all the mistakes that they need to and get it out of the system and they can be coached and guided in the right way so ok but how can any of that be shown in a show real or half the battle is just knowing that these the qualities studios are looking for and once you're aware of that you can make changes to how you present your work I'm also the kind of products you work on ideally big teams work into a high quality working to a fixed deadline also help showcase you as a production ready artist so getting back to that example the reason I know this looks better is because this is from one of my students after I firmly gently but firmly shooed him out for showing me the first oh no that the carbon copy of the Steve nipping shoes so iessons tried on a vertical cliff face and he did when he next showed me the timing was too slow and he fixed it the next version had had too much debris he turned it down and so on this still isn't final and in but even with that small amount of interaction the student got an understanding of what's required and expected in a production so let's see some more examples so I'll be sharing a bunch of stuff from our students from last year and this year I've tried to pepper it with with with lots of handy tips and tricks that I think you'll guys will all find funny useful ever Mike's a cool trick but that said this is also kind of a meta show real a lot of these clips are lifted directly from my students show reels because what makes for a good presentation also makes for a good show reel so try and keep that in the back your minds so look at these clips and think about how they're showcasing the tech I social stuff that I keep talking about up on it stuff now and then but look at these in relation to your own work students I know you're out there and how you might better present your tech I in social skills so this is the the color thief which was the short film we made last year I won't play the whole thing if you search for the colour thief online you'll find it this is a bunch of shots from it it gives you the vibe no it's destroy being sci-fi atmospherics saw that kinda jazz this was dictated right from the start with the concept art and previews and layout so how did our Houdini team help service this vision so here's a team most of them hadn't really used Houdini before and they came from backgrounds like graphic design and animation and soft development at the bottom of the upcoming slides I try to put the the primary artists or artists who did this bit but if I forget you can always come back to this slide in the video recording you should hire them all they're all awesome so Emma sterics and I said one of the first things I asked our team to do was to create a library of volume elements to give to lighting it's a nice nice easy intro start with some some basics shelf presets then gradually get into more app directed setups no using some volume knobs and bottom wrangles as the lighting department got deeper into their side that asked for more Pacific shapes so maybe a long flat fog and a noisy Half Dome things like that but but at stage we had a few people who are capable of running volume sims or doing or doing stuff based tweaks so it became a nice little feedback loop across departments which and we end up with a great library of volumes we we had a try at making looping volumes but it was easier just to look at look at the edits find the longest shot we had double that and then run our volume sims for that amount of frames so we're all always covered we also did some some slowly evolving fog so this is using a tip from Paul Wagner who I've worked at back at animal this is to disturbance fields one one really low frequency one really high frequency both are very low strength and you just let it run for a while and you just get this lovely slowly evolving fog do any well so this evolved into some of our hero shots so this isn't the you're gonna play come on you can do it which way we going hey everyone so all that for a very minimal slide but basically saying that this that workflow evolved into into hero interaction shots it's not the show is set up but it was easy to tweak and easy to to react to regular feedback here it is in context some really nice interaction going on here but it's interesting to compare this to the version one which was this ready this is the kind of thing you see on a lot of student reels what's the difference No how did it get from from this rather lawful result to the first one it's just the number of iterations by submitting this 35 times and maybe you put up six or seven times in dailies Callum trained his eye he got an understanding of the technical process and also responding to feedback from lighting and animation made him a better effects artist and he delivered a better shot his library of some steam Jets were made again it was simple to run but it was developed in a feedback loop with lighting and layouts to deliver a pretty cool end result like that a feature of our character is she leaves a color trail in her wake so we had this really cool concept art that was backed by Lexus and again it was a very collaborative process working across animation and effects and lighting in comp so we'd start with the character and using USD runs continuously Gurgaon we could isolate to just her feet then we generate some interesting velocity vectors using using a wrangle and a trail stop so that the vectors went mostly away and back a bit and we'd feed that into a pop soon and so this would just emits a copy of the feets on every frame and push them along those vectors we would then convert that to a signed distance field and then in a volume bot so in a point fob on a ground plane we could then we could then query the SDF and ask where is each point relative to the SDF and if it was inside the Sun distance field it'll be white it was outside it'll be black so we were feeling pretty cool about that but then we remembered that the ground plane had been modelled at wildly different resolutions for different parts of set so if we wanted to get down to a really fine smooth edge we would have to subdivide them and we just didn't have the memory or time for that luckily we knew that we could use mantra and we could just copy and paste the same nodes from a point Bob and put it into a shading Network so now this working pixel rather than put points we could then add noise to the position at render time so it get in more detail where we need it so this is Brenda is a really fast black and white mass which the lightest could then bringing in to nuke along with Renmin renders generated out of katana to get these fan results so again collaboration know lots of iteration lots of feedback lots of problem solving for my students to deliver a result that we were really happy with now no here we go from a really wide shots to one where the camera is almost like sitting directly on the cobblestones and that ability for manager to execute the X shaders was really really handy for us so you can't do a sci-fi short without explosions the setup for doing exploitive things wasn't wasn't complicated but it was highly tweakable a thing that came out of the out of the collaboration process here was that what is what is a good shot for foreign effects real may not be good for a film you know we would love a no huge 30-second rolling no mushroom cloud but in a fast edit you have to be able to sell those ideas quickly and we were taking timing reference from animation and from layouts to then drive how we how we do our sins here you can see a small sample of the amount of takes that we went through we got up to about a version 103 in the end but in the end we actually backtracked quite a few versions for the final for the final shots that's pretty common in production that was a good lesson for our student effects artists who now what be fazed by those sort of things once they're in industry in the rule industry here is in context so we've got Callum's interactive smoke in the back we've got the power thing in the front pretty cool shop for these shots where we had a smaller drone hit hit hit the bigger drone we wanted to really over-the-top Michael Bay style explosion now if you've spoken to me for more than 20 seconds with any sort of feeding problems you have and you say pyro I'll punch you in the throat I don't like pyro it's complicated that's what that's what proper effects artists do I just wake my hands but obviously we had to get a handle on for this shot that's what we found was that if we turned the temperature diffusion really low and the fuel inefficiency really high we could get this silly cinematic fireball that just roll and roll and roll and roll which is what we're sort of after by the by the leads and again always check the version numbers lot of iteration what a feedback Kalen became a master of pirate by the end of this process another interesting aspect of this whole production was was getting stuff out of effects and into lighting and comp most sugars aren't using mattre for rendering maybe it'll switch to commerce soon so it's useful on a real to demonstrate that he can play nice with other departments so here's a single frame of one of the five elements one of the requests from the lighting team was to keep things simple and efficient so we would only export dense so there's no extra fields for temperature or for fuel blah blah blah so we would take density which has been normalized to a naught 1 range always means we would expand that out to valleys expected for a black body radiation node inside random and wrist nodes and use that to drive the emission based on feedback from lighting we added an extra exposure control so they could dial up and down the the flame brightness really easily so look at that shader graph there and then we then replicated it inside katana using the same kind of nodes so again there's more collaboration between us and effects and then going over to the lighting shading team and that sort of collaboration and feedback workflow really appreciated by Studios introduction which is why we pushed this so hard at our school and here's the finish results so we've got IBD we've got three pyro Sims we've got particles we've got trails message and Comp and a funnel and a fireball at the end that goes for about 20 minutes in the end we kept playing with how this shot was gonna cut to the next shot there's gonna be a cross dissolve or a fade to black so we so we said Callan can just run it a little bit longer maybe went a little overboard do oh my god the burning flames Big Valley Red Roof and last year was using Houdini outside effects at one point we realized that the shots were getting really close to the roof to the roofs of our buildings close enough to reveal that the textures but we were using weren't holding up at that point the modeling team went and they shrugged inside and they they went to my they made a single roof tile and went to the control T moved across control D moved across and we had about 20 buildings it was worth spending some time on a procedural system to grow roof tiles on top of our existing geometry this pretty cool setup made by Phil it had controls for the stagger style random rotations and tile dropout and interesting challenge here was getting even parameterization per roof section when the models had been created by lots of different in lots of different ways with with with wildly different movies we found of good work around here was to isolate each each flat-panel converted to NURBS patch which it gave us a nice clean UVs and then we could divide and generate tile locations from that and so that way we it's I was curious so here you can see some of the buildings and we could bang out all of those roof tiles in the morning and then the model equipment could keep modeling and we could just keep updating roof tiles as needed really handy old-timey cobblestone streets was a feature of the art direction and again we used sujini o play again there we go so Moe published a library of cobblestones and we then came up with a method to scatter and orient stones in patterns but stored as a USD point instances set up we have a video that explains is some more detail if you search YouTube for Houdini USD ala and mob logic in Academy Ala you'll find it it's pretty cool and super efficient so when when we heard about Solaris and wops were very excited because yesterday is working out great for us one shot required a city in the backgrounds and seems simple enough the terrain scatter instance buildings done now for a effects real that might be fine but when put up in context of the short film the leads thought it looked too sparse and too random so reference came back of of Brazilian favelas really tightly packed we had to do more work the spinal setup here involved setting up directions for the buildings so that they would be on a clean bill south east west direction carving in roads and more importantly overpopulating the buildings and then pruning it back so that it could be as tightly packed as possible the tricky was a little for loop that would try and put down a building and it would test for an intersection against all of the previously placed buildings if it detected an overlap then the building could be put down so in that way we got the that we were after of course shortly after this the new pipe healed scatter node came along that does this for you well we even use it for 3d printing so we took this model we made for our our real-time project called Explorer which is going to be part of a pee hour this afternoon but either to take a model that was optimized for real-time and convert it into a clean water sight mesh for 3d printing the idea is pretty simple it's a full loop in you know in a connectivity mode we would polyfill each elements merge it with its with the rest of sections video me from polygons so we've got a clean water side pity be we then take a copy of that and we shrink it down and we then subtract that from the outer shell to get a nice clean tin bin shell what was great was that we could measure the volume of the SDF and we could generate an estimate of the print cost and then we could keep scaling up the print until we hit our print budget we may actually got a printed we were within about 1 or 2 1 at 1 $2 of the actual print cost which was pretty cool so yes be sure to come along this afternoon try it out you can chat with Ben who is one of the primary developers on this app but he's also a pretty amazing food any person you're going to show some more stuff of his later Houdini also got a huge workout for the other real-time project we did it was called Sam bender Ilya did an incredible amount of work using he was using Houdini engine inside Unreal Engine for doing procedural vegetation doing hide fields OBD tons of things is a amazing one-man-army that I was blown away by the end result very cool I better say that both marathon projects from last year they won awards at the rookies this year which is pretty awesome that's a last year what are doing this year what cool tricks can we share and how am I going to present these if the students are working on stuff and I can't steal any clips from the show reels that's mine out so in the first few weeks my effects team they went on a spree just throwing themselves into every tutorial they could find spread stuff you know good to learn the basics but you know liking liking that I and it's trivially to see where their social skills were so all see how they developed over time an interesting observation was that when I joined the school last year there was one person who was doing Houdini and by the end we had about seven people this year we started with five and it's now July August and most about 30 students are using it in some capacity it's no longer just that package to do effects but it's becoming a tool everyone just expects to know like say Photoshop which is pretty cool the short film that they're working on involves this character this out of shape passes prime Barry hunter we want him to have nice-looking cloth in here so something from this amazing concept scope by Danny we started to working this problems like for us vellum had been out for a few months so we got to work so Nathan here tackled a few key things we knew that we'd need to solve one which is skin the intersection now the animation department going to do their best not to hit silly poses but the proportions of them of this character means we're always going to get issues Nathan created this really handy a visualizer for us so the blue model is what is coming out of animation and showing all of the intersections and you can see we're getting all kinds of issues in the arm when he said it goes into the pose which will cause issues for cloth the green shows one of his early attempts at a skin skin D intersect still a work in progress but even this early test shows substantial improvements no Nathan tried to explain the setup to me and my eyes kind of glaze over but I thought that the core idea it's running a Deltan wash and a detangle stop in a for loop to just gently try and ease the skin out of itself the big challenge that I sort of set out for him was was that we want to be able to run this on the farm using PDG so that when animators check in the work this will automatically trigger so the by the time it comes to effects it's already been D intersected now the character has this sort of I guess MO vest over his shirt proved more challenging and it's made it both rigid and soft components as well as that funny Eagle thing on his arm so this was testing using using stiffest components and then but then rivet that the under the under forming geo back on to the solve using a few key points it's working pretty well this is a test on an early shot from production like the skin intersects the real aim here is to automate the cloth solves so that we can let the farm run those using a shotgun events and then the effects artists can get on with actually doing the more fun stuff as an early test of the hair setup we've since moved on from this workflow but there was some some fun ideas in here that are worth worth checking out the initial idea was that we would use the shapes from the concept scopes we had this big lump of clay to represent the the mullets and whose ideas from the hologram stuff that I talked about in a in Anaheim a couple years ago that was spin recorded where we're using volume trails to grow curves through a volume so here we took a lump that lump for the mullet and he was fine shortest path to create curves that went from where the mullet is is being is intersecting on the scalp to the very end tip of mullets and we get the whole bunch of curves we then defined tangents for those curves which which flow flow down and we then we transferred those tangents into a velocity volume we can in scatter points on scalp and then use a volume trail to grow the hairs through I wasn't bad for in early tests but then afterwards we found the volume velocity from curve stop and the guided rec stops which do exactly all that work in about 20 less nodes it's always a thing the other character is this little alien so this is still very working progress but this is us diving into the process with none of us having had any hair or groom experience inside Houdini lizard on a bit of X gender Maya but she don't know us no Houdini let alone put any fur tools lots of feedback without Department about what's the closest animal style work we're matching is it messy or clean is it scraggly or smooth and it's still ongoing this is a test of some of some procedural win on fur which should work for some of Addison shots now I getting to this point is involved a lot of iteration what's it trying stuff out throwing it away watching tutorials throwing those away just doing hundreds of tests a little tour of the current workflow that this is has develops we started with the standards object level turtles but we quickly abandon that in favor of a entirely stops approach we went this way for a few reasons the models being constantly tweets as concept art and direction kept changing especially in the early days when it was a it was a constant zipper scope sort of our battle royale the polygons would change the UVs would sometimes be there or they wouldn't be there at all or even the entire topology would change to help someone in rigging so if we're using painted maps and mineral grouping that was just never going to work this ever so mostly procedural and gives a good base all that'll groom but still allows for the final manual tweaking for fine detail so this is using a stash copy of an old Amish where we were pretty happy with it and she will then edit and move it to roughly conform onto the shape of basically whatever our latest meshes and then we would actually transfer all of those attributes that she's looking that she's using so if you're safer length of her direction and straggle and whatnot those attributes get automatically picked up by the guide creation tools so it gives a good base regardless of what mesh is coming in but by the time we got this far Liz had a pretty modular system and we were able to incorporate new ideas as we found him so there was a a hive talk earlier this year in London where armored at frame store who shouted me out at his talk I'm shouting it right back ahmud if you're doing clumping for sheets for at least saw them and mmm I'm gonna try that out so this is a test of how fine detail we can get for clumping we can get pretty fun clumping so this might end up in the show as aren't we cool another tip we got which is from miles green who's the effects loop at animal as well as a visit from from Camille Chang who's a grew Madison animal though they revealed that using an undercoat and mid coats and top coats to get really dense looking natural fur for Peter Rabbit so this thought she would have a go at that as well so this is a single coat here's two layers where she's also pulled in more attributes for third direction and for clumping and that's applied to the rest of the body just looking looking pretty cool lots of work is on into this rig so work at 480 there's probably gonna be another 200 - this lots more to go but again that's why written pretty cool here's a bit of crazy concept art from Enya more from her in a bit which was translated into 3d by Tom now the initial sculpt for the tree was done in ZBrush but he put those those odd little blue orbs on using Houdini why well one part was just as as a learning experience and we were keen to see if we could get some subtle keepalive on this without needing rigging or simulation so if you look close you can see there's a there's a bit of a gentle breeze waving it around here's the setup which looks more complicated than you think it would need to be and that's because if you zoom in and the concept art the spheres are actually little alien exacts which Tom matched so the setup is so that we can have multiple modeling variations of these little exacts with little baby aliens in them and they've all got their own little subtle wiggle animation and the set of super fast to to tweak if we decide we want to go more movement or less movement want to add more or less baby variations to do that you know traditional sort of ringing ringing so I pipeline which involve a lot of work was this it's actually pretty trivial it's also a really fun way to to introduce Tom to Houdini and already based on this he's got a pretty unique challenge that he can put on his on his real shop his problem solving self as short features some space spaceship shots and then Bo did a take on the rocket engines back in February and that's the one in the top two left he then had another go in June which is the bottom left which is a huge improvement showing progression like this is a great thing for sure real and it's great to see that while Bo needed a little bit of hope for the first one the second attempt was was all him and it's a simple setup and it looks way better on those we doing some explosions because explosions but we're keen to see if we could stay away from that to the standard pyro look beau was doing some cool tests here using using some animated noise vel fields it's not rocket science but it something that looks a little bit different on a real you want to say to stand out a bit bowed been thinking about other kinds of explosions and out of the blue we came up with this really cool test immediately I suspected he was stealing this from another tutorial and he said no no this is actually from my wiki where he's my basic example of rain on a pig and he used that to seed paracin it's totally cool results it doesn't look like anyone else's work and it shows that very handy ability to take an idea and remix it and tweak it to your own needs very useful skill in this in this industry no we don't want to have a really rich alien world met birders here adopted his basis colonization tutorial from from in tag ma and repurpose it so that it could detect services so we get this cool to a creepy coral vein look so for this setup the basic idea is to start with a scatter node on the on the geometry then we use noise to to move those those positions we take an SDF of the geometry and we delete any scatter points that have gone inside then of the points that I've been left over we choose some selling points and we run the space colonization algorithm but tweak it so that if a branch will stop growing if it moves inside the volume is another weed so P maids so starting with some scatter points on the surface it would copy random sized spheres onto those locations and he would one run one frame of an IBD solve to 2d intersects the the spheres and then he would take those new three positions give them a villain pressure constraints to then try and expand each sphere while keeping them pinned onto the grounds so we get this cool creepy exact bug going on Sarah Cashman here was also getting into not just Houdini for the first time but but 3d for the first time we've been given some some other amazing concept art for the word there were referencing coral structures we wanted to so sarin I work together to adapt one of the one of the differential curve growth algorithms that were that were floating around and we'd have and would use that to grow coral shapes and so now we can generate loads of variations really easily to fill out assets this is another coral setup by Sarah we're starting with with with with the base branching structure that she modeled and Maya we would use a copy to point stop to put tendrils on the end of the branch again really simple but really really handy we're also referencing some crystal and mineral structures so he is Sarah adapted a differential growth set up for volumes that was made by in Farnsworth and then we use copper points to put you know sort of directives crystal shards on top now notice the nice layouts of the the network view it's got color codings got post-it notes this pleases me in ways I can't describe such a neat set up so this is an early test of putting stuff in shots so we've got crystals and the pink seagrass from Matt's seaweed from Xavier important tip here for Cheryl's not show stuff in context look at it in context there's a great tech test to see that that all about effects elements are moving down the pipe and we've got a few tech issues but saying in context with the early animation the movement is really distracting it's vital to get into a workflow where you can get lots of reviews lots of iterations lots of chances to spot things like that and fix it Matt also went a hand with some general assets addressing so we took this beautiful piece of concept art from Enya of a of a destroyed Space Station Steve Davidson did a beautiful job modeling a very clean space station which Matt then destroyed a big part of this process if you do this kind of work is we were running a split spoke workflow is pretty common in studios where the understood ship went to servicing and we were expected to to do this work to it but then afterwards in lighting it would magically apply all of those all of those materials map had to be very careful especially within our USD pipeline to ensure that all the path attributes the names and the Eevee's remained attacks so that it would shake correctly when it came into katana also lots of art direction on this we're by default when he shattered it it went into a billion tiny shards and we had to work out a way to get these nice big graphic chunks and here's some pipeline stuff that we've been doing so Ben was it was very quick to jump onto PG and see what it could do for us so we knew that we wanted to be able to do real-time things with our film and would need multiple levels of detail so what's coming on here is the first node that is querying shotgun to say what are all the assets in our film the next node then gets the far parts on disk the next node loads them into Houdini and it runs ten levels of a poly reduce the next node then defines assets paths back in shotgun and the next node checks them all in and they all become USD variants of the pro model so with one graph and one clique within ten minutes we've got LEDs for all of our assets the next step is we want to have this running off of a shotgun trigger so that's when any artist checks and model this will automatically fire detect any change models and we'll just run the multiple lods off of that here is viewing one of this models inside you'll see view and so you can see that he can go and choose the different blood variants and they're all there so because we're trying to do some virtual production stuff as well doing some other amazing work vendors down inside the unity and vibe trackers and iPads is amazing Pooky rig that's really cool we can take the shots which are all designed for higher purpose or high-end film assets and immediately load them up into unity and we can do lensing we can really takes and then we can publish it back into the pipe it's really really cool right now as I type as I talk the group are working on some real fun projects so jess is part of us all team working on a zero-gravity space sim now just as what a concept artists and felt that the demo could benefit from some zero-gravity water she dumped into Houdini having never used it before and with a bit of help from from Liz and the others look at the basics of flip and the game did export tools and got this into into unity within a couple of days it's pretty cool the self involves some work unity moving to this new call shadowgraph set up but we were a little bit constrained with the mismatch of versions between take issues with hardware and hololens and and where we're at with unity Ben Skinner was able to take some of the shader code provided by the subjects games team and convert it into a unity shader node if people are interested we can share that around because I think that's it's pretty handy Jess was able to leverage this tech for another purpose so in the game you have to watch your oxygen levels if it if it gets too low such as suffer visions and hallucinations the team came up with the idea of having certain of fireflies and odd creatures all in this case a bird so just found an animated bird rigged online pull that into Houdini converted into flip exporter that with the vertex animation textures and pulled it into unity which is a pretty pretty cool sort of last-minute result that she that she threw in I know a neo who I mentioned before is working on another projects where the art style is heavily influenced by so in Mobius and 70s prog rock album covers and you also started with no 3d experience and she's now across ZBrush Maya substance and now Cugini and katana so this is in an in-game screenshot of the environments that she's working on he's a setup it's very straightforward pipe filled noise masking erosion give her to Polly's and pull it into the game in terms of this sort of Cheryl stuff what's what sort of stands out about about what Enya and her team are doing compared to standards I guess she'll stuff showcasing terrain and high fields is how they're treating it and his amazing eye for color has remixed those basic terrain stuff into something unique something that'll sort of catch the eye over other people's Cheryl's you can see in this one that she's also incorporated other elements into the hunt field like like like giant fallen statues and canyons and weak happens and things so to do that and you created STF bounds for the shapes that you want to insert compared to Polly's bullying them outs while others were just a straight pipe filled projects where the ends where the edges were blurred and soften together it was pretty cool as if this next video plays it did a both been doing some really cool real-time stuff too he's working on a hololens project that'll scan a room and it'll then procedurally generate sort of a cartoony snowy environment over it now the prototype how that idea might work beau developed a set up that could take a simple input geometry and create these stylized cartoony snow-covered rocks that match the art style from our artists 10 it's worth pointing out to that beau has been jumping between several projects so while the work that the this going right now is a competition between five teams at the Uni in reality in a production you'll be asked to jump between products all the time and beau has developed a great skill set to to quickly task switch and adapts which will shine in a show rope because of the variety of work we'll have on his real which is pretty cool and that's a sue Danny we do a whole bunch of other stuff as well I'd say anything that I've shown here and soon as I'm moaning about don't look like it's toriel if you're doing if you're an animator and you're showcasing the same walk cycle that you get from whatever all my tutorials don't do that change it mix it up if you're lighter don't do what everyone else does and it show the difference you're doing your show will tell the story make it go on another painful analogy that I'll make is making sure I roll this like is like make Michelle to get into a studio is like is like being a guitar player and you're auditioning for a band if every auditioning guitar player plays the intro to stay with a heaven the one person who doesn't play that is the one who's going to get hired furthermore if you only ever played guitar in your bedroom and you've never learned what it's like to play in time with a bass player and a drummer you have to learn how to be be part of a team then this to me is one of the most important reasons for going to a school to practice being part of that environment if this school isn't giving me that environment of being in a big team of thirty people working on one project just do whatever you can to force that environment you know get a team going with the fellow students do some collaboration stuff online it'll make you better artists and that will come through in your shell and that's my talk if you want to know more about us you can search there any questions whoa yep looks like you have a solid production level pipeline can you talk about that who managed that in how is it created yep so our school is run by three three leads so there's there's myself taking care of our effects bit of real-time and the hand waving the hand waving is my most useful skill Chris who is in charge of animation and and and performance and their creative and then identified who's in charge of all the tech and the backend at the side of last year dan decided we should overhaul our pipeline and go on follow this new fad called USD 10 if you heard of it so he had a very broad idea of what to do and we had two students join us to be pipeline titties to be developers one was Ben Skinner another was went and and he just very loosely directed them but it was it was all Ben and Ben and when who built it pretty much from from scratch so a lot of it was how do we load and save assets through USD how do we then check them in through shotgun how do we build shots that are made up of hundreds maybe even thousands of separate USD assets and then how do we push that onto the farm efficiently and how do we track versions one of the one of the tools that they made was to have our own acet resolver so if we have no card coded pads on disks for all of our assets with we would spend most of our time manually updating paths that they'll be exhausting instead what we do is we have more of a generalized database query so we say give me house 5 version 7 or we can say give me house 12 version latest and this this little help kit in the middle it will read that query us shotgun hey where do I find that on disk and it says I'll you find it here and I'll then pass that back to whatever app is after it we also had to make sure that it was going to run very efficiently on the farm because if if if shotgun was getting hammered by hundreds of nodes asking for thousands of things every second it would fall down so they had to come up with this course and messaging system and catching stuff that hot library we had to give it a name so we called the turret that's all open-source so we submitted that last year for the shotgun Awards and we won it was pretty cool but yeah you can just grab that so if you search for turrets USD you'll find it anyone's nono heckles and the entire presentation I'm very disappointed in you people card tricks a few card tricks all right yeah go on why didn't we use people under somebody I and then and Chris actually we're all at dr. D where we did try to use Houdini for final assembly and layup that was going about six years ago that was train wreck its we found even with sort of experience lighting artists managing that kind of stuff is was difficult that build whereas whereas katana it's got a pretty solid track records for doing that and we try to be as close to industry as we can so most places are asking for people who do katana so we said okay we'll do katana for that the new lob stuff that looks pretty cool so I think we'll be we've already been playing without a little bits it's probably a bit too late for us to change this year but next year you know as well see any other questions primarily my Max and I'm we're trying to motivate into Houdini is it better for us to go to the pole Houdini package or do that would be neat engine within Maya in terms of getting our like automation sure depends which want to do the PDG stuff that uh has been hella been so so been over there in the back he can have a chat with him if he's so inclined by using PDG PDG as you saw from that little demo of making levels of detail it can talk the shotgun and it can and it can if can control Houdini but it can also control my if you want or it can control ffmpeg or imagemagick or whatever if anything can be controlled from a command line then PDG can can drive it and do stuff the Houdini engine stuff is more for if you make a cool tool within Houdini but you want those sort of controls exposed in your software you can then get to it that way so both the easel both are sort of four more slightly different tasks but what's kind of nice about it is you can you can put it in as gently as you want I think you know it in the past it would have been like a big ass to say you know what we're gonna scrap everything and we're gonna go this way whereas the way the tools are going you can you can just uh Fitness seep it in through the backdoor and then you can sort of do a little bit more and a little more and then eventually you're using bid any for everything in your life is better I've got one more minute any other questions going once going twice it's all we're done thank you very much Trey you [Applause]

Channel: Houdini

Views: 53,719

Rating: 4.7192984 out of 5

Keywords: vfx, visual effects, visual effects software, vfx software, cg, cg software, fmx

Id: Y22rYfJE64o

Length: 48min 51sec (2931 seconds)

Published: Tue Aug 06 2019