Houdini 16.5 Launch Presentation

[Music] welcome everyone it's my distinct pleasure to start tonight's event the kickoff the official launch of Houdini 16.5 that's a celebration right there it's a very special night because we get to reveal what we've been doing the last few months publicly so we're going to see a lot of great stuff our D as usual has been packing a ton of features and enhancements into this release you'll see something there for everyone I think of Houdini 16-5 actually personally I think of it more as a full release there's so much in it just masquerading as a dot release it's also special tonight because we're in Toronto yeah yeah let's hear for Toronto that's sort of we wherever we go around the globe we're always received with a warm welcome but Toronto is home home to headquarters in the majority of the staff so it's very special to be at home and doing a launch at home and there's another reason it's special this year we've been celebrating our 30th anniversary that's the group gathering earlier this year the staff gathering in Toronto earlier this year and what better way to celebrate the 30th anniversary than to have not one but two releases Houdini 16 in February and Houdini is 16 five today and speaking of Houdini 16 we are so happy with the recent the feedback we've been receiving the feedback has been amazing from the user community we've seen the results grow in games and Studios and freelancers and schools around the globe we've seen growth grow across the pipeline from modeling to lighting and rendering and we all this increased feedback actually inspires us more and that's why just nine months after Houdini 16 we're able to produce the R&D department that's been inspired to produce Houdini 16-5 here tonight so without further ado and to walk us through some of those features of Houdini 16-5 let me introduce our Director of Development Kristin bargain Kristin thank you Ken and good evening everyone it's wonderful to to be here I think him is right it feels to us like it was only yesterday that we released Houdini 16 time has gone by incredibly fast I think it's fair to say that it was the great response that he offered to to that release and the feedback that followed it that put us here today so quickly and with so many features ready to show to you and I should clarify that Houdini 16-5 is not just an iteration on the features many features that we introduced a couple of months ago it also contains a lot of tools that we had prepared for a Houdini 17 that are ready already and this is the way we are you know us we don't work with marketing sheets we don't wait to build a certain number of features before we say go if something's ready you know we did we put out builds every night every morning for you so if we have something ready for you that you can use today we will do it right away and that's what we did with a lot of the work that is actually still ongoing for Houdini 17 so yes Houdini 16 five is feature-rich but beyond that I hope that what we're giving you is something that's relevant to you in production because whether a feature is spectacular or small big or small if it doesn't serve a purpose to some of you or one of you or all of you then we haven't done our jobs and I also hope one more thing that this release expresses not just the love that we have for the industry and for the work that we do but also expresses the commitment that we stated to you already some years ago that we want to continue to lead in visual effects while also very seriously tackling areas outside of that like modeling and animation and you know you've ease and and so on and so forth so shall we take a look and see what we got Scott if you're ready alright this is Scott so Scott and I will be presenting to you tonight in fact it will be much more Scott and and not not that much of me and which is fine by me anyway these are the features the areas rather that we want to cover tonight character modeling rendering user experience and dynamics so let's begin with the first one character and you know it's not just a random coincidence why we take character first if you look at everything that we're doing for character which is crowds rigging animation muscle and flesh grooming it's a lot of stuff a lot of work that we put into this release for character so it kind of deserves to be the first one it also speaks to the needs of a growing community some you know new arrivals to Houdini some old but equally passionate about delivering great quality characters and and great quality animation and also there's a fine mix here between traditional rigging and animation and the things that we're really known for which is visual effects so we have a nice blend and hopefully a nice balance between them so starting with crowds our goal there is to bring the crowds as much into the foreground as possible in other words to have that degree of quality of realism and in the crowd system that allows you to be daring with bringing them more and more into the foreground with rigging in animation that's pretty cut-and-dry I just want to make it clear that we are very committed to that role and we might come toward that goal from two sides you know a traditional animation and then heavy-duty effects but in the end the two of them should merge together and serve you better than each of them separately and then muscle and flesh well we did talk about that when we released Houdini 16 those were the early days and we've been really busy improving what what we put out then so very happy tonight to tell you that these tools are now ready for primetime and more so we've actually been back forwarding a lot of our work into 16o while developing the features for 16 5 and 17 but well you'll be seeing tonight are some things that we've reserved especially for you tonight that have to do with physical simulation fvm and all that crazy cool stuff and finally grooming this is one of those items that were meant for Houdini 17 actually because we had put out a major new framework and and and well-endowed with features in Houdini 16 but in the end we got a lot done and it's all very good production savvy work which I think deserves to be released tonight so that's my intro Scott can you take it from here please sure thank you right so we're gonna start off with crowds and sort of as Kristin was just mentioning there and is in his intro you know we've been very happy with the adoption of crowds out in the industry and our goal is to go from you know what's essentially a background effect and push them more and more into the foreground and the way you do that is you make the tools more flexible allow you to do more things with agents so in this case what we're showing here is something called a partial rag doll but also an idea called hierarchical if I can get that word out agents which basically means the horse is an agent and the characters are agents and they're tied together but are able to obviously come apart like in this example and a partial ragdoll basically means kind of exactly what it sounds like it means instead of becoming a complete rag doll you see when he gets hit the upper body becomes activated becomes a physics driven character but from the waist down he's actually still following the underlying animation clip and then after a period of time he goes full rag doll and he sort of slides off the horse on to the ground and these kinds of effects the ability to blend not only animation clips but physics and animation and then all the way to full physical simulation is how you make crowds more believable how you can put the camera closer and closer to the crowd so just really quickly I just want to walk you through how you actually set that up so here just zombies they're falling down they're going full rag doll and essentially all you need to do to create a you know a partial reg doll is to identify which piece of the body you want to become a rag doll and the way we do that is with this node called the transform group agent transform group and it really is basically just a matter of putting in which part of your underlying skeleton the underlying rig is going to become a rag doll and we have some nice little guide geometry here to show you so for instance the lower body or the upper body and you can actually blend in as well so if you don't want the you know the guy to fold in half at the waist when he becomes active you can sort of blend in the physics using this depth slider or even this ramp that you see here for our example we're just going to turn that off and basically we're pretty much done we jump back over to our simulation Network here you can see we have the ragdoll state and all we really need to do is turn on our partial ragdoll input the group that we just created and basically we're ready to go okay play our guys walk and around here they sort of flop over and hang so you get this nice blend between animation and physics that sort of gives you a lot more variety than you could get by just trying to pump out you know hundreds and hundreds of different animation clips for every possible scenario so here's just another example of a rider and a horse and of course he gets hit with this projectile and then the horse hits this wall and falls over so these rag dolls can be applied to quadrupeds as well basically any kind of rig that you have driving your agent in your simulation and quadrupeds sort of needs some special attention so here we have something that's really interesting which is actually keeping something in a full ragdoll state but then driving them using motors so on the Left we have just full ragdolls they're just hanging there they have no animation at all on the right they're still full ragdoll but the lower halves of the body is actually using the walking animation that you saw earlier so if look closely you can see their legs sort of kicking and flailing there so imagine having a suite of animations that are meant to be physical animations right that are half physics half animation so you can get this nice blend of not just you know on or off but a blend of motion and rag doll together so as I was saying so quadrupeds requires some special attention when it comes to crowds because of things like this you have an animation of a horse running you need them to make a tight turn and realistically they need to sort of lean into that turn right and you obviously could make a bunch of animation Clips where they turn left and right and so on but this just helps reduce some of that overhead reduce some of the artists time and lets some of the simulation take over you know those small details another thing that's important is terrain adaptation we already had that for our agents our bipeds but for quadrupeds you need to pay a little more attention because obviously they have long bodies and more legs so they have sort of special needs when it comes to adapting to terrain but we have added that in sixteen five so moving on to character and rigging and animation something that we've done is we've taken something called the pose tool and post tool was something that allowed you get access to special animation handles things like our invisible rigs that you're seeing here in the viewport where you just click on a piece of geometry and move it and so we took it from this specialized tool we actually moved it on to our tool bar here so you can very quickly get access to these controls without having to sort of invoke this special tool but not only that we've also added at the top there you see something called motion paths so this is also a fast way to get to this idea of a motion path and a motion path is essentially a a space-time curve that follows the controls on your character so you can see here this path through space in this case it's attached to this wrist controller on the the - or there and you can basically in 3d space manipulate the path that that is going to travel through over time and there's various ways to to visualize this path widths we can color the curve based on acceleration and so on but you also have Bezier handles in the viewport to control this these animation curves it just becomes a much more natural way for an animator to tweak a result rather than going into this sort of graph editor which is a more abstract way of understanding how something is moving and we're also adding an auto reg system for heads now this isn't a full auto reg system that's not a facial rig you're not doing blend shapes and all that kind of stuff this is a very basic rig it's a module that plugs into our existing auto rig system but you can see it gives you a lot of nice controls right off the bat like these look controls you saw the jaw just a moment ago and things like this about like how much the eyeball influences the eyelids for instance so even though this is not like a super complicated rig it is a nice automatic way quick way to set up a face for basic animations but still gives you some nice nice control some nice deformations and nice control over the eyes of course in the future we're going to be working toward more complicated auto rigging for things like facial capture and so on but for now this is sort of a step down that road so muscle and flesh is something that we sort of introduced in a sort of a beta state in Houdini 16 and as Kristin mentioned we have been sort of actually updating those tools all through the life cycle of 16 so if you've been getting daily updates you've seen the new tools as they've been as they have been built but here we have some new features that you haven't seen before and that's this idea of an isotropic muscle contraction so this is not a de former we're not deforming these muscles to create this motion here this is purely in the simulation we're basically triggering the muscles to flex and what you're seeing here is the result of the simulation after we've asked that so we say flex along this direction and you can see they tighten up they stiffen up and they bulge because of all the nice volume preservation all the stuff that comes along with FIM and then of course we embed those muscles inside skin essentially and you can see that when the muscles contract and flex under the skin they naturally pull the skin with it and you get this nice sense of muscles beneath the surface flexing and then a more complicated sort of example here you can see our monitor now running with the full simulation of the muscles under the skin and the effect to pick particularly on the sort of stomach and the pectoral muscles in his back you get a nice sort of natural aliveness just by running this simulation one nice thing about our muscle and tissue solver is that you actually don't even need muscles you can actually just run the animation from anything that has essentially a skeleton and drive a tissue sim on top of it so you can actually work at any level you want full muscles just skin just driven by a skeleton and now grooming so here's a character that we've roomed using the tools in 3d 16.5 he's rendered modeled in Houdini obviously the grooming is Houdini and you can see you know you get a nice natural effect but really the purpose of the updates and 16-5 are to get you to the end result faster in a more sort of logical way so we've served changed how you approach fur so we're gonna walk through some of the examples here and one of the biggest ones is that grooms now can be sort of piecemeal so you can see we've added a groom to the head then we're going to add a groom to the sort of the chest area here and over here in the network editor you can see we've created separate groom objects now so instead of everything leaving in one room they can be broken out into individual groups and then merge together using this new sort of object level network and the nice thing about that is that you can still get your full hair generation for your whole character but then each section of your groom becomes controllable individually so you can see changing the length just of the head but still getting a nice blend between that head groom and that body groom and then a nice thing for for people who are doing these kinds of grooms a short hair and long hair is we can now have different segment lengths per guide hair which is something that we could not do in the past and it really speeds up the way you work and makes it more flexible so here's something that we're doing that's kind of interesting for laying down hair very quickly right so you want to get a basic room you're not doing anything detailed here but what you want to do is set your overall hair direction and what we're basically doing is with these curves that you see being drawn directively there in the viewport we're essentially advective the hair through a field so as you brush you just affect the hair in the direction you want and let your very quickly lay down a base a base groom it's not super detailed even though you can get in and add smaller details in places that you like but it's a real fast way of building up a simple group another new feature is this idea of taking a 3d shape in this case these sort of bangs and the goatee and essentially filling them with hair and this actually works in a very similar way to the last tool in that we invent their from the surface of the skin through this shape towards the outer skin so you can see you can basically model a 3d shape which is something that you probably have anyway and then fill it with hair in a way that makes sense actually using a velocity field just like we were using in the previous example and here you can see putting these sort of Tufts of hair on the arm of the character there and obviously all the same controls you would expect from any sort of grooming tool so something else you want to do is try and keep you as procedural as possible it's very easy when you're grooming a character to start thinking about I need to go in with a brush and brush every hair into place and eventually you're going to do that but we want to make sure that you have a lot of tools at your exposure disposal to do things like mask off areas and apply procedural grooming tools to those so in this case we're just adding a lift to this sort of chest hair using that piece of geometry the box that you saw and this means that you can mask off areas and see the results sort of procedurally in your network here and a nice sort of intuitive way we've also increased the ways we can use noise to mask out different parts of your groom which is a nice way of breaking up your groom making it less sort of even overall giving us some more variety you can see we've got nice lots of tools now for roughness and levels of noise and also you can paint the attributes to paint where the roughness is for instance and we also have some nice presets now for things like length mass so this is a mass that says any hair longer than a certain amount put into a mask or hair shorter than a certain amount and so again it just gives you a nice intuitive way to control which part of your groom you're working on which tools which styles are being applied to those areas and then mix them together in interesting ways like this random mask multiplies against the previous mass so you can see if you wanted to add gray hairs or white hairs of tired hairs this is a really nice way of doing it so this whole system now is a nice fluid way of working procedurally but also directly in the viewport using sort of standard kind of brushing operations another nice thing is that we've tried to tie a couple of these elements together so what's happening here is that obviously we have this fluid it's affecting the hair simulation but we're also doing a couple of things we're using the attributes from the fluid to drive the groom merge that you saw earlier to mix between wet hair and dry hair essentially so this is actually a physical simulation driving a groom which then in turn drives the actual shader as well so you get the sort of wet hair effect in the correct areas and that sort of ties into our whole philosophy of as much as possible making all the pieces of Houdini sort of communicate with each other so that you know we don't have to work really hard to get these things to play nicely together we're gonna bring Kristin out to introduce the next section but just before we move on to modeling maybe one more thought on character which is as much as it's important to move quickly on feature development it's just as important to have learning material prepared and and we know we've heard you with seeing the forums people who want to dive into Houdini animation and rigging are asking for them we've provided some and now that we're at this stage with our software we feel very good about starting some rapid-fire tutorials and master classes on on rigging animation and certainly the muscle and and flesh the FBM simulation stuff that Scott has just shown you so that that's coming up soon and in fact I know that we have a grooming master class an update on what you just saw here that will be ready for the with the release ok so moving on to modeling I've list terrain meshing and you know UV workflows as the main things that we've focused on for 16-5 it really is a continuum I mean software development is a continuum but but modeling in particular is very important to us because it serves so many needs you know we may not have been known for interactive modeling but people have been modeling procedurally in Houdini buildings of sophisticated things since day one so it is important because it is for traditional modelers it's for VFX people it's for generalist it's for everyone that these tools are built and we want to do not just a good job but a great job so our our long-term plan vision is to have a super competent polygonal modeler and with the top procedural compute you know computational geometry tools in the industry and and hopefully this next iteration will tell you that that's the direction that we're following so terrain was another one of those big entries into the sort of Houdini landscape with Houdini 16 and we have a nice addition you'll see and 16 5 and we're continuing with that meshing is another big obsession we've we've had for a while starting with the noir the particle remeasure and whatnot and we have a quad remeasure planned but but the big the big one this time around is a poly reducer and yes we've had a pall reducer for some time but this is a completely new algorithm and set of algorithms and we're as excited about it as we were about the boolean saw a couple of months ago this is its cousin and and not not coincidentally it was written by the same developer so the other thing I want to say is as we as we renew our software and we do that continually and in some cases we catch up and in some cases were purely innovating with every step that we take forward which we try not to just meet the status quo but beat it and is one prime example of that and Scott and I had so much fun putting together the slides we could have done this whole demo just just on poly reduce so so check it out and then you v's that's one of those things that's become more and more important to us as we've been addressing the needs of people in games but not just not just them of course and it's not just workflow you'll see there are some really cool tools and algorithms that we've prepared for you for 16 5 Scott ok so let's jump into terrain so again as Kristin mentioned terrain was basically put into Houdini and Houdini 16 and you know a lot of the features that were working on for the future were really planned for a Houdini 17 but this tool came along and it was ready to go so we figured we may as well put it in 16 5 and let people play around with it so let's take a look at an example here of something that is a very typical request you have a piece of terrain maybe that's a scan data maybe another artist already made it and there's a piece of it that you really like and would like to put in a different terrain if you just mask it out using the tools in 16 while you basically get is something like this you you do separate it from the landscape but because it's on this hill and it's wobbly it sticks up off the landscape makes it very difficult to cleanly blend into another piece of terrain so this new tool this height field patch tool takes this patch and actually flattens it it does some interesting work to take this sort of very distorted thing and make it into a flat patch that you can apply to basically anything so let's take a look of it on an actual example here so you can see that same patch now and how it seamlessly blends in with the terrain from this patch that we've carved out now you can imagine that one that was sticking off the landscape that would be much more difficult to blend in here right but because it's flat it can basically glide over the whole surface of your terrain and blend in like this and then obviously you put your shading and your texturing on top of it to blend it all together for your final image so this is going to be a it was already sort of a requested feature but I think it's going to be really helpful especially for people pulling in multiple data to create a brand-new landscape so on the modeling side again there our topology tools are retopo tools are just continuing to grow on every release so this is something that we're calling a skin operation where basically you just draw the curves you can interact with it you can bring a bridge across and you can also draw the if the flow of the edges so then of course smooth the result and get it all nicely blended in here but again this is sort of bridging that gap between somebody who wants to work in a really interactive way versus somebody who wants to work in a totally procedural modeling way something else that was often asked for is this idea of a geodesic or edge based soft radius so you can see on the left our older version which more or less ignored the topology there is some some following of the the face there but you can see on the right that the soft radius needs to go around the mouth before it can start spreading into the rest of the face and that makes operations like creating blend shapes or just modeling in general you know much easier it's often asked for idea but you can see here how annoying it would be to try and do this to the mouth if the soft radius sort of just spread all over the face a very typical example when you're doing blend shapes or things like eyelids right so you want to affect just the upper eyelid and not the bottom eyelid and of course you don't want to go in and create groups and all that kind of thing fingers are another big example here where previously this green mask would have spread to all of the fingers as you increase the radius but now it stays nice and topologically consistent making this a much nicer workflow for people who are doing this kind of character finally or prepping for things like blend shapes something else that we really wanted to focus on is this interactive modeling in the viewport but still you know giving you a nice tool so here we actually have a new volatile key that you can hold down and you see this little widget showing you where your geometry is going to go and basically you can quickly orient one object to the surface of another object just by holding down this hockey and of course you get this nice visualization there are multiple ways to do this so in this case we're doing the same operation with this different modifier key to just do the alignment without the move and then of course connected to that is the idea of just moving it without doing any alignments at all but really the whole point here is that you know we had a tool similar to this previously but it required you to go into a menu select a bunch of points before you can move the object and now this becomes a much sort of more alive thing that you can just do as you're working and just to give you something a little more in context of how you might use this typically we've got this bolt we just want to slap it onto the surface of this spaceship and then do a duplicate so again not a complicated modeling operation but rather than having to break out of your workflow of being in the viewport and using our radio menus which is what you used to have to do you can now stay in this workspace and work the way you kind of hope you know only leave the viewport when you absolutely have to so poly reduce so poly reduce is as Kristin mentioned completely rewritten from scratch this is a completely new tool and one of the interesting things about it is obviously it Polly reduces things but it also has some nice features built into it for instance caching so this video is not sped up or anything it's actually sort of a live video so you can see that once we've cooked the initial reduction down to like 1% or something you don't pay the cost to redo that every time you change this slider the results are cached all the way down to whatever level you did your reduction so you can see that I can go from 1% almost in real time scrub this handle and go through all of the levels of subdivision I think it caps out around 300,000 polygons and goes down to a thousand or something like that and again this means that you can quickly find the value that you want without feeling like oh if I don't get it right away I'm gonna have to pay this huge cost to redo the cook every time something else I should point out is that we are properly preserving vertex attributes like UVs and we can also get this to go ahead do something really interesting which is preserve quad topology so typically a poly reducer will only give you triangles as a result but here if you've given the poly reduction tool a quad mesh you can actually preserve that quad mesh as you reduce so this isn't a remesh operation right this is actually just removing quads as you go and preserving all your edge loops so you can see how low you can get on that arm and still preserve the exact texture map the exact you've ease but drastically reduce the number of polygons but keeping that really nice topology that you worked so hard to get in the first place when you built your quad mesh and this would be really important if you wanted to run the same animation for instance on a whole bunch of different levels of detail model so there's lots of ways to control this we actually don't have time to go into all the ways that you can control this this this note is very powerful have lots of ways to interact with it but one of the sort of clearest one is by painting a mask right so in this case we've said well we want high density triangles in this area and low density triangles in another area and obviously that works pretty much as you would expect and this is important right you've got a budget of $2,000 yawns you want more of them in the areas that are important to you of course this can be sort of adaptively done right so if you give it a soft mess you get a nice soft fall off so you don't get those hard edges between the areas of poly reduction and that's really important again just giving you lots of control this but this is about density you know I want to put more polygons here and less in some other area but something else you can do is you can influence the weight of this reduction with the idea that you want to keep a certain amount of information that you've given it so let's just walk through a simple example here it'll be clearer so here we just have a basic piece of geometry with some UVs and what we've done is basically baked out some lighting and then we've applied it directly to the points on our geometry okay and you can see on the right how dense the geometry is and that's because we have all these sharp edges and so on in order to keep that detail we need a very high res mesh because we're storing it on the points so what we can do is ask the poly reducer to say okay take this information and preserve as much of it as possible while reducing the overall poly count so rather than saying I want more here and less polygons here instead you're saying find where the changes in this is and preserve those changes so that the end result looks as much like the input as possible so here we have our two million ish polygon grid and if I go forward you can see now this is a drastically poly reduced version but you can see we've kept all of the sharp edges right if we look at the underlying mesh this isn't changing the density just where it happens to be darker instead it's putting polygons where the attribute changes right so you can almost see a perfect outline of the shadow on the ground but then in the perfectly black areas it's done a lot of poly reduction because there's very little change in the underlying attribute and a slightly more complicated example here we've just done a soft shadow instead of a sharp shadow just to show you how so I just changed the slide there if you didn't you didn't see it how well it preserves these change and attributes along the edges so that it becomes you know as much as possible that you can't notice even though we've drastically reduced the number of polygons and then here is the result so again in the middle where there's the darkest areas there's also the little the smallest amount of change and so the poly reducer is said I can remove a lot of polygons in there because it's uniformly dark but around the edges as it fades out it adds detail to preserve that sort of gradient from dark to light so this is going to give you a lot of power with how you shape your mesh particularly if there's attributes that you need preserve like for instance sometimes a game artist will embed attributes to trigger certain sound effects on a mesh and this will actually preserve those types of attributes as you go poly reducing it down to something more friendly to something like a game engine so something else that's important is you know sometimes you want to reduce towards a camera right you're not just doing an overall reduction you know this object is only going to be seen from a certain position and so in this case we're doing the second character here with the red it's just a straight poly reduction without any waiting at all it's just reducing it down there like 2,000 polygons or something but on the right we're doing some waiting for polygons that are facing the camera but also polygons that are on the silhouette so if you look carefully you can actually see that it's not just decimating the back of the model it's adding more polygons along the edge where the camera can see sort of at a grazing angle so just to make that a little more clear so here's basically the same model reduced and you can see how it matches up to the starting point and it does a okay job but it's very jagged it's very obvious that it's been poly reduced whereas if we look at this one with the silhouette preservation you can see that the face overall is dramatically poly reduced but the silhouette has far more edges on it and so it matches up to the original silhouette almost perfectly and that becomes again really important when you're talking about this is a background element I know it's only going to be seen from this point of view so you can go ahead and reduce as much as you want so long as you keep that silhouette there's actually a whole lot of other things involved with this for instance you can add a volume and say reduce based on a person who's gonna walk around this volume reduce based on a path that the camera might follow so anywhere this object can be seen along that path there's a lot of features but like I said we actually just do not have time to go through all of them so the nice thing you can do with this tool is obviously do many reductions like this take one character reduce it apply the same texture map to each one and you get a really nice retaining of the detail from the very high-rez down to just a couple of hundred polygons the other thing is nice is that we actually did this inside of a for loop so we just repeatedly reduced and reduced and reduced and because of that caching that I mentioned earlier when you do a an operation like this where you keep doing a loop and reducing and reducing and reducing the cache is put back into the beginning of the loop each time so you actually do less and less and less work on each loop so the the loops get faster as the more of these reductions you do so okay so we've got UVs we can preserve them now we want better ways of interacting with them better ways of dealing with them so here I've just got these two objects they both have their own UV sets obviously the green and the blue so the new UV layout tool has a really nice feature that rather than putting everything in to zero one space we now have this nice handle where you can select a region to pack into right so I can just interactively move this handle and say this object is more important for whatever reason so I'm going to give it more space and leave room for my you know in this case the hammer because it's maybe doesn't have as much high-res texture detail necessary and so what you can do is pack obviously the green into this sort of top two-thirds and then the bottom third leave for the hammer and that just gives you some flexibility about how you lay these out in sort of more of an interactive way the other thing is that these areas can actually overlap so you can see that I've set the green you v's first and now this blue area that's being added as I expand this region find spaces between the other UVs to continue doing the packing right so you can take things and sort of add add things into the holes between your previous level of UVs and so to tie it back to what I was saying about reducing these characters like this here's the actual network that we use to do this you can see it's basically just a poly reduce inside of a for loop but what we can do is every time we reduce this character we can also reduce the UV space that it gets and so you can see we can have one texture map to unify all of these materials all the different materials that are on this object and then pack them into one UV space so all of these different levels of detail that you see here can be using one texture map which is just a really nice efficient way of dealing with large amounts of geometry like this so something else you can do is pack this information as tightly as you can possibly get it because you've ease you know they benefit from having as much desert texture resolution as possible and UV layout now can really find every piece of available space and fill it with geometry and then of course we have some nice new capabilities in the UV viewport for displaying certain things like the boundaries of UV Islands if there's flipped Islands if there's overlapping so if you've done this you think there's maybe an error you can quickly find it and do another packing with that in mind and this is just kind of amazing I think it's packed like when you zoom in there it's really crazy looking it looks like a city or something so speaking of the UV viewport here we've got just some small advantage small updates that really just make your life a little easier basically what you're seeing here is that as I select one of these objects in the viewport we switch our UV modeling view to those UVs but something else we do is we check to see if it has a material applied and if it does we list out all the texture maps that are applied to that object so you can quickly go in and say I want to look at my rubber toy and I want to look at it with you know its diffuse texture or its specular texture so rather than having to go just in Moncks all of your texture maps and find them we sort of source them for you in this menu something else to point out there you see that it says UV attribute and Auto and UV so in a big update that we've made that a lot of people may not even notice but it can be very important when you work with UV a lot is that you v's now can be just arbitrarily named attributes right so it used to be that you had UV 1 UV 2 and so on now you could make you know pig and toy you just arbitrarily name your attribute and all of the tools will work properly with this new UV type and so we've also gone back and revisited some tools that we've added recently to make sure that they automatically add UVs as you're working so previously you'd have to make sure you created seams and edges and track watch geometry you used now these tools will actually add movies for you as you model you can see them showing up here in the UV viewport which again just reduces the amount of work that you have to do down the road and gives you a nice interactive way of seeing exactly what's happening to your UV s as you're building them and if you look carefully at the interface you'll see that there's some options for how these UV x' adapt to the modeling whether they stretch whether they stay uniform we're basically just showing the rectangular version here but you can see how quickly you get this nice UV laya for these three different objects even as you play around with the parameters and then obviously if you would like at the end just use UV layout to pack them into one zero two one UV space and we're gonna move on to rendering and I think we'll get Kristen back thank you as many of you or all of you know rendering has been part of Houdini's life since the very first days twenty-something years ago and some releases have big spectacular features on with respect to lighting rendering and some are a lot under the hood sixteen five is a lot about doing work under the hood and the main thing that we focused on was performance and that one has so many areas that it touches for tonight I'll just mention one or two vex which which runs mantra in a big way right our render is very much dependent on how fast vex executes and so we've done a lot of work on the vex compilation optimizations that we have a new set of algorithms for vex caching that makes things like the old unified noise for example run much much faster than before and so on and so forth we've looked at IPR making sure that it's reliable that it that it updates much faster there's IPR autosave right now and many many enhancements like that for something visual that we thought you'd enjoy watching Scott will show you the the rounded edge tool that's been added to to mantra well it's actually evolved as well as you'll see some new interesting noise metrics and for those of you who swear by crypto Matt it's in thank you Scott go ahead okay Thank You Kristen so the rounded edge this is a very common issue with modeling where you have a cube for instance it has a very sharp edge but things in reality rarely have extremely sharp edges but you don't want to have to go in and model all those bevel everywhere so we've added this rounded edge vault that gives you this rounded edge normal look but also can do neat things like for instance create a mask on the edges of things and in this case I've just thrown some noise onto that mess so you can see how we've broken up that edge now we've got this red box it's the same box before just with this different material applied and now we can mix between them so you get a nice rounded edge where those sort of gray metal bits show through and we get a nice shiny metal surface elsewhere so this is a really really useful tool when you're doing modeling especially if you're doing a lot of things using things like boolean to do your modeling because you often end up with topology that's difficult to work with right it's difficult to add poly bevels nicely over something that's been heavily boolean so you can see here very sharp edges on the wireframe where the artist has basically cut holes and interesting bits into this object but then at render time you add all the nice bevels so you get those nice highlights along the edges and in this case he's also worked in a bunch of different shaders together using this edge wear technique that I just sort of walked through briefly and so this really is just a way of freeing you up especially if you're talking about doing concept art or a preview art or something where you know the mesh is never going to animate nobody's going to do anything like that to it you really are just interested in a final render so in this case you can imagine how if you did a boolean of all of these sort of holes into this tube how difficult it would be to actually add a correct poly bevel to something like that but here there's actually none happening it's purely the rounded edge shader doing this at render time something nice about this as well is that if you were to bake this after you've ease the rounded edge normals come into the UVs as well so you can actually store that rounded edge detail for a texture map down the road if you were going into games or anything like that so here we have these noise metrics that Kristin talked about now this seems like something that's like well it's just some new noise and to a certain degree that's true but this really allows you to create new types of looks procedurally you know this Manhattan and the Chev noise gives you this sort of interesting almost biomechanical look to things so if you imagine doing spaceship plating shaders procedurally or even actually used inside of the height field to create really interesting terrain detail the chebyshev noise gives you actually a really nice sort of broken up stoney look once you add some distortion to it so even though this is in a way a small feature it's actually a really valuable one to people who are doing as much procedural shader building as possible so they can avoid having to constantly create texture Maps when they don't have to something else we've done now in 16-5 is we've created this single sss so we've basically just given you a sort of this more simplified version of a subsurface scattering technique that you can pull out and add if you just want to add a small amount of subsurface to something you know because it is this single sss technique it's faster and so on but of course you can still mix and match them however you like really just another nice little under the hood change to give the artists a little more flexibility with how they approach hairless cats i guess so crypto met script them at is something that people have asked for a lot and it basically gives you the ability to pull out masks for scenes like this so somewhere where you have fog and volumes and a lot of different overlapping objects people will sometimes turn to deep camera maps and things like that but they have a massive data overhead deep camera apps are very large especially when you have things like volumes and so on and crypto mat is nice and light weight so in this case what we're gonna do is take a look at the crypto mat and what you basically get are these IDs for each object in this case as well as this volume but these aren't just IDs that you want to pull out they actually provide the mask directly so we also have a brand new cop to allow you to use the crypto mat directly so you can see here it's actually a little unclear in this slide but basically we're selecting the the fog layer at the bottom and you can see i'm doing it directly from the name of the object slash obj cloud from box object so this is actually a met you can access to create these masks and it will actually give you really nice masks even with these volumetric data so you can see with that one flat image you can pull out all these different mats it simplifies your workflow for compositing down the road and just gives you a lot of flexibility without the massive overhead of something like deep camera maps and for your convenience we've built this new comp to let you actually pull out those mats as well so it's a nice little system tied all together between the crypto mat rendering and the cop to actually give you access to these awesome masks and now we're moving on to user experience again user experience we know how important it is maybe in the early days we were more obsessed with with getting new functionality out well we still are but but now every work that we do every feature that we put out we run it through a very strict usability filter and and like like everybody I think we're learning and again I want to say how happy we are to see the the incredible adoption that we've seen since the release of Houdini 16 and how many great voices that has brought into the fold and a lot of that has contributed and will contribute years to come in making our software more usable but an equal thank you to those of you that have been with us for so long because you provide the right balance between what maybe used to be two worlds which we'd like to bring into one and and make Houdini not just you know a very productive tool but a tool that you're happy to work in not just with the procedural nodes and that the tools that we offer but in the viewport you know whether you're doing rigging animation whether you're modeling imperative ly to feel the same flow that that you used to enjoy in your other packages that's very important and it does take a few steps to get there but this is one big step along that way network editor well this was a big release in Houdini Steen but we didn't stop there there were some great pieces of feedback that came after that we still have a fairly long list but we've taken a couple of good bytes right into those for this release the private dialogue that's very much the cousin of the you know of the network editor for Houdini users and that one is seeing some really nice enhancements and I have a feeling that one of the features that Scott's going to show you that you'll be using on a daily basis maybe one of you more favorite ones in this release and then a hot key manager you could see a glimpse of it here on the on the left and then the viewport and again there's a lot of work that we've done there in the past couple of years some of it has been you know catching up with with the level of fluidity that that you guys expect and some of it is some new work that we are hopefully innovating with and all of them are coming together really nicely Scott can you show us please okay so yeah the network editor so you know obviously again a huge change in the network editor and 16 these are not as huge but they're very important so something that happens very often especially if you're modeling in Houdini as you get these long lists of nodes and it's a nice to be able to put them in a net box and hide them so in Houdini 16-5 we now have an adaptive net box so that when you close it rather than leaving this huge gap in your network it sort of compresses everything together something else that was often asked for once we introduced colors and shapes into Houdini is to be able to create a theme so your own custom set of nodes and colors shapes notes shapes and colors to apply to objects so in this case we're just making our light yellow our camera black changing up a couple of the shapes and because we've created this new theme now when I go ahead and put tab camera and drop down my camera it picks up my new theme so it's very easy for you to generate your own custom node library shapes and colors some other things that are just you know a little more subtle is the ability to for instance take multiple connections from and tops and plugged them into corresponding inputs on other nodes so in this case one it's a little hard to see there so what I did there was selected three parameters and connected all three automatically in one click to the node that's behind the bar there which again is really a small thing but when you're working a lot involves or you're working a lot connecting things together it's a really nice fluid way of working instead of having to constantly click click click so the parameter pane so first of all you can see we have these nice highlights on tabs and parameters and those are parameters with non default values but we also have this new search bar so you can see if I switch to parameters with non default values it hides all of the other parameters and only shows you the ones with non default you can see we have parameters with expressions time dependent parameters you can search for headings so for instance if I was a type shaped you can see I we just get our shape tools and no other parameters in our pane there I can search for specific parameters in this case OpenCL but it could be just about anything disturbance for instance and this means it's very fast to find if somebody has changed something something that you're interested in that you don't quite remember what it is and then this is a really interesting one where you can actually search for a specific value so in this case point 1 and every parameter with point 1 is shown in the parameter pane density in this case temperature and so this really makes the parameter pane suddenly a much more flexible tool to work with and you can even do neat things like create a list of search items and do sort of a quick customization of your parameter field the other nice thing is that this little icon that brings open the search bar can be stowed and opened and it maintains your search items it doesn't clear them and these persists across nodes so if you do a search on one node and then go to another node they both have their own search parameters there which now really just makes this a lot more intuitive to work with especially if you imagine somebody has given you a scene and you have no idea what they've changed you know in the multitude of nodes simply changing it non-default values you can see immediately when you click on a node what things have changed why haven't what things require your attention as a new artist coming into a scene for the first time so I think this combined with a feature that we added in Houdini 16 which is the ability to search your network so you can do a network search to a particular node and then continue to refine that search inside the node itself I think this is really going to change the way people work with nodes in Houdini so here's something else that happens a lot so let's say a very typical thing to want to do I have this pig head I want to move it to the centroid of our rubber toy and people who have used Houdini a lot they are very familiar with this expression called centroid it's extremely useful but it is tedious to type this in to all three parameter frames it's nice once it works it's ready to go it's great but through some feedbacks through from new users we had the idea of like well can we make some of these things these repetitive tasks that you do a little more easier to discover you know if you know the centroid expression exists it's very easy to type it in there if you've never heard of it you don't really know how to find it we now have something called get reference or the reference menu so I can go ahead now and right-click on that translate parameter and you can see this new item called reference I can grab scene data I can actually use this tree to navigate or just click on the node or the object you can see I'm going to my pig head or my toy here and I can grab the bounding box and the centroid and hit accept and it fills in all three parameters for you and fills in the expression right so this again just makes your life that's a bit easier right if you do this a lot now it's much easier to do it but this is a very simple example imagine instead of wanting to do something complicated like I want to take the worldspace rotation of our toy here and I want to apply it to this node called clip which basically slices our pig head in half so in order to do that it's actually a complicated series of expressions it's not one expression so I can go and get the relative world direction and you'll see that what it's actually done is filled in a very complicated expression that would have been very tedious to do multiple times but you can see now how quickly I've set up this nice little procedural network using tools that are readily accessible to you even if you have no idea how to convert a matrix into another math thing another thing to point out here is that this reference can also reference local variables which again just gives you a list so if you're not familiar with what local variables exist you get a drop-down list and also a very useful one on local attributes so whatever attributes exist if you want to reference another node and get its attributes you can directly go ahead and grab them something that we're not showing here again just for time is that you can also grab it from specific points specific primitives specific objects so they're not just object level references they can be right down to you I want to get something from this specific point you just select it in the viewport it's very interactive so in Houdini 16 you know for working in the viewport we added these things called radial menus and the idea is just to make your life a little easier right you can sort of focus on what you're doing you can stay in the viewport you can quickly get access to these different operations and we bound them through some hotkeys that you could switch to get different types of radial menus depending on what you're doing but very quickly you know it's very obvious that I may have a lot of these I may have customized lots of my own radial menus and we'd be nice to be able to just assign them to any hockey that I happen to like not just a list that we happen to provide it for you so now I'm going to our new hockey manager which shows you the radial menu you want to create a hotkey for what keys are available or taken at this level you can very quickly apply a hotkey just by clicking that button and now I just press that hotkey to bring up this menu so the hotkey manager you know it's always existed you've always been able to go there and do things but it was very cumbersome it was complex to use and it was difficult to understand because you had to go through lots of menus and so on we've changed into this much more visual way of working and you just click on the thing you want to create a hotkey for you bring up the hot key manager and you can immediately set it using this nice sort of graphical user interface again I think we're suddenly going to get a lot of people setting their own hotkeys while they're working now instead of ignoring it because it was very difficult to use in the past something else that's very useful especially if you're coming from the gaming side is that we can now drive a camera in Houdini using a controller input in this case it's the PlayStation 4 controller but it could be an Xbox controller and so this is a camera a new camera type in Houdini but it's actually being driven by a chopp network so this is not just a specific camera it's actually a chopped and that means that you can do procedural effects on top of this controller input you could record your controller input add noise and jitter and so on you could use this just a preview a level like in this case but you could actually use it to drive things because it's purely just a chop and so in typical Houdini fashion rather than just give you a controller we gave you a controller and a giant toolbox full of other things you can do with those controllers and then something that's really amazing is some of the enhancements we've made to the viewport when it comes to dealing with massive amounts of geometry so this is a cityscape you can see you know all the stats there above I won't bother to read them but the point is you know Houdini's been great at dealing with large amounts of data generating large amounts of data but when you're working with it in the viewport it becomes you know cumbersome it slows down your viewport but by changing some of these parameters we can sort of live remove geometry from your scene to give you a nice interactive experience if you look in the bottom right you can see the FPS counter here and so what's happening is that packed geometry or Alembic data is being loaded or unloaded based on your viewing angle and so when it swaps out the buildings it just swaps in bounding boxes so you still get a feel of your scene but without paying the price of showing you know millions and millions and millions of polygons and so when you're dealing with you know these giant cities or any kind of large amount of data it becomes really valuable to have this kind of interaction in the viewport to set up things like lights and camera or shots or any other thing else that you might need to do and keeping it things sort of nice and light but you know if you want to show hundred million polygons you absolutely can and we've been moving on to dynamics this is last and definitely not least dynamics we've put in a lot of work into physical simulation this release and we will in every minor or major release that we put out it continues to be our bread and butter and the place where we want to lead and in fact increase our lead compared to to everybody else out there so a couple of things this time around optical flow well that's our foray into computer vision and we all know that there are so many uses in our industry or industries related to to computer vision and we can't it we can't wait to explore them in this release we're using optical flow purely for VFX as Scott will show you in a minute then Aaron compressibility well that's dealing with air entrapment bubbles but doing that in a you know in a physically correct way that will help you implement more realism in those scenes that require it and then narrowband which is it's just I was gonna say just an optimization to flip simulation but it's a very important one because while it may be you know prosaic in what it does what it buys you in terms of productivity and you know and memory use is is fantastic so without further ado let's talk about dynamics please okay so optical flow so the idea in case you're unfamiliar is to take a source 2d or 3d source and create a flow field from it based on the sort of implied motion something that we're not going to show today but it's very important to point out is that this also works on 3d sources so if you had moving 3d things that had completely different topology on every frame and you needed to generate a velocity field from it you can actually do that now much in the same way we do it for 2d images so the idea is once you've taped created that flow field you can apply it to effects effects like this for instance and really this is just a fluid simulation with the color from the video put on the particles but every all the motion that you're seeing here is driven from that flow field so the the image itself is actually driving the simulation and pushing the particles around and I think this is gonna open up a lot of really cool interesting effects it's also going to be something that people who are coming from film or broadcasts who are working in motion graphics are really going to probably love because you can see it creates this really dynamic interesting look while still maintaining the look of the 2d video and here's just an example of what that actually looks like from sort of a three-quarter view so you can see how it's creating these large splashes as the flow field moves through the fluid and just to give you an example of how this is actually working and how easy it actually is to plug into your sim you can see we just bring in our footage through cups we use our new cop to network from 16 to pull that into soft land then we have this node called volume optical flow these are pretty much the defaults and we can go ahead and visualize the flow field you know not quite real-time but very quickly and so the idea here is that you can work that fast bring in bring in your footage tweak the results using a standard no pop effect by volumes we can pull in that velocity field into our dynamic network and immediately start pushing around our fluid simulation and so again not only is this a new tool that gives opens up these new interesting possibilities we also made sure that it all sort of plays nicely with each other because we want you to be able to go you know from cops to stops adopts or in more human speak from compositing through geometry to dynamics and that sort of seamless blending between those contexts is really important to us as we move forward with new versions of Houdini so here's just another example now again of this fluid simulation pushing around being pushed around by the the underlying flow generated from the image and then we have actually a completely different type of simulation using the same effect so this is actually a grain simulation and you can see how it's using the exact same flow field the exact same data but creating a completely different impression the feeling that there's almost something underneath the ground pushing these particles around so you can very quickly start to imagine not only really cool effects you could do with something like this driving effects using you know 2d animation but also anything involving motion graphics or anything like that where you want to do it you know I could very easily envision you know a title sequence using these types of effects so really just broadens the possibilities from effects all the way to more traditional 2d style rendering effects and it just looks really cool so err incompressibility and I'm not going to say that anymore so I'm going to say bubbles from now on on the left you know this is how the flip solver and Houdini dealt with air which is to say it didn't really deal with it at all so you see the flu it just pours out and on the right you get that typical of water cooler glugging effect as air that's in the lower sphere needs to move up into the upper spear in order for the water to change containers something important to point out with the system that we've developed here is that we're not using air particles if there's not there's not a separate group of particles to track where the areas we actually track the surface of the fluid itself and discover air pockets and then maintain the pressure within those pockets and that might seem like a you know a distinction that doesn't mean anything but if you imagine something something more like this where you're generating lots and lots of particles that whole empty area above the tank there would also need to be filled with particles so it's an optimization on top of everything as well as a simplification of your workflow so in this case in order to create this effect all you need to do is cut holes in your source right so you just use a sink to cut holes in the surface and it automatically will be created as a bubble and maintained in the fluid simulation so it's not just an interesting technical detail actually directly affects your workflow simplifying basically the other thing that's nice to know is that it behaves correctly with the other tools that we've built in our fluid solver things like viscosity you can see you know the bubbles behaving very differently on the very high viscosity fluid here and on the left working basically like a water cooler I was actually going to record a little workflow demo to show how all this stuff works but it literally is just turning on the button called air in compressibility and then bubbles work so there wasn't really worth showing you a video but the results are very cool this is maybe a little harder than just pressing a button but it's still a very cool result that would have just been absolutely impossible in the past so narrowband so what we're showing here with this shark jumping out of the water this is a pretty typical type of effect you have a large tank of water and there's something in there splashing around or doing something interesting the problem with these effects is that the reality is the interesting bit is only at the surface right you to basically don't care about what's happening below the surface but you're paying a pretty significant cost to simulate it you need a deep tank of water to get the correct type of motion at the surface so what narrow band allows you to do is say well I'm going to simulate that just using a volumetric approach and I'm going to maintain the surface using this narrow band of particles and that way you maintain all of the interesting detail at the surface you know all your splashing it's everything you'd want from a fluid simulation without paying this heavy cost of filling this entire tank with particles and simulating that transferring data from particles to volumes back and forth on every frame so saves you a lot of memory because you don't deal with that many particles and it also saves you simulation time because for the same reason you don't have to deal with as many particles you don't have the transfer data back and forth as often and the nice thing is it doesn't really have an effect on the result of your simulation here you can see you know a billion particles this spaceship rising up out of the ocean and apart from the fact that we're telling you this is narrowband there's nothing to actually tell you that this is narrowband the result of the simulation and should basically match what you had previously but as you can see with about a third of the memory in three times the speed and so this is a massive improvement to dealing with these enormous tanks fulls of tank full of water and the nice thing is that because you know it has much less memory if you're a smaller studio and you're trying to compete on these fluid shots you don't need a giant computer with you know 300 gigs of ram or something to pull off these simulations you can do it at a much lower amount of RAM and still get a really nice highly detailed result and again this just looks awesome and so that actually believe it or not comes to the end out of our presentation on new features [Applause] great job Scott thank you Houdini sixteen five releases on November 7 in a couple of days we're just getting a couple of kinks out finishing our Docs we have those master classes to finish up as well but in a couple of days it's all yours there's lots more to discover there then more than what we've shown you we hope you enjoy it and we hope you do great things with it and we'll see you in a couple of months to talk about seventeen and just before we end I'd like to give big thanks to our R&D team that's all our developers are TD's including Scott here our documentation nqa and our insurance and the amazing thing is that some of the features that you've seen today and that you'll get to enjoy in 16-5 war developed by our co-op students the majority or of them or all of them from the University of Waterloo and a couple of them just joined us a couple of months ago in September and some of the features are up here so big thank you to all of them and to all of you [Applause]

Channel: Houdini

Views: 113,756

Keywords: Houdini, SideFX, Houdini 16.5, Release, Preview, Sneak Peek, Overview, vfx, gamedev, visual effects, game development

Id: 26CTJ6kEWDM

Length: 73min 32sec (4412 seconds)

Published: Sun Nov 05 2017