 I know a few people in the second row. So my name is Fiona, I'm a producer at Blender now, my first conference with a red liner. So today we're going to talk about two short films actually, because Hjalte was like, yeah, but we didn't talk about George last year, so really like I should have my talk also, my moment of glory. I mean kind of another guy. Enric is not going to be here today, he's director of Wing It, he has also some family duties on the weekend, so Hjalte will have all the limelight. So it's not a fight, it's presented as more of a fight, it's a comparison, because this is what we do at the Blender Studio, right? They create and we share. And lots of things have already been posted on the Blender Studio.org and other tweets, et cetera, but all the marvelous things that those people are doing. But here we're going to get it made into some more details, some things that they don't really want to put them black and white, you know, on the website. And hopefully you will learn a thing or two, like I did in the past few months working on Wing It. So without further ado, let's give the mic to Hjalte, yeah? Hello. So, you know, you're an arrow key, okay, thank you. Yes, my name is Hjalte, and I have had a lot of different jobs at the studio. So, and for these two films that I'm supposed to compare, I also had kind of different kinds of roles. You might also be familiar with me as the hype man in the photo booth yesterday. I have lost my voice, but it was a lot of fun. When it comes to the different things, it's so kind of different regarding my role that I wanted to latch on to like a thing that actually is the same, which is the editorial. So, as I was going through the vault of old versions of one version, one film and the other, I realized that, you know, this is like the beginning stages of me slapping something together for charts. And then I realized it's, maybe what is the more interesting comparison is not something right from the beginning, which is not a fair representation of what the full thing looks like in the editorial, but also not at the end, because there's been too many cleanup passes at that point. So it looks too squeaky clean. So I grabbed two edits that are like just past the halfway point. So it's really messy and disgusting, you know. And yeah, a lot of good stuff going on there. And I thought, okay, maybe an interesting thing to highlight here is the hell is this? Because there's a lot of little bits and pieces and there's like, there's a pattern to the madness, I swear. And this is also just to tell people, this is also to tell people that when you're doing editorial and such, it's not that there is like this one guiding principle that just everybody, it's an industry standard and everybody should do it. You can kind of wing it. Hey. So just to figure out, especially if you're at a halfway point and you haven't done like cleanup passes, whatnot. So that is temporary sound effects. A lot of that is me just in a microphone. Or like, you know, with a balloon or whatever it is. We have some temp vocals. That's usually also me like me, whatever on the microphone. Then we have this got a little jumbled up at some point, I guess, but I kind of have the renders over there. And then I like having kind of an archeological dig where as you go down, you find like old remains of kind of the layout or whatever it is. But here we got the metastrips and that's a common feature for our studio because we use those basically glorified color strips that have a beginning point and end point and then that hooks up to our kitsu add on and there the information flows and we can use that for cool stuff. That's the layout slash animation. I think this is like right after a cleanup pass where some of the layout had kind of been taken away. So just imagine this being more amazing. Then there's the previous animatic. That is usually a jumble of mess which is fantastic and a lot of fun to work with. And then we got the temp score. Now, why is there so many? It's because I wanted to make sure I was constantly rotating different temp stuff so we would never get married to any one kind of a temporary score because otherwise we don't give the composer like full reign of making something and not everybody just gets sad. Oh, what about that one score that you kept using for 20 versions or whatnot? Yeah, this is just temp credits or whatever. So a fair comparison to rickety rocket as it was formerly known before we wang it. Similar stuff going on there. This is a version right after the halfway point of all the different versions. Very similar kind of structure. Temp sound effects, temp vocals. That's really where I did my best dock work. Metastrips again. And then we got the previous layout animation. So this actually was slightly different because we were using more sketches for it. And then I was handing the editorial back and forth between me and Rick a couple of times. And everybody has their own way of working so I think it got a little messed up at some point. But it's more or less there. The animatic and then the temp score. And we tried also to do a couple of versions. So here's the comparison. And honestly, from an editorial point of view, I mean, it's kind of, you know, the same way of making a sausage. So I have nothing interesting to say. I'm sorry. I was tasked to do this. But I guess one primary thing, which I just mentioned, which is in wing it, we were relying in the beginning a little bit more on sketches. And in one segment of charge, I did this kind of floor plan thing that I animated. And I did kind of, you know, time that... I totally know what I'm doing. Time that roughly up with, you know, the stuff here. And it was great for spatial awareness and tried to figure out where everything is. Also because things had to be designed later. Spatial awareness here was a little... a little bit loose. Still, there was some, but it was a little bit loose. So anyway, let's move on. So characters. Speaking of characters, who is it? Who does the thing? It's Julian Kaspar! It's hard to remember who made the characters. Okay. Yes, I'm Julian Kaspar. I'm a character artist at the Blender studio. And I'm gonna... I want to show a bit of like the differences between where the challenges were lying between like charge and wing it. I'm actually going to go the other way around, starting with wing it. So here's like one of the original concepts that were like made really early on. And that sort of like summarized the dynamics between the characters. Like, I really love the exploration process that we did with the characters during that movie production. We had usually the tendency to kind of like rush it a little bit and do like some concept thing. But then once we go into modeling the actual technical execution, we kind of dig our way into that. But this time around, there was actually like a lot of interplay between concept artists and sculpting and also the other departments. Our art director and concept artist, Vivian Rolkovsky, she was like on board for the entire thing and constantly doing drawovers, which helped a lot. So even early on when we were figuring out the character designs, we rarely had them in a default pose. They were constantly in an expression or some sort of pose to figure out the character designs. Because at the core, the character designs were pretty simple. You just like, you can draw a bunch of variations and then you pick which direction you want to go for. There's a lot of possibilities. But then you kind of stick to it and we mainly just focused on trying to figure out how to make this work in 3D. And that's where the fun part lies is that we were previewing a lot early on with sculpting and using Eevee just to get a sort of a sense of how this could look like, but also how the characters can deform because we wanted to really have this 2D animated feeling to it. So we were testing out the deformations and keeping it very rough only putting like preliminary shapes, anything that sort of sells what we're going for. And some of the cases we kind of took it too far. Like overall we tried more than we could implement in the end. The timing and the budget just couldn't quite afford this. But in theory, this is totally possible. I would say we could have gone for it where we could just really tailor the character designs to a camera angle and just go do crazy things. Like sometimes the experiments were really fun just to see. Can we disconnect the mouth? This really toony style or even just completely have the character head differently shaped from certain angles. And then this is still like sculpting and then we didn't even usually finish the sculpt. The more finished version was usually just the overpainting. So it was constant back and forth and we were getting to a final character design mostly in painting and then only really wrapped it up and polished it up once we needed it for retopology. And yeah, even just when it comes to expression and post-testing we did very little in 3D just to prove if something is possible and then a lot more variation was done in 2D expression drawings and paintings. So there was a lot of emphasis on the whole design aspect and how we want to do things. The retopology part in comparison with the actual modeling was fairly simple. We just needed to make something that is tailor made towards the two or three characters that we had and just kind of focused on focused on stress testing and seeing if it's if the topology can perform and that can even be done in very simple deformation tests and then handing it off to the rigors and seeing if they can come up with some clever tech to do it. And then you hand it over to the rigors and animators proper and even then there was further paint overs and draw overs constantly to keep the style consistent and push it further. So like the whole concept thing permeated throughout the entire project. Charge on the other hand it was the entire other way around and this is one of the original concepts which kind of like summarizes the character design in a way and we didn't really need any design pose or expression test to see how the character would look like because we figured out early on what the character was supposed to look like based on some references that we were looking at, some actors and then it was really the design aspect was done, it was really just down to executing it in as much detail as possible making it look realistic which was a lot of fun it was challenging in its own way and it was fun to keep referencing various photos and videos that we made I wish we could have used something like scan data but it was a lot of manual sculpting and modeling which then needs a shit ton of references this was like a lot but it's interesting how at some point we needed to go more into detail and because we were sculpting a lot of the details on top the retopology was actually completely locked we had to figure out the topology early on so that a lot of the final sculpting which actually happened in layers we used an add-on just to figure out and test a workflow that we wanted to implement before any features could be implemented properly but yeah most of the time I would even say went into figuring out that workflow and making it actually usable so a lot of the modeling happened in the detailed areas so there was a lot of non-destructive layers of sculpting and that was used then to figure out the deformations because we had a realistic face needs to deform realistically so we were looking up a system that is called FACS it stands for Facial Action Coding System and that is basically a way of sorting and categorizing all of the different muscle movements that the facial muscles can do and then giving them a number like I was starting to figure out how we could implement that and sculpting it but we realized early on that that would mean way too many displacement textures and shake keys that we could possibly implement in Eevee as only so much memory so I started combining them into first of all culling the amount of actions down to whatever we need actually in the movie he has a beard so we don't need that many deformations around the mouth and then combining the ones that actually are really working well together because they move in a similar direction and then those ended up in just six deformation layers essentially which we could then implement selectively with painting and masks where certain deformations should fire up so that can then be handed over to further rigging and shading and they just implement the shape keys and the displacement maps there was also a bit of experimentation with ABC maps for wrinkle deformation for the small details that I think mostly just ended up being experiments because we ended up just making it mostly procedurally it's like a bit of a mixed bag of like all the different workflows that we ended up using but it was still really fun to try to figure out how to do this in a proper sculpting workflow because one of the tricky things that was really is to test what you're actually sculpting like a really important aspect is the sliding of the skin because you can basically just sculpt whatever expression and make it as realistic as you want but if the skin doesn't actually properly slide and deform along those deformations then it will look weird and everything looks like it's swimming so we actually like I typically started first just sculpting the general skin sliding and the broad deformations and for that we also used geometry notes thanks to Simon for helping out making some of these like preparing these setups to give some live feedback on where the stretching on the face actually happens and then it's a bit easier to debug the deformations and smooth out some of those shape keys so that the skin sliding isn't as much of an injury then in the end an important aspect is also the baking there was a lot more effort needed on the baking process and to get the most accurate results out of that we actually also ended up using geometry notes so there was typically like because you can read the differences between like the source and the target mesh to attributes if you have a high enough resolution you could just bake that emission out from the shader and just bake that to a texture that gave us a lot of flexibility of baking accurate height maps or even vector displacement maps with the some downsides of it generally not supporting smooth shading and also using a lot more RAM but on our computers we kind of made it work and that was then for baking displacements and bump maps for not just the face but also the clothing we didn't really have the time to do a lot of well cloth simulation so that was also handled more manually and for every major bone rotations we had basically a shape for the jacket and the the pants and all of that was manually sculpted based on reference and then again triggered based on masks to apply that deformation specifically and this was just like for example just testing environment after I exported the shape keys and the displacement maps to actually see if it's working properly the yeah there was a significant effort put on figuring out that workflow and we have like I tried to invest put all of that information online document the process live as I was doing it not just as a cheat sheet for myself if anything goes wrong or I need to make adjustments but also that other people can actually read it as we were making the movie so that's actually on this Blender studio the studio.blender.org website as a course like a training well it's more like a written documentation and that's also like it's documenting the successes, the overall process, how to troubleshoot things and also our shortcomings because this was after also relatively short production and there's also like a bunch of notes on what we should tackle next on optimizing the workflow further automating some of the things and ideally also implementing some of the features proper with the help of the developers to figure out how this should be in Blender right. Also again like I want to give some big thanks to Kent Tramell, Daniel Bystead and Chris Jones, they were like also actively involved and just gave a lot of feedback and even some resources like brush textures and that helped a lot during the process, saved us a bunch of time and yeah, just wanted to thank them again and yeah I think I would just hand it over to the rigging department hey guys I'm Demeter, I rig things and so when it comes to pets or wing it and charge there's obviously not a whole lot of overlap between the two productions but that in itself I guess is interesting so I'm going to start with charge and I'm going to try to sort of I guess point out the differences in working at first different eyes so this is Anar, that's the name of our main character in charge and obviously he has these realistic eyes and I think something that's worth noting is that from an animators point of view these eyes are quite restricted and limited all they can really do is rotate and blink and that's kind of it and so that means it's easy to rig right but so for one thing and this is kind of for charge the eyes and eyelids are part of an intertwined system of facial areas and so for example while you're rigging a realistic face you might ask yourself should the cheek control affect the eyelid because I mean when you set that up and you try it out it feels kind of right but then you ask yourself should the nose affect the cheek and you try it out and it feels kind of right as well like yeah when you move your nose you move your cheek makes sense so should the lips affect the nose because when you know in real life your nose kind of moves so now your lip is moving in the nose and that moves the cheek and so now when you're moving your lips you're moving the eyelids and then your animators are sad so we went back and forth for a while to figure out an ideal control scheme that has a right balance of like automation and also freedom and control that was a bit tricky but we got there eventually but that's what made a realistic eye challenging I would say other than that there's also the fact that it's very detailed of course like do we have a slide about that though? Yes so realistic eyes are quite detailed there's a lot of little things that you have to take into account such as like this I'm just going to call it the third eyelid that slides in as the eyeball moves left and right the eyelids of course stick to the eye sort of and even that was a question at first like do we want to automate that and you kind of try it out and then you just kind of listen to your animators in this case we decided yes it's better if it's automated and of course there's the little thing that everybody loves with the cornea pushing the eyelids forward and finally the little moisture mesh that is something that's mostly Simon came up with and I just had to rig so yeah realize very detailed whereas what you'll see later on the pet's eyes they're not they're absolute opposite problems but so also interesting thing is that when it comes to like figuring out what real eyes look like you don't really need concept art obviously so instead we just looked at a lot of stock footage of older gentlemen's eyes which you know it's one way to save some budget on concept art money but it's a way to lose artist sanity but hey at least we got a nice wallpaper out of it feel free to use so yeah when it comes to finding the style on charge it was just about observing reality and converging towards that sort of fixed point in style as far as we can and you know we don't have to go all the way it doesn't have to be hyper realistic but we knew that we wanted to approach realism but anyways real humans are kind of boring you've all seen real humans you can just look to your left so that after this production we decided we would like to shake things up and that's why we I guess ended up with wing it and so instead of having a fixed style that we are aiming for throughout the development of wing it we just knew that we wanted to push things whatever that means and we didn't really know where the limit was or where different limits were until we hit them for example looking back at the eyes again I think I might need to hit play yes it's working yeah so these eyes as you can see it's just kind of a 2D plane that is just a curve that controls like a sausage and then there's a black dot in the middle so it's very simple in a way and in a technical sense it's quite unrestricted compared to the realistic eye that you saw earlier because you can shape this into any shape that you want and you can disconnect it from the face and make it fly around but so the process of figuring out these eye controls is more so about figuring out the restrictions that we want to apply on ourselves and finding the limits of the style like what actually looks good and what doesn't look good like what I'm doing there currently does not look good so you wouldn't do that but that wasn't obvious until you try for example a small detail here is that the top area is a bit thicker than the bottom like the top rim is thicker than the bottom rim and that kind of stuff okay and I want to just bring back this slide from you lens section because I just want to point out that it was interesting when like imagine that you're me but you haven't made this movie yet and you have a team that is counting on you and they're bringing you these drawings and they're like so you know can you do that and I'm like I don't know I think what I used to say but I'm pretty sure like no one really registered it what I would say is like I'm pretty sure I can get you 80% of the way and in the end I think that's pretty much what we got and I hope that we're happy with that I'm pretty happy with it but yeah because I mean yeah some of these things that you see in these tests I think you can't really do with the rigs unless you I guess try really hard but hey the movie still looks good so yeah that's another kind of limitation which in this case was like technological limitation that we didn't know what it was until we found it right so the hardest part of the winged characters was obviously the mouth because of this because it was crazy the guys wanted to push it crazy far and if you want to learn more about that then you can check out my talk from last year which obviously I don't have a screenshot of because I mean not yet last year yesterday so yeah if you want to learn more about the winged mouth rigs check out my 15 minute classroom talk okay and now moving on to the body rigs again the two productions had completely different challenges I'm sticking with winged now for now so these body rigs were actually kind of not that hard arguably to create and part of the reason for that is because it's sort of modular as you can see here that you can disconnect everything and when things are that way it not only gives animators more freedom to do stuff it also makes my life a lot easier because throughout this whole production I didn't have to rig a single waist and shoulder area and if you've ever tried doing that then you know that this was pretty great so that was kind of a pleasant experience I guess to rig these guys another interesting thing that is unique to winged is sort of the squashiness of the spines in particular and now when you look at it and the final result it looks kind of so obvious like of course this is how it should behave but it actually wasn't crystal clear to me and I actually worked a lot with one of my animators, Pavlikko and he sort of helped nail down the shapes of this squishy squashy spine and I think I'm pretty glad with how it turned out so thanks to him and we love to recycle old tech of course because it's efficient and so we ended up bringing back a feature that we use on sprite fright which is this sort of automatic auto rubber hose elbow tech thing that you can see here where you just kind of scale a bone and you get an automatic nice curvature in the arm if you choose to so whereas in contrast going over to charge which had these sort of realistic robots which are very much not flexible and noodley one of the goals that we had for these designs of the robotic elements was to have these single axis mechanisms so you can see how each sort of different area only rotates on one axis and so to get three axis of rotation you have three different points of connection we wanted to avoid ball joints because ball joints are lame, everyone does ball joints and we wanted to make our life difficult and soon after we found out why people avoid ball joints because indeed our life became difficult because it turns out when you have this kind of Aki setup you can very easily run into a situation like this where you're just trying to move from one post to the other and because you have these single axis points of rotation the only way the Aki constraint can resolve the angles that it needs to reach a certain point is well to go haywire basically and so the same issue was present on the main character as well of course the robot arm and had similar design principles of these single axis rotations and so you might wonder how did we fix that well we didn't I just kind of just one of those times sometimes you just got to look your animators in the eye and just be like I'm really sorry it is what it is so that's that to be fair part of the issue was this like I can blame blender a little bit because when it comes to ANR we could make the the IK elbow pole target work but on the robot we couldn't do that because it turns out when you have too many locked axes in an IK chain the pole target doesn't work anymore but anyways here's the setup for the robot's foot we put quite a bit of time and brain cells into this guy as well there was some back and forth between like rigging animation and then we even asked for some modeling changes to make this mechanism work and in the end it works pretty nice I'm quite happy with it and let's see let's see its appearance in the film shall we are you ready look at the foot that's it so moral of that story I don't mean to make fun of us too much so I mean you can look at it in different ways great attention to detail is it a waste of time and effort it's up to you but so the reality is that when we were developing this we didn't have a layout yet we didn't know how much different parts of the character might actually be visible in the film so you know we just got to make sure that everything works and it's just something that can happen when you're kind of laying down tracks in front of you as you're going which we're trying to you know improve on as we mature okay and finally I just wanted to mention that all of the charge characters are now free to download on our website studio.blender.org you don't even need to register you just go on the website you click on characters and these ones are free and then if you want to support us so that we can torture ourselves even further by using alphabet versions of Blender and trying to make movies on it then you can become a subscriber and get access to our to the assets of our latest film which in this case is of course wing it and that's it for me and then I would like to welcome Pablico. Hi, I'm Pablo I'm suffering I'm trying to animate Riggs that that guy is doing for me I'm gonna go fast because animation is animation and it's always the same but what I encountered was I had different challenges within the different parts of the animation so in this case I'm gonna show you what sequence was looking like this is the first layout sequence that we get before we had anything to animate I did this by myself following some healthy notes so it's like a really just few poses putting the cameras and having all the feeling ok how it's supposed to be yeah, this one is something that we did after that and it's a we did a pass of mountain sequence but just with video references that's something that we did tons of video reference a lot of video references until we get like the the text that we wanted and we put them together so like this we were able to make our blocking pass this one having a solid base is what we wanted to do and after that is the normal process we just continue doing the spline and the polishing for a project like charge the polishing pass it takes a bit longer because you need to make more interaction and you have to animate more frames that you do something like wing it it's on 2 so it's easier to skip some process but the only thing this kind of polishing makes the animation really expensive and then you have to fight your battles in a way that was it now I would like to focus a bit more on one of the shots this is the layout of the shot as you can see they are like everything is not still not modeled we have like a provisional props and it's everything it's really rudimentary after that we did we did a pass on different takes of how it should be taken the prop we put all the cameras how they supposed to be following the layout we put all the tools that he was going to use we recreate everything we put the table at the same height that he was going to use it we bought a car jack we 3D printed the pieces for the car jack so everything is there to be able to have the most realistic takes that we could do so we tried different stuff we always recorded different animators video reference just to have different different different video reference that we could point at what we like it or not so this basically is the main difference between the two process between the two shots after that comes the normal blocking the normal stuff from here it's the same like you will do after that you have a polishing pass and then you saw it you saw the blocking you saw the spline normal process of animation and after that we get more notes and this is the final the final polishing pass when you already make you animate all the other parts like the hanging thing from the car jack and everything is perfect and you ship to the next department and now for wing it we base everything more on sketches and drawings because we need to record ourselves as a fat dog so it was more like a 2D creative process instead of being having video sequences all the time it was just drawings we asked Vivian to do poses for us even before we had rigs so we had something to aim at when we were having any test for example here at the bottom we had these poses I wanted to recreate and then we have the first version of the rig and then it was like just try to do it how far can we go and then we had to have a lot of interaction with the rig because it was not the best but it was later the best but you know that's what we animators and rigors do we love each other and hate each other so same all this really helped us a lot to develop the shape language that later we will use on the rig like for example all the torso the formation that Matt was talking about we figured out in this stage of the pre-production we did the same with the cut before we had anything we had already like drawings that we could replicate and in this case we also took a storyboard from rig and tried to explore further how the character could be and then really helped with the attitude and the emotion of the characters in each part of the movie and later the same we found out like we didn't want to have any angles on the arms and the legs also in these stages it was helping us to have clean lines and rubber whose limbs and then I want to show you a bit the same process I did with Aner but in this case in Winged so this is what rig gave us as a storyboard this is the first pass of layout then it was notes over that and then it became something like this so I took this I started to push all the timing and the rhythm following the same principle that he was going to smash everything and after that when I had this I showed to rig I was like hey rig what do you think and he started to do scribbles like no this is like this and this is like that and it started to give me feedback but in this case the feedback is a bit different as you can see every bread dot is a note that he was giving but we didn't have to record references again it was just like push this push this arm hub push this timing here so everything was a bit more artistic in a way you need to go back to the take your camera and do stuff so it felt a bit more organic in a way you know it was like more just draw and after that then I took my blogging I polish it I added all the extra stuff that is happening here all the movement and I was done so yeah next up is Simon so please hello I am Simon the shading and effects on both those films when talking about the differences between those two projects in terms of shading and effects it's a bit tricky because basically the sum of all of the shading on both films is the differences well let's try anyways so start with the similarities we rendered both of them in EV and that's about it so I want to talk a little bit like more broad scope of the differences in like a more global picture and then go over a whole zoo of things that we came up with to fix some of those things so in terms of general style obviously like charge is living more towards a realistic style while pets is going bring it sorry into the cartoon territory and there is like a certain like story emphasis on that as well like with the realism you're trying to create like a certain believability and an emotional connection for people to connect with the story to add like weight to every single action and with the effects also to make them look realistic and make it look like the character is actually being impacted by what is happening on screen and with the cartooniness we're kind of trying to push a little bit more that kind of exact duration to push the humor and just allow for some of the gags to be like a little more of like a physical humor that you couldn't really do with a realistic style and then in terms of overall look to achieve that kind of stuff you want to go for more of like a depth so with the whole shading idea and also the general rendering you're trying to create more depth and dimensionality to also bring the viewer closer into the action and with the more cartoony style of going at trying to go in the opposite direction where we create more of kind of a flatness to emulate that cartoony feeling and create very clear shapes that can support the exaggerated poses and stuff like that and then in terms of detail it's also basically the opposite where for the realism we're trying to push a certain level of complexity to make it believable and add like a certain level of richness in what we're seeing and for the cartooniness we're trying to go in the opposite direction of pushing more of a simplicity by removing certain detail that is not really necessary to just creating additional noise to set like a very clear focus on characters and actions and also emulate that hand drawn look and then to achieve these kinds of things for rendering of course for realism it's very clear you just go with PBR it's just conventional, you just do that you're going to be good and then for the cartoony look it's NPR which it's a little bit more tricky to figure out what you actually develop for that thing and it's a more artistic expression when you are in that NPR workflow and that also means for the focus that we needed to set in terms of shading and effects what we want to focus on in terms of how we achieve these kinds of things for charge that meant a little bit more focusing on the execution of okay we know where we're going to go it's going to look somewhat realistic how do we achieve that how do we work to get to that goal and then for Wingard it's a little bit more on the technology side like what can we even do and then developing that at the same time as we try to find the look of the film so it was like also a very different approach in terms of the process of what we're doing alright so now bear with me I'm going to try to go through as many like things that we came up with to achieve these kinds of solutions as I can for charge in general we're trying to well I was trying to go for a procedural approach as much as I could because that's something that I'm personally also comfortable with and it helps me to be quite efficient with iterating over different variations of textures so here you can see a procedural wood material but it obviously comes with a lot of restrictions also in terms of performance for example like we're using EV and doing a lot of like back and forth also seeing how we can push the performance to make it as real time as we can which well spoiler alert the files aren't actually real time when you play back but I mean but to be able to push it a little bit better into the performance I started baking out textures but not instead of baking them at the very end to bake the different PBR layers I started baking out tileable noise textures that then I use for the procedural approach which basically the notary looks the same it's still a complicated notary with a lot of flexibility we can change all the parameters and it doesn't have the issue with the resolution that you lock yourself into by baking the end result but it still has all the benefits of the procedural approach and for that we had some setups like something to bake we had a lot of tiling noises like some utility note groups that I just made we had a library of stuff that we could just plug in and it would just work very easily and then an odd group to until it again because I mean if it's just perfectly tiling it also doesn't look great so for example for some of the materials in the factory floor here it was just very easy to take some tiling textures slap a note group to until it again in there to get some variation and just use that for like a random rotation offset and then something that I started to do a lot also for Charge was to use geometry notes to create some tileable patterns very easily by just floodfilling UV tile with some geometry like here I'm just adding just a bunch of very simple curves meshing them with geometry notes and then making them more complex and adding some additional fuzz to create this kind of textile pattern just the canvas texture that we had and that's very easy to just bake out the normals, the height map everything that you just get for free out of the 3D result into a tileable texture and it's also very easy to create variations for different type of textile and you get a lot of information for free without doing photo scanning and still having a lot of high resolution results and then also using that in different materials like here I have this bully hat texture and one thing that I utilized this technique for was also to be able to receive additional meta information about the structure of this kind of pattern to maybe I go back like this UV map here you can see makes us able to remove the tiling in terms of like the noise that we get on top of the individual like cells of this texture and then in general for the texture painting I also went for a very procedural approach where it just had mostly procedural setups that would be a PBR material that is mainly based on one single input texture that can just be used for painting in wear and then you could very easily do PBR painting on that single texture and because it's using EV you can actually see the preview in real time and it was actually really fantastic to use which of course the setup takes a little bit but if you know what you're doing it's not too bad and then this was one thing that I like to show because it's like a nice little thing a geometry node setup that is using the UVs of a mesh to basically stick a geometry on top of a surface like a texture so here I have a setup where I just draw in a simple curve and then with geometry nodes turn that into like a stitching pattern and then it just projects it onto the like NainSIME geometry and then just using the UVs and attaching it directly to the surface you don't really need to worry about deformation it just behaves like a texture but it's geometry it just works oh yeah by the way a lot of the things that I'm showing you can also go to our website and download the files so yeah and then another thing where we're using geometry to add some of this shading information rather than actually using textures is like for the detail weird things like that that would be a little bit more difficult to pull off in a convincing way just using shading it's just sometimes easier to just actually use geometry and you don't have the issue of like getting to a good resolution in terms of textures because it's actually just rendered as geometry for the facial deformation there was one thing that I invested quite a bit of time into which is creating a procedural wrinkle map based on the deformation of the character's face and it's a little bit more complicated I'm not going to go into too much detail of how that works but at the end of the day we didn't use it to like that much of a high capacity but we did use it in the final film and I'm pretty happy with the approach like it gives you a very very flexible way of using the data of the deformation of a mesh with some directional information in the shader I created like a demo file that you can also download on our website where you can see how this actually just dynamically creates this kind of wrinkle map based on the deformation of a mesh I also brought a relatively long blog post about this which you can check out and then in terms of effects something that we did a lot was rather than using simulation to create these kinds of dynamic effects just create geometry nodes assets or well assets using geometry nodes to generate a mesh that is animated based on the scene time and then adding a whole bunch of parameters on top of that that you could change to create more sparks or less like different seeds and then just the timing all sorts of things which is just a very nice way of iterating over a lot of shots but just drag and dropping the asset into the shot and just adjusting it to the where it needs to be and we had our spark guy healthy who is also the director, layout department editorial and everything else just scattering around the whole movie having a lot of fun another asset that we made was this kind of smoke which is also real-time in EV it's using an approach using planes which just meshes with a bunch of sprites as textures map on top which is great because it can receive lighting very nicely it's really terrible for performance but yeah it creates a relatively nice result and it's pretty cool to just like drag it in and use it as a background asset without having to worry too much about setting up a simulation another thing that I made for the movie which we kind of also underutilized a little bit but it was still fun to have is a dynamic lens flare using a reflection shader to kind of simulate how the image that you actually see would be deformed into a lens flare so it's like trying to replicate what the actual camera would do in EV which it's not accurate let me tell you that but it's nice and then here in one of the shots you can see how that asset can also just be used with a bunch of parameters to create different effects alright let's move on to pets so one thing wing it one thing about wing it is that we for the style decided to go a little bit for a different style between the characters and the environments which is also a little bit true to the sell animation that it's this kind of leaning towards and there was also a little bit of a technical reason to that for the environment we had this setup that we very easily be able to shade different different faces of an asset in a different color to just like lift or lower the values very easily to create this like fake 3D effect which kind of also creates these like these patches of color in 2D that you would also draw a lot of the time to create some sort of rims or something like that so we just had like a custom attribute that was just used in all of the shaders that you could set very easily from the edit mode another fact that we had to stylize reflections or trading in general was to facet the normals which is something that we're doing not for stylized trading in general to create like patches of color and breaking of the gradients like you can see on the on the round joint like if you turn that off it's like super smooth but with that normal faceting on it just creates this breakup in color and these integrated and just these patches of color that you also draw in are other reflections like the main reflections on some of the assets we actually just used drawn in 2D textures that were just projected from the camera view but locked to the position so it wouldn't just like stick in the view space and then something that we had on top of basically everything mainly on the characters but also on the environment was just in the instead of using a diffuse shading model just using subsurface gathering which I know it sounds a little bit crazy it creates this very nice effect of blurring out the details like losing a lot of this depth information they would have in the mesh because of some sort of deformation that the rig is creating by these crazy poses and you just get rid of that entirely and it kind of flattens out everything a little bit and then to glue everything together we also had a paper texture on top of everything it's like a very subtle thing in the YouTube compression you don't actually really see it that much but it does help a little bit especially because it is actually just a 2D texture that is in camera space it counteracts this effect that you get from like looking at a surface at an angle that looks like it is perspective if you have like this little bit of texture on top that is actually flat it does help to make it look a little bit more cartoony and drawn outlines well ok so we also had outlines on this film I am not going to go into too much detail how we did that I am just going to say we used the custom geometry notes solution for that there is also the grease pencil stuff we looked into it but there were some things that we couldn't do with it that we wanted to do so I just tried and well it kind of worked I have a setup on our website in the blog post as well I am going to talk a little bit more about it in another one but yeah if you want to try it out you can also go there so yeah we had like a little bit of a different setup between the outlines for the environment which we actually created as geometry in 3D for each shot and in the shot to actually get an outline to one of the objects it would be as easy as placing it in the right collection and then giving it this line properties modifier where you could then set up the color thickness and stuff like that for the outline and just specify that way and an additional thing that we had was the ability to just draw in lines and then we rendered in a very similar way also towards the camera and then it would stick to the surface so you wouldn't have to worry about the formation too much and that could also be used on the shot level to just draw an additional detail which is something that we wanted from the very beginning but weren't quite so sure how much we could actually do it because of budget but it's a very nice way of differentiating different shots that are with a different focus on a different POV where you're like more in a close-up shot you want a little bit more detail so like on the close-ups we really used this to a lot to just add a little extra detail and then for the characters a lot of the effects that we wanted to have in lighting which Andy will talk a little bit more were done in the shader itself so part of that is the outlines for the characters as well which use the same base setup as the environment outlines to identify where the outline should be but then the rendering was not actually done as a mesh that you just have on top of the rest like with the line art modifier and with the environment outlines like I just showed but instead it was rendered inside of the as part of the surface with the shader and then additionally using that same information we could also create these kind of fake rim lights that you can see where we change in the color of the triangle and this additional layer of this cell shading kind of tune shading effect where you could just orient it on the character as you want and the way that is done wait this is different you can see how this is actually just on top of the shader and obviously it's only oriented towards the camera which right now I'm going into a different view than the actual shot camera so it doesn't work at all looks right and the way it's done is by just getting a bunch of information out of geometry nodes as attributes like here I'm using distance fields that I generated from the identified outline position in camera space and then using those to set up the outlines with a certain thickness that I can just easily choose because I have the distance field in the shader itself and then some additional normal information to make those fake rim lights deciding where and what direction towards the camera the rim light should be placed and then these different effects that were actually done in the shader yeah well not to clean up it's a little bit difficult sorry are just stacked or layered on top of each other like we have like the base color and then the outlines the fake rim lights, the cell shading and additional effects like the paper texture and those actually just used as the color information for the subsurface rendering shader that we used for the rendering and there's also a blog post about a lot of this stuff I'm going to follow it up with another one that's more specific about the character stuff okay effects, last thing for wing it the effects were a little bit less tricky than the shading for sure but we had like a similar thing we had a bunch of effects assets that were just set up with geometry nodes which is like a simple setup here like a rocket engine and then just like a very simple rig to be able to pose the effects asset in the shot and just put it somewhere and make it work and then hey the sparks are also making a comeback just this time as hey particles is very nice if you're able to just easily reuse some assets that you created for another film just add some more turbulence make it a little bit slower and it looks like hey and then this I just wanted to show we also underutilized this it was really fun simulation nodes made it into blender and we wanted to try it we wanted to use it in a film because why not just came out so probably not going to break and it was really fun to play with this but it's not super controllable as you can see like you give it the initial shape but then what if you want to have it go in this direction okay I mean it's a bit difficult to control so most of the time in the more important shots I didn't set up but did it in a different way but some of the shots still do this and what this is basically just using simulation nodes to like a fake kind of 2D smoke simulation based on curves so because it's 2D we don't really need volumes or like dimensional meshes so instead I just use the curve expand it a little bit randomly and subdivide it and like split it up here and there and yeah with the nice simulation nodes you can create a very complex setup like that and let's move over to lighting and comp by Andy wow that was a lot of information okay I'm going to make it a little bit more simple so I'm going to talk about lighting which I did on both of these projects together with Beau Gerberns at the studio and so first for charge it was a fairly realistic approach in lighting because the film was supposed to be in as real time as we could make it so the lighting itself was very cinematic it was driven by the environment mostly there was this juxtaposition between the main character and this evil factory so the main character has these very warm colors and the factory is kind of cold and we wanted to create this cinematic effect by having a lot of depth and layer separation and kind of work with more cinematic lighting methodologies where you drive you start out with a set lighting and then you use that to drive the character lighting so in overall the look was kind of muted colors sci-fi semi-realistic and as Simon said we rendered this with Eevee so the really cool thing was that it was very immediate to work with we tried to make the viewport setups as responsive as possible with very lower geometry, not all the hair but we got it pretty close to the final render sorry and in general that meant also because we had a fairly limited time budget we wanted to make things a little bit more streamlined so for example all the factory shots we started with the practical lights that were already built into the set so we had some reflections that we can toggle on and off and then like you would do on set lighting on a real time on a real production you would emphasize your characters and separate them out so I'll give you a short breakdown of a very simple shot and so we would start with the base HDI lighting which is basically giving us a reflection and this was actually one of the HR IMAPs that are baked into Blender it has this kind of cold look and next up we would include the light sources from the set and then you could see already that we have some fog already built into the world which helps us with this layer separation and on top of that we would add more background lighting to add interest or to mask out certain areas for example like some of the reflections would show up in the background and we didn't want that and then we would focus on the foreground lighting which was mainly done with area lights and to mask out the characters make them stand out like you would do on a live action production with big big light blockers and that kind of stuff and then finally to add focus on top of everything we used shadowcasters we call them effects that don't show up in rendering but they just block out light in certain areas and that makes us really focus on the characters we want to be very selective about that and that's the final shot likewise with the compositing we wanted it to be fairly simple back then we didn't have the real-time compositor working that well but we made several base node groups and these effects were always kind of very camera-like effects so we had some bloom some lens dust which I'm going to go into some film grain, some chromatic aberration vignetting, sharpening and desaturation in the end and we plunked those into every shot so since those were node groups they would propagate into all the different shots and we had some values to tweak them but overall we used the same for everything so all the shots ended up fitting into the continuity so for lens dust sorry, we actually took a shortcut we had a procedural kind of solution first but we wanted to give this camera a very realistic, gritty look so we wanted some dust in the lens in almost every single shot and we did that by just taking a couple of pictures of fairy lights with lots of bokeh and then using that in the compositor through this node setup and it's hopefully not too disturbing, it's kind of a subtle effect but you can see it in a lot of the shots just helps us make the whole world feel a little bit more grounded next up there were some shots that were more complicated in terms of compositing where we had to combine different layers of render engines, we had for example this matte painting shot which the background was rendered in cycles because lots of complexity and the foreground was rendered in EV so we have the different layers this is the matte painting, we rendered it we constructed this whole environment and then rendered it and painted over it in Krita and then projected it back into some simple geometry into the same scene using the same camera but with that we get a very realistic sort of base set we added real-time fire as you can see also at the same place in the distance so cameras would line up and everything we would add the road, the cars and the reflections on the roads and that was rendered with EV but they were fairly simple models that Kalti actually created in layout they still made it in the film but yeah we had all the reflections we baked into that pass that's actually how the pass looked like just in real-time in the viewport and then we would have some environment bloom which is correct based on the distance that we're observing it and then we just rendered the foreground layer in EV so that's what you can see here and then finally on top of that we added some bloom and lens dust effects and grading, vignetting and that's the final shot fairly straightforward on wing it on the other hand it was kind of different so the whole movie was supposed to be yeah you can keep the audio down it's not supposed to play that long so the whole movie had this 2D cartoony style completely different from what we've done before and also I've hardly worked in that environment before so it was a little bit a learning curve the great thing was that we had Vivian who would come up with a lot of reference for everything so there were some color scripts and every shot basically had all the colors locked down basically so it was super great to have all this reference in the end we did a lot of color matching by eye to make sure that the colors in the renders are exactly or kind of similar like the colors that we had in the concepts yeah in terms of lighting what that means is that generally we have flatter colors we have lots of really big gradients we have sharp shadows and very thick outlines and the characters are simple in a kind of cell animated style and the style is a little bit different from the backgrounds that are painted the characters are in a cell more like with ink so how did we do that well Simon showed these shader setups which made the environment pretty blocky and the general approach was that we would have only one light source that would cast a shadow and most of the objects were actually disabled for shadows so there was no really complicated shadows they were fairly stylized and simple so only the cockpit windows for example would have shadows and the levers and light bulbs and all that didn't have shadows and then we placed around lights and mostly the lighting from the environment was different from the characters so if I placed the sunlight to light the whole scene it wouldn't affect the character that much we would do that with the lighting rig that Simon made and there we tried to match the environment light kind of but mostly the concept art and we had a lot of control over it for rim lights and outline color and all that kind of stuff and speaking of outlines as was hinted before we rendered the environment outlines in a separate pass so we could light it separately so we could get different effects for it like rim lights we could place lights into the outline collection and light it separately give it more emphasis or not and then we would composite those back onto the beauty renders and the compositing was fairly automatic we tried to make it build in our shot builder so every shot would start out with a setup so we'd merge the base layer down with some sort of painterly effect and then we would render out the outlines and we composite them on top of that these different layers are all using these preset node groups that we built and all of that was connected from start so the artist wouldn't have to worry about connecting everything but we would have control using the properties on the node and what enabled us well what that let us do is also to give a certain depth effect to the outlines and the environment separately and then on top of that the animators would make these amazing grease pencil drawings that would give us certain effects and we also composite them into the shots mostly they integrated fine but we had to do some material adjustment the animators didn't see the final renders so we would have to go in and tweak the material and it's actually fairly cool like we got the we got it to match the concept art pretty closely in certain areas so it's super cool on top of that we also did smears not only the smears that the animators gave us but motion blur smears in camera we made those with grease pencil and then just composite them on top and then as Simon mentioned we also had these lines everything every asset had these line layers built in and they would be separated out into separate passes but we could also we made this little add-on that would let us select any object and then just draw lines on top of it and it would take the normal and the base color of the surface as a preset already and you would be set to go in some shots we even managed to project that line into a deformed character which was pretty cool and it's actually super fascinating because we went into a lot of shots and actually scribbled in these lines and to give them a little bit more interest and a little bit more of a hand-drawn feel which was super rewarding which we didn't have on a lot of productions to have that extra time to go back in reflect and see and push to quality and that was really super great in general we got a lot of at some point we did the CVB pass, CVB stands for could be better so everyone does their notes of what could be better in the shot and we managed to address a whole bunch of them so things like make the pencil look more like a pencil because it looks weird we could actually do that so I'm super proud of that and that about sums it up so I'm back to tie this all up in a neat bow just a quick point to say that we will be live streaming the closing ceremony here so you don't have to move, you don't have to run down the stairs anyway there's no seat left in the theater, I can already tell you because I didn't work on charge myself so I just want to give a bit of feedback about what was just said because I can't do a comparison so I'll just do my producer job, give retakes so we only have one rigging artist someone implied that we have riggers that's not true so if anyone is ever wondering about making a career out of rigging you can, there's definitely studios looking for people like you so Matt is the only person with the rigging department on his shoulders and I mean he also has quite a selective memory because I remember some complaining about rigs that he just called fun I don't know so it's maybe a quality that's needed for that job Pablico also who said that he was getting a lot of notes as any animator would say obviously you were also giving notes to Matt a lot it's a love and hate relationship you might have seen it in the vlogs by Haru and Simon I'm sorry but I'll just wing it maybe he's still in character from the whole production so those are two very different projects both made on Blender obviously but also rendered on Eevee so we stretched everything real far and obviously it's not just making the movies that the whole team does so they also participate in the development of the software by actually testing the features that come up today we work on the daily build joy nothing ever breaks so it's sometimes really rough because what used to work on the day before doesn't work anymore and you would think that we have lots of technical support at the studio because we have the developers just next door actually they don't have that much so they are very technical artists also working out their solutions by themselves and then bringing it upstairs to the developers it's not like a hierarchy thing obviously they also write articles and make videos to share all of that we're doing not to say that this is the way but to say this is the way we do it here which might be great for you if you're a small team usually so please like feel free to go to the website look at what they've been doing for the past few years look at the older films and their files and you can also obviously follow what we're doing with the next movie Gold that's directed by Jerika who is in the room right now so gold is very different again you might have seen it through the technical talk that Serga and Brecht gave this morning so it's going to be another challenge rendered in cycles this time so I hope you can follow for the news supports if you can and see you next year for more fun