 Is it all good? So hello, everyone. My name is Odillon. And today I would like to present my vision of what Blender could be in the automotive industry. I say vision because for now I'm still a student. But this is how I would like to use Blender in my future job. So a quick overview on the content. I will start with a quick intro and then talk about how you can shape vehicles with Blender, how I moved then to use Blender as a data preparation tool in order to export stuff to, in this case, Unreal Engine. Then Blender as a visualization tool with EV. And then I will talk about what I see coming next and finish it with a conclusion and the final process I ended up on. So is there anyone here using Blender to do vehicles? Not that much. OK, that's what I thought. Basically, I do. And as I said, for now I'm still a student. But I would like in my future job and what I'm studying right now is what they call digital design. So in the automotive industry it means that you help someone to express his vision into something he can visualize, prototype, and then produce. So basically it's most of the time the link between the sketch part and the clay modeling part. But this is clearly not a defined process and each company has a different process which involves a lot of loops between sketch and 3D and 3D and clay. And so today we will talk about this virtual creative phase. And so to start with the beginning, I've been using Blender for nine years for now and my first encounter with it was with the glorious Blender 249B. And I couldn't get any of the features. I was just struggling so much and playing with things I didn't understand at all. And my goal was to do cars. So I was putting a lot of effort in it and it was really bad, really, really bad. And until I thought it was maybe the polygonal approach and Blender were not suitable to do cars because I couldn't get a nice reflection and details on surfaces. But still one doubt was keeping me trying and it was this guy with his Opel. And this was on the Blender official website which means I would see it like almost every day. And I was really thinking that if this guy was able to do it I just had to work harder and so I continued to try a lot and to be more and more precise. And then I entered a design school, continued to use Blender until I finally entered as an intern in digital modeling in Kiska. Kiska is a design studio in Austria who is responsible for KTM and Ustvarnam motorbikes. And there are like hardcore Blenders, really they script a lot of stuff and everything and they told me, they taught me how to use Blender in a much more professional way. And what I'm going to show you today is my evolution from last year as using Blender only as a modeler and up to using Blender as a Swiss Army knife I'll always carry with me for everything I do in every project. And I will start with Blender as a shaper. So this is for me the most common use automotive companies could do off Blender but the thing is for now the main tools are NURBS and most of the time LES from Autodesk. But still in some companies they're switching to Polygonal and some as Tata, we talked about Kiska but Tata as well are already using Blender. You can see two really nice talk of Matilde Ampe and Pierre-Paul Andriani who did pecans in 2015 and 2016 if I remember well. And so this is really encouraging but what I think is really cool is to see that some designers are individually learning Blender and basically they use it for a personal project but then they bring it to the office and work with it. And this is the case of Sidney Hardy. He's doing crazy stuff with Blender and he's right now in OD in California and using it also for his work. And so what would process of creation of a vehicle what it would look like in Blender? For this I will illustrate with a project I had to do with my friend Antoine de Salaberie for his diploma. And so basically it has nothing to do officially with the North Face or anything it's just his prospective vision of it but basically we would start on sketches and he would bring this to me and together we would create the basic volume no subdivision surface or anything just the rough volume and defining the silhouettes and then based on that I would take screenshots on which he could define the next parts we would work on and so back and forth like this and for this Blender it's perfect since you have a non-destructive workflow and modifiers which is great and which lets you go from the volume definition part to the more surfacing part really easily and you can just shut down the modifier to see what it gives in terms of just volume and then go back to surface so it's really cool and bit by bit we were able to end up with a final model but then this was kind of the let's make compromises project because we had really short deadlines so we had to go fast and I wanted to train on something maybe more challenging in terms of reflections and everything and see if I could find tools that could help me to do that and that's why I modeled this existing concept car it's the Genesis Essentia, it's a concept car I love and so basically the main process was the same rough volume definition and then you end up with a final thing but in between something changed and it was the use of the shrink wrap I guess you all know what the shrink wrap is but basically it was really useful just taking the vertices I wanted to be affected and it helped me make door cuts and everything and prove my younger self I was completely wrong on the polygonal approach because you can clearly keep the reflection and still have a lot of details which is really really cool but what also changed on this project for me is the scripting side of Blender which I was completely new to and basically I wanted to try it for years but finally I did it and my first exercise was to do this kind of menus you can find them in Autodesk LES it's called marking menus and basically it's just selection and modification tools but accessible really fast and Blender has five menus so I just tried to script the same I used to work with this in Kiska they have it but then when I left at the end of my internship I didn't have those menus and so I did them again and it changed clearly my modeling life and I was able to get this car at the end and to think about the next phase and so I had a friend at this time he was doing an internship in a visualization company and so we thought together like yeah let's go for a visualization project together he's called Valentin Becker by the way and so we started to use data sorry Blender as a data preparation tool and first thing we ended up with when we started to import is the UV problems and so two things appeared first is that for some part as this one we want clean UVs I know there is cuts but it's normal it's carbon fiber and but we want clean UVs not the auto unwrap and basically there is some stuff you want to do by hand but for most of the parts we didn't really care because we knew that we would never see these parts and so just an automatic unwrap would do the job and also Unreal Engine asks for two channels one for the texture and the other one for the light map and this just sounded as really boring and repetitive tasks so what I did was just another script which would help to just take an object check if there is a channel 1 UV if there is not it means you don't really care about the UV itself so project 1 for you and then it does the same for the channel 2 and rename the channels it doesn't really it's not really useful to rename the channel but I prefer to keep a clean file and then for 70 objects it was really better to have this and not make mistakes and then we thought like okay let's go VR now we can import data into Unreal Engine but we already knew that we would have some limitations and by the fact it's not retracing and mostly with reflection and shadows approximation so we started to do Bakes in Blender and first things we did was to bake a ground shadow so that the car is on the ground and then we thought okay maybe we can go a bit further and we didn't like the light on the tire so we did a proxy which was just a simplified offset of the tire on which we would bake an ambient occlusion and the idea of having a proxy is that it would let us spin the wheel we didn't do it on this project but we were already thinking about it and then we went more into cheating and we knew that it was easy to reflect an HDRI on Unreal Engine but it's way more difficult not to reflect it so what we did is using ambient occlusion as masks to say to kill the reflection or the specular on some parts it wouldn't bring more realism but just more contrast and would change the old stance of the car itself and we were doing fine with Bakes but then oh no sorry I forgot to mention that we already did a script to do the Bakes because obviously it's also boring to do Bakes for each object and so then ray tracing appeared on Unreal Engine and we thought like okay maybe it's a good opportunity and it's the next step into automotive visualization so we tried it but we also thought that it could change our ambition and instead of just having a steel car on the ground we would just animate it but we were like completely new to animation and so it was really really bad at the beginning because basically we did lay a whole rig with just empties everywhere and constraints and everything and we thought like yeah if it's moving in blender there is no way it wouldn't move the exact same way in Unreal Engine but it was completely wrong and it took us a lot of time to understand that the best thing was just to do one armature and not animate the armature itself but the bones and yeah we just did something simple with one bone on each wheel a driver which is connected to the main bone and then two bones for the steering and an empty in front of the car just to drive the steering and this is really simple but we could go further and bring like suspension and everything and something else appeared that we didn't expect which is the fact that you have only one object when you import it, when it's parented to an armature and basically it means that you have to shade every different component if you want to be able to shade them inside of Unreal Engine because well you have just one object but this is easy we just did it in blender it was fine then we did a quick studio in blender and we were finally able to have fun with Unreal Engine and the real-time retracing and I just want to credit NF Studio for the screens because he did like crazy artworks that we took from his email and it really gives a good mood I think on the video and it was really interesting to do this project and we learned a lot but we also understood that we were getting a bit far from the creative process I described at the beginning and this was not really suitable if you want to go fast like you can be really efficient to present stuff with Unreal Engine but sometimes you need to be more than fast and this is when I started to use more blender as a visualization tool and so for this I knew that I wanted to go experimental and try some stuff so I just did a sandbox project which had two main focus the first one was to use alias for the modeling and not blender because it's easy to bring mesh from blender to blender it's really not a problem but an industry standard is alias so I would use alias just to try but also to keep an open window on the creation phase and not fix things by doing exports and to Unreal Engine for example in terms of animation and everything and basically in other words this was just using EV and so I started with those sketches and did the modeling on alias but I wanted for each of those steps to be able to visualize it in EV and the big problem was the import part because basically alias is nerves and as you probably know blender and nerves are not really friends and so you have to tessellate it but if you do it you basically have two options if you do it in alias you take your group you often work with group of surfaces for each of your parts then you have to stitch it but the stitch will basically make one object out of connected surfaces which means that if you have surface not connected you will have two different shells and they will be named shell, blah, blah, blah and then you have to tessellate it this can be done easily during the export but still you have to do it and it makes a lot of steps for each of your object but if you import straight away without stitching in blender then you get this really bad outliner because basically you have one empty corresponding to your group with the name of your group and inside you have all your surfaces which is really bad because in the viewports you just select your object and you get one surface only and it's really annoying to shade so basically what I did was just doing another script which would take each of the empties get all the surface parented join them, rename them and remove the empty and this was helping me because I could just import each of my step in the evolution of the project really easily and really fast and so I was able to visualize finally my model in blender but still I needed one thing to be able to get this which is materials and I was kind of new to blender materials so I started to build this is the actual material shelf I have and it's really new so I have to bring a lot more but the basic goal is to have something only procedural from car paint to carbon and wood but I guess I won't be able to keep it procedural since for leather and stuff like this kind of pattern it's really really difficult unless someone here has the solution I'm really interested in and the other goal was to do something simple out of something way more complicated and bring this kind of tree into simple settings I could just tweak really quickly and not spend hours just to make the materials for the visualization and that was it I was able to have this and use it as an underlay so that I could sketch on top and think about the lines I wanted to give and everything and if I would have worked with a designer transportation designer he would have made really coolest thing not just lines but I'm sure he would prefer to work on the EV visualization rather than the alias basic visualization one but still I wanted to go further with this and I thought that if you have a review and you want to present your project maybe you don't want to get to the settings and the interface of Blender and I also it happened to me that I was just here to change the color and the designer was like a bit more blue, a bit more green and I don't really like that I prefer just someone to give his vision directly so I did just a small pie menu that would take the settings of your material whatever the material is and take the color on one side take the slider values on the other side and a preview on top and that way you can work in full viewport and change your color easily as well as changing the material itself because this is basically the visualization you have in the shader editor and if you want to do a full version full carbon version you can in this directly in the viewport and so yeah at the end I think Eevee is really playful and it can bring a lot in design reviews since you are able to show not only the project itself but also you can better define your vision with an environment and everything if you want to play with it so it's really cool and this is straight out of Eevee I don't like to do post-production stuff so this is just Eevee and what I see coming next well there is some stuff I would love to see implemented in Blender like VR I know there was something in the spotlight but and I followed with a lot of interest the Google Summer of Code but like a dedicated VR part in Blender using Eevee would be just great and would help not to use third-party softwares that this is what we do for now and sometimes it's really annoying in the API just having loading bars would be perfect because I always think that my computer will just burn when I try a script and interactive mode I heard some stuff about this and this would be great for design reviews just to be able to start some animation with keyboard events and to say like we open the doors we close the door stuff like this as well as an asset manager implemented on with which you could switch like environment and stuff this would be really cool but there is also existing tools that I have to try out like tissue that we saw this morning and it's so impressive I really have to try it out for like front grill and procedural design as well as Vershock it's kind of the same idea of parametric things and the conclusion is the process we ended up with my friend Valentin which is basically this to go from an ID, model it and then visualize it and what I like about this is that if you remove LES you basically have a completely free process for internal use and so further is the modeling part a cool thing is that LES recently added a feature which just like polygonal but it translated into surfaces and so what is cool is that you can import low poly OBJ and then it will translate it as surfaces so that you can do trims and fillets and every other thing you can do in the softwares I didn't talk about HGRI mapping but this is also a really, really cool thing you can do in blender to create environments then you go to blender just to bring all the things together and then you have the visualization part sorry with Unreal for the VR and retraced experience and when you have more time and EV if you want just quick reviews on the go and yeah I think this could be suitable for the work I'm going to do and I will try to bring this in my future job for sure because what I like is as you can see blender is at all step of this process and it really brings creative freedom that we don't have in other softwares so thanks a lot for your attention if you have any question you can drop an email and if you have more question on the Unreal Engine part you can ask to Valentin to help yourself thank you, thanks a lot