Texturing Uncharted 4: A Matter of Substance (presented by Allegorithmic)

[Music] good all right hello everyone uh my name is Sebastian I'm the founder and CEO at algorithmic and I'm today very proud to um uh start this session about how the people at noty dog used the substance Tool Set uh which we develop to texture and sharted for one of the most anticipated game ever basically um it started a a about 3 years ago um and they uh they will they will explain the the the journey because it's been a journey um and uh but it's working so today without further Ado basically I want to introduce you to Brad and roio uh from noty dog here uh they will explain how they uh first discovered the tool set um then did uh a lot of tests uh and I say a lot and um uh decided on using it to texture their uh next

project at the time which was in chared for so it's my great pleasure to welcome Brad on stage to talk about that thank [Applause] you thank you Sebastian and thank you to Lego rithmic for hosting us it's really a pleasure for us to be here um you know as we speak the team is actually still working really hard to to finish up the greatest Uncharted game ever made and we can't wait to get it into everybody's hands um so to kick things off we'd like to share the story trailer with you and and minor spoiler warnings uh if you haven't seen this you you may want to close your eyes and ears but ratings will be coming soon we were meant for this Nathan you me together

we were destined for something [Music] [Laughter] great for those who proved worthy Paradise [Music] awaits took a long time for him to get out of this game he's meant for this life oh oh crap last I checked we're all a bunch of Thieves digging around where we shouldn't Sam we're not on the same side I am a man of Fortune and I must seek my fortune so are you ready to seek your fortune hang on L murder the artif that's I made a promise that I was done with this life how long we've been chasing this thing long time real long time just you and me

no offense to these guys but they don't get it if you're done lying to me then you should stop lying to yourself the biggest Pirate Treasure of all time is within our grasp out look Nate I'm going to make you a onetime offer drop everything go [Music] home live your life or we can just end it right here he got to another way I need you on this [Music] one ever wonder like different choices how we might have ended up no I like the hand we've been done [Applause]

pleas gets me every time that trailer the team has a lot to be a proud lot to be proud of and as Sebastian said it really was quite a journey and and one that substance designer was a huge part of of from a texturing perspective uh so I'm here to talk a bit about the evolution and the pipeline that we we created at at Naughty Dog along with Lego rhythmic to integrate substance designer into our internal and external tools my name is Bradford Smith and I'm a texture artist at naughty dog so some of the artwork that our artists generated this is Sam's young Sam's dirt bike Here by Brian Kenny these are all substance materials and a city car by Tate messian and Jacob

Norris uh and the hero 4x4 that you might recognize from the trailer and it's arch nemesis the turret truck by the fantastic inol and here roio has some great fully procedural tiling materials from substance designer and there's another one some rough cobblestones and here's an example of mostly fully procedural Textures in the context of an architectural level and then again but in an organic context so what we could achieve with substance designer was very Broad and diverse so our journey is going to take us through the pipeline and the history of or the history and the evolution of the pipeline and we're going to look at how we really utilize substance designer to handle our Outsourcing and our content management we're going to come back home and we're going to look at how we used substance products internally and just a few bits on what the Horizon looks like moving forward so it's important important to

understand when we're evaluating tools naughty dog has a culture that's very open it's flat and it's extremely creative it really lives and dies on collaboration and flexibility is a key component into what really makes everybody successful there and ultimately creates successful products so we we really were jumping into Uncharted Territory at the beginning of the project so we were of course rolling off of Uncharted 3 and this was 2012 The Last of Us was in full production and resources were very tight so it was really up to the art Department to find Solutions and like most of you are probably already familiar with we were adopting physically based renderings of course consistency and Library materials were very important to us and as were material layering and inheritance we wanted robust libraries that artists could utilize quickly but the biggest thing that we knew we were going to run into was the

scope and the scale of the environments and and the props if if you're familiar with the Uncharted franchise you know that it's as richly diverse as it is large and we knew we had to be able to work efficiently and quickly and respond to changes on the fly so revisiting our culture and the needs for our tools we really had to find that marriage between a standardized common workflow and still allow our aist our artists to be be free and creative and do what they do best and substance designer really fit the bill there so of course we were looking at the time at substance 2.5 and it was really the start of an amazing relationship with a Lego rythmic that we were thrilled to be a part of you know not only did they help Define our internal pipeline but we helped inform some software design as well so it was a very special relationship and to have tools that really focus on the the games industry and texturing in general was a a big win and very well timed too but of

course we started off with these these very early questions of well can we really get high quality results out of a procedural and node-based system as as many of you are probably wondering or have encountered and can we prototype our pipeline because again our resources are tight so we need to prove this out on the art side and can our internal artists and our vendors effectively use it because there is a learning curve right and it's node based and not everybody responds to that so naturally as you may know you know Uncharted is a very stylized but hyperreal environment and we hand sculpt and paint a lot of things there's a lot of Pride and work that goes into every asset so it was natural to start in our zbrush Pipeline and we simply took our hyp Maps out of zbrush we brought everything in a substance designer generated all of our material maps and then Blended our materials in engine and it was very straight straightforward we

could see that was very efficient said okay well that's great now let's take those maps and now let's prototype our material pipeline so let's look at how we can combine materials together not just blend them but find interesting ways to achieve material effects like this rusted paint you see on the right so this was a very elaborate early prototype of a material layering system it's actually much simpler now um but it proved to be very successful and of course now in practice we have our height maps that are plugged into one of our early generators this woodwork node and as we move through the graph to the right you'll see we're blending a lot of materials and generating our final game textures and that was beautiful because we could simply swap the height map and tweak a few parameters whoops too soon tweak a few parameters and you've got a whole new texture set so the iteration and the speed and the power once you get things set up became really apparent so we said

okay let's take this one step further let's do a small scene let's really prove this out and so we did that we took a bunch of architectural and organic elements made a lot of Base materials a lot of these are fully procedural or using seab brush hyp Maps we also did a few props here as well and we thought it was largely successful at the time um and this is the result of that while we were also doing that it was important for us to gauge whether or not our vendors would respond to this so with very little training and very little information we actually did a test and we're pleasantly surprised that it was a little bit of a hurdle but the results were were far exceeded our expectations and we said okay this we can move forward with this so we were ready to take the leap I don't know about that though I'm afraid of heights but um so now we have some implementation pipelines some implementation questions we need to turn this into a

pipeline and we still need to really get the artists to Rally around the procedural workflow there was still some hesitation but we also needed to standardize our asset packages and this is really important because we we utilize Outsourcing heavily for a lot of our props and objects so things had to come back consistent and a lot of times these assets would go through changes internally and it's important that we utilize the non-destructive texturing aspects and our artists can quickly turn around changes so the next natural place to start would be to make a bunch of awesome nodes and the important thing about this these Library nodes is that they emulate existing workflows and applications that artists are already used to so the tools that you would do in in the tools that you would use in software that you're familiar with you would have access to right here inside substance designer and then we created some even more advanced ones that could do some

material effects like the the chipped paint and then some topology mask blending Rose is going to go into more detail a little bit about that later so we said okay well the workflow seems pretty straightforward this is great we're going to use substance in the middle in the red here as our sort of content management system for a particular asset all of the required meshes um the Maya file any bit Maps uh the high poly meshes they're all going to get fed into substance and we're going to Output our game textures from there pretty straightforward so we put a few restrictions on our sculpts so you can see on the left here that primary and secondary forms are very well established and a lot of the surface detail in the sculpts are largely absent because we're going to add that in designer later on in the process and ultimately that gave us the most flexibility we could take this asset we could have a stone version a marble version or a wood version and so material IDs became really

important and we used to have an old method on the left here that we would have a a zbrush plug-in that would do mesh IDs for us or you could use material IDs but it was kind of cumbersome until designer released their their name pair baking and their their their mesh ID and their poly group ID Bakers it became very flexible now we could just organize our assets nicely and and automatically bake all of our IDs so we said okay let's just say we're going to use substan as our Baker and it gave us great results um everything was reproducible inside the substance package we got all of our bake Maps here in the center from World normal to curvature tangent space all the necessary maps and we again would output all of our game assets from there so we couldn't call a particular asset package complete until all those components were together into the delivery package the game asset the high

poly any little bit of source and the important bit is that they were all tied together through substance that was essentially the little mini content management bit for each asset so great mission accomplished right not so fast um we have a lot of nodes now we starting to get a lot of assets management overhead became an issue as did debugging and training because we have a lot of standards in place and then trying to stay on top of technical issues and updates became a challenge and remember at this point we're still just an art team so it was time to Learn Python so I rolled up my sleeves right and uh luckily I have the logic side but not the math so the first step there right we want to automate this so let's compile the library and right away this fixed a lot of the issues that we had and by compile I mean take the substance Source the SP SPS file if you're not familiar with this and generate a sbsar file

which is a blackbox compiled node this right away solved a lot of the dependency issues we had with our our external partners and it also meant we were Source protecting our substance designer files now which was great because we built a significant number of them and some of them very complex and it might be slightly faster I don't know Sebastian might know maybe by a millisecond or so so here you can see on the left hand side what a very simple asset would where we're taking the bake Maps we're blending a couple materials and you see that since it's red that's an artist using the in-house nodes and on the whoops and on the right you see that all the nodes are green so that's going to be a vendor using the compiled nodes same results same workflow just different libraries so to manage those different libraries substance designer allowed us to create custom configurations and project files and this became this made it really easy in order for us to

control what our in-house artists see and what our out of artists outof house artists see but we also had a significant amount of software applications and configuration files to deploy it seems pretty low-fi these days but bat script was the best I can do and it worked really well actually with one click you got all the substance applications painter designer bitmap to material and everything was updated and there's a smiley face in the end too so that always makes you feel good uh this is one of my favorite um Shader customizations that we're having some issues with the video but the are we yeah I'm not even going to go there um but the idea is that even the Shader was customizable and what we're doing here is we're actually picking material IDs off of the model in the viewport which for complex models you know would save significant time um so that was a really nice win as well and I've been me

I've been talking about substance being our little content management system and this is something kind of unique and essentially the substance file is XML and that was awesome when we found that out right because now you can parse into it it's very open you can check for errors and you can fix errors so you can automatically technically review and fix a lot of things so okay let's take this automation thing a little bit further so we had a fairly robust python package uh that was all just a bunch of like substance related functions and then we also used substance batch tools which which ships with substance designer and this automated everything from our library maintenance our spsr conversion process we even automatically deployed a lot of our Outsourcing tools so that burden wasn't on any of our uh in-house management team and then we also had some batch XML editing tools and then I also would collect certain analytics

unfortunately the UI didn't make it because I had to switch to Art rated tasks so it was largely me just you know running it in the IDE but so we were in a pretty good spot to say okay you know let's lock down production this is really good so a bit about our vendors and this is where the amount of volume really comes into light so we work with about 7 to 10 Studios and we're still working on these numbers as you know we're finish in up production we're looking at about 3,000 assets that were done with our external partners and we estimate about 2third of those are background and prop related so that's a lot of assets that are going through the substance designer pipeline so training became really important and ultimately we tried a lot of documentation but really who reads documentation right um and example files and then videos row and I did a lot of demos and timelapse and proved helpful but the important bit you can see here

on the right is that the example files are all browsable in the library they are just assets but ultimately what was most successful is we did a lot of In-House vendor training we actually flew some senior and Lead level artists um for a week or more and they spent time in the studio we were doing classroom sessions we also had inhouse assignments and reviews and that proved to be incredibly successful and you can see here an example of what a typical asset package might look like this is a weapon set done by original Force so this is what the artist would see upon return and this is a little more on the complex side but the flow is the same materials are being Blended masks are being output and there's a lot of flexibility in that graph so what makes our back home or our internal workflow is unique well it's about the same really just a different Library um and again we we did a lot of

internal training and at this time now that we have a lot of tools our artists really embraced the workflow and our usage is is I would say a majority of our artists have used it but at different capacities uh a legar rithmic also their own Jeremy he would come up fairly often and do training sessions for us and email updates actually ended up being a big win just weekly email updates here are the library updates and changes that constant communication during this growth process was really important and you can see our texture pipeline is is actually very similar to our asset pipeline our our 3D tiling sculpts they get baked inside substance designer and all of our textures get output from there and here you can see these are actually all fully procedural base materials so not everything is sculpted um a lot of stuff can be done fully procedurally and our character and our prop departments are actually starting to largely use painter um mainly because

of just how awesome it is to make masks and then do all of the runtime material blending so what does the future look like for the pipeline these are predictions by the way I'm supposed to say that um but we really don't know but we know that we need more base materials right we wanted flexible tools that artists could use but we didn't invest heavily in base materials because we don't want to block we don't want to blackbox anyone but we know that this is going to be a critical component component to to getting even more efficiency we're also looking at completely moving our vendor workflow over to painter and we're going to continue to to automate the internal workflow provide more Deployable uis and and really get these these python tools into the hands of the artists so I want to thank you for listening to the evolution of the pipeline and I'd like to introduce my talented colleague row here who's going to go into more depth and I hope you enjoyed it

[Applause] hi my name is Rell and I'm a texture artist at na dog and to get right to it sub designer and its basic form is like Lego blocks and you have a different notes and you just attach them together and get something magnificent so here's an example that of a seab brush workflow and what happened early on is that theity of um artists that came in were kind of scared of the initial tool because it's not like a traditional painting tool or like SE brush where you end up being more artistic with your with like a pen or a wome tablet for example and so being a node-based system and initially was kind of scary to deal with so we end up making examples uh like this one H where we showed how TimeWise you were able to do a color map

in a painting program the color map that you see there in on the left side is uh it took around like an hour or so while the substance variations uh all these three were around like 10 15 minutes and here's a few more examples and as you see this was when we were baking with xn normal we bake now exclusively with substance um so in this case the we are also trying to see how we can do procedural materials and if it would fit our uh quality or painterly quality that we have in naughty dog this was one of the uh one of the procedural materials stucco and basically you can get enormous amounts of variation and this was part of the of the vetting process so the this the same stule uh generator and as you can see you can get plenty of variations from

it and then going back to what Brad was saying about this um car we actually ended up doing this layering process and vetting it because we are also working with our internal uh what is a material system H to figure out how we're going to layer all these materials into one asset as well as a we sent this same car to the Outsourcing vendors and so they can see the the results and at the same time we gave them this test and we were surprised with the cars that they sent out to us and while looking into doing this um test we also found out some interesting tools within the node structure of substance and within the node structure of substance we found a you know adaptive masks like we're showing here and these masks what basically they do is they grab information from like curvature AO uh World space uh Maps or

position and you're able to move around the dirt or the grime and um like raindrops and stuff like that on top of it using these masks you can mask away the the dirt and or rust and or like um what is it Edge cracking and stuff like that this video doesn't work so I'll move on basically I was going to show the uh the Adaptive masks and in this case case well we ended up having a lot of discoveries things that we didn't initially even think were going to happen is without Our intention this is an example of an asphal flat material and uh it has an asphal breaker attached to it now what the asphal breaker does is basically it breaks up the asphalt flat and makes it into have all these kind of cracks uh so one thing that happened was happy little accidents H we attached this asold uh was a breaker to one of

the Brad's substance um rough material and we were able to get a totally different texture that was not even intended and as we went on these are the kind of things that just kept on happening artists would come up with interesting things and we would end up getting different results uh without even our initial push and you know knowledge and that's what makes a substance amazing it's like Lego blocks it's a it's fun to play with and the same time you get some amazing stuff just putting it together and well greatness of Weights this is uh a note by Jonathan Smith a very talented texture artist at NAD and um this is a note that basically has a pattern node and it goes into a stone Carver node and what this does is that we can get multiple type multiple variations of this kind of er material this this node as far as I know from

talking to the artist at nauy dog it was used in seven different levels in a very different context so when we see these kind of you know notes just pop up and it's just gives an enormous amount of variation and flexibility and the iteration process gets shorter and shorter because of it so we got to a point of no turning back and so there next images are from uh the E3 demo and if you guys remember for the Sha sequence where the truck is going you know Full Speed Ahead Etc so Theory into action now this is a a huge Vista scene and if you've seen that demo the Vista scene is all over basically you can turn the camera around and it's going to be all around you the artist that worked on this vist scene is artum and he did amazing job of setting up the scene around 70% of this scene when it comes to textures was some

substance and these are various uh substance materials some of these went from SE brush workflow uh to the substance or are fully procedural and these h four are uh excuse me these four are fully procedural materials and the cloth you can change it up and get more threading less threading change up the threading if you like uh the same Stu and the same stucco combined with a brick and uh and a rust material here and all these you can change them up whichever way change the color and Etc so we got to a point also that we are making all these awesome tools and for this demo specifically since it was going to be a huge demo a we uh we had to do like patterns in this case uh we end up making these patterns they could be used in multiple situations this kind

pattern could be used for Cobblestone but it can also be used for like wood by combining these notes together and there's an endgame shot of that same Cobblestone and we also did this for uh the city Shake sequence it was um a different kind of tiles for rooftops H we only have practically used a probably two of these but there's more of these there's actually like seven or eight of these and they're all based like on actual like architectural tiles and you can make these look destroyed or you know clean Etc and here's a few examples of them in game H with the pattern showing where it is and here we're getting to the actual images of the level and in this case this is a market sequence where you're battling it out trying to

get to your Jeep so here what I want to point out is all the Outsourcing assets that we have to do to get them into our levels a lot of these all actually all of them are from H used substance and this scene is uh interior is seen and I know that the artist texture artists that worked on this Alice and as you see the names below are the artists that that worked on these scenes Alice um our texure artist here it did an amazing job of getting the quality for for all the materials in here and she exclusively works with substance this uh seene for the downhill sha section and this area that I worked with um sack Oliver and everything here is substance as well there's another scene for the staircase sequence I work with Jacob Norris and and it's a Outsourcing a full

substance and what's interesting about this one scene is that I don't know if you guys remember it's a scene where the Jeep is actually just pile driving through this staircase and there's it's just like amazing how much detail is put into a scene that you're really only going to see for like 6 10 seconds and there's uh the right after that staircase there some municiple area and there's like a road that goes downhill and it's um kind of like a community Road and then we get get into the uh into the vastness of uh the outskirts with more like Shanty Town areas and you can see that the our Direction starts to change and it becomes much more saturated these things are amazing actually you can actually go and literally inside our game play the character and she just walking around and it's this is

great there's a place where Nate and Drake is like dragging on the mud and hitting everything in sight an our scene from that area there's where there's a scene where um the Jeep actually tumbles over and then Drake is trying to get out of the Jeep and then shooting the bad guys there while the evil truck is after you and there's a a industrial Zone area where the grand finale for that scene and then it goes into the Cinematic uh sequence here H so yeah I mean this this was all done with substance uh the demo at least 80% of the assets when it came to textures were a used substance and I just want to say thanks to algorithmic for letting us use an amazing tool and also to the Outsourcing team and um props to mark our Outsourcing manager who has been

amazingly patient with all of us dealing with Outsourcing thank [Applause] [Music] [Applause] you thank you all right so um that was amazing hope you enjoyed it um at the beginning was not working I I have to tell you this so it's not it's working now but the I mean the tools 3 years ago were different than they are now and uh thanks to these guys and a few other people in studios we've been able to um um yeah make the tool evolve in such a way that it's really made for all the game developers um um I have to say a few words so first um let me see do I look yeah so it's not confidential obviously but a few words about road map uh for

substance painter so we you you've heard Brad say about painter and designer so we have mainly two tools substance designer and substance painter so substance designer has been the The Hub tool and the management tool that they the the guys at NAU dog I've been using a lot and painter came along came after this one um and it's a 3D paint application a little bit closer to what mudbox Mari uh or or or 3D code can can do um uh so it's it's newer and we just released a version 2.0 2.0 today a few minutes ago uh so uh if you want to have demos of both tools substance designer and uh substance painter too you can come to our booth and uh at our booth uh 316 if I'm not mistaken uh not not only will will you get some uh some demos from us algorithmic but also you will see more U talks like this from from Brad from Ro roio from Kristoff and and

glauco from nauy dog as well uh from uh Lucas from Ubisoft from other people uh uh um Josh from uh um Infinity W sorry uh so you will you will have some some more talks like this about how they these guys amazing talents Ed uh substance for for some of their creation so uh a few words about the the the road map uh what we will be added in 2016 or has been added already uh like scripting features in substance painter you will be able to um yeah script things and it's available already you cannot do much right now to be fair but you can do some uh i r integration uh we just presented it just a few hours uh also with Nvidia Ira I don't know if you know that Ira it's a it's a path Tracer from Nvidia uh what that means is that uh right within substance painter now you can use the path rcer instead of a a Ruster engine

uh to come up with a beauty shot so if you if you're working on an asset and you want to shed the best light on it and potentially export that to sketch Fab or or or or out station uh now you can do that right within the tool without having to export all the maps and uh reimport them in a in a rendering application so you can do that right within the tool substance pained so it's a it's a very very nice uh workflow look Dev look development workflow that you uh that you have now um layered material authoring and visualization it's uh the visualization part is uh already available again we haven't you can have demos about that so it's a little bit more technical so let's say you have a special type of Shader that you're using in your game engine uh you can load it up uh I mean you can twick it so that in the commands section if I'm not mistaken um you make sure that substance painter will only display what's useful and will remap things here there uh so that you can work within the context and within

the constraints of that uh specific Shader type you're using for your game so that's that's very very very powerful uh you can have demos on the on the booth as well so there is a new UI new shelf better quality content obviously uh it's it's better now um new painting tool like clone and blur and smudge that sounds stupid like having a smudge tool or blur tool but a clone Tool uh but the thing is that uh um substance painter is a fully non-destructive tool which means that it's when you actually use the Clone Tool uh whatever you cloned if you do a modification on what was cloned originally uh it will follow the steps it will follow the Clone step so if you say cloned a I don't know a part on a hard surface if you if you want to change the the screw if you change a screw it will be repeated over uh for you automatically so that was uh kind of complicated to actually Implement and uh

we're pretty pretty happy with it uh subsurface scattering so that's uh we we have a fake one uh at least you can do some and uh that will be coming these two items are important and uh 8K painting uh and EM support so we we currently have an 8K export it's hidden somewhere uh but we don't we don't talk about it and uh but it's an export function and it's because when you well when you want to paint in 8k to be fair it's a bit laggy and so it's not a great experience and we we don't like the fact to U not having our customers and and and and yeah users have a bad experience with the tool so we prefer to to hide it for now so 8K Painting full painting will will come eventually we're working on that uh and and we're not only hoping about uh the fact that Hardware would get better so so we're actually actively working on it and udm support I know if you know UD what are it's basically a way for for you to not uh having to care

about the um the amount of uh data and textured uh uh information you you want to paint on your object so you can get crazy with that it's a what uh uh it's a it's a standard in the U visual effects animation uh Fields uh where they where they really want to have the finest details on every nails of of King Kong so now if you want to have like the finest details like 8K on the nail each nail c will be able to do that so that's very happy um uh substance designer same thing scripting features uh are already partly here and we will get stronger um this is a biggie here um you I mentioned the fact that we've integrated Ira inside of substance painter Ira relies on something that is maybe not not more important but super important which is the mdl is material definition language so this is something Nvidia is trying to

push now it's a it's a open open um material definition language um it it's been missing in the industry for for for years to be fair and uh we we certainly hope and we support it at least and we we we think it has the potential to become a real standard uh when you want to describe full materials not only textures but material materials not shaders it's between the the textures and the shaders actually there is no lighting uh uh information instructions in the emdf but if you want to Define what makes a shade what makes a material mdl is the we think it's a good it's a good solution so we're turning basically substance into not only a a textured editor but also an author authoring system but also uh an mdl editor so we're adding one more type of graphs there there were there were a lot of them already so uh it's going to new to be a new type of graph that will let you define the materials on top of the the textures and this materials can then rely and refer to textures that have

been defied by substance designer so you can you can get crazy with that uh how surfacing tools uh we yeah we definitely want to to have something like this text tool again has been missing uh but we want to make it right again we want it to be really non-destructive uh the way it should be but um we working on that and enhance Viewpoint obviously having a better better viewport um you you already have one one parity with uh Unity unreal all the the best engines out there um uh because we basically use the same underlying principles uh and also we have been working closely with with these guys um so we we made sure that what you see in substance either painter or designer is Will what you will get eventually in in uh in unity or or or other all the other good engines um yeah that's uh pretty much it so I guess um some time for

questions and thank [Music] [Applause] you and before the questions I I want to thank you guys again uh it's been it's been yeah I cannot express how proud I am today and really glad to have you here so thank you again [Applause] any questions yeah the NAU guys we will have to repeat the question I guess so your your question was did we use it on the the terrain textures um we

did actually i i r can probably speak more about those particular terrains but we've used it on a number of different surfaces from hard surface materials organic terrain materials um really the the we've used it just about everywhere across the game yeah most definitely we've definitely used it ER everywhere we also use a WHL machine and I didn't do the a lot of the whites which are like the the the huge you know expansive Vistas that we see but the artist that did use World machine and um uh used the substance as well for the tilable materials that the mask were created from so we definitely did use that um there's a line so there are mics here so if you want to uh to ask questions please line up yes hi I have a question regarding the analytics you mentioned that as part of your automation package you were getting gaing analytics if you can elaborate on what kind of analytics you were

gathering and how they were useful eventually absolutely so your question was what kind of analytics did I gather and and how was it helpful um they were very basic you know at the time I was really you know curious about what type of nodes people were using um and what was working and what wasn't because we were really figuring all this stuff out and one important thing I found is that really complex Uber nude Uber nodes uh weren't being used heavily it was the small modular ones that allowed people to do one simple thing very well that were being used across a number of assets so um it was really just about how to better build a library of of nodes thanks hi um thanks a lot great presentation really enjoyed it um I was wondering the enhan closer to the mic sorry yeah the enhanced viewport is it possible we have our own lighting model uh and our own engine would it be

possible to adapt the lighting shaders as well the look at the what are can you repeat the lighting model of the the viewport itself the lighting model of the viewport itself you mean the changing the lighting conditions and adding lights and you support Onre and unity we have our own engine would it be possible to hack into the lighting Shader somehow and try to results uh I don't know actually well we I don't know well let's talk about the um I I might sorry might to add one thing to that it is just um it's open GL right so we were able to customize a very few things we changed some of the energy conservation um and then we did a lot of weird things like I wish I could have showed you we were able to pick material IDs from the viewport so anything you could do um in openg we were able to do inside of the viewport Shader okay yeah I managed to find a compiled open gel Shader but no no uh no source what's a question sorry can repeat the

question can you get closer I indeed found the view Port shers but I managed to only found find the compiled version and I couldn't find the source really they through the files yeah okay they they're actually in there's somewhere installed on the system it's all open you can actually open it up and and and see what's in there it's I think it's in the app data location but so it's open to a certain degree so this is why I needed to to to hear more about your your your issues all right we'll talk later sorry hi uh I I have to apologize for my English first uh sorry um uh two two new questions um what's the standard res resolution of your Textures in the game um well that varies depends on on the project and what we need it for uh most of our materials uh we we cap them out 10 24 and that's what we've gone for but

when it came to you know memory memory management and stuff like that we would have to bump them down to 512 or Etc uh and sometimes the quality was enough at 512 so uh but our standard stay at mostly 1024 okay and you can generate them at upper resolution because of substance at each time if you want to make more HD textures you just have to switch parameters or there is another you you know what I mean Oh you mean like runtime uh something not runtime if you want to in the next Uncharted if you want to use a I don't know the on the PS5 you want to to reuse all these materials because it's a very big problems you have to rebuild all your materials each time you CH switch uh your game and now you can make a material and then use it but during 5 years 6 years it's not deprecated because you you can

regenerate it with a better resolution yes I mean that's that's really the I guess the deepest core that we we don't even understand that's like the future of substance H we that's a good question because we we currently don't know how long how much it's going to expand I mean this the tools that we're making now are just keep on growing and I it's it's predictable that we're going to use many of these tools and or leur into the next game and and just because you you mention World machine it's a it's a software I used a lot and I love this but it's very close to substance designer uh finally and there is a a fabulous block called erosion to to make ter and I was uh many times asking to myself if I I have to rebuild it in substances to not use word machine anymore did you try to replace softwares

with designers um I actually it's funny because I actually did a test uh very early on it it's difficult yes you could um it I don't think we went there because it wasn't practical and why why make something why remake the wheel if it's already done well H for that example even though I got a really cool result from it and I was very proud of that initial node it was a a Terrain node that they could have expanded into something more but at the end the world machine did an amazing job for us and that's why we stayed with it okay yeah thank you very much yep Hey so uh my question is um for your foliage in world did you guys use substance for that or was that a more traditional workflow um that that's a separate workflow I I'm almost wondering I don't want to say too much about it because the people that worked on it are extremely talented I almost almost want to leave that information for them but

there was some poish that did use substance okay and some of them used a different workflow that's not necessarily SE brush either so I kind of want to leave that to those artists okay fair enough yes thanks what's up guys um I guess I sort of have a related uh question um and I'm just curious if um did any of these like procedural texturing uh stuff you did with with substance the libraries did that affect your character art at all did that uh or affect sort of like that part of the team so your question was was did the libraries and the workflow did that go into the character Department um yeah did the did the character guys using I'm supposed to repeat the question um so it it it did but to a different extent and you know I and row kind of touched on this with the foliage there was a lot of different pipelines within the art Department that were being developed in parallel so you know what we covered is largely

background and prop related um and the character needs while they're conceptually similar are very different in a number of respects they actually ended up going more toward painter as soon as that was released because it's it's just a much more fluid artist controlled program for their type of work um but that isn't to say that Designer still has a very real use in the content management side of things and the baking and and uh you know I would like to see more of that for sure hi um I'm interested to know if you've got say a metal shading substance and one day you go we want that to be more shiny and you've got shed loads of uh materials using it if you've got some kind of workflow that can change everything that's using that metal to be more shiny and get it into your editor so your question was how you know how do

we propagate changes to to everything in the editor if we're if we're using the source is that correct yeah um thanks brute force and many many times right and it sounds bad but the reality is is that um you know we did standardize a lot of things but often times when you get into the game environment lighting conditions T buffer Precision you name it there's a number of things that just just create different results um and it was easy in a lot of respects to go back to the substance package reexport a modified output and you're good to go or we could also override things in our proprietary Shader editor too so um both of those methods prove to be really flexible at the end when you're really polishing right to add to that we do have an inheritant based layering process in our Shader system so if we were to take that one metal material and

inherit it into another material then it breaks the it doesn't how to put it doesn't break the information the information goes with it but it doesn't affect the the hierarchy of the main material so I'm able to change that metal and make it less rust less um shiny on in that in that layer that's been put in h so that would help on some of those situations uh in our pipeline hi hi so um in the way to to build the library of materials um did you use a real reference with capture or have you a more artistic ways about the look of the materials I'm sorry can you can you repeat question did you use references like real life references for coming up with the materials yeah you know I actually cut those slides out what is the start of two materials it's a physical reference of a real reality or is more a vision of an artistic way it

it's actually a little bit of both we started with a lot of data in fact we went out and we captured a lot of color calibrated photo reference to try to get accurate albos um lot of material you know presets for different shellacs and paints and I could nerd out on that stuff all the time at the end of the day it wasn't getting the game done and ultimately it came down to just let's just make tools that allow people to make awesome art and ultimately that's what we did we had that data in reference to fall back on but for um artistic control and also game limitations uh we relied heavily on just our intuition and eye yeah nothing to add on that it's basically that you can well the the conclusion is that you can do ugly stuff with substance designer as well it's a very capable tool for this as well not naught a dog just ask me to do something for

instance hey guys I know that your style is pretty stylized and that you do a lot of your details pumping first in zbrush before you actually texture it did you find yourself having to compromise on your style so that you could keep the efficiency of substance designer when you TR to make the change or could you still do your style in substance designer that's a that's an interesting question I mean it's a with with C brush and and changing uh what that's the thing not everybody changed entirely to substance designer our our workflow is very open and we we enjoy seeing what the artist can put because we have amazing artists and they many have different techniques and substance being almost like a hub system where everything eventually will get to substance to get into our engine is uh is part of it and yes being a stylistic game uh we realized that we needed to test it further and as we tested it obviously

now at this point we're very happy with the results we were able to get that painterly look within substance and almost eliminate H the the sense that these materials look procedural yeah and that was quite a challenge initially sure uh the learning curve uh not all the texture artists uh have have um gotten to that point a there's still many texture artists that go through the C brush workflow and then finish it off in substance I I'll add one small thing to that too and that's when it came to the library noes you know that's really why it was important to focus on what our core workflow was so instead of creating like a you know make this thing painterly node it was here's a color scheme generator here's like like an an oil wash and like a direct you know opaque painting layer kind of like it was it's very traditionally based um and and it was really derived from existing workflow so it felt like a natural

transition that was really kind of the goal all right thank you thank you more questions all right one more hi um so I'm a modeler and I have some questions about how you handle uh smooting groups with while they texturing with substance uh something I get stuck with a lot is uh using either one s group for models or multiple sing groups I was wondering how you guys deal with that and uh when you use one Smo group and when you use multiple are you so your question is about how we handle smoothing groups are are you assume you're using 3D Max yes ah I can't weigh in on that well I haven't used Max in a long time you're you're talking about the baking and and as well so you got to make sure that that the tools are in line with each other and what we did in

studios is we use Maya so we made sure with various tests that Brad and I and many other artists did is to try to find the the gas of like getting a good normal bath bake uh and that was the situation that was really kind of in prepro and once we figured that out ER that that same workflow a step by step was sent to the Outsourcing team and obviously we have it inhouse as well and that'll be the most important thing to if you're trying to get a substance into your workflow make sure that the tools are talking to each other early on that way there's no like surprises yeah exactly I use like I watch a lot of the tutorials of algorithmic which are great by the way and they kind of teach that like uh one of the models they use is one entire swim group and that works uh out for a lot of my models but I was wondering when you uh uh when you decide a model doesn't have to be one Mo group

because the the normals will get weird the normal map for instance okay is there you can can you clarify anything of that U me I can't but maybe Jeremy who's here maybe and wave your your arm maybe you can answer to the resue come up come up yeah y Jeremy he's the guy who spent a lot of time at na dog yeah so uh but I'm a 3ds Max user so I can actually answer that um so so I mean it really depends on on on your preferences uh and and all the type of model you're using if you're doing something organic uh usually one smoothing room for everything is fine uh cuz even if you have some like messy gradients on your normal map it will it will can it won't it won't show in the game as soon as you're doing hot surface stuff um you may want to do do actually hard edges and and and kind of split things up because um especially when you're using first most math generators that kind of thing that are based on the normal map they will kind interpret the normal map to create a mask for example

if you get too many gradients on your normal map that just account for like the sing errors then your match erators may get you some weird results so it's really up to you to see what type of assets you're generating do you see any artifacts with with your current method and kind of adjust as you go there's really no no absolute rule about it yeah so it's basically like learning a lot and yeah it's really like learn as as you go and like there's no there is no easy answer that it depends I got one more question actually sorry uh before we get to our question there's actually if if you guys are familiar with poly count there's a Wiki there that shows uh a lot of different baking techniques and it also shows a lot of great tips on how to bake in general H so it might be one thing that you might want to look into is uh polyc count.com and they have a Wiki and they right there they have many PES with um many amazing artists dealing with this issue and they might have

actually have that information for you right there oh that's a good one um so my second question was um did I understand correctly that you only get hide Maps out of C brush or also normal Maps don't well well I excuse me I personally um I these days I actually just export 3D models out of zbrush um so tiling textures and props for me are completely baked in designer just because of the type of maps that I can output there and it's a much more flexible workflow thanks that's gone One S maybe I have like a a simple question or request since you're you're you're creating the geometry typically like coming from zbrush that's obj and then we're adding the substance for the material why don't you come up with a file format for simple shapes with their material and we're done with that thank you I like that idea

yes all right uh hi so I was just wondering if did you guys get to try using photo photogrametry into your uh texture Pipeline and if you did how effective was it h I know we H in our prepro we then h test it out ER we looked at it and I think when it come to naughty dog's artistic ER technique for for how we arrive to these textures and the painterly approach a and especially the the type of environments we make a these materials are not easily found uh a lot of these materials are things that you probably won't even see in the actual world so it becomes harder and harder if you're going to approach especially with stylistic games um but there is a lot to gather from that I mean there's many amazing games that

have done it and we have mentioned it and it's something to look into the future for sure it really is a labor of love and and you know that was one thing we really wanted to stay true to is is just the the Artistry behind um not only the tech but but ultimately that hand painted hand sculpted feel so we wanted to stay true to that uh one more question we don't have no more questions sorry it's getting sorry all right okay thank you thank you well thank you Sam come on sorry