Star Citizen: CitizenCon 2949 - Terra Firmer

Transcript — every row opens the video at that moment

1. 00:00:11

   everyone my name is Marco corbata senior technical director my name is Misha Colburn lead environment artists on the organics team and I'm Allie Brown director of graphics engineering we're going to talk about the evolution of planet tech today initially our goal was to achieve a seamless transition between space and planets so fly to a distant planet land on e and explore its entire surface with no loading screens so looking up the planets in the sky and fly up to them it's cool but what are the planets supposed to look like and how can we create them so initially I just started doing some test mixing procedural terrain with buildings here

2. 00:01:01

   is another early picture show in procedural a terrain with no mountains why we were getting so some very interesting results during our early test a key factor in star citizen is that we have a specific universe with pretty deep laws so we cannot just generate you know random planets we needed artist input to match the reference and low so we had multiple layers of noise some generic elevation maps shading and colors were based on procedural rules elevation and slope we had an initial editor integration we real-time editing of basic planning properties and atmosphere so after a lot of hard work we got our first version v1

3. 00:01:53

   working here you can see our first man on the planet here is the first spaceship on the planet successful landing so after we won we introduced procedural and artistic improvement during v2 we introduced the concept of ecosystem so the ecosystems a combination of terrain maps with vision objects and gameplay properties each ecosystem had a three channels code of textures use for blending texture layers and we use the same channels for object scattering as well using ecosystems we can make each planet unique matching the

4. 00:02:42

   law but still generating planet size content for players to explore so we to ship with Alpha 3.0 we had the new moons and landing zones we had seamless transitions deters visible from space all the way down to the surface we basically rewrote the entire 3d engine terrain system to work at scale and on a spherical planet alpha 3.0 was multiplayer ready you could already explore planets online with your friends you go to fly formations and planetside battles the planet generation is synchronized between client and server the physics

5. 00:03:29

   collision mesh is generated on demand we will talk a bit more about this later this picture is showing the sense of scale we achieved with alpha 3.0 as you can see here resuming him from a solar system down to a planet through phase you just plant a demon planet view from space while zooming in with the seen high-level formations and ecosystems getting closer objects are starting to appear next it is the space the Granta space transition

6. 00:04:21

   for yella a showing additional elements like icy surface from space next slide please okay and this the other way around zooming down from ground to space okay we version three we decided to introduce more artistic control here we can see the artistic improvements made to yellow in version 3 version 3 shipped with alpha 3 3 5 we are the new moons

7. 00:05:12

   and new planets including a city planet cover with buildings we introduced rear day/night cycles so planets are actually rotating in real time we made a large amount of tools improvements and built a new vegetation system in v3 we increase from 3 channels to 16 channels using alpha values to encode the material IDs but when it is 16 text layers for each pixel would be way too expensive so we found a very fast solution by blending only the closest surface using temporal differing so here we have several layers combined which with each other but without smooth blending and here we're blending 60 layers at once using

8. 00:06:01

   temporal difference so it's basically at the cost of blending only one day so what would be the next steps of the version free we had the design for something quicker to generate with less direct artistic control like we had previously in v2 but still being able to influence terrain shape and colors like in v3 we wanted more ecosystem to get more details improve blending and transitions Bastien we need to take into account special planets like our coal so enter planets before our final version we decided to go for a more physically based approach we wanted to improve on ecosystems blending and transitions and now I'm handing over to Mission to give

9. 00:06:52

   us an overview of before Thank You Marco I hope this clicker holds up because it seems to be some interference so like I said I'm the lead artist on the organics team our team works with the planet tech tool day-in day-out all the planets you've seen so far have been created by the team in de so we had some thoughts and demands when whenever we talk about tools and what we want to improve so wifey for let's go over some of this stuff so wifey for up until now for a long time our focus has been on moon landscapes and although these are visually distinct from each other they're essentially a single biome planetoid the one is desert one is icy one is a lunar landscape but biggest

10. 00:07:42

    differences is actually like the type of SS that you see the shapes in the terrain last year when the team started to work on Hurston we were actually confronted with our biggest challenge yet hjerson is a very diverse planet in terms of what we've done so far so we have dry wastelands we had our mining pits we have trash yards we have hot acidic areas and even very lush green areas so there's on a single planet we had to cover a lot of variety and this clearly proved a challenge based on on this experience working on on Hurston we we started to think about okay what can we do to improve or where are our bottlenecks and how can we proceed so we wanted to make sure we future-proof the technology because up until now it's

11. 00:08:30

    been Stanton but there's this we want to go somewhere right we want to go to other locations and the only way to do it is to work faster and more efficient so v4 is a fundamental change in how fast we can build planets keeping what worked well we work in the things that slowed us down and building a tool that's in engine and is a very intuitive artist friendly tool to use and we'll show some examples of that coming up yes all right cool it seems to work so here's an example of some of the stuff that we wanted to achieve or at least wanted to do but we felt we couldn't really do so you see here a Google image of the Africa continent and what you can clearly see

12. 00:09:18

    is on a large scale the transition from a desert landscape to a savanna to a thick jungle and if I don't know if any of you ever liked the joint like enjoy to look around on Google Earth I personally like to do that a lot reference gathering etc and I always enjoy seeing these these beautiful transitions and they seem quite like strong but if you go up close you can actually tell that these transitions are like incredibly complex so there's all these features in the terrain that inform why these transitions happen in this case where the vegetation gets more lush you see that the mountain escapes starts to block the drier desert winds the water paths are the first areas where trees start to grow and when you

13. 00:10:07

    zoom into these areas you can actually see how the terrain and how the terrain features actually start to inform these transitions another example I'm just have a few examples to go through another one where the terrain creates these beautiful patterns of transitions in between different biome types and lastly there we go European Alps where the transition in height actually causes a transition from like a regular green area to like the snowy mountains and these transitions happen over incredibly large area so let's have a look and go back at Hearst to Hurston and I have a bit of a chat about the challenges we had so we were

14. 00:10:57

    not really unhappy with the results that we were getting especially up close I think that was pretty good especially considering the scale of our game but we needed a wider range of biomes and we could only imagine like if you think about Terra Prime or anything that's that's very complex that we still want to do we needed a system that worked smarter and better so it became clear that our tiled approach was making it really hard to to scale and create convincing transitions now next slide alright so this is a very unflattering look of hjerson I turned off the atmosphere and some of the effects that usually go on top of this but what we usually do or attend did before was paint a global texture it

15. 00:11:45

    sort of represents the different areas on a planet so what we have here is on the pole it's a bit d saturated but that's where the Savannah area means the wasteland then although it's it's not super crisp it kind of gets the point across especially with the effects layered on top of it when you start flying close to the surface especially with all the effects turned off you start to see where the global texture starts to fill and you get this blurry look the textures that are loaded locally for each patch of terrain or each tile is not fully loaded in and you hit this dead zone in between where it just looks a bit blurry but muddy and it's not really getting us the results that we saw on Google Earth and then when you get even closer you start the CD individual patches coming in the texture of the biome becomes visible and

16. 00:12:34

    although up close this this is a cool result especially when you're walking around it these transitions yeah they can be quite hard so what you see here is a few wasteland tiles met like meeting savanna tiles and the artist and the art team had a really hard time just trying to alleviate these areas to get them as good as possible by creating additional tiles to sort of fit in between and it became quite complex to set up and maintain especially with more biomes and a more variety coming in and the drawback from the collar map up close is also that it didn't quite have the resolution that we wanted to so more complex men planets meant more

17. 00:13:22

    individual files to maintain all of these collective assets informed the final look so if you wanted to change a color on the planet you had to go through all these individual files and it's abused the later so we had all these terrain files these terrain steps they had color information as well so if you wanted the red on the planet to be a bit less red we had to tweak all of those individual files could mentor back over 20 or even more the overall look and final look at the planet was easily made up of 500 files that we all needed to maintain keep track of and made sure that they were all in sync so this you can imagine that this makes addressing feedback changes whenever I'm talking with Ian about art direction stuff it becomes quite tedious to get all that stuff polished up and very

18. 00:14:12

    time-consuming so we had a to do a fundamental rethinking of how we want to approach planets and the first thing that we wanted to do was separate the process or the idea of a biome temperature and when its lush or or dry from the actual terrain shape and by this we could keep our terrain library clean and we don't have to create additional color maps for every time we want to have a different color or a different type of terrain showing up so that way you can get large changes on a global scale this is not Google Earth anymore this is v4 and what we see here are natural transitions that globally occur globally due to gradual change in

19. 00:15:01

    temperature humidity and weather conditions and because these are occurring over a large scale informed locally by the shape of the terrain every area where it transitions or where a biome appears is informed by the shape of the terrain and it's truly unique to that area and we're no longer seeing a patch of Dempster savanna trees and then next to it is the wasteland you sort of see that stuff repeat it's a large-scale transition and everything flows into each other so we're no longer fixing a biome to a specific tile and that way across the server's over the over the entire globe every area truly becomes unique and this is a example of how extreme it could go with colorings just

20. 00:15:49

    as an example of some of the stuff you might see in the future just to reiterate this this another thing that we wanted to change actually it's the blending cross fading [Music] transitioning between areas and on the one hand on the left side we see v3 zooming out on the other side we see fee for and Burma free for is not this one is not entirely done we would like to paint a bit more but we had to come up with an example and what you see here is a seamless zoom out with no visual artifacts with things changing or or fading or popping whereas in v3 we still had a lot of transitions between all these layers of textures going on and you can even see it at some point

21. 00:16:38

    the v3 planner if you zoom out far enough this is a global texture kicks in it's the only one that that remains and it might actually have a completely different color from where you actually start it and I'm pretty sure people have not like experiences as well you see a spot on the planet you want to go there because it looks cool you go there and it's actually a very different color experience that you expected whereas at the v4 you can actually still make out the pattern the exact terrain shape that we started with and I think this is a like a big big difference okay let's move on so those were the things that we really wanted to improve and yeah this one is

22. 00:17:29

    another one that is especially important for the team like I said managing and keeping track of 500 different files doing color tweaks and less stuff it's become quite tedious especially with complex planets so a very clean file structure with library items feeding into one file that is edited authored and and maintained inside the editor in a very artist friendly tool so proof in the pudding Chris already spoiled it v4 meant that all all the planets that we've done in the last I guess two and a half years since we started shipping planets had to be updated to fee for which was a bit of a you can nervous from it right that's a lot of work that

23. 00:18:18

    we put into that and it was the true test of how efficient this new planet tech would be and how efficient this tool would be so I'm happy to confirm that the team is updated every single planet for 3.8 to v4 and it will be there so just to get a bit of a start into the more technical areas I just want to share a little bit what we do on the art creation side that feeds into what v4 does now already for the earlier version in v3 we use these little simulation tools that one of our in-house artist Sebastian Schroeder who had a big part

24. 00:19:06

    in the planet tech v4 he couldn't make it but essential input he made all these cool simulation tools that help us with terrain simulation we see erosion which is one a very common simulation that we that you would do with building terrain water flow simulation we see displacement of sand and soil based on wind input and we see terracing so these are just a few examples of the small little tools that we use in-house and see it helps up until now these tools stay we use them just as like a personal thing like you would have a brush in inside your your toolbox you just use it and then you have an end result an output that comes from it and then we

25. 00:19:52

    feed that into the engine so we were able to get like if you're at the right distance you get these beautiful terrain colors and terrain maps so we're not unhappy with that but every time we wanted to change something or I wanted to make a different combination of colors of vegetation you would have to redo it or make it new one so instead of that making final outputs and specific masks we decided to get those tools and simulate stuff and use the simulated data and feed that directly into the engine so this was the old one we had masks we had color maps specific patterns where where certain assets would show up and we started to rethink about rethink this and okay how can we simplify this and make this easier for everything else to use so we decided to go for humidity and temperature it's

26. 00:20:45

    there's the hundreds variations of things that you can think of that they will cause transitions of areas and biomes but I think ten sure whether it's something is hot or cold whether it's dry or very humid or I think those are the two most important things and if you look at some of the references that we looked at even up close the transitions are always happening because it gets colder because there is water in the area and even in desert areas where there's no water those flow lines they still are visible because if there is water it will trickle down into into those pathways stuff will sort of follow and get left behind in those areas so I think with those two maps we can capture a lot of interesting details so this is the new set height normals just due to shading

27. 00:21:35

    necessities flow maps and temperature so you can see in this one point where the flow maps where the water would flow where water would accumulate in the valleys and on the height it's a relative simple one I think high the temperature is more important on a global scale but you see the differences in temperature if you go all the way up the hill or mountain to make sure that every terrain set every height map represents the exact same values with those with those flow maps and temperature maps we have to unify all of our library height maps basically our whole library of height maps and we made them all exactly 4 by 4 kilometers with a maximum height of 1 kilometre up and

28. 00:22:25

    in a maximum of one kilometer down and the reason for doing this is that if we run these simulations on each height map the data that comes out of that represents the exact same value for each height map and this way in the engine we can actually blend these perfectly together and create this seamless canvas of data that that we can paint on and with that I'm gonna hand it over to my colleague Allie who will go into how that actually is handled inside the engine thanks a lot Michelle right so yeah but talk about how we make use of this climate data first want to step take a step back and talk about the

29. 00:23:13

    different approaches to how you can build a planet so I've got a line here showing all the various scales we need to accomplish and a typical first-person-shooter annual average small game I would typically made of maybe two layers of textures so you would have like your your small ground textures and might be representing millimeter accuracy details and that what type of take you up to a meter or a couple of meters and then they'll have a next layer of texture which would be their terrain and you typically offer that something like a one meter resolution and then that would cover your entire play space so say a kilometer for a small game and that would typically account for about five megabytes of memory depending on the tech here using which is awfully nice and cheap in this generation but moving on to something a bit larger a large

30. 00:24:00

    open world game will typically stretch ten kilometers by ten kilometers or that type of scale so 100 kilometers squared over something like this you're looking for 100 megabytes if you use the same per pod so that's still quite sensible amount of memory something you can easily achieve on a cone gen console and definitely on a PC that's fine that's not what we're doing though so what do we take the same approach and push it up to star system scale about five hundred terabytes for a planet so unless everyone's got some nice meaty SSDs you might be struggling there so clearly that's not the way that anyone builds a planet so what about a marker talked about procedural planets like v1 was almost a fully procedural approach had limited our input so if you build a fully procedural planet generation what do you tend to do is feed it of some

31. 00:24:49

    very high-level data it could be as simple as just the distance from the Sun and the size the composition of the of that request or it could be more detailed like a map of the continents things like this and then what you typically have is lots of layers of complex noise and algorithms that would procedurally determine all the details you would actually get on the planet so it'd all be driven by algorithms basically there were no artist has ever hand-painted a mountain it would just come from the system which sounds amazing photos sure here's a very limited quick example to happen to show what type of a small twins am and how you start off something lumpy and then it turns into something bit again so that's a very small scale example so this sounds good so if that

32. 00:25:38

    type of approach you can generate infinite planets which is good yeah we get the whole solar system or whole universe filled but actually it's not as good as it seems you the variety you get is actually just limited by the complexity of the algorithm so I could write an algorithm to say where rivers should spawn and how they should flow and you might get a million different rivers B will never see a waterfall and not until I go and then adjust the algorithm at a waterfall so you end up adding more and more layers of complexity which is fine but at some point especially with a game like ours and you're trying to simulate hundreds of different planets that are very unique you know we don't want 100 earths or 100 you know lunar moons we want lava planets we want you know all these radically different planets so this approach doesn't really work for us and what we actually want is something that's very art directed you know art

33. 00:26:27

    director Ian will tell us you know shows exactly what he wants like a cliff to look like it might be the slope of it the shape of it the colors and we have to match that as their as the tech team so doing something fully proceed well doesn't really allow us to do that and it becomes a very technical non artistic process which is quite limiting so what a start isn't do well up till v3 we we combined four layers of textures so typically would use the first few layers you would get in a normal first person game but then we had this two extra layers of textures that we were stretches all the way up to the global type of scale and we blended all these together in the shader and it cost us about 300 Meg's of memory which is something that you know most GPUs can easily handle so that that's fine but

34. 00:27:16

    like Michelle alluded to it can be difficult to manage all these different layers and to blend them to try and achieve the art direction at the ground level but then still to have a a seamless transition to space and there was always these hard decisions we had to make where if we won the too much color version on the surface then on the from space we maybe couldn't achieve it or there was this constant balancing out so we need to do something here so the fundamental approach for v4 is to make use of climate data so we keep everything we had but rather than like Michelle said painting at all color maps we now paint this climate data or simulate the climate data and on top of that we apply a global rule set for each planet that says based on that climate what would this planet do so this can tell you FM form what the color of the

35. 00:28:04

    ground would be whether it's sand or snow whether you might get a tree placed there or walk it informs us everything about the entire surface and by using type of this top-down and bottom-up approach at the same time we type of overcome all of the limitations that we have previously here's a 2d chart of temperature versus humidity Michelle mentioned we use these two these two properties because they seem the most relevant for most variations on a planet's surface but we're quite loose with the term they artists have a bit of flexibility what they put into them and obviously some planets or moons especially maybe humidity isn't relevant for them we can we could vertically just use any other single measuring one for climate there but the point is we've got two axes two things we can control so what we do is then for each individual

36. 00:28:52

    point on this 2d graph we get two we get to pick exactly what we want to appear so down here we might have like you know obviously snow textures where we have different types of trees appearing in the temperate rainforest we get to control the exact appearance at every single location within this chart and we have a hundred and twenty eight graduations of temperature and humidity which leaves us with 16,000 very variations we have to fill of today what would appear on the planet at that specific condition which is quite a lot filled with so here's a quick demo showing a how a small terrain pass just by adjusting the climate data we can very quickly adjust the visuals and you can see that we always get like

37. 00:29:40

    logical shapes and colors coming in so yeah the problem with this now generates as a new problem we've got 16000 sets of conditions we need to set up somehow so for the artists that were presented quite a challenge our first approach was for them to have like they would manually create these rules where I want this tree to appear at this temperature range and at this altitude and it was it was too unwieldy so instead we moved to a system where we literally paint the surface of the planet using like a paintbrush so this is a quick demo of this being used here and because we don't wanna paint the entire surface of the planet when you paint on the surface you're actually painting what you would like to appear at them climate conditions so if you paint somewhere there's a 20 degree C and the humidity is 50% you're actually painting every single point on

38. 00:30:29

    the planet that has exactly those climate conditions so by doing it this way we can very quickly build up very interesting biomes and while it looks like you're just painting the small layer of land in front of you your painting huge areas of the planet at once and this was something that was immediately appealing to the artists because this is a very familiar workflow for anonymous worked on a game engine before like a smaller scale one but yeah it scales up for us so here we've just painted a few trees and bushes an instant you can see the whole areas now being I think everybody did a little happy dance when that dent moment came in the paint okay so when we when they paint all this data it goes into what we call a lookup table

39. 00:31:17

    which is just that 2d chart and we generate a whole bunch of different ones so here's a couple of them or three of them so we have the ground color the type of surface or wood maybe so snow work that informs the physics engine of what to do there and what textures we should place but then we also have things like tree coverage - how much how much trees would spawn there so that when we at space level we can still draw the forests and - all the vegetation where typically most games would have to call that stuff out they just couldn't afford to deal with it we can understand what the you know say if we're looking at the Amazon rainforest we can understand what the color should be because we don't really want the ground color from space like there might be brown in the rain forest but we want the green of the lush trees above it so we can generate all this data and we can use any altitude and it gives us a lot of power and also this rich information we can use for various other gameplay effects which you'll see a bit more

40. 00:32:05

    later today we saw a sneak peak of on a first demo of how temperature can drive things but there's a lot to come with the these planet climate conditions so here's here's a visualization of the climate on cell in the red and green are just visualizations of the temperature and humidity and one of the look-up tables are shown there which shows everything us the ground color so once we apply that to the surface it starts to look a lot more reasonable and then same for Microtech we've got all the climate conditions there it's mostly much snow and Microtech no surprise but yeah and then to build it up from the surface to see the other side of it so this is purely just the terrain plus the global lookup table so it looks pretty boring at the moment once we apply some temperature variations there's a little

41. 00:32:54

    bit more interest the humidity variations gives us quite a lot and then on top of that we have a medium scale type of terrain textures which are driven from the climate and the slope and then finally we add the the detailed textures now our climate data is only stored at four meter resolution so that doesn't give us individual stones and walks we have to add a layer detailed ground texture but we need them textures to look consistent with what the climate data tells us so if the climate table tells us we should be having yellow sand here then we may need to make sure we have yellow sand now we can't make a texture for every single scenario so the solution for that is we normalize all of our ground textures so that they will have an average of mid gray color which

42. 00:33:42

    you can see in top left and then we have an average amount of bumpiness which is the the middle one and right hand side shows the roughness so they're all type of normalized so you have this equal amount of stuff whatever it is they're storing in that texture and then when we come to apply it to the surface we type of use a physically based algorithm to rescale them to achieve what the lookup table told is it wanted or what the artist wanted but we preserve all the details there in the original texture so in the color map that might be slight hue variations and in the roughness map there might be like you know pebbles and stones and sand of different densities and roughness so this allows us to keep all of that detail but make sure it's consistent and that consistency is really important for us we actually apply the same concept that's Michelle mentioned at the four kilometer scale so here's a very basic example of a bunch of different terrain tiles next to each

43. 00:34:33

    other and after there's a very hard join between them but because we've made the climate data normalized and in the same range when we try to combine them together we get a nice seamless transition and then we apply the climate data will sit on top of that then it feels completely natural and even in this very primitive example it's very hard to see the transitions and we get a very natural and logical progression from the terrain which is really nice there is one slight flaw of the technique which is a saying this little highlight box I've got in the bottom right if I want to have a mountain range that's going to cover multiple biomes multiple climate conditions if i zoom out to space I don't want to just get the average color of that region I want to make sure to see all details that would have been in there so in this I've got a combination of shrub land and snow

44. 00:35:21

    and polar desert so depending on how varied that terrain was I should be seeing a mixture of them colors but just using the average color which is what you tend to get and the engine if you without any extra tech you just get the color that polar region in the middle polar desert so to solve this we came up with cystic Allah model which we call variants Maps which is shown on the right hand side here which is where we store the variation for inside a texture so the left hand side it shows us the humidity of a particular patch of terrain and the right hand side is showing us the brighter areas is where the the climate is varying highly in that area and then as we increase in altitude the hardware the GPU hardware naturally uses lower resolution textures and to cover the small scale so then the variation map then increases in

45. 00:36:14

    brightness to cover the amount of variation that is within one pixel now so as we go now it's a 32 by 32 texture you can see that it's much brighter now and it's still the areas that are white now are the areas that had a lot of variation previously and when it comes down to actually applying the climate will set rather than just applying the rule for the particular location of the particular climate data we have now we actually use these variation maps to sum up an area of the climate data which we use the GPUs and a topic filtering for that and that allows us to sample all of the details that would have been within one pixel and make sure it's faithful to the original image the when you get down to the surface and you may have like you know 50 percent snow 50 percent grass when you zoom out you will get the exact blend you would expect on approximation

46. 00:37:01

    of the blend of snow and grass so yeah that that really helps us make sure we hit that consistent view from space and my softness and it always the need to have a blend to some other texture when you're far away so that's that's the real key thing we get from v4 in my eyes so I'm gonna hand that to Marko now to talk about how we put all this together and how we turn this into a planet okay cool Thank You Ari okay cool let's have a quick look at the planet generation so we are starting from a cube it's cube face is projected onto a sphere the inter planar surface is generated and render at different levels of details and the amount of details increases as you get closer to

47. 00:37:50

    the surface here you can see a debug screenshot showing the surface level humidity as we discussed before in before we are blending or nearby ecosystems for altering vertices using a bicubic interpolation approximation here is the same view I was showing temperature climate data so here is another improvement as we mentioned that we made in v4 is that object scattering is driven by the same climate data that we are using for terrain generation here you can see the object presets IDs driving object spawning on-demand this is the same view while showing the colors driven by climate data and

48. 00:38:38

    textures and here you can see the y frame mesh geometry generated from elevation data the terrain geometry blending is done on CPU this is the same view as before we're showing the planetarium without any objects yet and this is the final view with objects generated on terrain so we have a separation between large-scale and ecosystems which are the larger rocks trees and so on and then we have the so-called ground cover objects which are the smaller objects that are generated at ground level only additionally we have improved the cliffside generation and yeah the objects are placed based on climate data

49. 00:39:27

    and object preset settings and we are using LOD clusters for large-scale forest rendering which you will see Microtech a ground level we have additional parallax and ground texture details so physics collision geometry is generated on-demand on client and servers when players and spaceships are interacting with the physics screen here in light grey you can see the physics proxies are generated for terrain rocks and objects so generating on-demand physics on-demand is tricky because the spaceships are actually moving at crazy speeds across the surface and obviously we cannot store the entire planet

50. 00:40:14

    geometry in memories otherwise would be terabytes of data so each terrain patch is built in parallel on CPU the workload is distributed by the job system to all available course and basically the server is building a physicalism mesh for each client so we have many other features on the planets for instance Caves here isn't here you can see an example of a cave assembled by the procedure layout to after generation the caves are turned into object containers and they're placed on the planet surface so using object containers we can take advantage of object container streaming in future and here you can see an example of a cave with an exclusion

51. 00:41:04

    volumes which is used to avoid generating rocks on top of the cave entrance here is another example of exclusion volumes you can see how the trees are avoiding generating inside the volumes additionally we have space stations orbiting around planets all the space station interiors are generated by the procedure layout tool as well and then are placed as object containers in space yeah another new feature we added in v4 is the frozen ocean which can be seen here on Microtech the frozen ocean is physicalized and players can walk drive vehicles on it in these pictures we're showing the frozen ocean collision

52. 00:41:52

    mesh generated on demand so regarding mining the way it works is with some of the procedural objects are tuned into my novel entities and the player can interact with them when extracting minerals and with persistence becoming online players are able to deplete planet resources affecting the universe economy so let's show some of the rework which is coming with three point eight I'm just gonna let this run a little bit obviously we wanted to at least close close this talk with showing you guys the current state of all the rework the team is like I said the team has been working hard and is still

53. 00:42:41

    working at it we want to make sure that we deliver the best possible planet experience for a three point eight coming up but we're already seeing a lot of the features that came in would be for making it everything every location and we've known like it were built up so up until now looking way better and if there's there's a lot of subtle differences that might not be apparent in at first glance but we just wanted to reassure you guys all the planets are updated and they will be coming so I guess we could already start thanking everybody who's been involved with this because it's been to summarize it's been like a full year working on this with some very high-level people and a lot of people involved with this so this was not a an easy easy undertaking it was a

54. 00:43:32

    team of four so thanks for all the people on the organics team thanks to Marco Ali Sebastian Strider other people on the graphics and engineer and we hope you have a great time with the new planets yeah I hope you guys have enjoyed this presentation thank you for listening and enjoy the rest of the show thank you thank you