2023 | vis.gl, the powerful framework suite behind deck.gl and kepler.gl - Alberto Asuero Arroyo

I work as member of the technical student committee at deck GL and also I'm the CTO at carto and we'd like to talk to you today about the BGR framework that is that we join it to the to the open GS Foundation last year and it's the framework that Empower and the framework that have all the libraries that make corporate year and next year works how many of you you are familiar with with Kepler any of you have used its capital right now good many people and with deck good almost the same people are using both of them so this year is a suitable composition framework that have inside many different libraries we are going to try to explain to you what they are more popular and the one that they are not more so popular but they are still open source and you can still use them in your own project the first thing is capital GL probably the ones that are not familiar this is mostly uh UI for for the mapping library of the GL and it allows you to to visualize and create a styling using drag and drop and using analytics it works mostly on the front-end side so if you have a geodation and your blog to to the to this website all the information inside of your browser and you can visualize you can change the Styles and and create beautiful masks to to analyze that it have around 30 000 weekly users in the Cabaret year website and one of the good things of Capper the airless that is highly customizable you don't need to make even a a complete for to to being able to customize this this is for example a sample that is come from from Foursquare Studio or for the studio that is a customization of capital gear so what you need to do when building corporate that is that you import the package of coupler and then you can inject some of the component and replace them to create your your own behaviors also the same for for car to Builder is also a version that we have built at carto and and it's the same we use the the package of cabbage Earl and we modify that to to adapt to our custom grinding to sustain adding more features and things like that also you can use Kepler if you are working with Jupiter notebook or with Tableau or inside of Visual Studio that is probably not so popular but this is also available so if you for example are doing a data analysis inside of jupyter nodule and you want to create a map that is a good solution for you and now let's talk about xgl the dexterra project has grown a lot during the last couple of years two years ago we have around 50 uh fifty thousand dollars weeks and right now during this year we have arrived to to pick off to 220 000 downloads and why that is important that why I think that deck GL is so popular I think that one of the points is that is have an open governance and and really open to collaborations and why is that I think that is super important for for an open source project that's companies that's large Enterprise companies like Google or Azure Capital can start using that internally Carter is all the visualization part of character is based on this um Foursquare uh is also using this and that is something that's really pushed forward this this project the other theme is that it is a framework and we have so many different packages and so many different models so the the entering the project and the learning curve is not very high and and you can start in the project with with so many resources in general and and of course that is have proper support for from the community the other thing is that is basement pattern lock this what that means is that you can use the GL uh with your own library that you are using for your basement for example you can use ductile on top of Google Maps or you can use deck gr on top of map liver or on top of map first year version one or version two the same with Amazon location or with Azure mask because they are based on a much more GL version one but one of the most important thing why the glacine is so popular is because it's provide a huge variety of different geodata visualization available out of the box apart from that Integrations like this that you can integrate your own backend that you can integrate and you can visualize your layers on top of style 3DS from Google on top of Google Maps and so on and also they we have a couple of extensions that we are going to talk a little bit more later this is a very basic example of the GL that how you can visualize uh aggregation using using XL another thing that is also buyable is that you can use all the layers that are available in digital inside of a python node we'll use in pyrec this is different of using kepler-gr and baby in a notebook it gives you more programming flexibility to control the different layers by code but it is available for that and I would like to talk to you right now on a story that is is about open source that I really like and I think that could be interesting for other companies that can be in a similar situation or either now or in the future when in in 2019 carto releases a rendering engine that does was called Carterville and at that point I I started at the CTO at Carter and one of the things that I was wondering is say why we don't try something different why we don't try to join forces with deck GL why don't we try to to push all of us in the same direction and that is what we try to do at the very beginning in 2020 what we do is to create an integration in Cabra for cartoon that is like you can have coverage here and you can store that information there and then we started implementing the MBT layer we collaborate with xgr project to to be able to use an imitate laser and why because bacterial was found at that time to work with a client-side data so it's only work with 10 megabyte of data and if you need to grow further you need a server that you need to do this and that it was the reason and after that we create and we go forward and we create the character module and then more and more integration the encounter Builder that was based on Kerber and and all the different things and that is what I think that is really interesting that try to put the E for the company in the things that can be open source inside of an open source project and get the benefits in both directions this is not so popular but it is super powerful it's we have there in Lotus GL uh parsers for for a lot of different uh geospatial data and also for three data and so on and it doesn't require to use the xgl imagine that for example that you have a shape file in the brochure and you want to process that J file if you use a Lotus GL you can just decompress that loaded here also will use white workers to avoid execution attack inside from the main thread and you don't need anything else from from the actuals only with this you can be able to use this and and we have many other data formats like the rain loader GPA so if you have time I recommend that you take a look on this is like the the rendering engine that we have uh inside of of deck GL and it's the the rendering Engine That Power deck GL another tool like in this case ABS audio that we will see a little bit after that and it provides a fraction for for data visualization apis uh using webgl right now and that is the kind of things that the project that has been built only based on Luma is this AVS project is using mostly low material to to visualize that is a software that you can put in in an autonomous vehicles car in order to to be able to to visualize the different cameras and the and the data that is around you more about this that is something that I consider that is is important regarding base maps and xdl one of the things that we we made a decision is that the gr not compete in in the base Master space because they are different solutions the architecture of the xdl has not been thought to create base Maps is is more that is the part of uh map Libre or Maple GL or Google Maps the GL has been thought to create data visualization Advanced Data visualization that cannot be addressed for those Frameworks and delivery as we said is is trying to also create different Chinese with base Mass provider and what does mean what does mean is that if we partner with with Google Maps for instance if we partner with with map library and so on we can take advantage of that let's go to the sample of Google Maps with Google Maps we we have a samples inside of Google Maps that say how you can create a band visualization if you are a Google Maps user and we have that documentation inside inside of Google Maps public talk so it bring a lot of users to the community of dexil they're saying we expect that should happen with with Azure Maps or with uh with Amazon locations that they are using mattress GL and we expect to be able to arrive to that point with math delivery of course we have a very good relationship with them and we we expect to collaborate more so the xgl is not going to enter in in the future that a map Liberty resolved and we will mostly a compliment for them now let's talk a little bit about the the scenes that we have built during the during the last year in the project that we have a couple of them the first one is is a mass extension that basically it's allowed you to draw a single polygon and and this polygon will filter the data that you have uh rendered in the application so it allows you to create different experience like this that moving your mouse around you you can do all this filtering using the the GPU that is super fast another thing is the Collision filter extension this was mostly implemented for for leveling purpose but this can be used for for other things so basically is try to detect collision with a inside of the GPU and you can create very basic leveling system we don't I'm here we don't try to to to do a very complex leveling system this is work that should be done in in a base Maps but like basic things that if you have a road Network that you want to visualize on top of your data and create things like this you can you can use this this filter in extension this one is something pretty interesting I think that I'm really proud of this because we implemented this first thing that is the drain extension and what does the terrain extension is to be able to visualize your 2D data on top of a 3D server phase so basically if you have a geodeso and then you have a two-way interface you can put that geodation adapted to the to the terrain and the set coordinate is adapted to that to Showcase this feature one of the things that we did last last one is was to to do an announcement with with Google team because Google releases the photo photo realistic 3D tiles all over the world that is available right now in in many countries and I think that it's there in 49 countries and 2500 cities around the world stock what yeah this is order a very interesting theme because I think that's one of the main bloggers for the people that's going to use 3D in their use cases is that they don't actually have 3D data um in this case what we have is just a layer of buildings I don't remember if this come from an open Street Mass but what we did is just put a ledger of buildings on top of the Google 3D range and and can be structuralized so the idea is that you you will take that geotif you can create the club optimized version you upload to be query and then you can visualize directly from bigquery and you can do also additional operation to analyze the Thruster there of course as many open source projects we need to work on on typescript and and tooling so the code base of capital and next year during the last year was adapted to type script because the process was growing so much so many people involved and we needed to to invest here to improve it regarding development also we have more for from webpack to buy do you due to perform and reason some of the building screws has also using ES build and the documentation which changes to use the gustavros from I don't remember what we used it before Gatsby and we moved to the Gus Service because it's faster in general and what is Communists what is coming next GL version nine the the first thing is that we we want to to support web GPU web GPU has been released in in Google Chrome during during the last uh during this year uh and we we would like to to start this implementation inside of luma GL so what expectation is to have this ported on new material and then we will support that in index El and Kepler and the rest of the project we also will create a new category extension if you have worked with AGL this is a feature required from our user that if you want to filter in the GPU foreign attribute you need to compare to ins and so on and if not very easy to do and right now we have a approved request for for this and that will be something that we would like to include another thing is widgets to map one of the is right now and the Apple request too to create the basic things of widgets like creating navigation for widgets full screen ingredients and so on that is also a very useful index here but one of the things that we have also under discussion is to include data widgets so imagine that you have a geojson and you want to create different data widgets like the revenue by store type like Instagram widgets like formal Aggregates using the data that you have either of all your geodation or with the data that's come from the tile and calculated basis on the viewport the other theme is just to move our experimental terrain station to bring stability also to support highlight that is not supported today to be able to select features and so on another thing is to create a react grew upper for for Google Maps uh using reactive that's all we have also more things that we we will be sharing after that but they cannot be sharing right now because they are under NDA but we will have more things coming for the next year and one important thing before jumping into the question we have an event at a collaboration Summit in in September so if you have any use case that you want to present there if you are an user of next year or whatever for free to to pay me later and we will we will organize that thank you

Channel: FOSS4G

Views: 1,134

Keywords:

Id: lXl7bWPz-M4

Length: 17min 24sec (1044 seconds)

Published: Thu Aug 31 2023