WMS-V: Animate Raster Maps

Video in TIB AV-Portal: WMS-V: Animate Raster Maps

Formal Metadata

WMS-V: Animate Raster Maps
Alternative Title
WMS-V - Video Map Tiles for the 3rd or 4th Dimension
Title of Series
CC Attribution 3.0 Unported:
You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor.
Release Date

Content Metadata

Subject Area
WMS-V - Video Map Tiles for the 3rd or 4th Dimension
Satellite Standard deviation Group action Service (economics) Open source Computer file Multiplication sign Source code Temporal logic Streaming media Dimensional analysis Machine vision Web 2.0 Data model Mathematics Different (Kate Ryan album) Term (mathematics) Natural number Videoconferencing Software framework Endliche Modelltheorie Implementation Extension (kinesiology) Data compression Proof theory Domain name Addition Standard deviation Touchscreen Mapping Interactive television Propositional formula Water vapor Dimensional analysis Control flow Cartesian coordinate system Tessellation Mathematics Category of being Proof theory Digital photography Visualization (computer graphics) Virtual reality Spacetime
Complex (psychology) Dynamical system Range (statistics) File format Bit rate Open set Data model Different (Kate Ryan album) Computer configuration Videoconferencing Data compression Satellite Touchscreen Mapping Software developer Structural load Bit Motion capture Web application Process (computing) Angle Raster graphics Internet service provider Buffer solution Artistic rendering Simulation Arithmetic progression Computer file Computer-generated imagery Streaming media Event horizon Element (mathematics) Goodness of fit Term (mathematics) Computer hardware Standard deviation Focus (optics) Demo (music) Content (media) Frame problem Population density Visualization (computer graphics) Library (computing) Satellite Standard deviation Building Code Multiplication sign Raster graphics Web 2.0 Positional notation Synchronization Semiconductor memory Visualization (computer graphics) Cuboid Electronic visual display Information Endliche Modelltheorie Data conversion Extension (kinesiology) Pixel Position operator Stability theory Area Presentation of a group Boss Corporation Temporal logic Open set Element (mathematics) Tessellation Vector space Navigation Filter <Stochastik> Surface Frame problem Digital filter Server (computing) Statistics Web browser 2 (number) Revision control Cross-site scripting Natural number Operator (mathematics) Green's function Program slicing MiniDisc Dialect Slide rule Projective plane Dimensional analysis Subject indexing Cache (computing) Event horizon Doubling the cube Synchronization Videoconferencing Extension (kinesiology)
Touchscreen Escape character Musical ensemble Address space 2 (number)
Server (computing) Presentation of a group Code Multiplication sign Tournament (medieval) Computer-generated imagery Streaming media Client (computing) Open set Web browser Revision control Mathematics Different (Kate Ryan album) Hybrid computer Videoconferencing Aerodynamics Endliche Modelltheorie Codierung <Programmierung> Extension (kinesiology) Error message Proxy server Form (programming) Mapping Server (computing) Moment (mathematics) Planning Client (computing) Bit Tessellation Mathematics Cache (computing) Error message Blog Cube Hybrid computer Statement (computer science) Website Right angle Ideal (ethics) Quicksort Videoconferencing
hi I'm Tim I work from an Aquafina Alenka Research in New Zealand let me take you on a journey with geographical videos and then longitudinal compression and a story about my experience with this exciting new technology what is the problem to solve spatial data does not only present a consonant space it includes the thing that is fundamental to a very nature change how how can we change to summarize I want to briefly explain the motivation for W miss V then I'd like to show you a food cue proof of concept applications the first of which was the visualization of our National Post model furthermore I attempted visualizing temperature change in the Antarctic in 2015 the challenges of that data led me to visually nicer dataset the Himawari satellite takes the photo every 10 minutes which I downloaded sample of and concatenated two videos there are more challenges that I'll get into meanwhile I'll try to summarize the story of WMS me and briefly cover some concepts and considerations and finally propose one or even more standards first I want to ask why visualization of temporal data over time such as weather patterns forecasts push fires storms or different properties such as soil profiles traveling through depths in the ground water or animal bodies could really help us see what is going on in very interesting ways across different domains in general with what we perceive as time we can visualize at least one additional dimension of data whether this is in 2d or in 3d or on a screen or potentially in virtual reality for now let's start with 2d and data over time in an interactive way w miss time is a standard in the web mapping service framework it is a standard which enables open source maps to consider time a simple map tiles here you can see an example using WMS time and PNG images as you may be able to see we are making a hundred and ten requests here for one layer and one extent I don't think I need to explain why this is a problem there is no transition between friends and no server-side filtering between images possible not only latency but also decoding compressed images is costly so why not have a stream of images coming in as a video and being decoded as a stream video technology has been researched and improved upon over decades now and we have a very we have very efficient video formats both in terms of speed and file size my vision lies rest the data a lot of data sources raster aerial satellite or your VA imagery to name just a few
models can be less costly to process using raster's since some operations are costly to calculate with big data and those vectors let me show you the national postal model i won't go into too much detail of the model itself the focus here is on the video map technology as you can see there is a navigable mac map showing the videos versus map tiles the extent and positions of video tiles are shown accurately on screen regional councils areas and their statistics are shown there's a slider to advance time in a live preview how did we do this initially with the passo model we replaced all the image elements with html5 video elements and openly as to at the time Dom running was used which was the only approach possible then Dom rendering is expensive and was later replaced by canvas and WebGL rendering both of which are hardware accelerated and HTML canvas element enables developers to do filtering and an advanced image manipulations which are not while not having to worry about all the graphics challenges my namesake Tim Shaw put together a proof-of-concept which I learned was inspired by my box post previously this was after we did the national postal model here's the example we put together you can see how this kind of visual could not be achieved with downloading separately compressed images as frames or was tomlin rang with this kind of performance and at an angle like that I'll let you appreciate this for a few seconds I tried other data the remotely-sensed and Arctic surface temperature from 2015 however the data was not really suitable so W dovey MSP is not great for visualizing data that has little commonality also some caching some caching should be done video tiles and compose compositing them on the server may be the way to go cropping composing and pairing works ffmpeg and complex filters make this possible at fast speeds challenges transparent videos are not fully supported at the time I checked with the Antarctic data I wrote a client-side filter that uses green the green channel as transparency which of course only works if you are not using a channel in your data visualization layers on top of video layer are problematic without demanding synchronizing Visio tiles is really a great challenge the browser api's are not reliable enough for this purpose I tried a lot of things having the videos pre computed and sitting on the server as simple files was best for now downloading partial content then gets handled by the web server I spent a lot of time optimizing the synchronization in the end double buffering and frame caching seems to work best and surprisingly also seems okay in terms of memory use performance is a pillar stability are challenging however having tabs open for a while can crash the browser at first for the 2013 my demo failed because I had opened it in the wrong browser I've been implementing a dynamic WMS be that work with any bounding box so that videos can be encoded on-the-fly this solved this and could solves the synchronization problem but increases the load on the server and server-side filtering such as blending can slow things down and not to forget in which loading ffmpeg supports GPU rendering but it makes the video quality worse there there may be faster video encoding options libraries or similar that can solve these problems encoding one video and streaming it was not the bottleneck though it is the image tile delivery from a WMS time so if you if one could look the one could use low-level code and access the rest of data with indexes directly I believe their on-the-fly delivery could be the way to go so now more to the story in 2012 Leon Kerr was involved with the passo model I remember taking talking colleague about this exciting new html5 technology video and canvas describing how these elements behave like images in many ways in the conversation the idea of using video tiles came up robert coined the phrase longitudinal compression a few hours of progressive working later I had crafted some be some map tiles some video mapped outs from the post model raster data and combine some on a small grid in my browser playing the videos at the same time the tech got some interest however mostly used my free time when I presented in I'm Tommy I'm SB and 2013 I was talking to the openlayers guys and Tim sharp came up with the post compose idea and building it as an extension there was motivation to present again here today now two standards I have a long and skeptical journey with standards it began when my boss gave me the immense job of reading the WMS spec and later punished me was the even more convoluted wfs to standard at the time as you can imagine I did not appreciate standard much however having spent time with Dobby fs3 at a hackathon and getting annoyed yet again with different power plugs in different countries I realized that standards are important so I spent a bit of WMS time pun intended I hacked it behind my WM SV and learned to appreciate the ISO time notation
standards since they are even implemented in JavaScript I had a look at the time range speck in WS time and decided to implement that in my dynamic WMS V and here's the time spec you can find the details and how to interpret them online more that could be standardized frame rate video formats and coding options filter spec all this as extensions or score so more on streams if you're not asleep yet if you're not asleep yet this might have helped more on the story finding finding good example data is more difficult than I thought I was not happy with what I had to show so I found the Himawari satellite imagery where which you can see in a minute as an aside you could imagine that you could make a zoomable parable NK video possible with this tech satellite video feeds could provide live coverage of world events such as natural disasters weather forecasts are shown in everyday TV all this and more could be visualized with remotely sensed or model raster data and WMS feed so here the promised Himawari footage as a web app and using video tiles getting them to synchronizing like this was a major challenge and here at 1080p slice of it as you can see the weather is nicely visible here just a couple of days ago I got a dynamic version of the hemorrhoid videos going where I simply crop a high-resolution video on the fly and basically projects a portion of the video for display and could stream at five times the speed of the video this possibly works so here you see that it's
dynamically generated because the seconds are advancing [Music] there's no escape anyway you can try it out it's up I don't have the address on screen today
okay so coming to the conclusions video streaming is fast and image fetching is slow terral's are good for servers and dynamic is better for clients finding data without errors is still hard hybrid caching seems the way to go composing cropping existing videos on the server works well what was the problem statement from the beginning change it's what we want to see change is what WSB can show so consider WMS V as something on your radar thank you hi I'm is this like if you already implemented that and is that accessible at the moment through open layers or something like it's not accessible at the moment but I'm planning to make it accessible at least on my account yeah yeah and sort of what what are you targeting a version of I panelized is I'm assuming is soon cooperating up analyze is that right or was it gonna be separate well it's gonna be an extension already created an extension but it's I wouldn't say I'm not proud of that code yeah we have to clean it up a bit to publish yeah hi thanks a great presentation I'm just interested to know and I might be showing my ignorance here a little bit but how are you generating the the video tiles on the server yeah I'd say so I did many different ways so first I I built a pyramid from the from just the JPEG images or the imagery the image form so I built first I built a tournament with map server and map cache and then I simply had different times as different layers and then I just concatenated those two videos and just basically hacked map server to to serve up videos instead of images just with a proxy in front of it and then with the dynamic WMS be I did implement like an ffmpeg filter that's on so that there's a filter spec on on the ffmpeg website you can look it up it's it's a bit complicated but exactly okay yeah so and then there was the dynamic and then then I did this sort of hybrid approach with caching and also dynamic generation by first I tried composing the video stars but that had a problem so but it was just a superficial problem I think like you know just me misinterpreting the documentation but I think but the speed was actually good to compose them the different tiles until one video go you could actually stream one video to the client quite nicely better than trying to synchronize on the client and the other questions yeah I'm sorry so the video that you showed before was that like in the blog you could drag that that globe or whatever it is around what what was that video of or was it like a 3d model that you overlaid those videos images and I created a 2500 by 2500 video and then I just simply drop it on the server and the fly and that actually works quite quickly but then you deliver less data and it it still comes as a stream and you know the browser then handles all the decoding and so for example in that video could you like drag it around for example like the cube like we had been usable I tried doing that in the last couple of days I had the tiling idea first because I thought it would be too costly to generate on the fly but actually the on the fly works quite well with the hybrid


  128 ms - page object


AV-Portal 3.21.3 (19e43a18c8aa08bcbdf3e35b975c18acb737c630)