Avatars and motion capture for virtual fit fashion

Video thumbnail (Frame 0) Video thumbnail (Frame 1181) Video thumbnail (Frame 3021) Video thumbnail (Frame 4312) Video thumbnail (Frame 5993) Video thumbnail (Frame 7224) Video thumbnail (Frame 9004) Video thumbnail (Frame 10330) Video thumbnail (Frame 10997) Video thumbnail (Frame 11990) Video thumbnail (Frame 12688) Video thumbnail (Frame 15425) Video thumbnail (Frame 16131) Video thumbnail (Frame 20253) Video thumbnail (Frame 24048) Video thumbnail (Frame 25420) Video thumbnail (Frame 26505) Video thumbnail (Frame 27539) Video thumbnail (Frame 28701) Video thumbnail (Frame 30551) Video thumbnail (Frame 31303) Video thumbnail (Frame 32602) Video thumbnail (Frame 35245) Video thumbnail (Frame 36066) Video thumbnail (Frame 36607) Video thumbnail (Frame 38135)
Video in TIB AV-Portal: Avatars and motion capture for virtual fit fashion

Formal Metadata

Avatars and motion capture for virtual fit fashion
Title of Series
Part Number
Number of Parts
CC Attribution - ShareAlike 3.0 Germany:
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 and the work or content is shared also in adapted form only under the conditions of this license.
Release Date

Content Metadata

Subject Area
This will be a talk on the use of body scanning and motion capture technologies for application in virtual fit fashion.
Weight Fitness function Virtualization Bit Motion capture Cartesian coordinate system Finite element method Goodness of fit Virtual reality Hypermedia Computer animation Meeting/Interview Curve fitting
Point (geometry) Standard deviation Port scanner Electronic mailing list Mathematical analysis Shape (magazine) Function (mathematics) Dimensional analysis Number 2 (number) Product (business) Measurement Avatar (2009 film) Different (Kate Ryan album) Representation (politics) Belegleser Point cloud Metropolitan area network Standard deviation Multiplication Software developer Point (geometry) Electronic mailing list Mathematical analysis Port scanner Cartesian coordinate system Measurement Shape (magazine) Product (business) Process (computing) Computer animation Software Point cloud Right angle Figurate number
Point (geometry) Polygon Surface Port scanner Usability Revision control Avatar (2009 film) Mechanism design Term (mathematics) Different (Kate Ryan album) Object (grammar) Representation (politics) Endliche Modelltheorie Lie group Point cloud Noise (electronics) Polygon mesh Texture mapping Surface Point (geometry) Motion capture Line (geometry) Port scanner Cartesian coordinate system Computer animation Network topology Point cloud output Endliche Modelltheorie Right angle
Context awareness Texture mapping Texture mapping Digitizing Software developer Scientific modelling Fitness function Virtualization Medical imaging Category of being Digital photography Process (computing) Computer animation Meeting/Interview Term (mathematics) Different (Kate Ryan album) Order (biology) Representation (politics) Endliche Modelltheorie Address space Social class Spacetime
Overlay-Netz Simulation Texture mapping Virtualization Function (mathematics) Category of being Avatar (2009 film) Computer animation Integrated development environment Different (Kate Ryan album) Term (mathematics) Function (mathematics) Videoconferencing Representation (politics) Endliche Modelltheorie Simulation Condition number
Computer file Motion capture Real-time operating system Open set Shape (magazine) Function (mathematics) Dimensional analysis Avatar (2009 film) Medical imaging Virtual reality Insertion loss Optics Meeting/Interview Term (mathematics) Different (Kate Ryan album) Visualization (computer graphics) Condition number Physical system Social class Area Simulation Texture mapping Gamma function Real number Virtualization Motion capture Cartesian coordinate system Category of being Moment of inertia Computer animation Visualization (computer graphics) Function (mathematics) Optics Virtual reality Simulation Form (programming)
Meeting/Interview State of matter Fitness function Virtualization Limit (category theory) Disk read-and-write head
Greatest element INTEGRAL Multiplication sign Port scanner Set (mathematics) Real-time operating system Parameter (computer programming) Usability Neuroinformatik Web 2.0 Insertion loss Meeting/Interview Computer configuration Different (Kate Ryan album) Metropolitan area network Simulation Touchscreen File format Fitness function Data storage device Computer simulation Virtualization Bit Hidden Markov model Surface of revolution Measurement Type theory Befehlsprozessor Order (biology) Pattern language Right angle Geometry Spacetime Point (geometry) Implementation Existence Image resolution Drop (liquid) Graph coloring Machine vision Surface of revolution Traffic reporting Form (programming) Mobile Web Information Bound state Database Approximation Software Integrated development environment
Point (geometry) Implementation Service (economics) Divisor State of matter Characteristic polynomial Frequency Bit rate Different (Kate Ryan album) Reduction of order Keilförmige Anordnung Metropolitan area network Service (economics) Graph (mathematics) Electric generator Data storage device Fitness function Approximation Factory (trading post) Order (biology) Different (Kate Ryan album) Normal (geometry) Game theory Quicksort Curve fitting Reduction of order
Polar coordinate system Electric generator Multiplication sign Content (media) Set (mathematics) Mass Line (geometry) Medical imaging Computer animation Insertion loss Meeting/Interview Different (Kate Ryan album) Personal digital assistant Hypermedia Analogy Order (biology) Smartphone Office suite
Ocean current Metropolitan area network Implementation Simulation Latin square Image resolution Multiplication sign Fitness function Data storage device Set (mathematics) Real-time operating system Frame problem Web 2.0 Meeting/Interview Order (biology) Energy level Quicksort Virtual reality Head-mounted display Social class
Existential quantification Polygon mesh Code Length Image resolution Multiplication sign Measurement Vector potential Tensor Curvature Process (computing) Meeting/Interview Quicksort Virtual reality Spacetime
Area Metropolitan area network Meeting/Interview Linear regression Quicksort Field (computer science)
Lecture/Conference Meeting/Interview
Point (geometry) Area Plotter Motion capture Coordinate system Control flow Insertion loss Real-time operating system Lecture/Conference Point cloud Endliche Modelltheorie Object (grammar) Physical system
e yeah and now mining in my receive and
but it OK and Hello Good afternoon and thanks for the introduction of quotas for that it's like technical mishap there and I'm Andrew them as my colleague Anthony at low and we're going to talk about applications for the virtual fit fashion so 1st of all and to be a little bit of an overview of body scanning technology the
technologies that we worked with I use infrared depth sensors so similar toward finding that Microsoft Kinect and operates in the size of changing cubicle and takes a multiple scans from awful different senses and of the subject the and the whole process takes less than 10 seconds to run and and the output from that is point cloud data and which is essentially a number of uh and three-dimensional coordinate points and along with that and the software that ships with the scanners um gives a list of body dimensions body measurements and and in so far Chellappa fashion applications and this is used for the body shape analysis and clothing size standards research and product development and things related to that but with the software and also that produces the measurements it uses these measurements and to re proportional eyes a pre-made avatar and again that's used
the the accuracy of the representation I can see above using 2 different scanners also ice cream uh 1 being the T C 2 the and you can see that the right hand side representation is and a solid representation of the point cloud data and the figure to the left you'll see is that the representative avatar uh what you will see and the with the per re proportional ization uh application to the avatar was the body measurement points that are accurate and the the overall and the shape of the figure is not necessarily true representation as is the point cloud data is and from the point cloud data as is quite noisy
when we're looking at animating and another top no 1 important consideration is the topology of the 3 D model now you can see on the left hand side there they the physical representation the point cloud data you can see that noise in terms of its topology and so I talk about topology and talking about the anatomy of what makes the 3 D model handed in the different points and this is really important when we come to further animate that model and as but topology of that defines the the deformation points so a hard surface modeling and this is a very different workflow to modelling for an organic deformable mesh and such can say the reconstruction the point cloud data is is noisy which makes it quite unusable and for animation what we have there and this is the shaded you of it with without the topology lines on
um and something that's been input integrated with and 3 D modeling applications is uh re topology mechanisms and tools set so the model that you can see on the right hand side is really to apologize version of the point cloud data and the difference between that and what we've seen in the previous slide and with the the re proportional lies avatar has is again much truer representation of that point cloud data the the actual scans of the subject so being able to read to apologize and and defying where those um lines of deformation going to be makes for a much more usable mesh for animation yeah in terms of texture
and what we've seen so far is that a self grain on textured model and what were able to do the is texture uh 3 D model and the workflow with that involves breaking up and and OR fold in the the 3 D model onto a two-dimensional coordinate space which can see down here on the left end corner and applied to the 3 D model you can see that enable us to take a 2 D image uh and apply that to the 3 D model so you can see that is is a painted texture and using a digital paint packages and in order to colorize texture and apply that to the 3 D model and developments all that so we could use and photographs of the subjects and applied that image-based data a subjects onto the 3 D representation enabling a much more realistic and true representation of the subject that's being scanned the in terms of class texturing
and developing the 3 D cloth model and simulation the now can use that same process certain model uh address around the subject and all we can take and an image of the flat textile the that can be made into a 3 D model and we can apply in the context that uh we can close the 3 D model in in the class in terms of then animation that and need to be able to develop physically accurate I representations of of how that cloth will behave and so think in terms of different textile properties how the reactor to the environment and and how that cloth would appear and in terms of virtual fit fashion and we can look at a couple of
videos of the output of that it can get out so the video is either a who taking the 3 D model and all of a garment and applied and a light breeze so that so you can see how the the cloth would simulate under various conditions and different textile properties how we would react
and environment if we then apply that an overlay it to the avatar of the subject in terms of virtual fashion we can have a look at that I think OK so I have here that is is rendered output using the the textures model and representation of of the scanned subject and overlay that with a cloth simulation and this enables us to to view what backgammon would look like on
the individual under various lighting conditions and using different gamma properties and so did it so virtual fashion and experience and again is going to now talk about and 1 of them is going related to talk about so it is different technologies involved in the application of that directly and so in terms of the cloth simulation were
saying that have seen applied to a static avatar and that's not animated or were blocked relative is combined that with a motion capture technology is to be able to animate the avatar as well as the cloth simulation so there's many different technologies and for motion capture and this is the connect systems a sensor-based systems as optical camera-based systems all the different and the pros and cons and so woman area of the technologies to allow for is is real-time visualization as you can see on the image that allows us to interview in real time the output from the motion capture data and this can storing dates files safe later applications and so in terms of personalizing the habitat uh as well as getting the body dimensions and the shape and the texture in you can also regard subject motion and stalled out then and apply it to to an animated avatar so enablers to really personalize that animation and not subject when applied to a across simulation you can then see what a subject would look like in a particular garment and for the application of virtual for fashion and visualization and so tell so the and avenues for animated fashion shows enabling class region of a global audience and at the opening of possibilities for for virtual reality experience which and 7 talk about in and very much so the light on the
speaker mens ons and and so where does wonders talk metal all fit with regards to virtual fit technology and about everything so go through the state of play as the is currently the birth of the technology talk where the successors to augment its current limitations and what problems that
poses and as well as the possibilities and head to we sort of see this will evolve over the coming years that X 0 5 to 10 years or so for the future of
what could potentially be and real so virtual fit experiences so so looking at the virtual fit as it exists currently and this a lot different options and that it is still a small sample of of what exists currently on the web and the 1st thing you'll notice is none of them necessarily to take information and from body scan data retreat apologized with proper cloth developed place on top like under just talking that which would be the ideal solution in in in in an ideal world we get to do that but none of these really did and this great steps from what was possible a few years ago and all these web and integrations and app-based integrations mobile integrations are great and what would we like to we have to do is to provide more of the more the personalized more depth more technical experience both for the ease of use of the user and city come a long way with regards to the original idea behind visual the fashion and and the you decide it isn't new all uh again back 15 years the exact same idea existence and you're incredibly limited by technology technology moves along in leaps and bounds we're talking various CPU power and GPU power in particular and so we can do wonderful that of fashion given you a week's notice and the body scan and all this information and we can make something that looks essentially real given of computing power and what you would do that all we need is the other you in real time and the someone's to interact with those players in real time and see how they would actually fits and both also we we're still limited by technology so so that the idea's been around for a long time so this is a very old implementation and from around 2000 2001 and a bit of software called see me um resistance this MIT was still trying to achieve now this idea of viewpoints approximate measurements and driving drops clothes over the top and you get you get together and what it was was a very good at the time quite good but now very poorly rendered sets of J takes you simply click through that were automatically generated and you your away quite a for implementation but it was sunny revolution full-time looks horrendous birth and here you we are going back 15 years the point is willing to buy technology that is still limited and while all the implementations are good there exist currently and they exist in the format
they do because there's not enough computing power still to to do what we like to do so if you look at these again parameters of highlighting the technologies they use the primarily a lot of these we where G technologies and only 1 that really takes a so that the very basic 3 diabetology this this low on the bottom here and with a a couple of very very good ideas and very very clever bits of software in there as well so you look at what the bottom right for example and she's a Kinect sensor which is the uh at a depth camera and so far bottom right we have uh and is of showpiece type technologies so pressing goes into the store and the standard forms of the TV screen we connect pick them up it creates a very coarse rudimentary 3 D geometry they drag and drop using gestures which close they want to try on and move around and mobile they is a reasonably or look of the clothes they're wearing appear on the screen the idea is very good bond for to become simulate movement of cloth fast enough of point of resolution in order for that to work and the computer has to be able to read in the database the person's motion which you can't do offline of right in order to better interpret the information pass the course simulation and then displayed in real time on the screen very very difficult to do and another very clever bit of work has been done this with this uh mirror which is very similar technology to connect implementation and which focuses more on been up to also in real time the patterns on the colors in their environment and there it might be a with the way things are going with every single normal things in your hands becoming smarter inverted commas it might be that the eventual solution is a smart mirror that in 15 years time I do mean 15 years at least you do this when the attenuating the eventual solution may be a smart mirror whereby information is fed through to you and from retailers and you'd do the whole gesture thing and you see on the source of minority report sci-fi type of futuristic vision for things but that that is a long way away that maybe the end a was and what we're interested in is what technology we can use the narrow in order to be a potentially shape the space and serving you 10 uh the current implementations of the day they have to
drive to improve services and reduce rates and essentially budget but putting approximations people get a reasonable idea and provided way factor for the store the brand in question things like that and that that's there and especially the install implementations it takes for the wide factories for this very expensive these kids and isn't it wonderful and that sort of thing and on behalf wedge between online and in-store shopping and what we could take technology from different industry has been developed over a long period time has an awful lot characteristics and we use the already existing nodes enhance the experience the graph for this sort of work and and that's clear is is the gaming industry take this is the perfect place to be talking about this and I hope everyone is of the order to this point but what if they all were to take off now on the you will
I actually don't think it will this generation and is wonderful talk earlier than if an on-site historic and I missed the last name of for some reason to the about and i man she was talking about uh the our implementations and for the year 360 using inside stores you can become and always so the not virtual fits which you also state and use of it has virtual thing and I don't blame purely because what just told me it's not there yet what if hi and they are ended up becoming the norm so as an awful lot talks going on elsewhere here talking about concentration that a lot yesterday to admit the are concentration and how we generate the
content and how we make the content accessible in order for it to appeal to the mass market and that's that's the key thing and the 2nd key thing is while there may be a lot of the inverted commas she'd be I headsets that due to be sold so this is the forecast the 200 million of the images we sold by 2020 in reality a lot of those will be the lower power and that will be the mobile phone headsets and we still can't even those command leaps and bounds since it apparent last 5 years and still we have to do that we need is and PCs of let's take off really high and the uptake of which requires
media content creators to get behind it in a big way so all I don't think it will this generation we're talking for 10 15 years down the line in future generations next the next set of offices in excess of others the other next faces survives that's when it will probably gain a foothold and because it's not quite ready yet the pair not quite there and a good analogy to use is the difference between a PDA smartphone PDA hadn't had all the right everything you could do is not fun but the technologies they it wasn't quite helpful enough and of the of the do it until we get smart phone to work and work brilliantly and took off fortunately the PDA got completely left behind the left some companies with but that that that that's the Histadrut 7 years to took the generation time this what we can use and
unwise in real time 30 frames a 2nd we can have a store presence uh and virtual environment that looks at how the fidelity of something like that which looks pretty good a nice actually so this year old and from a large gold then what is very good on real artistry the artist and that runs in real-time at the 60 Frantisek and this is the kind of reality we can get mad in VR headset so repressed all looking like that we can move around it look like that we have headsets proliferating at households and is a big if it does require and a particular set of the market to have these headsets within that household so that the particular demographic that would be interested in the sort of work and then this so things and the are experiencing virtual fits and needs to be in the households in order for this to work but we can get this level of fidelity now in the I had rather than the current web implementations of whether they are so to be honest it's kind of not a class this is
in my opinion this is because this is not going to present a viable when this Latin because we have the power it will happen and lead bend happen all the time it's especially since GPU pair will not very far away and it again exceptionally high quality real-time cult simulation which is kind of the the crux of where he sits at the apex of whether this will work or not um as stated in the constellation is is is the most difficult thing here we need it running in real time at very high resolution that can run very well in real time as reasonable
resolution and that works from far away camera shot when you looking ahead you want garments 6 uh on what it says your body free talking about body measurements and it's the exceptional needs to be what currently takes a few days to simulate you as the simulated been used during real-time at least the friends of 2nd and under the problems that when we talk rabbits out especially in the virtual reality space the problem then is uncanny valley and too much advance familiar with with this idea of if something's not quite right in the virtual space with your whether it's uh but whether it's not real it can cause a sense of proportion and that's not something that you want your users heating if your wanting them to buy clothes want them to enjoy the experience you do not want the sense proportion so it's discussion of what do you hear the yeah do you make is
perfect of on what has to be perfect because would have to be in unable to tell if it's then or not which is not impossible to achieve completely and would you go for it this is generic have at all or nothing at all to completely remove the 3 D mesh and just focus on how the clothes that fit on a person's body which is which would be easy to do but that's a conical because looking at this serving and tensor you dance is it could be very close replication of of the shopping experience if we wanted to be with a very quick very easy but that process to something creating stole in 10 years time and they can change the entire length of it from 1 day to the next it had direct contact with the consumer telling them the still debated new clothes now you can get that at any time of day it's always open there's no there's no problem getting there and because of technology the that code is already there and available to allow us to you could actually and you can do you have the whole experience with people at the same time you could use to shop with other people over there to the people of the shop and see if you wanted to you could see every mouse is there if you want to and is also the added value with with this sort of work with regard to people that may want to experience something but can't because is this semester with disabilities also with mental health issues and that's that social anxiety for example people may not want to get it still people may not like even leaving the harassment of negrophobic thing about that so etc. this potential advantages flat thing as well and as potential as stands in danger at Denny's experiments for a few years with regards and introducing monetization uh and then end you get people to spend higher amounts of money as it were with
so non-tangible things that you're going to miss that such in the field that's that's will become quite achieve yeah but and you don't have to come if we created and tested Modesta C over many years what's that happened in sensors and expenditure and what but it be very interesting like so well as technology develops this the sort of area we can push into which we're very fine and very entertaining and re-estimate there's good thank you very much and is any questions of thing
we talk proponents regressions questions yeah here where you
have a few minutes questions if had was going questions on any of that or any ideas yep I don't think I was looking for the loudspeaker think if what I queuing expected was that connects will you measure the people and so my and it where we're actually
not using that we we have some work going on and
on with the Kinect that's not I will work with with the connector nodes busses pre-existing with someone else's implemented and but it's it's a depth sensor and it generates a very costly model over the top of a person they can dry clothes and stuff themselves and move around in real time it is a wood perfectly but that's kind of a which is a depth sensing camera the creates 3 D models that already areas um very very similar to the Camex slows talking about the the body scanner system uses that that depth sensing and just to scam the subjective it outputs a point cloud data the 3 D coordinate points that make a plot point cloud data loss is not motion capture and interacting directly from connected uses the same technologists say um the representational subjective anybody else yeah no well then thank you very much guys here and now we're going to have a little
break 15 minutes and afterwards we're going to continue with the future of e-commerce yeah in the yeah with objection