I am teaching myself how to create web apps where user can easily interact with visual elements.
I remember there was a clothing website where users can virtually try on things (using mannequin close to their likeness), but I'm not sure if its online anymore.
Now I found article where Amazon may launch something like this.
My question is, what app can be used to create this?
https://www.thesun.co.uk/tech/8293349/amazon-app-try-on-clothes-virtual-mannequins/
You do not want especially little. Software which can dress some clothes onto a human-like 3D shape exists, but it costs a fortune. The price is formed as well as by the sheer complexity of the task as the value which a pro user can get when he checks and presents his cloth creations. Here's one for cloth designers: https://optitex.com This is another: https://www.clo3d.com
NOTE: The vendors of high cost software often select who can get a trial of their products. They are not freely downloadable for everyone. A person who is considered to be a potential buyer will get it. As often there's no public price lists, everyone interested must ask a quote.
Not asked: One can say "get Blender and pay nothing". True, but a designer cannot learn general 3D modelling and physics software, he needs one which operates within his zone of comfort. It must know the concepts, methods and materials which are common in his industry. He simply hasn't the time to convert them to general 3D concepts.
Clearly, there are few details available about this proposed app interface whose wireframes you link to, and so what responses you get will be somewhere in the intersection space between sheer speculation and SWAGs - including this one.
Nonetheless I will hazard a moment of extrapolation.
Points to consider:
1. Amazon has huge resources, both in terms of money and literal computing resources, as well as pretty deep experience with AI and image recognition applications, from facial recognition on up.
2. Deep fake technologies already exist which apply transforms to extrapolated rough meshes to make false images combining disparate elements - none of these would fool educated observers, but will suffice for 80% of the population to be fooled.
3. There are already plug-ins for After Effects et al which will take photos in and derive a medium quality 3D mesh for compositing work - admittedly not nice topology for geometry-based animations, but fine for projecting animated imagery onto to achieve a decent effect.
4.**Applications like Marvelous Designer (which is middle-cost software and does a damned good job of fabric dynamics) allow one to very quickly create decent looking drapery and placements. If you pre-simulate most of the common fittings, with a generic "average" mesh, one female, one male, this will get you to the 75% range of the population very quickly.
5. Colour treatments over the top are not hard to do (Overlay mode anyone?) and allow -just-in-time corrections at publishing.
6. Data Mining - the stated goal of the linked app is to aggressively data-mine the social media accounts of its users, to pull facial imagery / hair / body proportions to on-the-fly create a unique personalised avatar for their outfit-trying, VR clothes-fitting experience. Stepping away from the obviously deeply creepy privacy-violation, EU GDPR-non-complying aspects of this, we can extrapolate a range of postulates from this point alone, especially when combined with the preceding points.
Pre-canned generalised fits per generic mesh (possibly several variant generic meshes) with those meshes modified on-the-fly by pre-canned morphs with exposed percentages of effect to handle the minor changes needed to bring each into rough compliance with the data gathered from the data mining to make the fit "close enough" to pass in a lower-resolution, in-browser experience, combined with rough texture-mapping of the gathered images onto the generic meshes; given current state-of-the-art texturing tools easily available, this can include on-the-fly creation of bump, normal, displacement, and a delit version of a basic albedo map - they won't bother with sub-surface scatter - too expensive and unneeded at this scale.
My expectation of quality:
Either it will fall squarely into the uncanny valley and really creep people out due to clear visual mis-match of texture, or with Amazon's full funding and resources behind it, the results will scare hell out of all users by being so darn close that it could fool incautious friends at first glance.
It will be extremely interesting to see where this goes.
