What is “the model problem” in VR, and how can we solve it?

Bookmark and Share

VR may be the future, it is certainly making significant inroads into business, with creative marketing being developed with virtual reality at its core. But, before it can become completely mainstream, and simply be part of the mix, there is still one major hurdle it needs to overcome.

This hurdle has been termed by many VR developers as “the model problem.”

This problem stems from the fact that much VR content available at the moment takes the form of 360 degree videos material.  Whilst it is true that producing these videos is undoubtedly a complex and challenging process, the video itself is still flat and non-interactive. The image has no genuine depth or measurement information held within the file.

While a 3D video certainly represents reality far more effectively, it isn’t any more genuinely interactive than the aforementioned 360 degree video. On top of that, a 3D video requires far more data usage, the file size tends to be much bigger, but this doesn’t necessarily equate to a better, more integrated video experience.
Since we can’t easily create virtual reality experiences from ‘flat’ video material, this inevitably leads us the introduction of developing models. 3D models offer the best experience, are flexible, and hold the necessary dimensional data, but are more difficult to create, and are more time consuming, as a result they represent only a small portion of already produced available content.

This poses the VR developer with a tricky problem. How to combine the quality of model based VR with something that doesn’t take excessive amounts of time to create?

There are a couple of potential solutions on the horizon. One is investment in state of the art development tools, which will be especially useful for training purposes.

The other, however, involves incorporating something that is also at the very forefront of current technology, something akin to AI. Something that can review and interpret the available material. For this to work effectively and efficiently, ‘flat’ video becomes the core input for the AI. Pattern recognition software, like that that has been implemented by Google and Facebook, could be developed to scan the surroundings of the video, and then the AI, using models from a pre-installed repertoire, turns the video into a 3D, modelled, VR environment.

But this solution is not perfect. It only provides us with a way to turn videos into VR models, nothing more. It will still require, skill and experience to turn those models into something involving. VR Developers know that simply interacting with an environment, as important as that is, is merely the start. It is the interest, the involvement the understanding of the user interaction, that can turn a VR model into a VR experience.

Still the ability to walk through a realistic scene, to interact with the data on an almost physical level, is the Holy Grail for VR developers. We believe that it will be only when this milestone is reached will VR truly be able to fill its potential.

The common assumption is that the magic of VR will be in creating fantasy worlds for games, but it may transpire that the real magic comes from being able to accurately depict real world experiences.

This is why it is important to solve the model problem, combined with creativity and experience if we are to ensure that the success of VR. To be truly valuable so need to create VR experiences that users can seamlessly interact with, and importantly, want to interact with.