Royalty Free Stock Images

UX Testing For The Masses: Keep It Simple And Cost Effective

User experience design (UXD or UED) is the process of enhancing user satisfaction by improving the usability, accessibility, and pleasure provided in the interaction between the user and the product.
This nicely encapsulates what the design part is all about, but what about the other equally important facet of UX, the testing process? The former can be self-taught, at least to a degree. The latter can be considered as one of the more misunderstood, but ultimately necessary steps in UX design. It has to be effective and involve the most important people – your users/customers.
For the UX guru-in-training, testing can be a difficult and overwhelming topic to approach initially, due to its sheer scale and the diverse directions it can take. This can sometimes be confusing and misleading, depending on which area you wish to focus on and what your professional background is.
For the sake of this article, we’ll approach UX testing from the aspect of a web/app designer who wishes to extend their UI design skills and better understand the core User Centered Design (UCD) approach to an application that should take place before Photoshop or Axure are even powered up.

Understanding User Centered Design (UCD)

Before we proceed to testing, let’s start by explaining the basic concept behind UCD.
UCD places the user first in the design and development cycle of an application/website. UCD is based around an understanding of the application’s environment, tasks, and its users. It then addresses the complete user experience as a whole.
What this basically means is that the entire design process involves real users throughout, in order to ensure the end product meets its initial brief requirement as fully as possible.
To sum up the process in its most basic form (there are many variations of UCD), the following phases are as follows:
  • Context of use: Identify who will use the product and what they will use it for, and under what conditions they intend to use it.
  • Requirements: Identify any business requirements or user goals that must be met for the product to be successful.
  • Design solutions: This part of the process may be done in stages, building from a rough concept to a complete design through a number of iterations.
  • Evaluation of designs: Ideally through usability testing with actual users. This step is just as important for UCD as quality testing is to good software development.
Some of the techniques and methods used in UCD are:

Card Sorting

Card sorting can offer useful insight at the UX Design/Design stage.
Card sorting involves participants being given an unsorted group of cards, each card has a statement on it relating to a page or section of the website. The participants are then asked to sort the cards into groups and name them.
Card sorting is a simple and effective way of testing your UX designs on a range of different subjects.
This is usually a great way of learning what your website navigation and content structure should look like, and how they should work in a way that’s logical to your intended user base.

Usability Testing Session

A usability testing session involves collecting data from a group as they use the website/interactive prototypes. It usually comes at a relatively high cost, because it involves a lot of human interaction and legwork.
What does a usability testing session look like? People are invited to attend a session during which they will be asked to perform a series of tasks on the website, while you or the moderator takes notes. The user will often be asked to fill in a questionnaire at the end of the test, to ascertain how difficult it was to perform certain tasks, such as buy a product on an e-commerce site from a specific category page and proceed to checkout.
This type of testing is usually reserved for high-end interactive prototypes or interactive wireframes. It is a great way of gathering data on the most common issues real-world users will encounter.

Focus Groups

Focus group testing is more or less self-explanatory. It involves asking focus group members (which could be site users or the intended target audience) a series of questions related to the website, and being encouraged to share their thoughts and feelings on different related areas of the site design/wireframes.
UX tests involving user groups and questionnaires can cover a broad demographic, but both come with trade-offs.
It’s normally a good idea to have an experienced moderator during such a group session to ensure accurate notes are taken. Additionally, a good moderator should be able to identify the telltale signs of groupthink, and make sure that the whole process is not negatively affected by group dynamics.


Questionnaires can be a great way of generating invaluable solid statistical data – providing the right questions are asked.
A questionnaire can be particularly useful when you want to collect a much more varied cross-section of data than could be achieved through a small focus group. It can also be argued that people tend to be more honest without the immediate pressure of being in a small user group.
The risk of groupthink is averted, so individuals will make their own decisions.

Testing on a Tight Budget or Timescale

Don’t worry, none of these processes are set in stone. In case you are forced to operate on a tight budget or cut corners to meet a hard deadline, there are ways of streamlining the process without sacrificing too much.
If you have to UX test on a tight budget or on short notice, you will have to cut corners and think outside the box.
For example, you could organize part of these processes differently, or merge them together and use your friends and family as test subjects if needs be. What is important is that you are actively seeking involvement, feedback, and constructive criticism on the processes you design from other people.
If your budget and schedule won’t allow you to do everything you had in mind, you need to think outside the box and come up with new ways of obtaining usable test results. While this approach involves some tradeoffs, you should still be able to get a lot of actionable information from your test subjects.
This post originally appeared in the Toptal Engineering blog

3D Data Visualization with Open Source Tools

In his recent article on Toptal’s blog, skilled data scientist Charles Cook wrote about scientific computing with open source tools. His tutorial makes an important point about open source tools and the role they can play in easily processing data and acquiring results.
But as soon as we’ve solved all these complex differential equations another problems comes up. How do we understand and interpret the huge amounts of data coming out of these simulations? How do we visualize potential gigabytes of data, such as data with millions of grid points within a large simulation?
A data visualization training for data scientists interested in 3D data visualization tools.
During my work on similar problems for my Master’s Thesis, I came into contact with the Visualization Toolkit, or VTK - a powerful graphics library specialized for data visualization.
In this tutorial I will give a quick introduction to VTK and its pipeline architecture, and go on to discuss a real-life 3D visualization example using data from a simulated fluid in an impeller pump. Finally I’ll list the strong points of the library, as well as the weak spots I encountered.

Data Visualization and The VTK Pipeline

The open source library VTK contains a solid processing and rendering pipeline with many sophisticated visualization algorithms. It’s capabilities, however, don’t stop there, as over time image and mesh processing algorithms have been added as well. In my current project with a dental research company, I’m utilizing VTK for mesh based processing tasks within a Qt-based, CAD-like application. The VTK case studies show the wide range of suitable applications.
The architecture of VTK revolves around a powerful pipeline concept. The basic outline of this concept is shown here:
This is what the VTK data visualization pipeline looks like.
  • Sources are at the very beginning of the pipeline and create “something out of nothing”. For example, a vtkConeSource creates a 3D cone, and a vtkSTLReader reads *.stl 3D geometry files.
  • Filters transform the output of either sources or other filters to something new. For example a vtkCuttercuts the output of the previous object in the algorithms using an implicit function, e.g., a plane. All the processing algorithms that come with VTK are implemented as filters and can be freely chained together.
  • Mappers transform data into graphics primitives. For example, they can be used to specify a look-up table for coloring scientific data. They are an abstract way to specify what to display.
  • Actors represent an object (geometry plus display properties) within the scene. Things like color, opacity, shading, or orientation are specified here.
  • Renderers & Windows finally describe the rendering onto the screen in a platform-independent way.
A typical VTK rendering pipeline starts with one or more sources, processes them using various filters into several output objects, which are then rendered separately using mappers and actors. The power behind this concept is the update mechanism. If settings of filters or sources are changed, all dependent filters, mappers, actors and render windows are automatically updated. If, on the other hand, an object further down the pipeline needs information in order to perform its tasks, it can easily obtain it.
In addition, there is no need to deal with rendering systems like OpenGL directly. VTK encapsulates all the low level task in a platform- and (partially) rendering system-independent way; the developer works on a much higher level.

Code Example with a Rotor Pump Dataset

Let’s look at a data visualization example using this dataset of fluid flow in a rotating impeller pump from the IEEE Visualization Contest 2011. The data itself is the result of a computational fluid dynamics simulation, much like the one described in Charles Cook’s article.
The zipped simulation data of the featured pump is over 30 GB in size. It contains multiple parts and multiple time steps, hence the large size. In this guide, we’ll play around with the rotor part of one of these timesteps, which has a compressed size of about 150 MB.
My language of choice for using VTK is C++, but there are mappings for several other languages like Tcl/Tk, Java, and Python. If the target is just the visualization of a single data-set, one doesn’t need to write code at all and can instead utilize Paraview, a graphical front-end for most of VTK’s functionality.

The Dataset and Why 64-bit is Necessary

I extracted the rotor dataset from the 30 GB dataset provided above, by opening one timestep in Paraview and extracting the rotor part into a separate file. It is an unstructured grid file, i.e., a 3D volume consisting of points and 3D cells, like hexahedra, tetrahedra, and so on. Each of the 3D points has associated values. Sometimes the cells have associated values as well, but not in this case. This training will concentrate on pressure and velocity at the points and try to visualize these in their 3D context.
The compressed file size is about 150 MB and the in-memory size is about 280 MB when loaded with VTK. However, by processing it in VTK, the dataset is cached multiple times within the VTK pipeline and we quickly reach the 2 GB memory limit for 32bit programs. There are ways to save memory when using VTK, but to keep it simple we’ll just compile and run the example in 64bit.
Acknowledgements: The dataset is made available courtesy of the Institute of Applied Mechanics, Clausthal University, Germany (Dipl. Wirtsch.-Ing. Andreas Lucius).

The Target

What we will achieve using VTK as a tool is the visualization shown in the image below. As a 3D context the outline of the dataset is shown using a partially transparent wireframe rendering. The left part of the dataset is then used to display the pressure using simple color coding of the surfaces. (We’ll skip the more complex volume rendering for this example). In order to visualize the velocity field, the right part of the dataset is filled with streamlines, which are color-coded by the magnitude of their velocity. This visualization choice is technically not ideal, but I wanted to keep the VTK code as simple as possible. In addition, there is a reason for this example to be part of a visualization challenge, i.e., lots of turbulence in the flow.
This is the resulting 3D data visualization from our example VTK tutorial.

Step by Step

I will discuss the VTK code step by step, showing how the rendering output would look at each stage. The full source code can be downloaded at the end of the training.
Let’s starts by including everything we need from VTK and open the main function.
#include <vtkActor.h>
#include <vtkArrayCalculator.h>
#include <vtkCamera.h>
#include <vtkClipDataSet.h>
#include <vtkCutter.h>
#include <vtkDataSetMapper.h>
#include <vtkInteractorStyleTrackballCamera.h>
#include <vtkLookupTable.h>
#include <vtkNew.h>
#include <vtkPlane.h>
#include <vtkPointData.h>
#include <vtkPointSource.h>
#include <vtkPolyDataMapper.h>
#include <vtkProperty.h>
#include <vtkRenderer.h>
#include <vtkRenderWindow.h>
#include <vtkRenderWindowInteractor.h>
#include <vtkRibbonFilter.h>
#include <vtkStreamTracer.h>
#include <vtkSmartPointer.h>
#include <vtkUnstructuredGrid.h>
#include <vtkXMLUnstructuredGridReader.h>

int main(int argc, char** argv)
Next, we setup the renderer and the render window in order to display our results. We set the background color and the render window size.

  // Setup the renderer
  vtkNew<vtkRenderer> renderer;
  renderer->SetBackground(0.9, 0.9, 0.9);

  // Setup the render window
  vtkNew<vtkRenderWindow> renWin;
  renWin->SetSize(500, 500);
With this code we could already display a static render window. Instead, we opt to add a vtkRenderWindowInteractor in order to interactively rotate, zoom and pan the scene.
  // Setup the render window interactor
  vtkNew<vtkRenderWindowInteractor> interact;
  vtkNew<vtkInteractorStyleTrackballCamera> style;
Now we have a running example showing a gray, empty render window.
Next, we load the dataset using one of the many readers that come with VTK.
  // Read the file
  vtkSmartPointer<vtkXMLUnstructuredGridReader> pumpReader = vtkSmartPointer<vtkXMLUnstructuredGridReader>::New();
Short excursion into VTK memory management: VTK uses a convenient automatic memory management concept revolving around reference counting. Different from most other implementations however, the reference count is kept within the VTK objects themselves, instead of the smart pointer class. This has the advantage that the reference count can be increased, even if the VTK object is passed around as a raw pointer. There are two major ways to create managed VTK objects. vtkNew<T> and vtkSmartPointer<T>::New(), with the main difference being that a vtkSmartPointer<T> is implicit cast-able to the raw pointer T*, and can be returned from a function. For instances of vtkNew<T> we’ll have to call .Get() to obtain a raw pointer and we can only return it by wrapping it into a vtkSmartPointer. Within our example, we never return from functions and all objects live the whole time, therefore we’ll use the short vtkNew, with only the above exception for demonstration purposes.
At this point, nothing has been read from the file yet. We or a filter further down the chain would have to call Update() for the file reading to actually happen. It is usually the best approach to let the VTK classes handle the updates themselves. However, sometimes we want to access the result of a filter directly, for example to get the range of pressures in this dataset. Then we need to call Update() manually. (We don’t lose performance by calling Update() multiple times, as the results are cached.)
  // Get the pressure range
  double pressureRange[2];
Next, we need to extract the left half of the dataset, using vtkClipDataSet. To achieve this we first define a vtkPlane that defines the split. Then, we’ll see for the first time how the VTK pipeline is connected together: successor->SetInputConnection(predecessor->GetOutputPort()). Whenever we request an update from clipperLeft this connection will now ensure that all preceding filters are also up to date.
  // Clip the left part from the input
  vtkNew<vtkPlane> planeLeft;
  planeLeft->SetOrigin(0.0, 0.0, 0.0);
  planeLeft->SetNormal(-1.0, 0.0, 0.0);

  vtkNew<vtkClipDataSet> clipperLeft;
Finally, we create our first actors and mappers to display the wireframe rendering of the left half. Notice, that the mapper is connected to its filter in exactly the same way as the filters to each other. Most of the time, the renderer itself is triggering the updates of all actors, mappers and the underlying filter chains!
The only line that is not self-explanatory is probably leftWireMapper->ScalarVisibilityOff(); - it prohibits the coloring of the wireframe by pressure values, which are set as the currently active array.
  // Create the wireframe representation for the left part
  vtkNew<vtkDataSetMapper> leftWireMapper;

  vtkNew<vtkActor> leftWireActor;
  leftWireActor->GetProperty()->SetColor(0.8, 0.8, 0.8);
At this point, the render window is finally showing something, i.e., the wireframe for the left part.
This is also a resulting example of a 3D data visualization from the VTK tool.
The wireframe rendering for the right part is created in a similar way, by switching the plane normal of a (newly created) vtkClipDataSet to the opposite direction and slightly changing the color and opacity of the (newly created) mapper and actor. Notice, that here our VTK pipeline splits into two directions (right and left) from the same input dataset.
  // Clip the right part from the input
  vtkNew<vtkPlane> planeRight;
  planeRight->SetOrigin(0.0, 0.0, 0.0);
  planeRight->SetNormal(1.0, 0.0, 0.0);

  vtkNew<vtkClipDataSet> clipperRight;

  // Create the wireframe representation for the right part
  vtkNew<vtkDataSetMapper> rightWireMapper;

  vtkNew<vtkActor> rightWireActor;
  rightWireActor->GetProperty()->SetColor(0.2, 0.2, 0.2);
The output window now shows both wireframe parts, as expected.
The data visualization output window now shows both wireframe parts, per the VTK example.
Now we are ready to visualize some useful data! For adding the pressure visualization to the left part, we don’t need to do much. We create a new mapper and connect it to clipperLeft as well, but this time we color by the pressure array. It is also here, that we finally utilize the pressureRange we have derived above.
  // Create the pressure representation for the left part
  vtkNew<vtkDataSetMapper> pressureColorMapper;

  vtkNew<vtkActor> pressureColorActor;
The output now looks like the image shown below. The pressure at the middle is very low, sucking material into the pump. Then, this material is transported to the outside, rapidly gaining pressure. (Of course there should be a color map legend with the actual values, but I left it out to keep the example shorter.)
When color is added into the data visualization example, we really begin to see the way the pump works.
Now the trickier part starts. We want to draw velocity streamlines in the right part. Streamlines are generated by integration within a vector field from source points. The vector field is already part of the dataset in the form of the “Velocities” vector-array. So we only need to generate the source points. vtkPointSourcegenerates a sphere of random points. We’ll generate 1500 source points, because most of them won’t lie within the dataset anyways and will be ignored by the stream tracer.
  // Create the source points for the streamlines
  vtkNew<vtkPointSource> pointSource;
  pointSource->SetCenter(0.0, 0.0, 0.015);
Next we create the streamtracer and set its input connections. “Wait, multiple connections?”, you might say. Yes - this is the first VTK filter with multiple inputs we encounter. The normal input connection is used for the vector field, and the source connection is used for the seed points. Since “Velocities” is the “active” vector array in clipperRight, we don’t need to specify it here explicitly. Finally we specify that the integration should be performed to both directions from the seed points, and set the integration method to Runge-Kutta-4.5.
  vtkNew<vtkStreamTracer> tracer;
Our next problem is coloring the streamlines by velocity magnitude. Since there is no array for the magnitudes of the vectors, we will simply compute the magnitudes into a new scalar array. As you have guessed, there is a VTK filter for this task as well: vtkArrayCalculator. It takes a dataset and outputs it unchanged, but adds exactly one array that is computed from one or more of the existing ones. We configure this array calculator to take the magnitude of the “Velocity” vector and output it as “MagVelocity”. Finally, we call Update() manually again, in order to derive the range of the new array.
  // Compute the velocity magnitudes and create the ribbons
  vtkNew<vtkArrayCalculator> magCalc;

  double magVelocityRange[2];
vtkStreamTracer directly outputs polylines and vtkArrayCalculator passes them on unchanged. Therefore we could just display the output of magCalc directly using a new mapper and actor.
Instead, in this training we opt to make the output a little nicer, by displaying ribbons instead. vtkRibbonFiltergenerates 2D cells to display ribbons for all polylines of its input.
  // Create and render the ribbons
  vtkNew<vtkRibbonFilter> ribbonFilter;

  vtkNew<vtkPolyDataMapper> streamlineMapper;

  vtkNew<vtkActor> streamlineActor;
What is now still missing, and is actually needed to produce the intermediate renderings as well, are the last five lines to actually render the scene and initialize the interactor.
  // Render and show interactive window
  return 0;
Finally, we arrive at the finished visualization, which I will present once again here:
The VTK training exercise results in this complete visualization example.
The full source code for the above visualization can be found here.

The Good, the Bad, and the Ugly

I will close this article with a list of my personal pros and cons of the VTK framework.
  • ProActive development: VTK is under active development from several contributors, mainly from within the research community. This means that some cutting-edge algorithms are available, many 3D-formats can be imported and exported, bugs are actively fixed, and problems usually have a ready-made solution in the discussion boards.
  • ConReliability: Coupling many algorithms from different contributors with the open pipeline design of VTK however, can lead to problems with unusual filter combinations. I have had to go into the VTK source code a few times in order to figure out why my complex filter chain is not producing the desired results. I would strongly recommend setting up VTK in a way that permits debugging.
  • ProSoftware Architecture: The pipeline design and general architecture of VTK seems well thought out and is a pleasure to work with. A few code lines can produce amazing results. The built-in data structures are easy to understand and use.
  • ConMicro Architecture: Some micro-architectural design decisions escape my understanding. Const-correctness is almost non-existent, arrays are passed around as inputs and outputs with no clear distinction. I alleviated this for my own algorithms by giving up some performance and using my own wrapper for vtkMath which utilizes custom 3D types like typedef std::array<double, 3> Pnt3d;.
  • ProMicro Documentation: The Doxygen documentation of all classes and filters is extensive and usable, the examples and test cases on the wiki are also a great help to understand how filters are used.
  • ConMacro Documentation: There are several good tutorials for and introductions to VTK on the web. However as far as I know, there is no big reference documentation that explains how specific things are done. If you want to do something new, expect to search for how to do it for some time. In addition it is hard to find the specific filter for a task. Once you’ve found it however, the Doxygen documentation will usually suffice. A good way to explore the VTK framework is to download and experiment with Paraview.
  • ProImplicit Parallelization Support: If your sources can be split into several parts that can be processed independently, parallelization is as simple as creating a separate filter chain within each thread that processes a single part. Most large visualization problems usually fall into this category.
  • ConNo Explicit Parallelization Support: If you are not blessed with large, dividable problems, but you want to utilize multiple cores, you are on your own. You’ll have to figure out which classes are thread-safe, or even re-entrant by trial-and-error or by reading the source. I once tracked down a parallelization problem to a VTK filter that used a static global variable in order to call some C library.
  • ProBuildsystem CMake: The multi-platform meta-build-system CMake is also developed by Kitware (the makers of VTK) and used in many projects outside of Kitware. It integrates very nicely with VTK and makes setting up a build system for multiple platforms much less painful.
  • ProPlatform Independence, License, and Longevity: VTK is platform independent out of the box, and is licensed under a very permissive BSD-style license. In addition, professional support is available for those important projects that require it. Kitware is backed by many research entities and other companies and will be around for some time.

Last Word

Overall, VTK is the best data visualization tool for the kinds of problems I love. If you ever come across a project that requires visualization, mesh processing, image processing or similar tasks, try firing up Paraview with an input example and evaluate if VTK could be the tool for you.
This article was written by Benjamin Hopfer, a Toptal SQL developer.

Google Cardboard Overview: VR On The Cheap

Imagine if you will, standing on the surface of the moon, overlooking a crater from your lunar rover, listening to mission control chatter. Or don’t. Instead of imagining it, just order a cheap Google Cardboard VR set instead, stick your phone in it, and start exploring the solar system, museums, tourist spots, coral reefs and much more. Let the Imagination Technologies GPU in your phone live up to its name and do your imagining for you.
Google Cardboard is hardly a new concept. It was unleashed on the unsuspecting geekosphere at Google I/O 2014, roughly 18 months ago. Since then, Google has tweaked the Cardboard reference design, but the concept hasn’t changed; Google Cardboard was envisioned as the cheapest Virtual Reality (VR) solution on the planet, and at this point, nothing else comes close in terms of pricing.
Google Cardboard is significantly cheaper than competing VR platforms, so why is adoption so slow?Google Cardboard is significantly cheaper than competing VR platforms, so why is adoption so slow?
If you keep track of tech news, you are probably aware that Oculus Rift started shipping a few days ago. The news even made it to mainstream media, and CNN interviewed a few Oculus execs, who discussed the future of Oculus and VR in general. Demand for the Rift appears to be high because the pre-order website crashed hours into the launch, which coincided with the Consumer Electronics Show (CES) in Vegas. The Oculus Rift is priced at $599, and you also need a $1,000-plus computer to use it properly, but the high price obviously didn’t faze the loads of consumers who pre-ordered one.
I could try to explain what makes Oculus different and why it costs so much, but that’s beside the point. It’s a product for enthusiasts and connoisseurs, people who don’t mind spending a lot of money for a great gaming user experience or for some niche professional applications. Compared to Google’s VR platform, Oculus Rift is a technological tour de force, but for the price of a single Oculus headset, you can get more than 50 prefabricated Google Cardboard headsets. Mind you, I am not talking about cardboard DIY sets, but proper headsets made out of plastic, with soft padding and a few straps to keep the contraption on your head.
Considering that you can get a Google Cardboard compatible set for $10 to $20, you’d expect that loads of people are buying them, but that’s not the case. Let’s take a closer look at Google’s platform and try to figure out what’s going on.

2016: The Year Of VR? Not Really

In addition to being the ZIP code for Sarajevo, that’s the number of users that have rated the official Google Cardboard app so far. The number of downloads is in the one to five million range. That’s low by anyone’s standards, and for a Google product 18 months after launch, it’s a shockingly poor result. Granted, there are VR apps with more downloads, but even they are stuck in the 100,000 to 500,000 range.
Does this mean we should dismiss Cardboard as a hyped up geek fad? Does the user experience suck? What the hell is wrong with it?
This may sound a bit harsh and opinionated, but I believe Google simply can’t do hardware. Regardless of how good they are, Google sucks at marketing its own hardware solutions. In the interest of full disclosure, I am a Nexus veteran and I tend to like Google hardware, but most consumers don’t (many don’t even know it exists). The fact that people aren’t buying a dirt cheap product like Cardboard, and that more companies aren’t using it to build their own products and services, despite the fact that it’s free, might vindicate my position on Google hardware.
There is just one problem: Google Cardboard is a good idea and it works.
While Google may not excel at hardware, Cardboard VR is a sound concept and it works.While Google may not excel at hardware, Cardboard VR is a sound concept and it works.
Rather than dismiss Cardboard outright, I decided to try it out. I quickly realised the concept is sound and there’s nothing terribly wrong with the user experience, but once again, Google failed to market it properly and make it appeal to non-geeks. As a result, adoption is pathetic, at least for now.
What about VR adoption and popularity in general? Weren’t we told that 2015 was going to be “The Year of VR?” Or was it supposed to be 2016? I am sure CNN said it was going to be this year.
It all depends on whose marketing pitches you were listening to, but in reality, 2016 will not go down in tech history as the year of VR. Sure, it sounds good and investors like the idea, but it’s not going to happen. This is not my personal opinion. A few industry sources believe things will start moving in 2017, but it will take a while.
When I say “industry sources,” I am talking about GPU industry execs, people who know this stuff better than anyone. They’re pointing to 2017 and beyond, but they’re doing so off the record. This applies to cheap VR solutions like Google Cardboard and expensive sets like Oculus Rift: VR won’t get a lot of traction this year, don’t fall for the hype!

Google Cardboard: VR For The Masses

How does Google Cardboard work? What makes it different?
The really fascinating thing about Google Cardboard is its simplicity and low price. The concept relies on off-the-shelf hardware, you just need to stick a smartphone into a Cardboard headset and you’re ready to go, more or less.
As the low price suggests, Google Cardboard doesn’t contain any magic or expensive components. All you need is a couple of lenses, a plastic or cardboard body, and a couple of magnets which double as a physical button. When you push the button, the phone’s magnetic sensor, or e-compass, detects the changes in the magnetic field and that’s all there is to it.
The recipe for Google’s Cardboard VR special sauce: simple, widely available, cheap, open-source, based on prolific hardware.The recipe for Google’s Cardboard VR special sauce: simple, widely available, cheap, open-source, based on prolific hardware.
There are a few caveats: Google Cardboard won’t work on any phone because it relies on sensor input that might not be available on many devices (gyroscopic sensors aren’t very common on cheap phones). The phone also needs to have a high resolution display, but thanks to the pixel density marketing craze, this shouldn’t be much of a problem moving forward. Having a bigger display with more pixels simply makes everything look better. While 1080p on a 5-inch phone sounds like a lot, once you start using Cardboard, you’ll see individual pixels. I didn’t try it on a 720p display, but I am convinced it wouldn’t be enough. There are a few other problems, such as battery consumption and overheating, and let’s not forget that you could get a call or message while you’re in the middle of your VR experience.
In spite of these foibles, Google Cardboard has a lot going for it. For starters, it does not require consumers to spend a small fortune just to get a taste of VR. It relies on one of the most prolific software/hardware platforms on the market, so it’s within easy reach for hundreds of millions of smartphone users and developers alike.
Unfortunately, this vast potential has not translated into market success. With a few dozen thousand users in the wild, I could hardly blame anyone for dismissing Cardboard as a geeky curiosity, but I’d stop short of calling it a flop.

What Went Wrong?

Nothing, apart from the fact that Google can’t do hardware.
To be fair, Google Cardboard wasn’t envisioned as a product with mass market appeal and I personally view it as a tech testbed rather than a proper product. It’s not the only VR concept to rely on a phone for display and processing: Samsung’s Gear VR is similar.
However, this seems to be part of the problem, because it does not appear Google is taking Cardboard very seriously. Although Google Cardboard was released 18 months ago, a lot of building blocks weren’t ready for launch. Google is still dragging its feet, but there’s some progress: As of May 2015, Cardboard can be used on iOS devices, it has better OpenGL and WebGL support, and Google launched a few new VR initiatives, including Jump and Expeditions. YouTube also got a dedicated VR/360 degree video channel, and it could become the go to place for people searching for VR video.
Cardboard’s biggest problem is not technological.
The platform is too small to attract a lot of third-party development, and who could blame app makers for refusing to waste man-hours on projects that don’t guarantee a return. That’s one of the reasons I decided to try it out; I kept looking at Google Play stats, the terrible reviews and I started wondering whether or not Google Cardboard has a bright future, or any future for that matter.
I don’t intend to turn this post into a Google Cardboard review, but I think it’s important to review a few things, just to give you a clear idea of what to expect (in case you haven’t tried it, yet).
Let’s start with Cardboard requirements. I should note that these are not official Google Cardboard requirements:
  • Android 4.1 or iOS 8 device required
  • Gyro sensor
  • NFC or magnetic sensor
  • High definition display (1080p is sufficient, the more the better)
  • High capacity battery can’t hurt
  • Loads of storage
  • Fast network/broadband access
The good news is that there aren’t any software hoops to jump through. Since Google Cardboard relies on standard smartphones, designers and developers are unlikely to encounter many hardware-related issues. The biggest hardware compatibility issue is on the sensor side. A lot of inexpensive Android phones don’t feature some of the sensors that may be employed by Cardboard apps (namely gyro and magnetic sensors).
Google Cardboard can accommodate a range of different phone sizes, so it should work on standard 5-inch phones, as well as oversized 5.5- or 6-inch phablets. Display density isn’t much of a problem on 1080p, although it could be better. Resolution will eventually go up, as hardware-makers shift to 1600p and 4K/UHD displays on plus-size phones. Sony already has a flagship Android phone with a 4K display.
I’ve already discussed the more or less pointless trend of moving to higher definition phone displays in one of my Toptal blog posts, but VR is an exception. There’s no way you’ll see individual pixels on modern, high-definition phone displays, unless you use them in a Cardboard headset.
However, higher resolution displays don’t mean a thing unless you’ve got high-res content for them. Unfortunately, there’s not a lot of 1080p VR content out there, let alone 4K/UHD content.

VR Video Resolution Conundrum

Bear in mind that increasing the resolution comes with trade-offs. This brings us to the next problem: Even if we had loads of 4K VR videos, how would we get them on our devices? The problem I encountered was simple: I quickly started running out of bandwidth and storage, in some cases even on 1080p. Sure, you can stream 1080p video on even a slow internet connection, but you’ll often need to slow down and give your device time to buffer, which is always annoying, but it’s really annoying when you have a VR headset strapped to your cranium.
Some of you may be thinking that I live in a part of the world with terrible Internet infrastructure, and I’ll be the first to admit that Bosnia isn’t exactly Silicon Valley, but bear with me; my broadband is still faster than the average speed in the US, UK, Sweden, Japan, and a bunch of other highly developed economies. In other words, a lot of users in California and Tokyo still rely on even slower Internet access. Recent surveys indicate that just one fifth of US homes have enough bandwidth to stream 4K content.
I know. I’ll just download the videos and enjoy them off local storage, but it’s not an ideal solution. First of all, a lot of content isn’t available for download at all, it’s just streamed. Worse, you’ll need a lot of storage to pull it off. For the past couple of years, mobile services have been shifting to streaming in lieu of local content, allowing people to make good use of fast mobile broadband. Why keep gigabytes of music and video on your phone when you can enjoy Netflix or Spotify on the go? Resorting to local storage for high definition VR feels like a step back, but if you’ve got good 4G coverage or fast broadband in your home, it won’t be much of a problem.
Limited bandwidth and resolution are the biggest problems facing virtual reality video at the moment.Limited bandwidth and resolution are the biggest problems facing virtual reality video at the moment.
In addition to requiring more bandwidth, high resolution content also requires more processing power. At 1080p, this isn’t a problem because this industry-standard resolution has been around for ages and even cheap hardware handles it with ease. However, at 4K you simply need more bandwidth and CPU/GPU muscle to handle the data and decode the stream. This means more milliamps, more heat, more charging. Smartphones aren’t designed with this application in mind, they’re simply not supposed to be used for this stuff. With cranked-up screen brightness, high CPU and GPU loads, and a lot of data streaming in to ensure smooth playback, a standard phone will run out of steam in a couple of hours or less. On top of that, it will heat up in minutes. Bear in mind that there’s no airflow inside the headset, so the device will have a hard time dissipating the heat.
I tried it out on a Snapdragon 808 device. For those who don’t pay close attention to the silicon space, this is a one of the latest Qualcomm smartphone chips. It’s a 20nm planar part with a couple of ARM Cortex-A57 CPU cores and powerful Qualcomm Adreno 418 graphics. The same chip is used in Google’s new Nexus 5X. It’s fast enough, but it heats up in no time despite the fact that it’s one of few mobile chips to be produced in a node superior to 28nm.
However, video is not the only type of VR content out there. Let’s take a look at some alternatives.

Different Types Of Google Cardboard Content

I focused on video in the first section of the article because I feel it will be the most attractive form of VR content, at least at this early stage. However, I think people who choose to use their VR headsets solely for video will be missing out.
VR video is usually limited in terms of terms of field of view, and what you see is what you get: You can’t walk around a VR video scene, you’re just stuck in a single virtual location, either a front row seat at a Paul McCartney gig, or a cockpit of a Swiss Air Force F-5 jet performing an acrobatic routine. My problem with video and photos is that the user can truly enjoy this sort of content only once, and there’s not a whole lot of it out there.
Don’t get me wrong, these experiences can be good, but what about getting my glorious behind off the sofa and navigating a real VR world? What about generating a different environment every single time, and interacting with it?
The only way of doing this is on Google Cardboard by rendering the content locally and putting the user smack in the middle of a digital environment. We’ve been doing this since the early nineties, when games like Wolfenstein took the world by storm (and games like Descent made a lot of geeks experience motion sickness without moving, just by staring at their screen).
It’s possible to render 3D VR content locally on most smartphones, but quality is limited due to a range of technical challenges.It’s possible to render 3D VR content locally on most smartphones, but quality is limited due to a range of technical challenges.
This is what makes Oculus Rift fundamentally different: It relies on desktop hardware to render 3D content and display it on the VR headset. This is obviously a huge difference, and the official Oculus Rift requirements look like a gamer’s shopping wishlist: a powerful Haswell generation Core i5 processor backed by 8GB of RAM. More importantly, the list includes Nvidia GeForce GTX 970 and AMD Radeon R9 290 discrete graphics cards, based on Maxwell and Hawaii GPU architectures respectively. High-end PC processors, like the one listed by Oculus, usually have about 1.5 billion transistors. Big GPUs, like Maxwell and Hawaii designs used in the GTX 970 and R9 290, have five to seven billion, and they’re getting bigger. The combined power draw of a PC with such specs is a few hundred Watts, roughly 100 times more than the power consumption of an average smartphone chip. In other words, even if you still believe in Moore’s Law, it’s obvious that we won’t get the same level of performance on mobile devices for years.
Most smartphones have enough GPU muscle to render good looking 3D scenes in 1080p, although they don’t come close to the sort of overkill graphics you get on a high-end PC. You can forget about fancy shaders, advanced antialiasing techniques and many post processing features, but let’s not forget that phones have come a long way and that this sort of technology would have been next to impossible just five years ago.
This is not the only bit of good news; Google has two SDKs for Cardboard developers: an Android SDK using Java and a Unity SDK, using C#. Both rely on OpenGL, and Unity support was added to the iOS SDK earlier this year. Once you are no longer bound to video, VR starts to make a lot more sense. Done right, artificial environment can immerse users into a dynamic and interactive 3D environment, so even simple demos look and feel good.
Even if you plan to rely on video content or photos, you’ll still need a UI that works, and chances are it will use some form of 3D, or at least 2D objects placed in a 3D environment, per Google Cardboard guidelines. Most apps that focus on digitally generated imagery instead rely on Unity. There’s nothing wrong with that, Unity is a popular engine and it’s quite capable.
As I’ve already pointed out, Google Cardboard relies on standard hardware, hence there aren’t that many technical challenges to overcome. Make sure you follow Google’s Cardboard guidelines and best practices, and you should be in the clear.

3D Is The Way To Go, Sort Of

So what’s the problem with using Unity and 3D graphics in general? It sounds straightforward and offers people a chance to experience a true VR experience on a budget.
Let’s not get ahead of ourselves. Here are a few issues that come to mind:
  • Battery life
  • Heat dissipation
  • Limited GPU power
  • Need for high resolution assets (mainly textures for 3D models)
  • Different level of detail (LOD) approach
  • Motion sickness
  • Limited ability to control movement and interact with environment
I’ve already addressed the problem of heat and power consumption. Placing a smartphone in a small environment with no airflow and maxing out the GPU is more or less the worst thing you can do in terms of thermals and efficiency. This issue cannot and will not be resolved. Phones simply aren’t designed to be used this way.
This brings us to the next problem: GPU performance. While smartphone application processors have evolved at a staggering pace, they are not developed for sustained performance. A discrete graphics card or integrated GPU on your desktop can run at high loads for hours, even days, but your mobile GPU cannot. Once the device starts overheating, it will throttle the processor to stay within the thermal envelope, protecting the hardware and saving battery power. Sure, you can get good graphics out of smartphone chips, but running a VR app with a virtual UI in 3D, along with loads of core 3D content, will drain the battery and overheat any phone.
Mobile game developers already know a thing or two about optimising their creations for this sort of hardware. Unity has been around for years, so generating good looking 3D content should not be a problem, right? It depends on the type of environment being designed. If it’s supposed to be a photorealistic 3D environment with advanced lighting and post processing, designing for VR could prove a bit more challenging. This is the problem: Although we’re still using the same resolution, the field of view is much bigger. As a result, the VR experience on a 5-inch 1080p display looks a bit pixelated, and you certainly get to see a lot more details than you usually would. While these devices boast high pixel density displays, the real metric to have in mind is PPD rather than PPI.
This basically means the user gets to see more than you’d expect given the resolution, which means 3D models and textures need to be optimised for a wider field of view. For example, a few low resolution textures won’t ruin the appearance of 3D model on a 5-inch phone. It can still look good because of the small size of the display, but once you put the same phone in a VR headset, you’ll get to see all sorts of compression artefacts and other nasty stuff. If an object looks good on a phone with even with a low LOD, that doesn’t necessarily mean it will look good in VR; it might need more complex geometry and textures. It’s not solely the resolution, please keep that in mind.
Lastly, motion sickness and nausea remain a concern. One of main causes is lag. It takes a tiny amount of time for the phone’s gyro sensor to figure out the user is moving, and then it takes a bit more time for the phone to crunch the numbers and render the subsequent frame while taking the motion into account. If, for any reason, something goes wrong and you drop a few frames or experience stuttering, the VR illusion will break down right before your eyes. This process should be fast and automated to such an extent that the user has no idea what’s going on behind the scenes.
However, this is easier said than done in complex, heavily subdivided 3D scenes with huge textures. A standard phone will struggle with photorealistic graphics even when it’s not overheating, so trying to get a phone to render smooth, photorealistic 3D graphics is not a viable option at this time. In addition, a number of effects and features that could help improve the visual experience are not available. Sure, motion blur, depth of field effects, high-quality antialiasing and other techniques would help, but they’re still not an option on mobile devices.

Google Cardboard For Developers: Opportunity Or Waste Of Time?

So, Google Cardboard is not perfect, it suffers from a few teething problems, lack of content, lack of users, and lack of developer interest. By now, a lot of you must be wondering why I am convinced Google Cardboard has potential. After all, I listed a number of real and potential problems hampering mass adoption.
Why bother with Google Cardboard?
It’s a legitimate question, and considering the size of the user base, coming up with a good answer is not simple. This is still a very tight niche, and even if you manage to come up with a great idea and execute it flawlessly, you won’t make much of a difference (or much money, for that matter). The limited number of people interested in VR is a huge problem.
This lack of interest becomes obvious as soon as you start browsing the Play Store for VR content. There aren’t that many Google Cardboard apps around, and I can confidently report that most of them suck. If you don’t believe me, just check out the user reviews. In fact, many of these apps aren’t real apps; they’re tech demos.
Did Android developers drop the ball? Not really. A lot of these subpar apps are clearly a work in progress, or they are pet projects that allowed individual developers to play around with VR. Very few apps come from big publishers and this is understandable; with such a small user base, nobody can afford to burn thousands of man-hours to create an app that won’t turn a profit. Oddly enough, this could be good news. If you are confident you can do a better job, go for it. There’s not a lot of competition, and if you create something good, your product will definitely stand out.
Not all Google Cardboard apps are bad, so you could try out a few quality designs to get a sense of what makes them tick. I usually don’t list products and services in my blog posts, but I will go ahead with a few examples of promising Google Cardboard apps:
  • Jaunt VR is a highly acclaimed VR platform with one of the best user review scores of any VR app on the Play Store. Jaunt is a relatively big player in the small VR ecosystem and has a number of good products. I’d direct your attention to the UI layout and the quality of the content itself.
  • YouTube and Google Maps are an obvious choice, and happen to be the only Google core apps with Cardboard functionality. They will give you a chance to check out how Google does stuff, although I was not too impressed. Don’t underestimate the power of YouTube. If a lot of VR content is uploaded, it could tip the scales in Google’s favour.
  • Fulldive is an ambitious app with loads of features. You can use it to view panoramic photos, watch local and YouTube videos, take VR photos and more. There are a number of similar apps out there, but I feel the Fulldive team did a better job in the UI department. The UI is clean, fast and intuitive.
  • Sites in VR is a different sort of app and I think it’s a good showcase of what might be achieved by an individual developer. The app allows users to experience a number of different VR sites, ranging from the lunar surface to the Eiffel Tower, plus some good-looking examples of Islamic architecture. I appreciated the ability to tweak settings that aren’t available in most VR apps.
  • VR Roller Coaster is a good example of 3D VR, and the name is self-explanatory; roller coasters are a popular theme in VR apps. In addition, the same concept is used to create a VR tour of the Solar System.Titans Of Space and VR Cosmic Roller Coaster are good examples of this approach.
  • Shadowgun VR and Sisters are nice examples of VR games; the latter is spooky, if you’re into that sort of thing.
We intend to publish more content dealing with the finer points of VR design and development, so if you are interested in this emerging field, be sure to tune in from time to time.

The Elephant In The Living Room

It’s bad for SEO, talking about it is bad for tech publications in general, and it might not go down well with some of our team members or redears, but I have to get it out of the way. So what is it?
The adult entertainment industry played a pivotal role in the adoption of legacy video standards. Can it boost VR adoption as well?The adult entertainment industry played a pivotal role in the adoption of legacy video standards. Can it boost VR adoption as well?
There, I said it. And no, I wasn’t joking.
The adult entertainment industry was instrumental in the adoption of multiple video standards, from VHS over Beta, to Blu-Ray over HD DVD. Granted, these were physical storage standards, but they were around when physical storage mattered a lot more than today. Nowadays, content distribution is digital, on-demand and fast. Best of all, the same content can be distributed across multiple platforms with relative ease.
The adult industry played a pivotal role in the mass adoption of major content standards for decades. It can do it again, albeit not through physical standards. It can obviously make a big difference by generating demand for all sorts of VR devices. Google Cardboard looks like an obvious candidate for VR content distribution on the cheap, and it will undoubtedly be the first glimpse of VR for millions of users.
Does anyone doubt the adult entertainment industry will attract millions upon millions more? For many people, that first glimpse of VR could be described as NSFW.

Virtual World Of Potential

Google Cardboard is a good step toward mass adoption of VR. It’s not without its problems, but we can’t expect miracles this early on, especially not from the cheapest VR platform on the market.
But that’s sort of the beauty of it: It’s cheap and disposable, yet it’s upgradeable. You can get a better headset if you feel like you need one, and the occasional phone upgrade should take care of the actual hardware behind it. One could potentially repurpose old phones as well, provided they sport the necessary sensors and hardware.
Despite my optimism, Google Cardboard isn’t a very popular platform and I don’t think anyone expects it to become one in the immediate future. However, in the long run, I am confident it will attract a lot more users, and not just geeks. As always, mainstream users are the Holy Grail, and I can report that non-geeks are even more impressed by the Google Cardboard experience than tech savvy people.
It all boils down to content. There’s not enough VR content out there, use-cases are limited, and there’s not a lot of urgency to get involved. However, moving forward we are bound to see a lot more VR video, along with other types of content. As soon as we see more VR content being churned out, we will see more adoption. I suspect many people will choose to try out VR over the next couple of years, and once VR starts gaining mainstream traction, price will become even more relevant.
In a mass adoption context, the fact that people can enjoy Cardboard VR for the price of a decent lunch could make Google’s “no frills” VR concept a lot more attractive in no time.
This article was written by NERMIN HAJDARBEGOVICa Toptal Technical Editor.