“Why is making good AR displays so hard?” Article by Daniel Wagner, the CTO of Daqri

Back to Blogging

I apologize for the long break, there were a lot of personal things that got in the way, including moving houses, a bunch of travel, and our new startup RAVN ramping up. I then had a writer’s block over having so much to write about and not knowing where to start. I hope to get to things I saw at Photonics West, SID Display Week, and visits in the UK. I also want to talk about some trends in Augmented Reality.

So, I decided to ease back into the blogging with a recommendation with of excellent article titled, “Why is making good AR displays so hard?” by Daniel Wagner, the CTO of Daqri, along with his co-authors Louahab Noui, Adrian Stannard. I will add just a few comments.

Comments on the article

The article, in particular, makes some good points about the difficulty in supporting a wider FOV. In part, they use etendue to explain why the display and optics (what the article refers to as “the projector”) must be large to support a large FOV. He also shows how large waveguides must be to support various FOVs making for a given eye relief (the distance of the waveguide from the eye).

Figure: Waveguides sized 35mm (left), 60mm (middle) and 135mm (right)

When it comes to brightness and transparency they cite (with permission) my blog  (here and here). The article uses figures from this blog to illustrate the uniformity issues found with most diffractive waveguides. He also cites this blog when discussing the issue of the display device’s resolution versus the effective resolution after the optics.

The article goes on to make interesting points about frame rate and motion blur with respect to both frame rate and duty time (how long the image stays on within a frame time).

He also makes some good points about not blocking a person’s vision, something that is exceptionally bad with the Magic Leap One (ML1)among other so-call “AR” devices (something I commented on the first time I say the ML1).

The article goes on to discuss vergence-accommodation conflict and other issues of eye comfort. I wished they had gone into more on the subject of contrast with AR, specifically, problems with contrast against the real world. What you see in the real world is the “black,” so the contrast that counts is how bright the display is versus the real world after any blocking of the light (say by the optics and with tinting).

They cover a lot in a short article and make some good points. I plan on expanding on some of the points they made in future articles on this blog.

Default image
Karl Guttag

14 Comments

  1. Welcome back, and congratulations on the move, they’re always a bit of a trial. Please don’t feel rushed to write, your blog entries are worth waiting for.

  2. Hey Karl, I have a genuine question. If you had to build an AR headset of your own how would you do it?

    • I’m working on building an AR headset for military and first responder use. I think that there is technology that can support specific applications.

      This is very different from the overblown claims that AR technology is ready to be the next big thing after cell phones. It is an extremely hard problem trying to simultaneously solve for size, weight, power, comfort, and a myriad of image quality issues including but not limited to, resolution, uniformity, reflections/artifacts, contrast, and transparency of the real world. There are also many technical issues regarding control and input. On top of these technical factors, there are all the social issues with people wearing headsets from what the person looks like and their perceived behavior. There are also many safety issues with respect to isolating/distracting the user when they are out in the real world.

    • I’m not sure my first answer directly address your question, but it depends upon the application and many other factors as to the technology I would choose. There are pros and cons based on size, weight, cost, form factor, brightness, resolution, transparency to the real world, and many other factors.

      • Thanks for the reply. I like that you approach it with a logical and engineering thought process. However the headset (< 579g ) I had in mind was something that would be worn indoors (brightly lit to dim rooms) for maybe brief amounts of time ( < 30min) throughout the day. I was picturing a headset focused on productivity and information architecture.
        I've done some work with the Project North Star (https://www.pumori.io/) but the display tech isn't the most accommodating (stereoscopic) nor mind blowing. The larger FOV is very nice to work with and makes building an actual UI possible. But the constant mismatch with the depth of field is bothersome .
        I guess what I'm asking is do you know of any current way to build a light-field AR display that can deliver a 100° x 80° Fov?

      • do you know of any current way to build a light-field AR display that can deliver a 100° x 80° Fov?

        The short answer is no, you appear to be asking for something that is nowhere close to existing. Creal3D (https://www.roadtovr.com/creal3d-light-field-display-ar-vr-ces-2019/ and http://lightfield-forum.com/2019/01/creal3d-swiss-startup-presents-novel-light-field-display/) is trying to make a Lightfield headset but it is still a proof of concept and I’m not sure if there is a display device that will support their concept in full color.

        You seemed to be focused on FOV and depth of focus at the expense of everything else in user interface. If you have wide FOV but low angular resolution, then you can only have big blocky text which will make things like reading slow.

      • Hi Karl
        Genuinely love learning and reading from your blog, its always a pleasure to see a new article.
        May I ask what year you would predict/estimate an AR product will satisfy the mass market (as you mentioned phone replacement)?
        I think 2023 for the first product and 3years after for it to become mainstream.

      • Thanks,

        I believe there are many difficult challenges yet to be solved, so I think you are probably being very optimistic. Many of the problems are bumping up against the physics of light. Then you have all the human factor and social issues of what people will wear and what they will look like.

        I think people make a lot of very bad analogies to say the original iPhone. The iPhone took technologies that existed and put them together in an extremely good way, AR is not near that point today.

  3. Hi Karl,

    Would it be possible for you to prepare a table that summerizes the differences between the various waveguide technologies (Diffractive, Reflective, Holographic..) comparing FOV, EMB, Eye-relief, MTF, Image quality….this could realy help in understanding where we stand and what can be achieved eventually.

  4. Hi Karl can you please comment on Kura Tech ? Their Gallium glasses let look awesome. Would love to hear your thoughts on how their tech works.

    Thanks
    Sanjit

  5. Hi Karl,
    Would love to hear your thoughts on Kura (https://uploadvr.com/hands-on-kura-wide-fov-lightweight-ar/?amp).

    And many, many thanks for all the great reads.

    Best,
    Adam

  6. Hi Karl, Thanks for the awesome interview. I am HCI aspirant looking to develop applications for AR headsets. What platform do you suggest I start developing for? Most headsets seem to be using proprietary OS’s and expensive to buy.

    I recently came across the open-source Project North Star AR headset from Leap Motion (Now acquired by UltraHaptics) which despite being quite bulky offers a large FOV(100 degrees) and uses high res LCD displays (Exact one used in Valve Index).
    (https://developer.leapmotion.com/northstar/)
    (https://www.smart-prototyping.com/blog/Building-Project-North-Star)

    I would love to hear your thoughts on it and your opinion on whether aspiring HCI devs should look into it?

  7. Karl, I read a more recent article you posted that you skeptical about Gallium AR glasses (https://www.kura.tech/) being produced int 2020, but I would also be interested in hearing your comments on the technology they are implementing vis-à-vis pin hole mirrors and ‘structured geometric waveguide’.

    The specifications they are posting for brightness, depth of field, fov, and transparency are impressive if they are for real.

  8. Why people still use to read news papers when in this technological world the whole thing is existing on web?

Leave a Reply

Physical Address

304 North Cardinal St.
Dorchester Center, MA 02124

%d bloggers like this: