Hololens
17 views
Skip to first unread message
Jan Ciger
Dec 2, 2016, 9:50:16 AM12/2/16
to vr-g...@googlegroups.com
Hello,
Well, I think I need to take back few things I have said when Microsoft announced their Hololens.
We have received ours today and we have done some quick tests using the built-in apps. The tracking is really very good - not very brightly lit, cluttered office and I was able to pose objects around using few simple gestures no problem whatsoever, even walking from room to room and the system was able to keep tracking without any setup or adjustments from my side. The only issue I have observed was occasional short "swimming" of the displayed overlays after turning the head rapidly - probably the sensor fusion "catching up". There was also sometimes issue with the depth the objects were perceived at - the objects weren't properly occluded by obstacles in front. Both of these could be maybe improved by doing calibration/setup properly, we have only winged it rapidly. Seriously impressive!
A lot of people have criticized the poor FOV. Well, if you are used to the Vive or Rift, this certainly doesn't compare. However, a more fair comparison is with the Epson Moverio BT-3000 see-through HMD, which also costs $3000. The limited FOV on these is usually not that disturbing because it is a see-through display, so you don't see dark frame of the glasses or something like that, as in a classic VR HMD. I have found the horizontal FOV quite acceptable for AR, the vertical could be better, but it is usable. Compared to the Moverio which we also have, it is much better. That one only has a post stamp-sized image floating in front of your eyes. The Moverio also doesn't have any tracking to speak of (only an IMU and a very bad camera), very primitive "SDK", feels much more fragile and there are no apps for it. Absolutely no contest there.
What I have also liked is the feel of the HMD. It really feels solid in your hand, nothing creaks, there are no flimsy plastic strips or rods that could break, it stays put on your head too without being uncomfortable. It even works with glasses with a bit of fiddling - I am near-sighted and had to fetch my pair because I couldn't read the text, so this is important.
The last thing worth mentioning is the out of the box experience. The packaging is very Apple-like (or Oculus-like, if you want). Really really clean, nice design of the box and the HMD comes packed in a zipped oval hard (!) carrying case. Very nice! When you turn the device on, there are pre-loaded demo applications so you can try it out right away.
What a contrast with the Moverio which comes in a huge styrofoam box, loaded an old version of Android (I believe 4.4 Kitkat) and you can't even use the device unless you connect a Bluetooth mouse to it - the control box doesn't let you drive the pointer, you can only click with it! Of course, there are no demo applications on the Moverio neither, except of one app for the (non-Android standard) camera and nowhere to download them from, because even the Epson "store" doesn't have any - only applications for the retail variants BT-200 and BT-300. Oh and even if there were some, you would have to first figure out how to sideload them - there is not even an ADB driver shipped by Epson, they only provide a poorly translated PDF showing how to "hack" the generic INF files supplied by Google.
What remains to be seen is how easy (or difficult) is it to develop a custom application for the Hololens.
Overall, I am positively surprised with the device. It certainly won't make a dent into the Rift/Vive/retail market, both because of the price ($3000 dev kit, $5000 "pro" price) and because of the different focus, but it seems to be a very interesting tool for the professional market.
Regards,
Jan
TobyCWood
Dec 2, 2016, 5:39:25 PM12/2/16
to VR Geeks
Thanks for that early peek. I agree there's a number of professional use cases this device will meet that will make it worth the money. The first that comes to mind is the tech at the on site out in the field in connection with the expert back at the office. Or the nurse practitioner connected to the specialist MD. Also, the CAD design use case could easily make tedious and time consuming prototyping with 3D Printing unnecessary. Will everybody want one? Doubtful. Unless they can get the price down to around $300,
Jan Ciger
Dec 3, 2016, 8:14:26 AM12/3/16
to vr-g...@googlegroups.com
Hello,
On 02/12/16 23:39, TobyCWood wrote:
> Thanks for that early peek. I agree there's a number of professional use
> cases this device will meet that will make it worth the money. The first
> that comes to mind is the tech at the on site out in the field in
> connection with the expert back at the office. Or the nurse practitioner
> connected to the specialist MD.
Yes, those are common use cases being mentioned and asked about, but
these are quite a bit more complicated than simply displaying a few
overlays (that's the easy part, actually).
Both of these would likely profit more from a classic HUD (something
like the Google Glass) type device than a full blown AR - in most cases
the person in the field needs documentation, checklists, procedure to
follow to be displayed in a hands-free manner, not necessarily having
this information precisely registered in space - e.g. the nurse likely
knows *where* to push (<- type of info that spatially reg. overlay
conveys) when doing a CPR on a patient, but may need information about
their medical history while doing so.
Going for a full blown AR would be likely a gimmicky overkill that would
complicate things a lot for little gain - the system still needs to know
the "shape" and localization of the environment and then the overlays
have to be pre-placed by someone there for the field person to find
them. A repair tech could have that pre-prepared, because e.g. the
electrical cabinets he works on are all the same. But nurse? Every human
and apartment is different and she likely wouldn't have time to map the
space first so that the overlays stay put afterwards.
Also these assume that there is a live high speed connection available -
which may not be. In many places you could only use a mobile phone
network and outside of cities the 3G penetration is still very poor. May
not be an issue for the maintenance tech because it could be required to
install a broadband internet/wifi on premises in some cases, but the
nurse will certainly have to go to places where the connectivity is poor
or simply unavailable.
> Also, the CAD design use case could
> easily make tedious and time consuming prototyping with 3D Printing
That's mostly a wet dream of the companies selling these systems, but
there is a reason why 3D printing and prototyping is still used, despite
computer visualization, be it in 2D,3D, on a computer screen, projector,
stereoscopic, in a CAVE or VR HMD, being available since a long time.
Hololens or AR in general don't really bring anything new that couldn't
have been done before to the table.
The main reason is that simply 3D visualization does never convey all
the information a physical piece does. If you need to fit several
complex parts together, you may want to prototype them, simply to make
sure that the tolerances required are sufficient. The 3D model is
perfect and will always fit together. The actual manufactured part may
not. Also humans are very "tactile" creatures, we need to feel the part
in our hands, an abstract visualization doesn't convey the same thing.
Car manufacturers are still building the clay models of their cars
during the design stages, because it is so easy and quick to work on
them (just use a chisel!) and to communicate the result to the
non-technical (marketing, management, etc.) people. Anyone can look at a
physical piece and understands it, with an HMD you would have to first
spend half of your (often precious) meeting time to explain how to use
the device and to troubleshoot issues.
Also many companies will 3D print prototypes to do preliminary testing
of the part as well, before committing to full production.
So I don't really believe that CAD will be a large market for these.
There isn't really need for that there.
Where the Hololens type AR could be genuinely useful are things like
factory floor maintenance (known environment, so can be prepared in
advance), complex assembly tasks (e.g. electrical distribution panel
wiring) and training in general. That's where the ability to do a
spatial registration and to show precisely localized overlays has real
advantages.
> unnecessary. Will everybody want one? Doubtful. Unless they can get the
> price down to around $300,
That's only concern for a retail/consumer market. If a manufacturer is
buying a $50-100k training system that will save them several millions
in lost production time every year, the $3-5k price for the HMD is not
really an issue.
E.g. the professional HMDs from the likes of Sensics and Rockwell
Collins were selling for $20-30k+/piece. With no tracking, of course.
Companies and universities that needed them were buying them, because
there hasn't been anything else with the same capability available.
Regards,
Jan
On 02/12/16 23:39, TobyCWood wrote:
> Thanks for that early peek. I agree there's a number of professional use
> cases this device will meet that will make it worth the money. The first
> that comes to mind is the tech at the on site out in the field in
> connection with the expert back at the office. Or the nurse practitioner
> connected to the specialist MD.
these are quite a bit more complicated than simply displaying a few
overlays (that's the easy part, actually).
Both of these would likely profit more from a classic HUD (something
like the Google Glass) type device than a full blown AR - in most cases
the person in the field needs documentation, checklists, procedure to
follow to be displayed in a hands-free manner, not necessarily having
this information precisely registered in space - e.g. the nurse likely
knows *where* to push (<- type of info that spatially reg. overlay
conveys) when doing a CPR on a patient, but may need information about
their medical history while doing so.
Going for a full blown AR would be likely a gimmicky overkill that would
complicate things a lot for little gain - the system still needs to know
the "shape" and localization of the environment and then the overlays
have to be pre-placed by someone there for the field person to find
them. A repair tech could have that pre-prepared, because e.g. the
electrical cabinets he works on are all the same. But nurse? Every human
and apartment is different and she likely wouldn't have time to map the
space first so that the overlays stay put afterwards.
Also these assume that there is a live high speed connection available -
which may not be. In many places you could only use a mobile phone
network and outside of cities the 3G penetration is still very poor. May
not be an issue for the maintenance tech because it could be required to
install a broadband internet/wifi on premises in some cases, but the
nurse will certainly have to go to places where the connectivity is poor
or simply unavailable.
> Also, the CAD design use case could
> easily make tedious and time consuming prototyping with 3D Printing
there is a reason why 3D printing and prototyping is still used, despite
computer visualization, be it in 2D,3D, on a computer screen, projector,
stereoscopic, in a CAVE or VR HMD, being available since a long time.
Hololens or AR in general don't really bring anything new that couldn't
have been done before to the table.
The main reason is that simply 3D visualization does never convey all
the information a physical piece does. If you need to fit several
complex parts together, you may want to prototype them, simply to make
sure that the tolerances required are sufficient. The 3D model is
perfect and will always fit together. The actual manufactured part may
not. Also humans are very "tactile" creatures, we need to feel the part
in our hands, an abstract visualization doesn't convey the same thing.
Car manufacturers are still building the clay models of their cars
during the design stages, because it is so easy and quick to work on
them (just use a chisel!) and to communicate the result to the
non-technical (marketing, management, etc.) people. Anyone can look at a
physical piece and understands it, with an HMD you would have to first
spend half of your (often precious) meeting time to explain how to use
the device and to troubleshoot issues.
Also many companies will 3D print prototypes to do preliminary testing
of the part as well, before committing to full production.
So I don't really believe that CAD will be a large market for these.
There isn't really need for that there.
Where the Hololens type AR could be genuinely useful are things like
factory floor maintenance (known environment, so can be prepared in
advance), complex assembly tasks (e.g. electrical distribution panel
wiring) and training in general. That's where the ability to do a
spatial registration and to show precisely localized overlays has real
advantages.
> unnecessary. Will everybody want one? Doubtful. Unless they can get the
> price down to around $300,
buying a $50-100k training system that will save them several millions
in lost production time every year, the $3-5k price for the HMD is not
really an issue.
E.g. the professional HMDs from the likes of Sensics and Rockwell
Collins were selling for $20-30k+/piece. With no tracking, of course.
Companies and universities that needed them were buying them, because
there hasn't been anything else with the same capability available.
Regards,
Jan
Reply all
Reply to author
Forward
0 new messages