[FoRK] Re: [Wearables] So, what *is* wrong with HITL/microvision?
Eugen Leitl
<eugen at leitl.org> on
Sat Jan 12 08:57:50 PST 2008
----- Forwarded message from Ozan Cakmakci <ozan.cakmakci at gmail.com> -----
From: Ozan Cakmakci <ozan.cakmakci at gmail.com>
Date: Sat, 12 Jan 2008 11:51:43 -0500
To: Blair MacIntyre <blair at cc.gatech.edu>
Cc: "wearables-list at media.mit.edu" <wearables-list at media.mit.edu>
Subject: Re: [Wearables] So, what *is* wrong with HITL/microvision?
X-Mailer: Apple Mail (2.753)
On Jan 9, 2008, at 9:40 AM, Blair MacIntyre wrote:
>> I was hoping that there would be some set of wearables
>> applications that can benefit from 20-40 degree fields of view
>> range with compact physical size (within a few cm^3), full color
>> (450-650nm), high-quality (20% MTF + <5% distortion), cheap to
>> produce (<100 USD), decent pupil size (let's say 3 to 8mm pupil*)
>> & eye clearance (~15mm or greater), see-through optics.
>
> I'd definitely imagine there is ... just not for me. :)
>
>> >40 degree see-through + all of the constraints above within the
>> eyeglass form-factor (or as close as we can get) ? We probably
>> have to work harder or loosen some of the specs.
>
> Which ones? :)
I realized when I saw an announcement today that I forgot to mention
panel resolution, pixel spacing and diagonal size :-) Anyway, looks
like Kopin came out with a new microdisplay: http://optics.org/cws/
article/industry/32411. This will be interesting for new head-worn
displays. Looks like an 800x600 panel, with a .44" diagonal and a
11.25 um^2 pixel size. Pixel size sets the difficulty on the optics,
gets more challenging for smaller pixel spacing, diagonal size is
related to the size of the magnifier optics (magnifier forms scale
with diagonal size).
> All the alternatives are big, heavy and expensive, so loosening
> something isn't so bad.
>
> Heck, if you can make anything with > 40 degree diagonal that
> costs < $1000 and is descent quality (e.g., as good or better than
> the eMagin Z800) that would be great.
>
>> To answer your question though, I don't know the exact number for
>> how big the fov can get for a single mirror geometry. We submit an
>> abstract at the society of information display conference to study
>> this aspect.
>
> good luck!!
>
>
>>
>> Ozan
>>
>> *I would like to say that even 4mm is challenging with these specs.
>>
>>
>> On Jan 9, 2008, at 8:09 AM, Blair MacIntyre wrote:
>>
>>> Cool stuff, ozan. How big an fov can you hope to get? I do AR,
>>> not HUD-based wearables, so 24 diagonal is not really useful; 40
>>> is barely useful, much higher will be necessary in the long run.
>>>
>>>
>>> ---
>>> Blair MacIntyre
>>> Associate Professor, School of Interactive Computing, GVU Center
>>> Georgia Tech, 85 5th Street NW, Atlanta, GA, 30308
>>> (please excuse the terseness, this was sent from my phone)
>>>
>>> On Jan 9, 2008, at 7:45 AM, Ozan Cakmakci
>>> <ozan.cakmakci at gmail.com> wrote:
>>>
>>>>
>>>> On Jan 8, 2008, at 9:52 PM, Blair MacIntyre wrote:
>>>>
>>>>> On Jan 8, 2008, at 8:28 PM, Adam Oranchak wrote:
>>>>>
>>>>>> Robin Lee Powell wrote:
>>>>>>> I remember, and I'm sure some of you do to, Back In The Day when
>>>>>>> HITL said they'd have full-resolution 24-bit colour in
>>>>>>> something the
>>>>>>> size of a grain of rice or whatever for a few hundred
>>>>>>> dollars. IIRC
>>>>>>> "the day" was 1995 or so.
>>>>>> Well, you just stepped into a realm that I am intimately aware
>>>>>> of.
>>>>>> (Yes,
>>>>>> you cheeky ones, I said "intimately!") It just so happens that
>>>>>> all
>>>>>> those
>>>>>> specifications neglected to tell you that the image was crap,
>>>>>> that if
>>>>>> the display moved out of alignment from the wearer's eye by a
>>>>>> millimeter
>>>>>> off you lost half the image, that the corners were dark and
>>>>>> pin-cushioned and that the 8 bits of red depth of one pixel
>>>>>> appeared
>>>>>> over the 8 bits of green produced by a pixel 10 pixels away. Oh
>>>>>> yeah, it
>>>>>> cost $20K.
>>>>>
>>>>> Wow ... tell us how you really feel, Adam. Don't hold back. :)
>>>>>
>>>>> Of course, never having used the original one, and having owned
>>>>> and
>>>>> used a more modern $3.5K version of VRD (the "nomad expert
>>>>> technician", I would suggest that readers take what you say with a
>>>>> grain of salt. I found the Nomad to be very nice. Yes, there are
>>>>> issues when you move rapidly, but aside from that, I found the
>>>>> image
>>>>> quite good, the exit pupil large, the field of view better than
>>>>> most
>>>>> other displays in that price range.
>>>>>
>>>>> I won't bother listing my "credentials", though. I don't do HMD
>>>>> optics, I just use them.
>>>>>
>>>>> All that said, to actually answer the first question: rumor
>>>>> has it
>>>>> that they will have a small, full-color prototype delivered to the
>>>>> military this year. I'm anxious to see it, as I still believe
>>>>> that
>>>>> the VRD is the most promising approach to usable, high-res,
>>>>> wide-field-
>>>>> of-view HMDs out there. All other approaches seem to require
>>>>> lots of
>>>>> big crap in front of your face, which just won't cut the
>>>>> mustard with
>>>>> the general public.
>>>>>
>>>>> cheers,
>>>>> blair
>>>>
>>>> There is another promising approach which is not studied
>>>> extensively - limits of a single mirror. We are trying to look
>>>> in the direction outlined below with Adam's help on industrial
>>>> design and opto-mechanical design.
>>>>
>>>> Starting to believe that an alternative ideal solution to not
>>>> requiring big crap in front of your face is to use a single free-
>>>> form off-axis mirror as a magnifier. Free-form to say that
>>>> rotationally symmetric aspheres are unlikely to perform in this
>>>> geometry. Off-axis to say that there is no single axis around
>>>> which the system is symmetric.
>>>>
>>>> I'm unsure how the optics would have a lower element count than
>>>> a single optical element + the microdisplay. Given the lack of
>>>> space/volume to place the optics, lower element count designs
>>>> seem to be helpful. Additionally, such a configuration has no
>>>> moving parts which is desirable. Single mirror has no
>>>> dispersion so color correction is not needed. We just fabricated
>>>> a single free-form mirror that is designed for 24 degrees full
>>>> field diagonal with 20% light at the Nyquist frequency of the
>>>> modulation transfer function (MTF). It is being assembled right
>>>> now, we will know more soon. I don't know too much about molding
>>>> free-form parts but I've heard people say that it can be mass
>>>> manufactured for about a few dollars. Yes, there will be issues
>>>> with manufacturing tolerances but I believe they will not
>>>> require invention for their solutions. This would potentially
>>>> let people get into the <100 USD/hmd range.
>>>>
>>>> We have to optimize the mirror surface (coefficients describing
>>>> the mirror) across a particular field of view to get the best
>>>> image quality possible (<5% distortion + 20% light at Nyquist as
>>>> determined by the pixel spacing on the microdislay). The
>>>> geometry of a single mirror + microdisplay is highly constrained
>>>> from an optimization point of view. There are not that many
>>>> degrees of freedom to optimize to achieve the goal of good image
>>>> quality. Tilt of the mirror is set to an absolute minimum to
>>>> reduce the angles of incidence on the mirror. In such a
>>>> configuration, the shape of the mirror seems to turn into the
>>>> major optimization variable. The consequence is that how we
>>>> describe shape becomes important. Free-form usually means
>>>> Zernike or x-y polynomials for the description of shape. We are
>>>> finding out that we might need to change the basis for
>>>> describing locally anamorphic pieces along the mirror surface.
>>>> We have been experimenting with optimizing the radial basis
>>>> function network representation (RBFN) with a Gaussian basis to
>>>> represent the mirror surface. Initial results indicate that you
>>>> gain 20% MTF compared to standard descriptions of free-form
>>>> shapes such as anamorphic aspheres, Zernike polynomials and x-y
>>>> polynomials (we had a submission pending with optics express
>>>> which just got accepted this morning :). The surface will remain
>>>> manufacturable through diamond turning as long as the
>>>> rotationally non-symmetric sag from the best fit sphere is kept
>>>> within a few hundred micrometers. This seems to be the case for
>>>> the specific configuration of a single element magnifier
>>>> described using an RBFN description. Another concern is
>>>> alignment tolerances but we learned through fabricating dual-
>>>> element magnifiers that the alignment tolerances are not as bad
>>>> as people think they are. Anyway, what can we do with the gain
>>>> in performance (MTF) if we go to a different surface
>>>> representation? It may be traded off with a larger exit pupil
>>>> size or a larger field of view. Establishing the field of view
>>>> and pupil size limits of a single off-axis mirror seems relevant
>>>> to this problem. In terms of improving brightness, people are
>>>> thinking about new microdisplays that differ from LCD, OLED or a
>>>> laser based source. It seems that the brightness of
>>>> microdisplays might increase considerably within the next 5-10
>>>> years.
>>>>
>>>> Thank you,
>>>> Ozan
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>> _______________________________________________
>>>>> Wearables mailing list
>>>>> Wearables at mailman.cc.gatech.edu
>>>>> https://mailman.cc.gatech.edu/mailman/listinfo/wearables
>>>>
>>
>
_______________________________________________
Wearables mailing list
Wearables at mailman.cc.gatech.edu
https://mailman.cc.gatech.edu/mailman/listinfo/wearables
----- End forwarded message -----
--
Eugen* Leitl <a href="http://leitl.org">leitl</a> http://leitl.org
______________________________________________________________
ICBM: 48.07100, 11.36820 http://www.ativel.com http://postbiota.org
8B29F6BE: 099D 78BA 2FD3 B014 B08A 7779 75B0 2443 8B29 F6BE
More information about the FoRK
mailing list