Header banner

<< Previous Thread Properly designed DREAM stereo viewer? Next Thread >>

Subject: Properly designed DREAM stereo viewer?
Date: 2008-07-26 02:48:55
From: Bill G
Ahhh, nothing nicer than dreaming about the ultimate stereo
viewer...something many of us do on a regular basis....


I will make some quick comments on some of the items of discussion
recently, as stereo optics are quite unique vs. other visual optics.
Then in the next post, I will show some pix of my dream viewer optics...


EYE PLACEMENT - IPD - Unfortunately, all visual optics are highly
sensitive to eye placement. Stereo viewers are the most sensitive
optical instrument got eye placement, since it's one of the few optical
instruments in which the eye must swivel to see the entire field of
view (film). While in binocs, Telescopes, Microscpes, you simply move
the subject of interest to the center. This is the first huge obstacle
in designing optic s for a stereo viewer. Typically, the more
elements in the optics, the more sensitive eye placement is. All EP's
(Eye Pieces, aka film loupes) are designed for a specific eye pupil
diameter..... and considering most stereo viewers have very low levels
of light intensity, unfortunately our eye pupils are wide, which makes
lens design even more difficult.


The eye pupil diameter is the start of the lens design process... as you
have an entrance pupil (often called the Field Stop Diameter) and an
exit pupil.... a well designed EP will allow all the sharp rays to enter
our eye pupil. However, if the eye is not reasonably well centered on
the lens axis, the eye will never see the "design" rays. Design rays =
the sharp rays the lens designer provided to fit within your pupil,
WHICH assumes your pupil is in the proper location.


How tight is this tolerance? It's extremely tight, which is
problematic with devices that offer no IPD adjust. When the eye pupil
and the lens center axis are mis-aligned due to a mis matched IPD and
center lens spacing, the result is distortion, resolution fall-off, MTF
degradation and sometimes chromatic aberrations. A simple test to
witness this, take your doublet, and look at graph paper with a tight
grid over your light box. Using one eye, center your eye on axis and
view the grid paper...you will see the normal distortion inherent in the
lens (barrel or pincushion).... now de-center your eye pupil with the
lens center, you will witness much higher levels of distortion when you
de-center, the most obvious being distortion on the Z axis, the grid
lines will appear like swells in the ocean. You might also notice
chromatic aberrations of the grid lines if you continue to move your eye
further away from center.


The real kicker is, try this test in stereo, align the grid paper so you
have a nice merged 2d view with both eyes, now de center your eyes by
moving the lenses apart..... Each eye will experience a different form
of Z axis distortion, creating binocular rivalry. You may need to
create a jig to perform this task.


In real world stereo viewing, the brain fights through this rivalry, and
other forms of rivalry, to deliver an acceptable image, but at a price
of mental stress...some people react to this level of stress more than
others, in the form of light headiness, dizzy, etc. Of course the
amount of viewing time is the biggest factor which determines how
pronounced any side effects are. If you view for 10 minutes, the avg.
person can deal with it.... The bottom line here, all EP's require a
centered eye, to deliver the proper rays to the eye.... IPD adjust
should be mandatory on all viewers for this reason.


As a side note, I once had an EP designed that would enable +/- 4mm eye
placement leeway while holding the MTF values and distortion at
acceptable levels throughout the full 8mm range. To accomplish this
task at 60mm fl, the lens design required the glass to be 4" in
diameter... yep, certainly can't put two of them side by side.


Anyway, this is the "clear apt." mentioned throughout these
threads..... this is a relative easy test you can perform with the
graph paper.... you will find that ideal eye centering for a doublet is
within the +/-.5mm region... certainly not the numbers being tossed
around.... BTW, this is true of all visual optics, even $2k+ Zeiss
binoculars, hence why their IPD adjust is so smooth and can adjust down
to the smallest increments..... proper IPD setting is critical in all
visual optics.... if Zeiss requires it, I think 3d world lenses need it
even more :-)


F STOP of an EP vs. Camera lens: The f stop of a lens is.... fl
divided by its apt. opening. In the case of an EP, the apt. opening is
the usable diameter which rays from the image pass through. In the
case of the Saturn lens, approx. 80mm fl / ~44mm diam, or f 1.8. The
3d world viewer is about 75mm fl / ~32mm = f 2.3. These f ratios
would hold true if the doublets were used as a camera lens.....but they
are not.


In stereo viewing the doublet becomes part of an optical system, which
now includes the eye....so both optics (doublet and eye) now represent
ONE single working lens, as it meets he definition of an optical system,
as it has a focal plane (film) and an image plane (projected image on
the retina).


When the doublet is introduced into this optical system, now the eye
pupil diameter becomes the apt. opening in the system. Therefore, the
true f stop of the Saturn lens is 80mm / ~5mm eye pupil diam. = f 16
lens. The 3d world EP is about f 15, hence why these lenses perform
very well (on-axis), as optical systems, as at these high f ratios,
aberrations are well corrected. (assuming eye pupil is on axis)
There is no way an EP / eye can be designed which will perform well at
f1.8 .... Modern optics design is not capable of designing such an
elaborate EP's.... most camera lenses at f 1.8 have 15 - 20 elements
due to the number of aberrations they must contain with lenses this
fast.


The importance of this is...... everyone wants wider fl EP designs....
and this produces a higher degree of design challenges, specially in
35mm viewers, as a very wide fl would be in the 25mm fl
range....combined with a 6mm eye pupil (hard to funnel light into small
film), it produces an f 4 lens, which is very difficult to produce with
the image quality of an f10+ optical system. At best, even at f4, it
would require 5 + element design. This is one reason MF viewers have
such a huge advantage in optical performance, as they operate in the
75/5 = f15 range. And why occasionally, a surplus achromat can
perform reasonably well, considering the lens was not designed for
stereo viewing.... of course, this is hit or miss proposition. I think
Alan got lucky with his Surplus Saturn lenses :-)


DISTORTION - Assuming a doublet lens is designed for use in a MF
stereo viewer.... the amount of distortion is a function of the fl.
(assuming equal size film) The shorter the fl, the greater the
distortion will become... IMO, the lowest you can push a doublet is in
the 70mm fl range..... and even then, expect 5 - 8% distortion. As you
can see from this plot, distortion grows out on the radius. In this
distortion plot, about 6 - 12% distortion from a 4 element design


http://tinyurl.com/5sax4r


About 5 years ago, I had a nice 70mm fl doublets that perform very well,
they are corrected for MF viewing and have about 18mm ER.... their
resolution is consistent throughout the image radius.... but to overcome
the aberrations and control distortion, the doublet is quite big....
about 23 mm thick and about 48mm in diameter IIRC. Of course, i was
not happy with them..... we all want more...


Dons previous comment of:

> IMO, optical quality, (assuming good quality control), is determined by what the achromats were designed to be used for, not by their diameter, whether large or small.



This is generally true, unfortunately, in MF doublets, the very wide diameters are what produces greater eye placement leeway component.....and with large eye pupil diameters, the wider the lens, the better in this regard, hence why 3d Worlds doublets are much more sensitive to eye placement vs. the Saturn lenses.


A few other responses....


> I had search for a lens for years that had a larger sweet spot than the 10-15mm range nd never found one that was suitable for viewing.


If you do the graph test mentioned above, you will see that 10-15mm range is not attainable in any lens, at any price, no matter how many elements.....this assumes some image quality thresholds.... of course, you can subject your visual system to anything and your brain will make the best of it. So my only point is, it certainly is not ideal, but often workable....I have had many 72mm IPD people use my Saturn and they get by with it.... as well as 59mm IPD users.... some complain after 30 minutes, but not too bad all things considered..... so this is "ideal" vs. "whats available".




> Correcting this distortion always meant more elements, and more elements meant vignetting from the barrel edges.


Yep, the only reason a lens design uses more elements, is to correct problems that still exist with less elements. However, if properly designed, vignetting will NOT be an issue, certainly not in the 40mm + fl range. With shorter fl lenses, vignetting (which is NOT necessarily image clipping) does become a problem, no different the Cosine theta fall-off in wide angle camera lenses.



> As for shorter optics, the sweet spot grew even smaller, requiring even more necessity for interocular adjustment.


Yep again..... generaly speaking, the more complex the design, the less eye placement leeway the optic offers. In optics software, we always produce MTF charts of the lenses performance with the eye on axis, .5mm off axis, 1mm off axis, etc. The results are often dismal after ~ 1mm off axis.... Here is good on axis MTF chart for a 5 element EP...

http://tinyurl.com/66elyq



To make a good EP design for a stereo viewer, you need to run these MTF values at about 4 positions from exact on-axis, (eye placement leeway, which assumes an IPD adjusting system) then run the same 5 MTF charts for 4 different eye pupil sizes, then for the entire focus range of the optic, typically about 5 diopter increments, which can yield hundreds of graphs.... then, you will get a good look at how the lens performs for the avg. person considering all the variables at play when the viewer is being used by a wide range of users. When some parameters, often focus range, force image quality to degrade below your pre-set threshold, then, you remove eliminate those diopters of focus and force the user to wear his corrective eye wear.... Often the focus MTF charts shows % modulation at the different image radius positions like this....


http://tinyurl.com/5abvb6


This is why optics design is so tedious and difficult, even with modern design software..... an "ideal" lens can take years to hone in on.....




> The "sweet Spot" is independent of lens diameter meaning that larger lenses may not have a larger "Sweet Spot" than a smaller lens but it could.


Just so we are clear, this is false....hence why good stereo EP's are so wide.... my best viewing lenses, which have +/- 1.5mm of on axis eye placement leeway (to hold a given threshold of optical performance within all the variables above, - certainly not +/-10mm) has the widest lens elements at 57mm in diameter.


EYE RELIEF (ER).... is often most overlooked component in EP design for stereo viewers....specially considering the eye MUST swivel to see the entire image area... this forces very long ER. Much longer than what binocs or telescope require, where your eye stays on-axis. When ER is too short, when your eye swivels, it will not encounter the sharp rays, and you must bob your head around to find them....not desirable for a stereo viewers.... below is a simulation of this effect in optics software....it shows ER too short for a given user, when the eye is swiveled.... you can see how the periphery ray bundles can not hit the pupil, as the pupil is not in the design location due to ER being too short...... or visa versa.... the problem being, you can not hold the area of focus AND the periphery sharp.... instinctively, we move our head so the eye is positioned to see the sharp rays where we are focused.... of course, in real world viewing, your brain will fight to overcome this, or at times, the peripheral vision of the person is so bad, they won't notice it.... but this is the effect of ER which is too short

http://tinyurl.com/6zglb2



> The surface of the pupil is not directly on the surface of the viewing lens, but at least 5mm from it.

5mm air space would be ideal (glass to cornea distance).....but since ER is almost never sufficient, the user naturally slams his eye balls onto the glass, to see the sharp rays, with no vignetting.... hence why the glass is always full of eyelash gunk :-) 3d World ER is way too short, barely 12 - 13mm.... For non eyeglass wearers, when the eye must swivel, 18mm should be a bear min.... 22mm being more ideal, then you won't gunk the lenses. Eye Glass wearers in a stereo viewer would need 25 - 27mm of ER (based on the Apparent Field of View - AFOV) of the EP design - degree of eye swivel.... this will prevent vignetting also....


> An amazing wish list has been presented:

The lenses being sharp.

The lenses being distortion free.

The lenses being cheap.

A real possibility ?


The answer is YES, assuming you will accept a very small AFOV and FOV of the film area. But that is unrealistic, as it defeats the purpose of a MF stereo viewer, as we are trying hard to simulate unaided human vision..... so therefore, the answer is, NO..... except if you want 70mm+ fl lenses :-)


The two most critical factors that dictate the complexity of the optics design is the angular field the lens must see (film) vs. its fl. An EP with a fl equal to the films diag. is relatively easy to produce, hence why an 80mm achromat is quite effective (albeit very limited AFOV and often way too short on ER)...... as you increase the AFOV of the EP (IMAX effect) the lens design complexity starts to increase exponentially, as well as the size of the elements and the weight. Here is an example of a nice SWA (Super Wide Angle) EP design for a stereo viewer, notice the widest element is 5" wide....certainly not usable with a dual eye instrument.... and 4 lbs each.


http://tinyurl.com/6p2t93



Telescope users have a huge advantage over us, they can use IMAX like EP's , as they never are worried about that second eye :-) Have a look.....3.5 lbs, 3" wide and $620 each! These are GREAT optics... and have true IMAX experience! (albeit, not for stereo viewers, as their field of view (Field Stop Diameter in astro EP lingo) is much too small....


http://tinyurl.com/6argjs


Anyway, I will go post my dream viewing systems.....

Bill G
Subject: Re: Properly designed DREAM stereo viewer?
Date: 2008-07-26 13:43:55
From: John Hart
--- In MF3D-group@yahoogroups.com, Bill G wrote:
>
> Ahhh, nothing nicer than dreaming about the ultimate stereo
> viewer...something many of us do on a regular basis....

How True! (don't know about the "nicer" aspect though ;-).

I doubt ordinary MF transparencies (e.g. like the ones in MF folios)
will stand up well to the magnification required to get to 80 or 90
degrees FOV. You will see all the flaws in the core area, which can
often be seen already with ordinary 3D world viewers.

The only photography I have seen which really shines at 90+ degrees
FOV, are large crystal archive prints (like 4' high x 10' wide) made
from 4x5's or 8x10's or digital-stitches. You can walk up/into these
and they almost look 3D (although they are not). Such prints have the
advantage of going through the digital darkroom, where most flaws can
be removed. They also sell for $40K or more.....

So what about forgetting about complex viewer optics and finding a way
to display/view large-format 3D prints? This is my holy grail of 3D.

John
Subject: Re: Properly designed DREAM stereo viewer?
Date: 2008-07-26 16:19:53
From: Bill G
Hi John
>
> I doubt ordinary MF transparencies (e.g. like the ones in MF folios)
> will stand up well to the magnification required to get to 80 or 90
> degrees FOV. You will see all the flaws in the core area, which can
> often be seen already with ordinary 3D world viewers.
>


to produce 80 deg AFOV, for the diag. of MF
film, the EP fl required, assuming very minimal distortion, would be
approx. 44mm. Magnification of a 44mm fl lens, using the 10"
convention of standardized viewing distances, is 250/44 = 5.7x.


A MF color transparency shot with high-end
lenses, will hold about 50 lp/mm of resolvable detail (lots of details
here I won't present to keep this discussion on course), so 50/5.7 = 9
lp/mm projected on the retina. As you know, 9 lp/mm is beyond what is
truly resolvable by the human eye.... key word being resolvable, not
discernible. The human eye can discern higher resolutions. With a
very strong back light to drive the eye pupil to its maximum resolving
and MTF values, any shortcomings in such a system will be a result of
either human visual shortcomings or poorly designed optics. If you
design a well corrected 44mm optic, it will deliver approx 80% of this
resolution to the retina.... this is not theory, or pie-in-the-sky, I
have designed and built these optics - often the view is sharper than
what you see in the real world, it produces drool.....


Another way to evaluate your position
is..... a magnifier (film loupe, EP, etc) itself does not provide
magnification, instead, it's designed to function as part of an optical
system which includes the eye.... and as we know, the eye is no
different than any other lens, so its nothing more than half of a
compound imaging system. If the EP stood on its own merits, it would be
classified as minifier, not a magnifier. The method in which a
magnifier produces perceived magnification, is by allowing your eye to
move-in closer to the subject (film) and correct the ray bundle angles,
so the eye can accommodate (focus) on the ray bundle. The closer
you get to a subject, the larger it appears. So if we viewed the
film at 2" with the aid of the optic, we would be viewing a 74mm diag.
at 50mm distance. This is equal to a ~ .67 ratio, of viewing distance
vs. image diag. So as a point of reference, take any print you make
from a MF tranny and view it from a distance of .67x its diag. and tell
me if the print breaks up? That is the exact same thing we are doing
with the stereo viewer with 44mm fl lenses.


As a general rule, a good tranny can stand up to 10x
enlargement and still be perceived as VERY sharp when viewed at any view
distance, usually the limiting factor being how close we can focus
without corrective eye-wear....


Heck if we could focus our eyes at 2" from the
subject, we would not need optics to produce the desired 80 deg. AFOV at
ultra sharp resolution. Unfortunately, very few of us can focus at 2",
hence the need for optics.....



> The only photography I have seen which really shines at 90+ degrees
> FOV, are large crystal archive prints (like 4' high x 10' wide) made
> from 4x5's or 8x10's or digital-stitches. You can walk up/into these
> and they almost look 3D (although they are not). Such prints have the
> advantage of going through the digital darkroom, where most flaws can
> be removed. They also sell for $40K or more.....
>
These $40k printers, (Chromiras, Light
Jets, Dursts, etc.) lay down approx. 200 dpi, or 4 lp/mm (/25/2).
Film holds about at least 10x this, so you can get a lot closer to film,
then you can a print, hence the beauty and potential of MF viewers with
well designed optics....... as it produces the best views AND in the
smallest compact systems.


Also, you are comparing front lit prints to
back lit trannies, there is no comparison, trannies win every time.
Why? Because humans perceive the most sharpness, in the 2 - 8 lp/mm
range. A high contrast image presented at at 4 lp/mm will look waaaay
sharper than a 25 lp/mm image at low to med. contrast. This is called
the Contrast Sensitivity Function, CSF. You can read a bit about it
on Korens site, scroll to bottom......

http://www.imatest.com/docs/sqf.html#csf


Anyhow, this relatively low lp/mm and HIGH contrast fits the projected
retinal image of a MF viewer quite well, using the fl's required to
produce the 80 deg AFOV. This is why MF viewers have the potential to
hit the sweet spot of human vision in every regard. It's the contrast
in this small resolution range that makes humans say WOW! Back-lits
can produce contrast levels 100x+ greater than front lit prints, as
front lit prints absorb a very high % of the light that strikes them vs
trannies. Of course, all this is dependent on a near diffraction
limited optic which meets all the other criteria.... no simple task, not
cheap either.


> So what about forgetting about complex viewer optics and finding a way
> to display/view large-format 3D prints? This is my holy grail of 3D.
>
Ahhhh, I agree, why do you think I
embarked on this obsessive compulsive journey? In the previous post,
I have showed my dream viewer. The reason it's my dream viewer, is
because it's the only possible way to produce the HOLY GRAIL you refer
to.


I wasted a year in my early dayz,
trying to produce the HOLY GRAIL of viewers using no optics, instead,
using large images, vs. small film. Without boring you with hours of
details, I can cut right to the biggest shortcoming - human eye
spacing is too short to produce any significant AFOV's to rival an
optics based system. You can accomplish a very healthy AFOV with one
eye, but the mirror closest to your eye will be so wide, it will
trespass on the other eyes restricted space, unless you desire a ONE eye
SWA viewer (Super Wide Angle).... sad, huh..... Too bad we are not
hammerhead sharks, that would solve this problem.


This is where optics provides an eloquent solution, however, it too has
limitations in AFOV, as the wider you push the AFOV the wider the optic
becomes, (as seen in the previous graphics and the telescope EP's pix)
soon the limitation is the width of the optics, as they get so wide,
they exceed IPD spacing. Hence why 80 deg. AFOV is the highest an
optic can achieve in an optics based stereo viewer, at least with the
use of spherical elements, and even then, we are talking about 5 - 8
element designs, certainly not doublets. The take-home message here
is.... a viewing system with NO optics can not come close to the total
viewing experience of an optical based viewing system. Sucks, huh.....



Considering our unaided vision can only see 90 deg FOV with ONE eye, we
almost matched our unaided FOV - not so bad. However, with two eyed
unaided vision, we see closer to 110 deg Horiz FOV, as the two eyes have
temporal periphery unique to each eye.... so when you consider this,
optics will still fall short of real world Horiz FOV. But nonetheless,
improved optics, which provide 2x the FOV vs. whats been available up
till now, is one heck of an improvement and probably as far it will ever
be pushed in consumer priced products using spherical elements.


I must be weary of the word "consumer prices"..... as I refer to this
against other high end optical products such as high end binocs, which
sell in the several thousand dollar range. What I was differentiating
from is large dome theater viewing systems which expand the FOV to match
our natural FOV, costing millions $ to construct and used only for
public viewing in amusement parks, etc.


I designed and built a large optical viewing contraption that would
view 24" square images.... I even made some back-lits on a chromira /
Fuji Trans, to simulate film.... Fuji trans uses the same dyes as used
in Fuji film, so there is potential here. What you end-up with is a
monster sized contraption, and final image quality that can NOT match a
MF viewer with equal AFOV optics.


It comes down to this.... if the HOLY GRAIL of stereo viewing is defined as:


1) maximized AFOV


2) the sharpest possible images projected on the retina


3) the ability to accommodate human IPD range


4) the ability for NON Eye Glass (EG) users and EG users to get an
equal view


5) the ability to drive the eye pupil down to a diameter which
represents its best compromise of aberrations and apt. diffraction....
(this is a significant component of the final perceived sharpness)


6) NONE, or very minimized distortion of any type, keystone, barrel,
pin cushion, rectilinear, angular, etc.


7) No perceivable chromatic aberrations


8) The ability to swivel your eye to see the entire FOV without loss
of sharpness or added distortion


9) The ability to prevent divergence, and always have the eyes viewing
infinity objects with the eye axis being parallel (no convergence or
divergence) regardless of IPD of user...


10) a relatively small and manageable sized system...


Well, there is only one possible product that can meet
all these objectives, and it's a film based optical viewing system, in
which MF film is the ideal size. Life is full of limitations.....
such is life, huh....


Bill
Subject: Re: Properly designed DREAM stereo viewer?
Date: 2008-07-27 17:48:42
From: John Hart
--- In MF3D-group@yahoogroups.com, Bill G wrote:

Hey Bill,

Thanks for your long and well-reasoned analysis. Although I have a
few questions/comments (below) you've obviously thought this out
well. I wish you luck! Love to see your optimized viewer at NSA or
ISU!!

Analysis aside, what I said, based on my real experiments, still
stands, I think. Even with 3DWorld or Don's nice hand-made viewer, I
easily see distracting flaws in most MF's, even carefully made ones.
For example, granulation rivalry (in Provia100) in sky and low
contrast uniform areas. Double the magnification, and these will
really pop out. Not being able to correct such digitally is a
disadvantage. One can, of course, do this by scanning and then
making MF film-recorder trannies. In fact, since all my high-rez
stuff is digital I go this output route anyway. Trouble is, such
output, at least what I have been able to get (via Lightjet2080,
Solitaire, and a couple other machines) seems to have a real rez
limit of around 30lpmm (as do a lot of not-quite-perfectly-made
MF's ;-). Magnifying these more than what happens in a 3DWorld
viewer does not work well, at least in my experience.

As for the Holy Grail (HG) list, the single-mirror print viewer I
designed ( www.mirscope.com ), can do, in principle, all of your list
except 10. If you decrease the central angle a bit from production
units on the website, you can easily get 80 degrees diagonal FOV (~53
horizontal FOV, or 1:1 viewing ratio (distance to width), though the
mirror gets closer to the apex so the page turning book-mode is then
no longer as practical. In fact I once ran a prototype test with
about 80 degrees horizontal field of view using both prints (up to
19" wide, via Epson3800) and backlit transparencies (8" x 10" via
Lightjet 2080). Either of these has as much information as one might
find in a MF color original. Although the immersion is great, and
everyone gets the 3D right away without adjustments, and it's open-
frame (so others can see what you are looking at, etc.), and it's
trivial optically and quite cheap, etc., the device isn't perfect for
my intended application in high-end galleries and museums (big, a bit
gawky, suffers from whacks from linebackers, etc. :-(. But for these
apps, my personal opinion is that a microscope-looking binocular
viewer would not work very well either.

Thanks, John

> So as a point of reference, take any print you make
> from a MF tranny and view it from a distance of .67x its diag. and
> tell me if the print breaks up?

Don't know about MF tranny to print, since I don't do that. But in
general a large enough print won't break up if the source material is
also up to snuff. How large at 1:1 viewing ratio. 8.5x11 is very
good, 13x19 is great.

> well designed optics....... as it produces the best views AND in
smallest compact systems.

This is a good point. What is the enclosing box size for your viewer
system. Cost and weight? Just curious.

> Also, you are comparing front lit prints to
> back lit trannies,

Sure, but one can make larger trannies from digital files if one
really really needs to. This point seems a little nit picky to me.
I have seen some rocking print presentations.... My HG is to walk up
to a big print and see it 3D! Maybe high quality stereojets will
come back some day :-), it would be interesting.

> This is called the Contrast Sensitivity Function, CSF.

Yes, I know all about this. It's why movies on a Kuro 9G HD plasma
screen look a hell of a lot better than 35mm projections in a
theatre - contrast is essentially 100% over the wow range of the
eye. It's why even the lowly 1400x1050 projections at NSA clobbered
the perceived sharpness from the 35mm film projections (not to
mention registration errors, out of focus corners, dust, dirt...).

> because it's the only possible way to produce the HOLY GRAIL you
refer to.

Well, my HG would also exclude peephole type devices. Sorry, it's
just me. They are fine for personal viewing, but I don't like them
for general-public use in galleries and museums.
Subject: Printing on transparency film using inkjet
Date: 2008-07-27 18:25:45
From: Aaron Muderick
Has anyone tried to scan their MF transparencies, correct digitally, and then print onto OHP Pictorico transparency film using a high-end inkjet?

They claim to print at 2880DPI which, if my math is correct, would be 56 lp/mm.  That should deliver 9lp/mm on a 5.6x magnification.

Thoughts?

Aaron
Subject: Re: Printing on transparency film using inkjet
Date: 2008-07-27 20:02:21
From: Don Lopp
Correction:

In the real world, 2880 dpi only equals 28 lp/mm, not 56 lp/mm.

One white dot plus one black dot equal 1 line pair.


DON
Subject: Re: Printing on transparency film using inkjet
Date: 2008-07-27 20:47:11
From: Aaron Muderick
I thought I took that into account:

2880 dot per inch = 113 dots per mm
113 dpmm / 2 = 56 lp/mm

Yes? No?

Aaron

Don Lopp wrote:



Correction:

In the real world, 2880 dpi only equals 28 lp/mm, not 56 lp/mm.

One white dot plus one black dot equal 1 line pair.

DON

Subject: Re: Printing on transparency film using inkjet
Date: 2008-07-27 21:23:07
From: Bill G
I thought I took that into account:
>
> 2880 dot per inch = 113 dots per mm
> 113 dpmm / 2 = 56 lp/mm
> Yes? No?
>


Yes Aaron, you got it right, to convert dpi to lp/mm, just divide dpi
by ~50 (25.4*2)
Subject: Re: Printing on transparency film using inkjet
Date: 2008-07-27 21:23:07
From: Bill G
> Has anyone tried to scan their MF transparencies, correct digitally,
> and then print onto OHP Pictorico transparency film using a high-end
> inkjet?
>
> They claim to print at 2880DPI which, if my math is correct, would be
> 56 lp/mm. That should deliver 9lp/mm on a 5.6x magnification.

What is the purpose, to view a 2d
print on the wall?
Subject: Re: Properly designed DREAM stereo viewer?
Date: 2008-07-27 21:23:07
From: Bill G
Hi John

> Thanks for your long and well-reasoned analysis. Although I have a
> few questions/comments (below) you've obviously thought this out
> well. I wish you luck! Love to see your optimized viewer at NSA or
> ISU!!
>

I have optics, opto-mechanicals, but not a
complete stereo viewer.... It's expensive enough to design / build all
the optics and optical mechanicals.... .. :-) Hence the beauty of a
simple doublet. For personal use, my prototype loupes generates the
views, but it's a cumbersome process..... some day, I might consider
making a viewer....




> Not being able to correct such digitally is a disadvantage. One can, of course, do this by scanning and then making MF film-recorder trannies.

yes, I fully agree with you.... digital is a
much desired intermediate step to produce stereo views of any kind, film
is not perfect, subjects are not always as cooperative as we would
like. I would not consider it 100% mandatory, as every so often, we
can get some real winners straight from the camera....but digital
capture to viewing media, or film-digital-to viewing media is a very
desirable work flow. I am considering a few ideas for this now.....
but surely nothing consumer-friendly... :-( I think digital stereo
capture will become the wave of the future, but there needs to be some
good display methods, possibly a viewer like your Mirscope.



As for seeing rivalry quirks in the
trannies, even with 3d world viewer.... there is a lot of variables at
play here. It's possible you have exceptional vision and notice
certain grain rivalry that others do not..... or maybe it's rivalry from
the camera lenses, or the film processing, or possibly an aberrated
viewing lenses. For $39, I doubt the 3d World viewer lenses have tight
quality controls :-) I am not questioning your observation, just
pointing out all the links in the chain.... in my experience, the film
itself, is one of the more consistent links... and often when it suffers
certain oddities, they are "usually" the consistent in the l/r
views,...... but I agree, film is not perfect.


Using my Mamiya 7 lenses and my Digitar lenses, combined with optics I
mentioned previously, I have NEVER yet seen what you describe. But I
fully concur that as magnification increases, all errors are magnified
as well. With a pair of 25mm fl lenses I have (10x magnification),
with only a 4 element design, I have viewed these chromes in stereo and
remarkably, they still look excellent, of course, not quite as good as
the 44mm fl's, but still damn good. The 25mm fl lenses are not WA,
hence their remarkable sharpness, but they don't make great WA stereo
lenses. Then again, I have many off-the-shelf brand name loupes that
are horrendous.... film loupe design never benefited from the advent of
optical software era, as conversion to digital capture killed the market
for film loupes as well....



> In fact, since all my high-rez
> stuff is digital I go this output route anyway. Trouble is, such
> output, at least what I have been able to get (via Lightjet2080,
> Solitaire, and a couple other machines) seems to have a real rez
> limit of around 30lpmm (as do a lot of not-quite-perfectly-made
> MF's ;-). Magnifying these more than what happens in a 3DWorld
> viewer does not work well, at least in my experience.
>

Agreed, the digital-to-film
industry died too quickly for products to mature to the level we desire.
However, there is some options here too, but all of them involve
obsoleted equipment, darkroom tricks, etc. Certainly nothing
user-friendly. Also remember, to transfer 30 lp/mm targets with min.
losses, you need a printer (film recorder) that can print at 3x (6x even
better) the originals resolution, otherwise, you suffer 2nd generation
degredation. I would consider many good MF trannies hold closer to 50
lp/mm, at least with Velvia, making the process even harder. And yes,
if you double the magnification of the viewing lens, you just blew 95%
of these options off the table. Right now, I only view film which came
direct from the camera.



> As for the Holy Grail (HG) list, the single-mirror print viewer I
> designed ( www.mirscope.com ), can do, in principle, all of your list
> except 10. If you decrease the central angle a bit from production
> units on the website, you can easily get 80 degrees diagonal FOV (~53
> horizontal FOV, or 1:1 viewing ratio (distance to width), though the
> mirror gets closer to the apex so the page turning book-mode is then
> no longer as practical. In fact I once ran a prototype test with
> about 80 degrees horizontal field of view using both prints (up to
> 19" wide, via Epson3800) and backlit transparencies (8" x 10" via
> Lightjet 2080). Either of these has as much information as one might
> find in a MF color original.

This is very interesting John....
First, I went and checked out your web site, you did a fabulous job on
the web site and the MirScope. It has been quite some time since I
remember this product being announced. Kudos to you. Has this
mirrored viewing system gained any legs from stereo enthusiast? It has
a lot of advantages, as digital stereo capture is the wave of the
future, and your Mirscope compliments this market nicely. It allow for
the avg. consumer to convert stereo images into a nice viewable format,
as the digital-to-print process benefits from mainstream technology
advances. The price of all the components, including your Mirscope is
very appealing also. I would imagine a back lit version would really
increase the appeal, albeit, you loose the book approach, which is
probably a big selling point.



As for the 80 deg HFOV you suggest with
19" wide prints. Working with the left eye only, as it views the print
directly, you would have to view the print at a 11" view distance,
right? This brings back horror stories in my non optical viewer
attempts....cause not many of us in the 45+ crowd can focus this
close....now, we are forced to use corrective eye wear. Herein lies
the problem, as reading glasses are not designed for the eye to swivel
in such extremes angles. They are designed to have the subject of
interest in the center of the optic.


Their also is DOF issues at such extreme angles. With the 19" wide
images, the edges are 3.5" further from the cornea vs. the center of the
image.... and the corners are even further. As you know reading
glasses in the 1.75 - 2.0 diopter range have distinct planes of sharp
focus. I can see this being an issue. In my previous attempts, I
purposely made the prints further away from the eye to avoid this close
focus issue, which is how I always ended up with ridiculous sized
contraptions. So when I presented my "issues" with non-optical viewing
devices in the previous post, I should have been more clear as to the
limitations I set. I am sure you have considered the close-focus
issue, how do you deal with it?


Also, is the right image printed
normal? Or is it printed with an anti-keystone pre-set, to offset
keystone distortion?


I also fully agree, you
can gather all the resolution of a MF tranny and bring it to, a back lit
print in a viewer like this..... assuming the original capture method
was sufficient. I am with ya on the NON optical viewer, as they
certainly have a lot of potential benefits.... physical size not being
one of them :-( Our only difference of opinion is on the
perceived resolution issue. Mainly due to the eye itself... front lit
prints can't drive the pupil diam. small enough, and front lit can't
generate the contrast as a back-lit can, but you can solve this by using
back-lits, so it's not a limitation. All things considered, the
Mirscope seems to deliver tremendous price / performance ratio as many
of your satisfied customers mention on your web site. Very nice job
John....




> Although the immersion is great, and
> everyone gets the 3D right away without adjustments, and it's open-
> frame (so others can see what you are looking at, etc.), and it's
> trivial optically and quite cheap, etc., the device isn't perfect for
> my intended application in high-end galleries and museums (big, a bit
> gawky, suffers from whacks from linebackers, etc. :-(. But for these
> apps, my personal opinion is that a microscope-looking binocular
> viewer would not work very well either.
>
Hence, my basic premise - a
film based MF optical viewer, solves more problems and brings the max.
possible visual experience vs. any other method I have found.
Granted, the MF viewer I refer to has never been a salable product, but
all the bits n pieces exist for its creation. However, designing /
building them is no simple task, as the there is a lot of design
expertise required with so many elements of glass, as tolerancing and
alignment issues become ultra critical.


And I also agree, assuming the non optical system
could come close to matching the film based MF film viewer, the size of
the images make the entire contraption too big for most
applications.... and then, you have the "how to change views" issue,
which becomes complex and cumbersome with large images.... Small
film solves this problem via magnification at time-of-use....


>
> This is a good point. What is the enclosing box size for your viewer system. Cost and weight? Just curious.
>
As I mentioned above, I only have the
optics and the opto-mechanicals, so i view on a high powered light box,
a bit cumbersome indeed.... But if a viewer was to be made, it would
be just a bit larger than a Saturn, maybe 4" longer for extended light
box.... maybe 2" wider. Design / build costs for a working prototype
would cost as much as a house :-( Production unit costing would be
very dependent on volume... in low volume, under 100 units, approx. $4k
- $5k each is my best guess.... this does not account for amortized
design / prototype costs.



> My HG is to walk up to a big print and see it 3D! Maybe high quality stereojets will
> come back some day :-), it would be interesting.
>
Well, this too is a problem with precision
optical instruments, you have to set IPD and focus, no different than
binoculars. A $4k pair of binocs are useless, unless they are set to
the particulars of your vision.......it's not as plug n play as a non
optical system, so clearly there are trade-offs. Even if Stereo Jets
came back, can you imagine the physical space 50 images at 30x40" would
take up in a gallery? IMO, this is the benefit creating AFOV through
magnification of film.


>
> Well, my HG would also exclude peephole type devices. Sorry, it's just me. They are fine for personal viewing, but I don't like them for general-public use in galleries and museums.
>
What is peephole type devices? Like viewing
lenses on the wall?



> As for the Holy Grail (HG) list, the single-mirror print viewer I designed (www.mirscope.com ), can do, in principle, all of your list except 10.


From a price / performance standpoint, your Mirscope blows away any high-end precision optical viewer, I can say that without reservation. But like many technology thresholds, gains are often measured in the 10% - 30% range, and the costs to achieve those gains is often in the 500% - 2000% range. Optics clearly falls into this category, hence why you can buy a pair of $20 binoculars and a pair of $2k binocs, both, very similar in the task they perform. You are paying for small gains in optical performance and build quality. Reminds me of race car engines, the last 10 hp costs a few million $ to achieve :-)


With that said, I would like to comment on the list of HG items.. First, for a non optics viewer, we must consider the limits of close viewing by an avg person. If we add optics, such as reading glasses, we introduce new problems....so assuming a long enough focus distance to achieve the true plug-n-play benefits for an avg. person, then the non optics viewer will have a much restricted FOV, vs. what is possible with an optics based viewer (not whats available). And IMO, AFOV is a big part of the HG. When you add the "size" issue, for both the viewer / media, it also creates huge gap in the comparison. So although I do agree your viewer accomplishes many of the HG items quite eloquently, the comparison becomes moot, as we are comparing apples and oranges - hey, they are both fruits :-).....


More importantly, your non optics viewer solves a lot of the shortcomings of the 3d world viewer, as it relates to focus, IPD and lighting, digital work flow, etc.

Just one mans opinion.....

Bill
bglick@rconnects.com
Subject: Re: Printing on transparency film using inkjet
Date: 2008-07-27 21:30:10
From: Michael Kersenbrock
These calculations all assume that every color in the image (both in terms
of color and shading of that color) are exactly represented by the inks
in that image.  So if you assume black ink for the black lack of dot for
the white then you might be in the ball park for a line-drawing's image
(meaning NO gray scale, purely solid white or black).

For the typical image that's not 100% that of the half-dozen
ink colors (with single uniform brightness) the printer needs
to dither multiple colored dots and "non-dots" (bare paper "white")
to produce the array of colors and shades that one might expect.
Much like what's Seurat did in his painting (although he may have
had more "ink" colors).

In other words it takes a LARGE area (compared to dot-size)
to represent a full color and shaded image-pixel.  In other
words the actual image resolution is tremendously less than
either of the values that have been presented, and as such it's
not worth bothering to argue which is right because both aren't
close to being so.

Unless perhaps, if the 2880 dpi printer is a dyesub printer.

Mike K.


Aaron Muderick wrote:
I thought I took that into account:

2880 dot per inch = 113 dots per mm
113 dpmm / 2 = 56 lp/mm

Yes? No?

Aaron

Don Lopp wrote:



Correction:

In the real world, 2880 dpi only equals 28 lp/mm, not 56 lp/mm.

One white dot plus one black dot equal 1 line pair.

DON


Subject: Re: Printing on transparency film using inkjet
Date: 2008-07-27 22:38:11
From: Aaron Muderick
The purpose is to create duplicate transparencies with a digital intermediate step.  If I can avoid having to pickup a Polaroid 8000 film recorder and instead can use an inkjet printer, that would be great.

We'll see what results the experiment brings.

Aaron

Bill G wrote:


> Has anyone tried to scan their MF transparencies, correct digitally,
> and then print onto OHP Pictorico transparency film using a high-end
> inkjet?
>
> They claim to print at 2880DPI which, if my math is correct, would be
> 56 lp/mm. That should deliver 9lp/mm on a 5.6x magnification.

What is the purpose, to view a 2d
print on the wall?

Subject: Re: Printing on transparency film using inkjet
Date: 2008-07-27 22:38:25
From: Aaron Muderick
Very good point Michael.  Nonetheless, I think I will give it a try with a friend's printer.  I think you are right and that the results will look pretty bad.  But, until you experiment, you never know ;-) Stay posted.

Aaron

Michael Kersenbrock wrote:

These calculations all assume that every color in the image (both in terms
of color and shading of that color) are exactly represented by the inks
in that image.  So if you assume black ink for the black lack of dot for
the white then you might be in the ball park for a line-drawing' s image
(meaning NO gray scale, purely solid white or black).

For the typical image that's not 100% that of the half-dozen
ink colors (with single uniform brightness) the printer needs
to dither multiple colored dots and "non-dots" (bare paper "white")
to produce the array of colors and shades that one might expect.
Much like what's Seurat did in his painting (although he may have
had more "ink" colors).

In other words it takes a LARGE area (compared to dot-size)
to represent a full color and shaded image-pixel.  In other
words the actual image resolution is tremendously less than
either of the values that have been presented, and as such it's
not worth bothering to argue which is right because both aren't
close to being so.

Unless perhaps, if the 2880 dpi printer is a dyesub printer.

Mike K.


Aaron Muderick wrote:

I thought I took that into account:

2880 dot per inch = 113 dots per mm
113 dpmm / 2 = 56 lp/mm

Yes? No?

Aaron

Don Lopp wrote:



Correction:

In the real world, 2880 dpi only equals 28 lp/mm, not 56 lp/mm.

One white dot plus one black dot equal 1 line pair.

DON


Subject: Re: Properly designed DREAM stereo viewer?
Date: 2008-07-28 07:58:15
From: John Hart
--- In MF3D-group@yahoogroups.com, Bill G wrote:
>
> (re www.mirscope.com) Has this mirrored viewing system gained any
legs from stereo enthusiast?

I think most people who bought one like it. I have heard reports of
many successful uses in shows, galleries and stuff (like the 3D
Center in Portland - to whom I donated one), where lots of people
walk off the street and see good stereo right away. Surprisingly,
to me, reports are they don't need re-adjustment very often.

I should add that being in a business gets really old fast!!! I
have a few left from the original production run, and when they are
gone.........

> As for the 80 deg HFOV you suggest with
> 19" wide prints. Working with the left eye only, as it views the
> print directly, you would have to view the print at a 11" view
distance, right?

Well 80 HFOV is pushing it, I agree. But 80Diag (or 53 HFOV, say)
is entirely within the realm-of-reasonable for the 19" viewer (e.g.
19" viewing distance). I admit these are monsters - I only built 6
of them, but they work quite well for public display. Ron Labbe
mounted LCD panels on one and plays movies. I have used backlit
trannies in one, etc. Books are great, and 19" prints off the
Epson3800 are very good.

> Also, is the right image printed
> normal? Or is it printed with an anti-keystone pre-set, to offset
> keystone distortion?

The nice thing is that there is zero keystone distortion. I did a
mathematical proof of this once, but it's easier just to run your
eyes around a normal print pair (with right image horiz flipped of
course) and see that all the edges line up exactly.

> and then, you have the "how to change views" issue,
> which becomes complex and cumbersome with large images....

In the "book" form, it's pretty easy. For gallery display I just
leave single prints in several viewers.

> Even if Stereo Jets came back, can you imagine the physical space
50 images at 30x40" would take up in a gallery? IMO, this is the
benefit creating AFOV through magnification of film.

Not many artists hang 50 images in a gallery show. Maybe ~10, but
hey, if they are really really good, they'll fill the available
space. The stereojets I had made at San Fran Imaging about 10 years
ago were not very good (low contrast, poor color saturation - but
good 3D). Since then I have seen quite good backlit ones made by
Ron Labbe and Peter Sinclair. One can view these well at 53 deg
HFOV (e.g. 16" or 20" viewing distance). There is some talk of
reviving them with circular polarization (so head tilt is not
critical). Personally, I am quite interested in this.

Would one propose an exhibit with 50 optical viewers? Would people
pay $20K for one with a small set of images (dupes?)? What will
people pay for large high-qual stereojet backlit stereo art?
(Retorical questions.. :-)

> What is peephole type devices? Like viewing
> lenses on the wall?

Or on a stand. Anything that's so personal, nobody else can see
what's in it while it's being used. (my def).

>then the non optics viewer will have a much restricted FOV, vs.
what is possible with an optics based viewer (not whats available).
And IMO, AFOV is a big part of the HG.

I agree, but 53 HFOV (80 diag for square format), is not that
difficult. Like you say it comes down to size and presentation.
Personally, I really like size and open-frame (like a print on the
wall). I have seen some 4'x10' 2D prints of the grand canyon that I
would pay the asking price of $40K for (if I had the money ;-) ;-
) ;-). But even the same scene 3D in a bino-head optical viewer
wouldn't motivate me much. It just wouldn't do the same thing for
my home, etc. If one could do 30"x40", say, in stereojet..... That
might be my choice of fruit.

Finally, if I had a good and easy way to make "original quality"
MF's from my digital files, I would probably invest in a high mag
personal viewer in order for me to look at them at the max.

Keep up the good and interesting work,

John
Subject: Re: Properly designed DREAM stereo viewer?
Date: 2008-07-28 10:37:41
From: Bill G
Hi John

> I think most people who bought one like it. I have heard reports of
> many successful uses in shows, galleries and stuff (like the 3D
> Center in Portland - to whom I donated one), where lots of people
> walk off the street and see good stereo right away. Surprisingly,
> to me, reports are they don't need re-adjustment very often.
>
Interesting, thanks for sharing.... Kudos
to you, it must be gratifying..... I do like the plug-n-play aspect of
your viewer.... from the pix, it looks very intuitive, which is great
for the applications you discuss....I also think the turn-the-page
concept keeps the user focused on the task at hand.....vs. ONE IN, ONE
OUT, which tends to distract people, they must change eye focus
positions often, etc.



> I should add that being in a business gets really old fast!!! I
> have a few left from the original production run, and when they are
> gone.........
>
heee heee heee....... Isn't that the truth,
maybe this is why so few stereo products are ever brought to market.....

>
> Not many artists hang 50 images in a gallery show. Maybe ~10, but
> hey, if they are really really good, they'll fill the available
> space. The stereojets I had made at San Fran Imaging about 10 years
> ago were not very good (low contrast, poor color saturation - but
> good 3D). Since then I have seen quite good backlit ones made by
> Ron Labbe and Peter Sinclair. One can view these well at 53 deg
> HFOV (e.g. 16" or 20" viewing distance). There is some talk of
> reviving them with circular polarization (so head tilt is not
> critical). Personally, I am quite interested in this.
>
I do agree, a good stereo printing method is
what the market lacks.... I think the litho (holography) field is making
great progress in this area...soon we will see some new products as they
continue to refine the lens on the optical panels. The are close now
to making them smaller than we have ever seen, increasing the effective
resolution. However, there will always be certain shortcomings of
these displays...... the biggest one being, IMO, the closest viewing
distance is always limited to the image diag. So AFOV will never
exceed 50 degrees.... This is due to the angles the L/R lenses
project.... being too close produces l/r interference.


> Would one propose an exhibit with 50 optical viewers? Would people
> pay $20K for one with a small set of images (dupes?)? What will
> people pay for large high-qual stereojet backlit stereo art?
> (Retorical questions.. :-)
>
Great questions..... in general, IMO, the
market would be small for any of these high end products...hence the
reason there is little incentive to invest heavily into their R&D and
production....




> Or on a stand. Anything that's so personal, nobody else can see
> what's in it while it's being used. (my def).
>
Ahhh, gotcha, sorry I missed that.... For
serious WOW, IMO, only personal viewers will produce the levels of WOW
that excites people today. (Again, not including multi million dollar
type applications) I doubt in our life time, a method of stereo wall
display will ever match what a personal viewer can deliver.....sad
huh.... because I support your wishes. It's interesting, the first
thing many people say when they see good stereo is, can you print these
and see them on the wall..... it's natural for even novices to desire
this.... bummer.
>
>
> I agree, but 53 HFOV (80 diag for square format), is not that
> difficult.
And I would consider this impressive, as a
current 3dWorld viewer only produces about 38 deg. HFOV. 35mm
viewers, much less of course. And since your product allows the users
to wear their corrective eye wear, and accommodates all IPD's, I think
it represents a very nice mix of WOW, optical accommodation and
ease-of-use.....vs. anything else that is available today.




> Like you say it comes down to size and presentation.
> Personally, I really like size and open-frame (like a print on the
> wall). I have seen some 4'x10' 2D prints of the grand canyon that I
> would pay the asking price of $40K for (if I had the money ;-) ;-
> ) ;-). But even the same scene 3D in a bino-head optical viewer
> wouldn't motivate me much. It just wouldn't do the same thing for
> my home, etc. If one could do 30"x40", say, in stereojet..... That
> might be my choice of fruit.
>
My thoughts are the same....I have sold
several 10ft - 20 ft long prints, (shot on 8x10 cameras and rotational
cameras) and while many people ooooh and ahhhh at their size and
immersive effect....it still does not match peoples reactions when I
show them stereo views with my jury rigged system. I think the
perceived resolution, the beauty of back-lit film and powerful AFOV
combined with stereo effect, .... it overwhelms our visual senses.



> Finally, if I had a good and easy way to make "original quality"
> MF's from my digital files, I would probably invest in a high mag
> personal viewer in order for me to look at them at the max.
>
Agreed again (sorry for being
repetitive)....but this is a huge shortcoming for film viewers....the
ability to get an image back to a high quality film image at a
reasonable price...... is something that has always bothered many MF
stereo shooters. I know I can do it, but the amount of time, energy,
equipment and R&D to perfect it.....argggggggg......


Bill