Google cardboard impressions and lens math

This is for discussion and development of non-commercial open source VR/AR projects (e.g. Kickstarter applicable, etc). Contact MTBS admins at customerservice@mtbs3d.com if you are unsure if your efforts qualify.
Post Reply
Heffle
One Eyed Hopeful
Posts: 5
Joined: Tue Jul 29, 2014 12:59 pm

Google cardboard impressions and lens math

Post by Heffle »

So I just registered to post this.

To build Google Cardboard, I had bought a pair of double convex lenses from surplusshed with a focal length of 45 mm like Google suggested, but I picked one with larger diameter to get more FOV in theory.

Now that I have it all built, I can actually see the edges of the screen and the cardboard itself with those lenses. In addition, the effect is cool, but not very immersive. Definitely not what someone like me who has not experienced virtual reality before was expecting. There was absolutely no feeling of "presence" at any point in time. I made some calculations with this equation (from wikipedia):


a=2*arctan(d/(2f)) where a=angle of view in degrees, d=dimension of object, f=focal length


Since I was using an S5 as the display, d=64.77 mm for the horizontal dimension of one side, so I got 71.5 degrees FOV for each eye, theoretically, and if I'm interpreting the equation correctly. That seems perhaps a bit narrow, so maybe I'm wrong in my math. Please confirm anyway.

Now, before I even tried figuring out stuff mathematically, I tried stacking a few reading glasses together and the magnification improved, at the cost of lower image quality and slightly smaller FOV. However, what that did for the feeling of presence was immense. Sometimes, but not at all often, I could actually feel like objects were within reach and could be touched. In addition, I got slightly motion sick and slightly unbalanced after playing a few demos for a while, compared to before where it never happened. It's still no where near what I want though.

So now I'm making more calculations to come up with an optimal design. I am going to use the above equation, and two more (also from wikipedia):


M=(0.25*p)+1 where M=magnification, p=diopter strength, 0.25 is the approximate nearpoint in m

M=0.25/f where 25 is in m, f=focal length in m


And I get some results. I need focal length 25-29 mm for 104.67-96.31 degree FOV or 10x-8.62x magnification or 36-30.48 diopters. 18.2 diopters is 45 mm focal length biconvex lens so I need to add 17.8-12.28 diopters or 5.45-3.07x magnification or 45.87-81.43 mm focal length. I am also assuming that stacking lenses right next to each other will add their diopters.

All of these calculations and results seem to make sense and are intuitive in my opinion, but I'm not entirely sure everything I have done is correct. If I am correct, I could either order lenses to my specification needs, or find some to stack on, whichever is cheaper/easier. I'm guessing I could just get a pair of those 5x loupe magnifiers. What do you guys think?
budda
Sharp Eyed Eagle!
Posts: 388
Joined: Thu May 22, 2008 9:13 pm

Re: Google cardboard impressions and lens math

Post by budda »

Hi,

This formula you found on wikipedia is a good basic formula:
a=2*arctan(d/(2f)) where a=angle of view in degrees, d=dimension of object, f=focal length
Its really the same thing as:

a = 2*arctan( (d/2) / f) where a=angle of view in degrees, d=width of display, f=focal length


This formula is based on simple geometric optics and takes no account of distortion which can magnify or minify the field of view.

The other magnification formula serves no useful purpose as far as I know.

Thanks.
Heffle
One Eyed Hopeful
Posts: 5
Joined: Tue Jul 29, 2014 12:59 pm

Re: Google cardboard impressions and lens math

Post by Heffle »

What type of distortion could affect the FOV? I know of barrel distortion, but does that really change the FOV that much, even when it is corrected in software?
User avatar
Fredz
Petrif-Eyed
Posts: 2255
Joined: Sat Jan 09, 2010 2:06 pm
Location: Perpignan, France
Contact:

Re: Google cardboard impressions and lens math

Post by Fredz »

Heffle wrote:What type of distortion could affect the FOV? I know of barrel distortion, but does that really change the FOV that much, even when it is corrected in software?
It's more complicated than this. You need aspheric lenses to correct for spherical aberration and you need to put them at the focal distance to have the light collimated (focus at infinity).

The display of the Galaxy S5 is too small (5.1" diagonal) to get a good FOV. The DK2 uses high magnification lenses (10X I think) to get a nice FOV (~100° horizontal), but with a 5.7" display.

To calculate the FOV correctly you need to write a ray-tracer or use an optical software (like Zemax or Oslo) that can handle aspheric lenses.

Something like this :

Image
Heffle
One Eyed Hopeful
Posts: 5
Joined: Tue Jul 29, 2014 12:59 pm

Re: Google cardboard impressions and lens math

Post by Heffle »

OK, so it is quite complicated. Well, I don't have to be exact on the sort of design I want, but it would certainly be interesting to find the most optimal design that might fit the display on the S5. Actually, I had some suspicions about this before, but I measured the length of the display where things were actually being lighted, and found it to be 11.5 cm, which is less than the supposed 5.1". I guess this means either the view is clipped on the left and right edges, or pixels are wasted. I really just want something on par with the DK1 in terms of immersion, so I think I would make a trade off this time for FOV instead of pixels, as the screendoor will be noticeable, but not as much as on the DK1. In the end, I can't fix the obvious latency built in with the smartphone either, so this might be the best.

I also have another idea. What if I used a mirror so that the images are slightly displaced from the center to fit the average IPD? I imagine what that might look like would be the S5 placed either on the top or bottom of the box, tilted towards the mirror, which is actually split in half and tilted both slightly outwards and in towards the S5 so that the image is directed accurately towards the lenses. The only problem with this might be that the lenses will focus unevenly on the mirror, but I'm not sure since I'm not entirely certain that's how light reflection works.
User avatar
Fredz
Petrif-Eyed
Posts: 2255
Joined: Sat Jan 09, 2010 2:06 pm
Location: Perpignan, France
Contact:

Re: Google cardboard impressions and lens math

Post by Fredz »

Heffle wrote:I measured the length of the display where things were actually being lighted, and found it to be 11.5 cm, which is less than the supposed 5.1".
Actually it's a bit better than expected, a 5.1" diagonal gives a 11.3 cm width with this calculator : http://www.prinds.com/tools/screenDimensions.htm
Heffle wrote:I really just want something on par with the DK1 in terms of immersion, so I think I would make a trade off this time for FOV instead of pixels
Without additional optics I guess you won't have a choice anyway. I've bought 10X lenses to try them with my S4 (5" diagonal, 11.1 cm width), I'll report back the results when I get them.
Heffle wrote:In the end, I can't fix the obvious latency built in with the smartphone either, so this might be the best.
I think the latency can still be reduced on some smartphones.

Sensors have a slower update rate than the Rift (1000Hz) but not THAT big. My S4 has a 100 Hz update rate and the Nexus 5 200 Hz for example (discussed here).

That would give an added latency of 10 ms or 5 ms, still higher than the Rift's 1 ms but probably manageable if the rest is not too slow. You can get the refresh rate of the sensors for your S5 with this app : https://sites.google.com/site/dantasse/ ... erTest.apk

Also, although most phones use tiler GPUs which add a frame of latency, there are hybrid GPUs on some phones (like the Galaxy S4, I don't know for the S5) that allow direct rendering without added latency. I've put a page with some infos about tiler GPUs here : http://vrwiki.wikispaces.com/Tiled+rendering
Heffle wrote:What if I used a mirror so that the images are slightly displaced from the center to fit the average IPD? I imagine what that might look like would be the S5 placed either on the top or bottom of the box, tilted towards the mirror, which is actually split in half and tilted both slightly outwards and in towards the S5 so that the image is directed accurately towards the lenses. The only problem with this might be that the lenses will focus unevenly on the mirror, but I'm not sure since I'm not entirely certain that's how light reflection works.
I guess it could be an idea, yes.

With two sets of mirrors (first surface preferably to avoid ghost reflections) you could also try a layout similar to beam splitters like it's done in stereoscopic photography. Could be feasible with stick mirrors for small sizes.

I think using acrylic prisms could also be helpful to extend the FOV on the sides, but you loose clarity when you add optical elements.

you can also try OpticalRayTracer if you want to test some optical layouts, a bit limited but much less expensive (free) than pro softwares : http://arachnoid.com/OpticalRayTracer/
Heffle
One Eyed Hopeful
Posts: 5
Joined: Tue Jul 29, 2014 12:59 pm

Re: Google cardboard impressions and lens math

Post by Heffle »

Wow, that's a lot of information. It's more than I could probably handle given my expertise with computer science, which is next to none.

At this point in time, I'm not really worrying that much about latency, but more about the pure image that you're getting, as I feel that may be more important to the feel of immersion. The apps currently available in my opinion are not all finely tuned so that the latency isn't noticeable, but on the best ones they are adequate as long as you don't move your head very fast.

As far as getting the displacement of the image in the right position, I've been rolling around the idea of using mirrors or prisms in my head for the past day. I have access to mirrors, actually a lot of them, but I don't have the tools to cut them, or the expertise to design and build a platform to place them on so they are accurately positioned at the right angle measures. That also kind of goes for prisms too, though I've thought of something slightly different that I may just be able to do given my current situation. I have these thin pieces of transparent plastic that looks almost exactly like glass. I could design a prism that uses the plastic as the case, and water as the actual prism to refract the light.

In fact, I've done that already and came up with an approximate angle measure of 8.7 degrees from the plane parallel to the screen of the smartphone when it is 4.13 cm away from the center of the lenses. This assumes an index of refraction for water to be 1.33 and air to be 1, which is why it's approximate, and I do know that the different wavelengths of light will bend more or less. What I don't know is what other sorts of distortion could arise from this, other than simple distortions caused by imperfections in the material (plastic and water). It couldn't hurt to try this out so I'll be experimenting today or tomorrow on this.

Finally, I would like to address the issue of FOV. If you magnify 10x or more, you get somewhere around 100 degrees horizontally assuming the Note 3 display size. However, you would have to have a setting in the app itself for about 100 degrees wouldn't you? With an in-app setting of less than 100 degrees FOV, you would essentially be viewing an over-magnified scene, so it'll feel like everything looks too large, or as if you're looking through actual binoculars in real life, although probably to a lesser extent. I think it would be good if app developers considered this type of variance in HMD design and accounted for it by giving you options in the app to change the settings of the FOV. To be able to change the amount of distortion correction and IPD adjustment would be wonderful as well, but so far only one app I know of has hit on giving you only one of these options.
User avatar
Fredz
Petrif-Eyed
Posts: 2255
Joined: Sat Jan 09, 2010 2:06 pm
Location: Perpignan, France
Contact:

Re: Google cardboard impressions and lens math

Post by Fredz »

Heffle wrote:At this point in time, I'm not really worrying that much about latency, but more about the pure image that you're getting, as I feel that may be more important to the feel of immersion.
I'm a bit biased because I think latency (and the lack of positional tracking) is what gave me simulator sickness with the DK1 (to which I'm very sensitive), whereas I only experience this during fast accelerations with the DK2.

I also think that minimizing latency helps to make the world feel more "solid" and is a very important factor for immersion/presence. But as you said it's necessary to generate a correct image in the first place for this to be effective.
Heffle wrote:I have access to mirrors, actually a lot of them, but I don't have the tools to cut them
I've played a bit with glass cutters to build an anamorphic prism for a projector (that I didn't finish in the end) and it's not really complicated to use nor expensive (less than $5). You'll also need some oil for lubrication that is cheap as well (I think it's called kerosene in English).

You can also probably find stick mirrors than can be easier to cut (basically plastic) but I didn't try that yet.
Heffle wrote:I could design a prism that uses the plastic as the case, and water as the actual prism to refract the light.
For my anamorphic prism I've used glycerol (also called glycerin) which has a bigger refractive index than water (1.4746) and quite close to glass, doesn't evaporate as much and doesn't create mold in the prism. It's also quite cheap and can be found at a local pharmacist. You can look for fluid based DIY prisms in this forum or Google to find more info about that.
Heffle wrote:What I don't know is what other sorts of distortion could arise from this, other than simple distortions caused by imperfections in the material (plastic and water).
I've read a bit about this in the past months to write my ray-tracer and from what I think I understood the main aberrations that are corrected for lenses are the five Seidel third-order aberrations :
- spherical aberration : taken care of by using aspheric lenses instead of spherical lenses (elliptic profile)
- coma and astigmatism : I think it's taken care of by the aspheric lenses as well (higher order terms)
- distortion : taken care of by the de-warping algorithm
- curvature of field : I don't think it's taken care of by the DK1 lenses (the edges are blurrier) but IIRC I've read it's the case in the DK2 (I can confirm that the edges are in focus), but I don't know how they did it though.

And lastly there are two types of chromatic aberrations (longitudinal and transverse) which give a total of seven optical aberrations. I think both of these can be corrected in software by modifying the pre-warping depending on the R, G and B wavelengths. It's not perfect as can be seen in the DK2 with higher magnification but I think it's very acceptable for now.
Heffle wrote:It couldn't hurt to try this out so I'll be experimenting today or tomorrow on this.
Nice! Keep us posted.
Heffle wrote:However, you would have to have a setting in the app itself for about 100 degrees wouldn't you?
Yes, the de-warping distortion algorithm should be parametrable depending on the eye relief (distance between the lens and the eye) and the IPD to get orthosteroscopy (ie. 1:1 view of the virtual world).
Heffle wrote:I think it would be good if app developers considered this type of variance in HMD design and accounted for it by giving you options in the app to change the settings of the FOV.
I think the FOV should be calculated depending on measurable parameters (eye relief, IPD, lens focal length) like it's the case in the Oculus SDK, the FOV value is not enough. It's a bit unfortunate that it's not the case in the Google Carboard app since it'll give a bad reproduction of the scene for most people.
User avatar
Fredz
Petrif-Eyed
Posts: 2255
Joined: Sat Jan 09, 2010 2:06 pm
Location: Perpignan, France
Contact:

Re: Google cardboard impressions and lens math

Post by Fredz »

Some feedback about the Bausch & Lomb Watchmaker Loupe 10X I bought recently.

These lenses are no adequate, way too small a diameter and they seem to be spherical and not aspheric. There is a lot of distortion on the edges and the size of the visible zone with clear focus is very small (2.5x2.5cm).
Menas
One Eyed Hopeful
Posts: 11
Joined: Sun Aug 10, 2014 7:30 am
Location: Metz, France

Re: Google cardboard impressions and lens math

Post by Menas »

I've finally ordered a pair of lenses at Surplushed. Focal length 30 mm (magnification 8.33x ?) and a diameter of 44.1 mm. My display will be a Galaxy Note 3 (5.6 inch).
My wish was to use the same lenses as the DK2, which have a shorter FL according to Fredz.
Does anyone know the DK2's lenses diameter?
@Fredz: Your guesstimation of 10x is based on the FOV announced by Oculus and a simulation with your ray tracer. Is this consistent with the FL measured, e.g. using the autocollimation technique?
User avatar
Fredz
Petrif-Eyed
Posts: 2255
Joined: Sat Jan 09, 2010 2:06 pm
Location: Perpignan, France
Contact:

Re: Google cardboard impressions and lens math

Post by Fredz »

Menas wrote:I've finally ordered a pair of lenses at Surplushed. Focal length 30 mm (magnification 8.33x ?) and a diameter of 44.1 mm.
Nice find, made of Pyrex also. A refractive index a bit higher than acrylic and less dispersion, could give good results. Should be a bit heavy though with twice the density of acrylic.
Menas wrote:My display will be a Galaxy Note 3 (5.6 inch).
It's 5.7" in fact, the same display than in the DK2.
Menas wrote:Does anyone know the DK2's lenses diameter?
Around 4cm measured with a ruler.
Menas wrote:Your guesstimation of 10x is based on the FOV announced by Oculus and a simulation with your ray tracer. Is this consistent with the FL measured, e.g. using the autocollimation technique?
In fact I just tried different lens magnifications until the display was entirely visible.

It gave a 10X magnification and the horizontal FOV ended up a bit lower than the DK1. I thought it was incorrect since IIRC it's been said that the FOV would be a bit higher for DK2 but it's been confirmed recently that it's indeed a bit inferior to DK1.

I didn't measure the focal length yet, thanks for the info about the autocollimation technique, very useful.
Menas
One Eyed Hopeful
Posts: 11
Joined: Sun Aug 10, 2014 7:30 am
Location: Metz, France

Re: Google cardboard impressions and lens math

Post by Menas »

Hello,

Hopefully these lenses will give good results, but I've just realized that they are plano-convex, while the DK2 has biconvex lenses. If Oculus (should I say Palmer?) decided to put biconvex lenses, there might be good reasons. I've read application notes from Edmund Optics telling that the plano-convex lenses are usually used in pairs. So even if such lenses can collimate and magnify, a bicknvex lens may be a better choice because of the fact that both the object and the eye are close to the lens. Indeed, I'm nearsighted, so the screen will be placed so that the image is formed at my punctum remotum instead of infinity.
Moreover, plano-convex lenses don't seem to be appropriate for polychromatic light. Hopefully, the fact that they are aspherical will minimize chromatic abberations in addition to the spherical ones. Edmund optics advises to use specific kinds of Loupes for imaging : Hastings or Steinheil. Have you ever heard about them?

Does anybody think that a similar immersion to the Rift can be achieved ? Apparently neither the Google Cardboard, nor other smartphone based HMS provide an as good feeling of immersion. Apart from the latency issue, I think that there is no reason why this cannot be dkne. DK2 owns a Galaxy Note 3 screen, so the only remaining point is the lenses.
budda
Sharp Eyed Eagle!
Posts: 388
Joined: Thu May 22, 2008 9:13 pm

Re: Google cardboard impressions and lens math

Post by budda »

Hi,

It would not be surprising if Oculus created a dedicated lens for their commercial head mounted display.

The lenses being used now are off-the-shelf mass produced items that were appropriated from loupe magnifying lens technology.

These work well for prototyping purposes at short notice and low cost.

However, the lenses could well be optimised for a specific head mounted display and should be economical to produce in large numbers as envisaged by Oculus.

For instance, the aspheric properties of the lens could be be better used to correct for curvature of field rather than for distortion, the lens could be made from lower dispersion material to reduce chromatic aberation, and the optical surfaces could be designed with tailored astigmatism to reduce the screen door effect. They could even incorporate some prism in the lens to open up the field of view and the central 3D overlap region as well.

These lenses could well provide a distinct competitive advantage in a crowded head mounted display marketplace.

Thanks.
User avatar
Fredz
Petrif-Eyed
Posts: 2255
Joined: Sat Jan 09, 2010 2:06 pm
Location: Perpignan, France
Contact:

Re: Google cardboard impressions and lens math

Post by Fredz »

Menas wrote:If Oculus (should I say Palmer?) decided to put biconvex lenses, there might be good reasons.
I think they used biconvex lenses because that's what was available in acrylic. Biconvex lenses seem to correct for oblique astigmatism if the front surface is chosen appropriately, so maybe this is the advantage over plano-convex lenses. Also they allow for more space for the eyelashes and they may allow a bit more FOV.
Menas wrote:Indeed, I'm nearsighted, so the screen will be placed so that the image is formed at my punctum remotum instead of infinity.
If you're nearsighted you should probably modify the lens/screen distance so you can focus at infinity, like the B/C and B lenses do in the DK1 and DK2 resp.
Menas wrote:plano-convex lenses don't seem to be appropriate for polychromatic light. Hopefully, the fact that they are aspherical will minimize chromatic abberations in addition to the spherical ones.
Chromatic aberration is not corrected by the lenses in the DK1 or DK2 but in software.
Menas wrote:Edmund optics advises to use specific kinds of Loupes for imaging : Hastings or Steinheil. Have you ever heard about them?
Hastings are triplets, they can offer a big magnification but I don't think you can get a wide FOV with these and the image quality is lowered when you add lenses. The Altergaze HMD is using 3 lenses as well, I don't know if they're Hastings though.

The Steinheil seem to be made from glass and look like achromat triplets so I guess they should suffer from the same problems.
Menas wrote:Does anybody think that a similar immersion to the Rift can be achieved ?
I guess it should be possible by finding the appropriate lens, but only for quite big displays, such as the 5.7" used in the DK2.
Menas wrote:Apparently neither the Google Cardboard, nor other smartphone based HMS provide an as good feeling of immersion.
These projects have not been built with the same attention to details that Oculus brought to create the Rift.

As you said the latency is a problem and it's not clear if it can be corrected on mobiles. Tiler GPUs add a frame of latency, sensors are not fast or precise/accurate enough and you'll miss the positional tracking and the low-persistence. And on top of that you can only get a wide FOV with higher diagonals (>~5.6") which severly limits the possibilities.
User avatar
Fredz
Petrif-Eyed
Posts: 2255
Joined: Sat Jan 09, 2010 2:06 pm
Location: Perpignan, France
Contact:

Re: Google cardboard impressions and lens math

Post by Fredz »

budda wrote:It would not be surprising if Oculus created a dedicated lens for their commercial head mounted display. [...] The lenses being used now are off-the-shelf mass produced items that were appropriated from loupe magnifying lens technology.
I think the lenses for both the DK2 and CV1 are custom made. See these posts from Palmer :
- http://www.reddit.com/r/oculus/comments ... om/cjiqiy7
- http://forums.modretro.com/viewtopic.ph ... 82#p155882
Menas
One Eyed Hopeful
Posts: 11
Joined: Sun Aug 10, 2014 7:30 am
Location: Metz, France

Re: Google cardboard impressions and lens math

Post by Menas »

Thank you both for your replies.
As I'm nearsighted, I'll have to change the distance between the screen and the lenses, you're right. Thus, the screen won't be placed at the focal plane, and the light won't be collimated. The image will be formed at my punctum remotum, i.e. the furthest distance I can see without glasses (not infinity, as th3 rays will not be parallel any longer if the screen is not located right in the object focal plane).

I have never tried any HMD yet (that's why I'm pretty excited to receive the lenses and to experiment. Even if I'm less excited when I read that the Google Cardboard or the Durovis Dive are funny, but have nothing to do with the Rift. And that's exactly why I intended to use the same lenses as the DK2.

Regarding the positional tracking with smartphone based HMD, how about using the rear camera and one or several big QR codes?

Now that the lenses have been ordered, I might buy ski goggles, and thinking about a design. I've thought to use use bottles caps for holding the lenses. Thus the focus could be independently adjusted, even if it's just a bit.
User avatar
Fredz
Petrif-Eyed
Posts: 2255
Joined: Sat Jan 09, 2010 2:06 pm
Location: Perpignan, France
Contact:

Re: Google cardboard impressions and lens math

Post by Fredz »

Menas wrote:Thus, the screen won't be placed at the focal plane, and the light won't be collimated. The image will be formed at my punctum remotum
Ah yes, thanks for the correction.
Menas wrote:I have never tried any HMD yet (that's why I'm pretty excited to receive the lenses and to experiment.
I'm curious, why didn't you buy the DK2 ? Too expensive ? Waiting for CV1 ?
Menas wrote:Regarding the positional tracking with smartphone based HMD, how about using the rear camera and one or several big QR codes?
I guess that should work, just like Valve did in their VR room, but then you'd end up with something like this :

Image

Another solution would be to use a laser projector like in the Kinect so you don't have to plaster your walls, something like this :
http://www.amazon.com/Uhome-Fantasy-Pro ... 00KT441S4/
Menas wrote:I've thought to use use bottles caps for holding the lenses. Thus the focus could be independently adjusted, even if it's just a bit.
You could also build cup holders from cardboard or plastic like this and stack two of them with the lens inside to secure its position :
Cup holders.png
You do not have the required permissions to view the files attached to this post.
Menas
One Eyed Hopeful
Posts: 11
Joined: Sun Aug 10, 2014 7:30 am
Location: Metz, France

Re: Google cardboard impressions and lens math

Post by Menas »

The reasons why I don't directly buy a Rift are multiple.
Actually, I'm not a gamer, and except Portal which was downloadable for free on Steam, I don't have any PC game. My laptop doesn't have a good GPU.
I have a Xbox360, but I rarely play at the moment. Sometimes Street Fighter, Soulcalibur or Mirror's Edge (which must be awsome with the Rift).

However, I spend a lot of time on my smartphone, much more than on the XBOX+Laptop+Tablet combined. I used to play Modern Combat, NFS, Nova, Virtua Tennis and PES when I went to work by train. So I can easily justify the cost of a state-of-the-art smartphone, since I also use it for professional purposes, as a camera, camcorder, etc...

Regarding the HMD, I'd like to show-off and demonstrating my family and friends what can be achieved with a phone and lenses. It may not be as good as the Rift, but I hope that it will be better than the cardboard, as its lenses have a too large focal distance. I think. I'd like also to learn cool stuff. I've just discovered this hobby, and I love it :D

The day when I'll want something that works, the Rift will be there, and fortunately at a fair price.
Menas
One Eyed Hopeful
Posts: 11
Joined: Sun Aug 10, 2014 7:30 am
Location: Metz, France

Re: Google cardboard impressions and lens math

Post by Menas »

Hello,
I've received the lenses. I don't know if it's because of my sight, but the FOV is not as good as I expected. I can see the edges of the screen. The edges of the lenses are out of focus. Apparently some users of the Rift have reported this issue too. I cannot say if it is as much or more than the real Rift, as this is the first time I'm experimenting VR (or simply strong magnifiers).
The barrel warping seems to be useless, as the picture through the lenses still have this distorsion ( whereas it is supposed to be cancelled). The lenses were described as being aspheric, still I can see a distorsion, and what is weird is that this distorsion looks like a barrel distorsion instead of the pincushion. My hypothesis is that the lens has not been designed to be used as a loupe, but to effectively collimate a laser beam (?).

I am going to mount the lenses on a frame, hoping it will be better. I'm hesitating to buy new ones.
User avatar
Fredz
Petrif-Eyed
Posts: 2255
Joined: Sat Jan 09, 2010 2:06 pm
Location: Perpignan, France
Contact:

Re: Google cardboard impressions and lens math

Post by Fredz »

Menas wrote:I don't know if it's because of my sight, but the FOV is not as good as I expected.
I've guesstimated the characteristics of these lenses and I got a ~93° horizontal FOV with the Galaxy Note 3 display, a 6.35 cm IPD and interaxial and a 2 mm eye relief :

Image

I used a 1.474 refractive index, lower than the one for acrylic (1.4919), which I incorrectly thought to be higher previously but I found several references for this value. For the back surface profile from the 30mm effective focal length and the 1.4 cm thickness I estimated a 1.425 cm radius with a -2.1 conic value and zeroed aspheric terms.

With the same display I calculated the DK2 horizontal FOV at 109°.
Menas wrote:I can see the edges of the screen.
That was to be expected I guess. With a small eye relief you can also see the edges of the screen on the DK2.
Menas wrote:The edges of the lenses are out of focus. Apparently some users of the Rift have reported this issue too.
Yes that was to be expected as well, it's the same with the DK2, although IIRC it was much worse with the DK1.
Menas wrote:what is weird is that this distorsion looks like a barrel distorsion instead of the pincushion. My hypothesis is that the lens has not been designed to be used as a loupe, but to effectively collimate a laser beam (?).
That's weird, could you take a photo of a grid taken through one of the lens ?
Menas
One Eyed Hopeful
Posts: 11
Joined: Sun Aug 10, 2014 7:30 am
Location: Metz, France

Re: Google cardboard impressions and lens math

Post by Menas »

Hi, sorry for the delay.
After a few days, I finally find the lenses OK. I've understood that a single lens cannot eliminate the aberations, but 9nly minimize them. The solution would be more complex optics I think in order to get rid of the blurry edges.
I've also understood that a wide FOV, therefore immersion is OK for gaming, not for watching videos. I've tried SBS videos and the result was awful. Too blurry and no real 3D effect. I don't think that biconvex lenses would solve this issue.

Regarding the barrel-shaped edges, it's actually normal, as the plano-convex lenses are slightly concave at the edges (like a bell). I've shot a photo of a grid viewed through one of these lenses. One can notice that first the distorsion is a p8nscushion, then a barrel.

I've bought ski goggles, so I'll be able to start the fun part!
You do not have the required permissions to view the files attached to this post.
User avatar
Fredz
Petrif-Eyed
Posts: 2255
Joined: Sat Jan 09, 2010 2:06 pm
Location: Perpignan, France
Contact:

Re: Google cardboard impressions and lens math

Post by Fredz »

Menas wrote:I've also understood that a wide FOV, therefore immersion is OK for gaming, not for watching videos. I've tried SBS videos and the result was awful. Too blurry and no real 3D effect. I don't think that biconvex lenses would solve this issue.
The distortion is easily corrected in software like it's done by the Oculus SDK, there are also several video players that use this to show SBS 3D movies.

The blurring can't be corrected in software unfortunately, but I've found the DK2 much better than the DK1 in this regard.
Menas wrote:I've shot a photo of a grid viewed through one of these lenses. One can notice that first the distorsion is a p8nscushion, then a barrel.
Thanks for the photos. Nice to see I wasn't too far off for the general surface profile, but I didn't expect what seems to be a dual focal length. Makes sense for collimating light for laser cutting though.
Menas
One Eyed Hopeful
Posts: 11
Joined: Sun Aug 10, 2014 7:30 am
Location: Metz, France

Re: Google cardboard impressions and lens math

Post by Menas »

Are you thinking that it's a dual focal length lens because of the barrel distorsion? In this case, I don't think that my lenses are similar to what is described in the website dealing with laser focusing lenses. Indeed, the lenses which are described don't have any concave part, but 2 convex parts with different curvatures.
I think that the concave part in my lenses are not intended to be used for optical purposes. That's only a side-effect, and this "bell profile" is necessary for mounting the lens.

Edit: or maybe you were talking about what can be seen on the profile picture?
User avatar
Fredz
Petrif-Eyed
Posts: 2255
Joined: Sat Jan 09, 2010 2:06 pm
Location: Perpignan, France
Contact:

Re: Google cardboard impressions and lens math

Post by Fredz »

Yeah the link I posted is not really relevant, but I guess the effect that can be obtained with these lenses may be similar.
Heffle
One Eyed Hopeful
Posts: 5
Joined: Tue Jul 29, 2014 12:59 pm

Re: Google cardboard impressions and lens math

Post by Heffle »

I'm back, after being busy doing a bunch of stuff and moving. It's good to see you guys continuing discussion.

Unfortunately, my prism idea didn't come to fruition as something urgent came up and ever since I haven't had enough time to continue. In any case, the google cardboard I had is now broken, crushed, and pretty much ripped apart, so I'm building a new one, this time with adjustable focus. It won't be adjustable as you are using it, but it's certainly better than nothing at all. Also, it will only take minor modifications to the original design, plus one more piece of cardboard, and some more velcro, meaning that it's not going to be much harder to build than the original.

Also, about the lens that google uses. I think they picked that lens precisely because it has less magnification. Larger lenses would make the distortion more apparent, and finding a biconvex lens that is also aspherical with that high of a magnification is relatively hard for the average consumer. Of course, it's also the lenses that the Durovis Dive uses.
User avatar
cadcoke5
Binocular Vision CONFIRMED!
Posts: 210
Joined: Mon May 24, 2010 8:43 pm
Location: near Lancaster, PA USA

Re: Google cardboard impressions and lens math

Post by cadcoke5 »

I am curious if there is any sort of method for determing the field of vision for when you move your eye to one side. In other words, if you want to look at an object towards the right side of the screen, you turn your eyes to the right, and your pupal moves to the right somewhat. Then the pupal is no longer centered on the lens. How does this affect the resulting view?

-Joe
metronome
Cross Eyed!
Posts: 114
Joined: Sat Mar 16, 2013 12:42 pm

Re: Google cardboard impressions and lens math

Post by metronome »

Cheap google cardboard kit here under $4 lol

http://www.dx.com/p/assembling-google-v ... aki-339632
User avatar
Dilip
Certif-Eyed!
Posts: 692
Joined: Sat Dec 13, 2008 9:23 am
Location: Ahmedabad//INDIA

Re: Google cardboard impressions and lens math

Post by Dilip »

@Fredz

i have came across these marginally bigger Aspheric optics here.
http://www.ebay.com/itm/Donegan-Aspheri ... 5d53324737

Do these have any merit over these well known?
http://www.ebay.com/itm/5x-pocket-loupe ... 3a79987a19

I think donegan being 2.14" can offer marginally wider fov with all goodness of 5X.

What are your thoughts can they be better over long used 2" ones.
apart from being slightly economical whats your opinion on these.

Thanks in advance... :D
User avatar
Fredz
Petrif-Eyed
Posts: 2255
Joined: Sat Jan 09, 2010 2:06 pm
Location: Perpignan, France
Contact:

Re: Google cardboard impressions and lens math

Post by Fredz »

The previous model was already too large to fit your nose and the limiting factor was not the diameter of the lenses but the size of the display, so you won't get more FOV from these lenses.
User avatar
Dilip
Certif-Eyed!
Posts: 692
Joined: Sat Dec 13, 2008 9:23 am
Location: Ahmedabad//INDIA

Re: Google cardboard impressions and lens math

Post by Dilip »

@Fredz
Thanks :)
konstantin_lozev
Cross Eyed!
Posts: 192
Joined: Fri Jul 04, 2014 1:43 am

Re: Google cardboard impressions and lens math

Post by konstantin_lozev »

Heffle wrote:I'm back, after being busy doing a bunch of stuff and moving. It's good to see you guys continuing discussion.

Unfortunately, my prism idea didn't come to fruition as something urgent came up and ever since I haven't had enough time to continue. In any case, the google cardboard I had is now broken, crushed, and pretty much ripped apart, so I'm building a new one, this time with adjustable focus. It won't be adjustable as you are using it, but it's certainly better than nothing at all. Also, it will only take minor modifications to the original design, plus one more piece of cardboard, and some more velcro, meaning that it's not going to be much harder to build than the original.

Also, about the lens that google uses. I think they picked that lens precisely because it has less magnification. Larger lenses would make the distortion more apparent, and finding a biconvex lens that is also aspherical with that high of a magnification is relatively hard for the average consumer. Of course, it's also the lenses that the Durovis Dive uses.
Hi, are you sure Durovis use the same lenses as Google Cardboard? I have experimented with the Open Dive lens kit and the lenses seem much smaller than the Cardboard blueprint holes.
My thoughts after experimenting with the Open Dive lens kit and Galaxy S3 Mini (4" screen, small, I know) are that:
- the focal length is very short, so when the phone's screen is on focus, it is literally up my nose, my eyes are on the lenses and I cannot see the full height of the image (wasted pixels even on 4" screen?), I just wonder then what is the benefit of a 5-5.5" screen...
- on a 4" screen the screendoor effect is really noticeable, but that is expected, no surprises there
- the much bigger problem with the 4" screen is that there is simply no way to watch the 2 images on the screen without forcing your eyes to cross (similar to trying to look at your nose, not pleasant and gives huge eye-strain), the distance between the centres of the 2 parts of the screen is simply much smaller than my IPD
So after many tries, I have established (I think it's not fixable) that even though Durovis list the S3 Mini as compatible, you need a much bigger screen size, but then comes the conundrum - with the fact that the screen stays so close to your face, would a large portion of the pixels on the 5.5" screen just get wasted...
jumpjack
One Eyed Hopeful
Posts: 3
Joined: Tue Apr 05, 2011 5:01 am

Re: Google cardboard impressions and lens math

Post by jumpjack »

konstantin_lozev wrote: - the much bigger problem with the 4" screen is that there is simply no way to watch the 2 images on the screen without forcing your eyes to cross (similar to trying to look at your nose, not pleasant and gives huge eye-strain), the distance between the centres of the 2 parts of the screen is simply much smaller than my IPD
Doesn't it exist any app capable of IPD tuning? I think it should be a must in any VR app!

I found this store for lenses, which lens do you guys suggest to use with my Samsung S3 Neo?
http://www.cliped.it/lenti_varie.htm

I also found this interesting explanation of how lenses work:
http://online.scuola.zanichelli.it/amal ... Amaldi.pdf

As a final note, I think planar-convex lenses should be avoided, because I'm quite sure they cause color distortion if not used in couple, due to different, asymmetric light path.
Dimon Hell
One Eyed Hopeful
Posts: 1
Joined: Wed Apr 29, 2015 9:52 am

Re: Google cardboard impressions and lens math

Post by Dimon Hell »

Greetings community!
Beforehand apologise for my terrible english...
Don't want to create topics, so gonna ask here.
I've ordered chinese "glasses" Baofeng 2, for use it with my 5.2" 1080 smartphone.
I know, those have not proper lenses and i gonna experiment a lot with some different lenses.
As i understand i need some short focus lenses, that i gonna buy from Surplushed.
Can you explain, which lenses suits better for my needs? Exactly diameter and FL, and lens type? PCX, DCX, BCX, ASP?
There's ASP Dia 44mm and 30 FL, also BCX Dia 35 and 40 FL, and PCX Dia 35 and 40 FL.
Can you advice something? I don't want to buy a bunch of lenses...
HUGO2345
One Eyed Hopeful
Posts: 1
Joined: Wed Nov 23, 2016 4:21 am

Re: Google cardboard impressions and lens math

Post by HUGO2345 »

Fredz wrote:
Heffle wrote:What type of distortion could affect the FOV? I know of barrel distortion, but does that really change the FOV that much, even when it is corrected in software?
It's more complicated than this. You need aspheric lenses to correct for spherical aberration and you need to put them at the focal distance to have the light collimated (focus at infinity).

The display of the Galaxy S5 is too small (5.1" diagonal) to get a good FOV. The DK2 uses high magnification lenses (10X I think) to get a nice FOV (~100° horizontal), but with a 5.7" display.

To calculate the FOV correctly you need to write a ray-tracer or use an optical software (like Zemax or Oslo) that can handle aspheric lenses.

Something like this :

Image

Hi,

Could you please explain why did you put the lens focus point in that location ,is not the eye 's retina ? I 'm just thinking about the right design maybe like this
You do not have the required permissions to view the files attached to this post.
Post Reply

Return to “VR/AR Research & Development”