Increasing the depth of field of human vision

[brianbenchoff]

A lot of Hack A Day readers sent in their comments on this post.

Since that article isn't on the front page anymore, we might as well move everything over to the forums.

Post your ideas on how to build this:

[Reggie]

Not sure which part of the question to look at but I suppose it would be the astronomy side of things.

Amateur astronomers use motorised mounts that have large databases of objects on a handset or can be connected to computers via serial, they use apps. like stellarium or starry night pro to control what the telescope points at and can be made to be very accurate. It's possible to use what's called a dual mount bar and mount 2 of the same size telescope + 1 extra scope and 'co-align' the pair of scopes onto the same object (I'll explain what the 3rd scope is for in a minute). The dual mount bar would put the telescopes about 24"-30" apart. That would be a cheap way to do it, up to you what you use for cameras, you could use something as simple as an SPC880pc philips webcam.

You could also get 2 of the same mount, 4 telescopes (I'll explain why 4 in a mo, 2 pairs) and place them pretty much any distance you like apart. Although they've obviously got to be able to see the same object and in the night sky preferably, unless you want to look at the sun, do not *ever* look at the sun with binoculars or a telescope or your mates will be calling you cyclops, your eyesight will be gone within seconds!!

So, what's the extra telescope for in each method? Astronomers sometimes use what's called a 'guide scope' which is a dedicated telescope that takes a 'guide camera', this is a dedicated camera that is used to lock on to a target star, to keep the telescope aligned perfectly with the object you are tracking. This is necessary because the gears in at least amateur astronomy mounts have inherent errors, which will show up every now and again by your guide star moving, the guide camera is connected to software that looks out for these movements or 'periodic errors' and as soon as it sees your 'guide star' move 0.25pixels it will nudge the mounts motors and put the guide star back in position. Using a guide camera also cancels any misalignment between the mount/telescope and the celestial pole (about 0.75degrees off the pole star, polaris).

You could also adapt a pair of pan/tilt mounted cameras to do a similar job, just set the alt axis at the same angle as your latitude and align the centre of the latitude axis with the pole star, it won't be as accurate but would track an object well enough for a while, as long as you could set the camera to do 1 revolution per day.

Here are some links to the types of kit you might use:

PC controlled GEM mount:
http://www.firstlightoptics.com/skywatc ... nscan.html

A reasonable/cheapish telescope, although I'd prefer a 66mm as it will give a bigger FOV:
http://www.firstlightoptics.com/william ... g-apo.html

http://www.firstlightoptics.com/equinox ... o-ota.html

and maybe something small and cheap like the startravel 80 at the top of the page:
http://www.modernastronomy.com/telescop ... actors.htm

For the guide camera I would choose something simple like a long exposure modified spc900nc/spc880nc philips webcam.

and a dual mount bar to strap telescopes to:
http://www.astro-engineering.com/images ... 2%20sq.jpg


For a main imaging cameras I would look at something that can do hd video, so probably a canon 500d DSLR (Amateur astronomers regularly use DSLRs for long exposure imaging, particularly canon). This might seem like it's quite an expensive project but in reality this is probably around the lower end of amateur astronomy gear, that having been said it's quite likely that you should be able to find a couple of people with the same mount/telescope with only the guidescope/camera being different.

All of this would need some software to control it all, you would control the skywatcher HEQ mount using free software called 'EQMod' which improves the resolution of the motors way beyond what the skywatcher handset can achieve. http://eq-mod.sourceforge.net/ It gives you 9024000 steps per revolution, which works out to 0.143617 per arc second.

To tell the mount where to point the scope you would use something like stellarium or starry night pro. The former being free.

And to control the guide camera I would use PHD (push here dummy) Guiding from stark labs. http://www.stark-labs.com/phdguiding.html




I'm still not certain that this would give you any kind of depth perception for stars, I think it might well work on the moon and the sun (Don't look at the sun with a telescope, it burns quicker than you can say 'why's it gone dark?!!), jupiter probably, but I just don't see it happening with being able to see a constellation for instance, in 3d. I think the distances are too great for our eyes/brain to be able to process the information, the closest star is something in the order of 26 trillion miles away. The other issue you might have is our atmosphere being a dome will likely be more likely the kind of 3d effect you might get, which combine together to make it naturally look the stars are being projected onto the inside of a dome.

BTW. looking at the burning man stereoscope thing, I believe it's making whatever you are looking at appear a lot closer than it is, as you go up in diameter of the mirror the bigger I expect it to get a zoom effect (given the focal length).

[brianbenchoff]

I'm imagining that this could be fairly easily simulated in Celestia or Stellarium.

Have 4 instances of [astronomy program] running at the same time. Every one of these simulations will be centered on a corner of a square and ~50 AU from the sun (about the same as Pluto). Here's my MS paint mockup. Red are the cameras, yellow is the sun:



With that, you could simulate a 3D view that (I think) would have about 50 times the depth of field of annual parallax.

I'd like to point out that the constellations would probably start to look really weird. This is from the margin notes of Cosmos:



So, either really weird or really cool.

[Reggie]

Not sure I understand your post brian, you've represented the sun and cameras on a 2d plane, I take it you would like to create a complete 3d view that you could insert into a virtual universe and essentially walk around the sun to view it? That certainly sounds doable, although I don't think a simulation would give you much information, as stellarium is going to give you the same static jpg? image. I believe the stereo project uses a pair of telescopes to image the sun, I think they recently did a mapping of the 'dark side' of the sun.

With the constellation, I think that's going to be entirely the problem, until you see the 2nd image you have no idea where in space in relation to their distance from you the stars actually are, even looking at them through a telescope you can't see more than a pin prick of light, so there's not even size to give you brain something to relate to, there's nothing for them to reflect off etc. Anything that you can see that looks more than pin prick in the night sky is either a planet, a nebula or a globular cluster.

Also with the constellation they don't label the stars, so even given all the different images of the how the constellation would look we still don't have enough imformation to try and visualise it in 3d, a top down view would probably have done it

[brianbenchoff]

Using 4 cameras gets around the problem of the xkcd comic - if you turn both cameras 90 degrees, they'll still be looking at the same thing but when they're fed to the smartphone, you'll lose the 3D stuff.

To do it 'right' in the comic, you'd need a static bar that can turn 360 degrees, and two cameras at each end of the bar. If you're going to simulate it, you might be able to interpolate the entire sky in 3d with 4 cameras. 2 for the x dimension and 2 for the y dimension.

I just remembered that it would be in 3D, so you'd need 8 cameras.

[Haku]

Here's a quick pic from the bullet cameras I set up yesterday (left it setup overnight):

(click for fullsize)


You need to cross your eyes on the pic to see the stereoscopicness of it.

Until I saw both images side by side on the same screen I didn't realise one was slightly out of focus and there's a slight colour difference between them, I'll have to investigate getting correct focus on them as I just grabbed them out of the box of 10 without checking.

The clouds are out in full force today and no rain so I'll be doing some more testing, hopefully get some good pics and pics of the rig.

[DeadlyDad]

Rather than go to all the hassle using high tech, how about using something like this setup, with telescopes between the mirrors? Alignment would be a problem, but that is what lasers are for. (Come to think of it, a double mirror on each end, with perfect right angles, should return the laser back to its point of origin.

Code: Select all

/--------------------------------------------------\
\-------------------------\/-----------------------/


I would mount it on buildings, as they are going to be the most precisely lined up N/S/E/W.

[Haku]

It appears the devil is in the details.

Went out late in the day on top of the hill next to me and this is the best pic I managed to get:



I rotated one image by 1.2 degrees to match the other, and I think they're the right way round to cross your eyes and get a picture, but the cloud is so far away that even with the cameras 22 feet apart it's almost impossible to see any depth.

The hardest part was getting the cameras aligned just right, and my right eye not having perfect vision doesn't help when looking at the cameras outputs on the LCD glasses. That and I could do with a couple of identical tripods to put them on, being low to the ground meant avoiding the long grass getting in on the picture.

I'll come back to this another day in the future sometime, I think I've got another headphone extension cable somewhere to extend the distance between the cameras further.

[Reggie]

Rather than go to all the hassle using high tech, how about using something like this setup, with telescopes between the mirrors? Alignment would be a problem, but that is what lasers are for. (Come to think of it, a double mirror on each end, with perfect right angles, should return the laser back to its point of origin.

Code: Select all

/--------------------------------------------------\
\-------------------------\/-----------------------/


I would mount it on buildings, as they are going to be the most precisely lined up N/S/E/W.

Fyi. The mirrors are not too far removed from a Schmidt–Cassegrain telescope, you'd just need a curved mirror and a corrector lens.
I just wonder how you'd get the images from the telescopes onto the DSLR sensor, as the hole you look through on a telescope is usually 1.25" or 2".

In reality though, your method isn't low tech as such, it uses exactly the same sorts of math to produce an image at the end of the procedure

[Reggie]

Haku, I think the trick is that the cameras need to be lined up with each other, whether the plane they're on is at angle or not, just like your eyes are, even if you tilt your head, your eyes are aligned.