We're sorry but this page doesn't work properly without JavaScript enabled. Please enable it to continue.
Feedback

Add depth!

00:00

Formal Metadata

Title
Add depth!
Subtitle
Stereoscopic imagery for everyone
Title of Series
Number of Parts
47
Author
License
CC Attribution 4.0 International:
You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor.
Identifiers
Publisher
Release Date
Language

Content Metadata

Subject Area
Genre
Abstract
Stereoscopic imagery (photography and videography) is a fascinating way to create 3-dimensional images of landscapes, unmoving and moving objects, and of course, people.
25
Thumbnail
06:06
29
VideoconferencingLimit (category theory)Field (computer science)SurgeryMachine visionMedical imagingCASE <Informatik>Natural numberComputer-assisted translationView (database)Similarity (geometry)Extension (kinesiology)Order (biology)Computer animationMeeting/Interview
ArmMachine visionWaveSpacetimePoint (geometry)Direction (geometry)Faculty (division)outputProcess (computing)SynchronizationLeast squaresArrow of timeBlind spot (vehicle)First-person shooterLine (geometry)Degree (graph theory)CausalityField (computer science)View (database)Vulnerability (computing)DistanceMeeting/Interview
AreaMedical imagingPixelRight angleImage resolutionException handlingRow (database)Disk read-and-write headBitFreewareMachine visionTouchscreenLatent heatSlide rulePoint (geometry)SpacetimeArithmetic meanVideo projectorLogistic distributionInterpreter (computing)Sound effectSet (mathematics)Flow separationoutputLetterpress printingThree-dimensional spaceNeuroinformatikLeakMobile appView (database)Projective planeObject (grammar)Polarization (waves)DistanceReflection (mathematics)Different (Kate Ryan album)Level (video gaming)Real numberOpticsField (computer science)Error messageEquivalence relationResultantVirtual realityDreidimensionales VideoBlock (periodic table)AngleFilter <Stochastik>File viewerMeeting/Interview
Medical imagingView (database)LaptopFile viewerTouchscreenLetterpress printingArithmetic meanPoint (geometry)Closed setMultiplication signSurfaceFreewareDistanceTotal S.A.Field (computer science)Machine visionRight angleProcess (computing)Row (database)Sound effectAngleSource codeRevision controlBitLine (geometry)Image resolutionMeeting/Interview
Sound effectQuicksortDistancePhysical systemBookmark (World Wide Web)Arithmetic meanWebsiteMetrePrime idealMacro (computer science)Product (business)Ring (mathematics)Medical imagingCASE <Informatik>Combinational logicMeeting/Interview
Focus (optics)MetreArithmetic meanSound effectDistanceMedical imagingPlanningDegree (graph theory)AngleRight angleMeeting/Interview
Physical systemProduct (business)MetreThread (computing)Ring (mathematics)Meeting/Interview
FreewareFrame problemRotationFile viewerDigital photographyGroup actionView (database)Meeting/Interview
Frame problemRight angleMedical imagingWeight
Gamma functionOpen sourceSoftwareMedical imagingComputer programmingDefault (computer science)Computer configurationRight angleMeeting/InterviewComputer animation
RectangleMedical imagingPosition operatorView (database)Decision theorySound effect2 (number)Moment (mathematics)Right angleFrame problemCASE <Informatik>Greatest elementContent (media)Selectivity (electronic)Clique-width
Correspondence (mathematics)Right angleMetreVideoconferencingFrame problemView (database)Medical imagingCASE <Informatik>DistanceProjective planeDifferent (Kate Ryan album)Streaming mediaVelocityDirection (geometry)Shooting methodLinearizationLogical constantObject (grammar)Sound effectCausalityDreidimensionales VideoSequenceQuicksortMultiplication signMoment (mathematics)Lecture/Conference
IcosahedronGEDCOMMoment of inertiaMedical imagingVideoconferencingHeegaard splittingRight angleStandard deviationPosition operatorCASE <Informatik>Asynchronous Transfer Mode2 (number)Filter <Stochastik>PixelFrame problemKeyboard shortcutView (database)Video trackingDistanceSequenceSet (mathematics)AngleMetreClique-widthDegree (graph theory)RectangleDifferent (Kate Ryan album)Computer animation
View (database)Medical imagingSound effectComputer animation
Branch (computer science)VideoconferencingGoodness of fitParallel portView (database)TwitterTouch typingMedical imagingClosed setBitMeeting/Interview
JSONXMLUML
Transcript: English(auto-generated)
Hi. As I want to start this talk on an important accessibility note, some of this talk assumes that you have functional vision in both eyes. You will of course be able to see the images and the examples in this talk if you need to wear corrective lenses. I certainly need to wear them too. But
if you happen to have very bad vision or blindness in one eye then by the nature of this very topic you won't be able to follow some of the examples and I do apologize for that. Likewise if you have nystagmus or strabismus similar limitations might apply and possibly also if you're using an artificial lens implant as is common after cataract surgery. Again I do
apologize for that but in this case by the very nature of my topic there's very little I can do to alleviate that unfortunately. However even if you yourself cannot see stereoscopic images and videos this talk does contain suggestions for you how to make them which you can totally do even
if you happen to be making them for somebody else. So let's first get started with what stereoscopic vision or stereopsis as it's called really is and how it works. So first of all what we need in order to visually perceive depth is two eyes and more specifically two forward-facing eyes. So
one thing that we humans have in common with say cats and foxes and owls is that our eyes sit on one side of our skull facing what we call forwards and they provide two fields of view that overlap to a great extent and then of course comes at a cost namely that our field of vision is comparatively narrow. So if you ever sit or stand like this and you stretch
out your arms you bring them to your side you're only going to be able to bring them about as much as this before you lose your peripheral vision. So our complete arc of about 180 degrees and beyond that you can't even perceive
motion. So if you wave your hands but you go behind your shoulders you probably won't be able to see that anymore. If you compare that to the eyes and a horse for example those sit on either side of the skull and they give the animal a far greater field of vision. So a horse can almost see a 360 degree arc it only has a small blind spot directly to the rear of the skull and that of
course is because the horse is an herbivore and its best defense against predators is flight. So the best thing you can do is outrun its predator and then live to graze another day so the broader its field of view is the higher the chance of detecting a predator early even when it's in vulnerable circumstances such as foraging. So get a head start on the
predator, sprint, run away from them as quickly as possible. We humans are predators ourselves and omnivores and we have evolved to sense depth which greatly helps us in things like prey hunting particularly since we've also evolved brains that enable us to make tools. So if we make a spear or a lasso
or a Kylie or a bow and arrow and we have an animal to throw a shooter at then it greatly helps us to be able to sense pretty exactly and yet intuitively how far that animal is away from us. So how exactly does that work and what in our eyes and brains actually makes us sense depth. So in principle our eyes produce
completely separate inputs for our brains to process and the fact that the input that we get from our eyes is usually locked and in sync is really not innate to us. If you look into the eyes of a small child as in an infant you will sometimes see that their eyes can go off in completely separate directions which can look quite amusing when it happens but
eventually at some point in our early childhood in fact so early in our childhood that none of us can really remember ever not being able to do it we acquire the faculty of virgins and virgins is the ability to make the axes align the axes of our eyeballs so you can imagine an imaginary straight
line from your pupil through the center of your eyeball and onto a perpendicular intersection with your retina you can align those axes symmetrically and at a particular point in space so you can align them in such a way that they intersect any point in space in front of you. You
might be able to make them not intersect at all such as when you kind of stare aimlessly ahead but you kind of we almost need to consciously try to do that and without that effort they'll always converge at some point even though perhaps that's a point that's very far in the distance you
really won't be able to make them diverge which would be able to have your eyes point outward but you will be able to make your eyes converge almost arbitrarily which of course you can test by stretching your arm out in front of you holding a finger up looking at your fingertip and then bringing the finger close in while you're still looking at it and most
people will be able to even touch the tip of their noses while keeping the the eyes verged on the finger it'll make you extremely cross-eyed and it'll cause some discomfort so you won't be able to do it for long but you should be able to do it normally. So vergence gives us the ability to point both our
eyes at one spot in space and then the other thing that we need for depth vision is accommodation which is the ability to stretch or relax the elastic lens in our eye the thing that sits right behind our pupil in such a way that a specific point in space in front of our eyes produces a
sharp image on our retina in the back of the eye and commonly the accommodation point in space in front of us is also a vergence point meaning we point our eyeballs at a specific point in space and then we simultaneously adjust our lens so that the image of that very point is also
sharp. So that means that when we're looking at something our eyes produce two images two images that are pretty much identical in one spot the vergence and accommodation point and then slightly different anywhere outside that spot and that is called binocular disparity and through what has to be an
inordinate amount of trial and error in our infancy our brain learns to interpret these disparities between the two optical inputs as cues about depth so if an object in a field of view is at nearly the same spot in both
eye images then it's comparatively far away and if the difference between the inputs is comparatively large then the object is relatively close by and it's by these differences that our brain essentially builds a depth map of our surroundings. Okay so now that we know how real depth perception works
meaning how our eyes signal or collaborate with our brains to signal that an object may be at a greater or a lesser distance from us we can talk about how we can trick our brain into seeing depth when objectively there
is none so taking two two-dimensional images and making our brains conjure up a three-dimensional image from that now for now let's just assume this we have two two-dimensional images and those correspond exactly to what your
left eye and your right eye would see so here's an example of two such images don't worry for now about how we make that happen how we create these images we'll get to that in a jiffy and then what we need to talk about is this thing called stereoscopic projection so making your left eye see only the left eye image and making your right eye see only the right eye
image so let's start with something very low-tech a stereoscopic viewer or simply a stereoscope I happen to have one here let me swiftly explain how this works you take a print of your stereoscopic image you slide that print
in here you close the device so the print is jammed in place and cannot move about and then you bring the whole thing up to your face and you look through the eyepieces like this and if you look at this little device from this angle like that you'll notice something important namely there is this little
black separator that ensures that your left eye can't peek over at the right image and the right eye can't catch a glimpse of the left eye image and here are a couple of others that follow much the same principle except they're designed for larger images so here's one for viewing larger prints it's a
simple cardboard device and look through it like that and here's one for viewing images on a computer screen okay so these are all very simple and effective devices that have been in use for stereoscopic viewing for a staggering very nearly 200 years so this thing or his predecessor was
invented by a chap named Charles Wheatstone in 1832 this device is a very modern interpretation using ABS plastic and whatnot but it's still the same principle even a modern set of virtual reality goggles is the same principle except what you have is a screen right in front of each of your
eyes but this kind of device so having to use this kind of device has a bit of a flaw it doesn't fare so greatly in the social aspect of art right so of course you can pass this around with a small group and appreciate the pictures together but you'd be hard-pressed to actually do an exhibition with this or show anything to a larger crowd so in comes the next
step in the evolution of stereoscopic viewing the stereoscopic slide projector and the principle is this you load two slide projectors with a magazine of slides each one magazine holds the slides of the left eye images
the other magazine and the other projector holds the right eye images and you advance the slides in the projector simultaneously and you project the slides over one another so that they fully overlap and then here's the clever twist which is that in front of each projector you put a filter namely a
polarizing filter suppose for the sake of discussion for now use a vertically polarizing filter for your left projector horizontally polarizing filter for your right projector and then you distribute goggles to your audience of glasses to your audience that have corresponding filters in the glasses so they let through virtually polarized light on the left horizontally
polarized light on the right and then the filter will block all the light coming from the right projector from entering the audience's left eye block all the light coming from the left projector from entering the audience's right eye and the result is that your audience's left eyes only see the left eye image right eye image only the right eyes only sees
the right eye image voila stereoscopic vision or 3d vision in reality things are just a little bit more complicated which is to do with the fact that if you do things this way your audience have to keep their heads level all the time because once they tilt their head some of the left eye
image light leaks into the right eye and vice versa the 3d effect quickly collapses and very strenuous for someone to sit somewhere forward like an hour and not be able to ever move their head so you can address that by using circular polarization instead so for example for the audience's left
eyes you would use clockwise polarized and right you would use anti clockwise polarized light and then it's okay for your audience to move their heads around a bit and they'll still retain the 3d effect this is the stereoscopic technique that most stereoscopic movie theaters 3d cinemas
use today and then there is the at-home equivalent of that which is 3d TV where instead of having two projectors you've got alternating rows of pixels so one row of pixels emitting light that's polarized one way next row of pixels emitting light polarized the other way you wear goggles and effectively each of your eyes sees a picture that is at half the
nominal vertical resolution and then your brain again pieces those eye inputs together for a three-dimensional depth effect but of course that too has a downside which is it's pretty involved on the technology and logistic sides you need projectors you need a silver screen because
otherwise you'd mess up the polarization on reflection you can't project polarized light on a white wall your audience needs goggles the whole bit so it's all very much non-trivial so what if we could instead employ a technique that requires none of that no special projection equipment no goggles no nothing where you could literally hang a
stereo photography in a gallery and your visitors could stroll through and look at your images and see them at full depth so you could make a picture book or you could send 3d images to a friend on your phone and they wouldn't need an app to view it and of course such a technique totally does exist and we're going to focus on it for the rest of this talk it's called
free viewing and it comes into flavors so free viewing in general means that we can take two stereoscopic images and look at them using a certain technique that requires no stereoscope no projector and no nothing
all we need is two functioning eyes and for the two flavors of free viewing we've got one that's easy to explain and understand but difficult to do and another one that's a bit more complex to comprehend but somewhat easier to do at least for most people and I'll start with the one that's easier to explain so what I have here is a print of a stereogram
it's the same image as the one that I showed you earlier with the stereoscope which is blown up a bit because that makes things easier to demonstrate left eye image on your left right eye image on the right just like with the one I put into the stereoscope earlier and I put the
image up here on my little easel and I use these two sticks to illustrate where you would be looking so if you were to now look at this print like you normally would and you were standing in front of this easel your virgins and accommodation points would both sit right on the surface of this print okay
so if I were to illustrate two straight lines from a viewer's eyes to the point that the viewer would be looking at it would look like this right and that's what our eyes do naturally it would just come together on the surface
of the print however that is of course not to say that your eyes must always do this your focal your accommodation point and your virgins point don't necessarily need to be in the same spot what you can do instead is instead of looking at the surface of the print you can look at an imaginary
spot behind the print which would look roughly like this so if you do it just right and if you look at it from the correct angle your left eye would end up looking at the left image and your right eye would end up
looking at the right image kind of like this okay and then what you see is one image in your brain that overlaps that's made from both of those source images and magically has depth and you can try that with the image that's on
your screen right now I don't know if you're watching this on a full-size monitor or a laptop screen or a big screen TV or your phone so you may have to play with the distance between your screen and your eyes a little bit but the principle is the same you kind of want to look over the top edge of your screen add a spot on the wall spot in the distance say
maybe you have a picture on the wall or something like that and then you want to try coming down just a little bit with your eyes and then change that spot and not change the spot that you get that you're focusing on in the process so you might be looking at something behind here if we're taking the easel image again and then drop your view just a little bit and then what
you're going to see is rather than two images you're going to see three a left one that's blurry a right one that's blurry and then a middle one from the other two overlapping that's sharp and that magically has depth so if you're watching this as a recording now might be a good time to stop the playback give this a few tries but if it doesn't work for you don't fret
because like I said there's another method that we can use that is easier for most people including myself incidentally so let's go ahead with the other method namely cross-eyed stereoscopic free viewing across view here's how that works so to create the cross view version of the same picture we're just
gonna have to swap them okay so the left eye image is now on the viewers right the right eye image is now on the viewers left so how do we view this so you you probably already guessing that if in a wall I image of
virgins point is behind the print in a cross-eyed image it's in front of the print so it looks like this okay like that and of course that means that the viewers right eye is now looking at the left half of the print
and the left eye is looking at the right eye of the print and you do that by simply crossing your eyes a little bit and the interesting bit about that is it's a lot easier for a lot of people to voluntarily do that and like I said that very much includes myself so what you want to do to see if cross view works for you is this you look at the image on your screen
and then you deliberately cross your eyes which probably means that you're initially going to be overdoing it which doesn't matter and then you settle your eyes back let your eyes settle back to normal and the chances are that the stimulus that your brain produces when the image is suddenly clicked into place and you see depth is so strong that your eyes don't want to go back to where they were so go ahead and try that with the
image that's on your screen now and do feel free to pause the playback again and try a few times until you get the hang of it and when you initially try this it might help if the image that you're looking at is rather large relative to your total field of vision meaning you might want to
get your face close to the laptop screen or hold your phone up in front of your face and once you get the hang of it however it won't matter anymore and you'll quickly be able to snap into any cross view image at will no matter how small it is it's probably a good idea to get started on a large ish
image and chances are that you will have a much easier time getting a depth effect from cross-eyed free viewing because that works at any image size a warlight preview intends to work well only in relatively small images and once you go larger you probably need a stereoscope of course I can't assume
that you have a stereoscope at hand right now as you're watching this talk which is why will default from cross-eyed previewing from here on out ok so now that we've talked about how to look at stereograms to get the desired 3d effect let's talk about how to make them and as you can imagine there are multiple ways of doing that for example you could be using a
stereoscopic lens that fits on your camera camera system that I prefer happens to be the micro four-thirds system and sure enough you can fit a lens that just fits onto such a camera only specific cameras but generally fits on a micro four-thirds camera and that's for example this is
this product the Panasonic Lumix 12.5mm f12 3d lens now this is a discontinued product but you can still buy it used on your favorite second-hand photography equipment website and you can use it just like any other prime lens with autofocus auto exposure and whatnot and it'll create two
stereoscopic images in one frame and by the way there will of course be others for other sensor systems you'll quickly see the big flaw of a lens like that the two lenses are very close together which makes the pictures that you take heavily hypo stereoscopic and that's what we call images where the stereoscopic distance the distance between the lenses is a lot less than
the distance between your eyes the distance between your pupils is approximately 70 millimeters and in this case the lens has a stereoscopic distance of maybe 10 millimeters and what that means is that the stereoscopic effect that you get from pictures taken with the lens like that is a lot less pronounced so you'd get a lot less depth than you'd
expect so this sort of lens really makes sense to use only in combination with a macro ring for stereoscopic macro photography which can be very interesting but it's less useful for shooting say stereoscopic portraits or stereoscopic landscape shots and things like that then there is this
device which is a manual lens meaning you have to manually adjust focus and aperture but it produces a much more naturally stereoscopic image with a 90 millimeter stereoscopic distance and that makes the image slightly hyper stereoscopic meaning the depth effect is slightly more
pronounced than it actually would be which you can see if I hold it up to my eyes it's slightly wider than that. Up next are mirror lens attachments now be warned this is where it gets really bulky and what you can see here is a thing called the cooler deeper made in Iceland now the
way this works is through an arrangement of several mirrors and the mirrors are all at 45 degree angles to the plane of the sensor so the path of the light is deflected twice at 90 degree angles and in a way what it does it just offsets the light hitting the left half of your sensor a little
further to the left and the right half of the sensor being hit by light that comes from a little further to the right and what you end up with is again a moderately hyper stereoscopic image now with this thing the effective stereoscopic distance varies somewhat based on the focal length of the
lens that you attach this to so you have to do a fair amount you might have to do some cropping or some post-production and now to be perfectly honest this is something that I bought out of pure curiosity to see how well it will work with my camera and the answer is unfortunately not too well but that's not the fault of the of the product but this thing is made for a
really big DSLR camera it attaches with a 78 millimeter thread which is something that you typically find in the Canon EOS system and it's generally meant to be used with a wide angle lens and what I'm undoing here is a stepper ring that I'm using to attach this it kind of works but for
me it's not the ideal way to take stereo pictures but if you do shoot with a big camera with a big DSLR and you want to get into stereophotography let me know and I'll be happy to rehome this. Now on to gadget free stereograms
creating stereograms with any regular camera because we want to go low tech here so I'm about to show you a stereo photography technique that works with literally any digital camera at all. A DSLR without any stereoscopic attachments, a point-and-shoot camera, a GoPro, your phone, anything. Now this
technique won't work for action shots or candid shots but you can still take absolutely stunning 3D views of buildings, parks, statues, sculptures, landscapes and even pose portraits with this. So what do we do? So first as in
any and all digital photography we pick a shot and a frame. Now we're working with digital we can of course crop and straighten and rotate in post-production that's easy but there is one thing that we'll want to do as we take the picture. We want to select something in the shot that can naturally attract the viewers attention something that their eyes would be
naturally drawn to and we want to put that somewhere near the center of the frame. So what I want to do here is take a 3D image of just a detail in my garden and from where I stand there is one thing that clearly sticks out and that's this little garden ornament here. So what I want to do is
put that garden ornament somewhere near the center of my frame and now what I'll do is this is I'll hold my camera up with both hands and I'll frame the shot with the ornament in the center and then I'll put all my weight on my left leg and take my first image and then I shift my weight over
to my right leg center the ornament again take my second image. Alright now we've got the two halves of a stereoscopic image how do we turn them into an actual stereogram? Now we can do this in a few very simple
steps using exclusively free and open source software and I'm about to show you how. Enter the Gnu image manipulation program or the GIMP. So here what I'm doing is I'm opening my images in the Gnu image manipulation program and I use that open as layers option so I get two
layers that correspond to my two images and I'll rename the layers to make them more self-explanatory so they're going to be called left and right. Now the first thing that I'll do here is align these two images on the selected center of attention so the first thing I'm doing is I'm taking my
right eye image and reducing its opacity to 50% and then I'll use the move tool which by default moves the whole layer so that the garden ornament in the right layer which is currently in the foreground aligns or overlaps quite precisely with the one in the left layer in the background and now
what I can do next is I can do a simple rectangle select to select the portion of the image where I have content in both layers exactly as their position now and then I can use image crop to selection which slices
through all of the layers and I end up with two perfectly aligned and identically sized layers that correspond to a right eye and a left eye view. Now comes the big moment where I need to make a decision do I want to make a wall light or cross-eyed stereogram in this case I make a cross-eyed stereogram so I'm moving the right layer over to the left
and then I resize my canvas to fit both images and then I finally add a new background layer in white so I get this nice little bar in the middle this vertical bar in the middle and I found that that just makes the
makes viewing the image much easier because as you free view that one bar in the middle becomes two bars either side of the image which frames it quite nicely and it really helps things click into place easily and you can of course extend that idea even further and also add a bar at the top and at
the bottom of the image preferably of identical width as a one in the center but in my opinion that's entirely optional. So I'm going to keep this up for about 30 seconds what you want to do here is cross your eyes slightly so that the two images become three a left center and right image the white bar in the middle should become two framing bars the left and the right of the
center image and you want to keep your attention focused on the garden ornament in the center and then everything should click into place and you should get a 3d effect. And now stereoscopic motion picture or motion
stereograms. I think it's probably self-evident that everything that I talked about in still photography that relied on using special lenses or special lens attachments all of that also goes for videography cause your DSLR or any other camera including your phone by the way can do video
just as well as it can do stills obviously so anything they can put on there put on your camera or your phone that creates a stereoscopic image for you will also create a stereoscopic video. But a lot of people will probably be interested in using a different kind of camera for 3d video namely a
drone camera. I hope you agree with me that aerial videography can be absolutely amazing to begin with but now imagine shooting aerial video that's stereoscopic that actually includes depth which makes it all the more interesting. Now of course to address that you could design and build your own
drone perhaps something like a fixed wing UAV that has a wingspan of two meters and has a GoPro embedded in each wing or something like that which would result in seriously hyper stereoscopic images which would give you the impression or the feeling of being something like a Pteranodon sized
creature soaring over the earth. But not only would that come with rather significant expense but you'd also probably need an airstrip to take that thing off from and land on. I don't want to go into those details although you know it's a wonderful engineering project feel free
to go ahead and do that. Now there is a way for you to get 3d footage with a single camera. A single camera that's mounted on your drone on your car or inside your car on your bicycle wherever. So let me swiftly explain the principle of how that works. Obviously as you know by now every stereogram
any stereogram consists of a left eye and a right eye image and in the case of a drone we have only one camera like this one like this GoPro which I'm going to use to illustrate. So I'm holding this GoPro up to my right eye so that's obviously my right eye view. So where's my corresponding left eye view that I get with just one camera? I could of course use two
GoPro's but no I want to do this with one camera. Well once I add some sideways motion relative to where my lens is pointed. So motion the lens is pointed this way and I want to go that way. If I do that you quickly realize that if I demonstrate it like this my left eye view at any time is
simply where my right eye view was a moment ago right like that. And we can harness that we can harness that fact in video post-production to create a 3D video from a 2D video that includes constant lateral motion. So again
consider that a natural stereoscopic distance the distance between our pupils is about 70 millimeters. So if we're shooting video at 50 frames a second every frame is 1 50th of a second or 20 milliseconds in duration. Thus if we're
moving the camera perpendicular to the direction of view even at a speed of a meter per second which is roughly the speed of a slow leisurely walk. One frame covers about 20 millimeters sideways displacement. So if we were to
duplicate our stream offset one of them by four frames project them side by side then we get stereoscopic video with an effective stereoscopic distance of approximately 80 millimeters which is already very near our natural stereoscopic distance. If instead we're doing the speed of 10 meters a second
which you can totally achieve on a bicycle or on with a drone then even a single frame offset gives you a stereoscopic distance of 200 millimeters or 20 centimeters which already makes such a video solidly hyper stereoscopic. And you can also use something like a 5 frame offset and you get this nice giant's eye view from that one meter stereo distance
where you're where you're rather heavily heavily stereoscopic. And this becomes especially interesting when you're doing either linear drone flight along a roughly linear feature say coastline or in general flight that's
parallel to some sort of feature or when you're doing orbital flight where you're orbiting above the object and ideally you're doing this when you're when you're orbiting above the object and your camera is not only pointed inward but also slightly downward makes for a very nice effect.
So what I'll show you is how to apply that technique to a sequence that uses linear flight along a linear feature with constant velocity in this case it's a drone flight along a beach. So what I have here is a drone
flight sequence that's been shot at 30 frames a second with the camera flying the camera pointing forwards and the drone flying sideways to the right at about 40 kilometers an hour or about 1.1 meters per second or put
differently the drones flying forwards at 1.1 meters per second the cameras pointed 90 degrees to the left which is exactly the same thing. Note that the camera wouldn't have to be pointed at a 90 degree angle to the heading of flight we could also be pointing the camera directly short and then fly to the beach at a 45 degree angle doesn't matter what matters is
that we have some lateral motion and we're able to quantify it which saves us some guesswork so in this case we have the camera pointed directly sideways so that means that every 90 milliseconds the camera is displaced by one meter laterally displaced by one meter which is a nice hyper
stereoscopic distance to work for in this panoramic shot so I open this in shortcut I have I have to open it actually twice I have to add it to the playlist twice because I cannot apply two different sets of filters to the same clip if it's not in the playlist twice but that's just a little
quirk of shortcut so the first thing that we're going to do is we're going to create we're going to add a video track and I am going to make that the left eye image and you'll see soon that since we're making a cross eye video
that's actually going to the right so I'm going to pull the clip in here and there's no difference between these two they're exactly the same thing I'm just going to pull the clip in here and then I'm going to apply two filters the first one is going to be a crop filter just a standard rectangle
crop and what I want to do is I want to select the middle half if you will of the image so the center which is half the width of the whole image so
what I'm going to do is I'm going to create a crop of a size 1352 and I'm going to offset that crop by 676 pixels so that I get the center here the
center of the shot and then what I'm going to add is a position filter and I'm going to take that and move this 676 pixels to the right this is
a cross view video so it goes on the right now we're going to add another video another video track rather we're going to name that right because that's going to be the right eye image and now we're going to take the
second playlist entry exactly the same clip so far and we're again going to apply some filters to it namely the exact same crop filter that we have previously so that means we're going to create a 1352 pixel crop offset 676
pixels and then we're also going to add our position filter and in this case we're of course going to move this to minus 676 so that stays on the
very left of the picture and as you can see we now have this weird little overlap here which we still have to fix by changing the blend mode I'm going
to change that to add and voila now I get these two images side by side okay and now comes the big trick which is we now need to move one of our clips by a small offset and precisely by an offset of 90
milliseconds so we're going to select that here I'm going to go pretty exactly with this one so we here at nine hundredths of a second and now what we're going to do is we're going to shift this one down here by
just that amount like this and then we still have to trim the rest of the video so we're going to do a split of playhead here remove this and then
we're going to go all the way to the end place the playhead up here go there we go split the play heads again or split at the playhead again cut that do the same thing here and then we're going to go back to the
beginning of the video and then shift these to the beginning there we go and
now we can render this and once we render this we're going to get a full cross view 3d drone flight across this Brazilian beach here this is the
resulting clip which I will play twice back-to-back again this is a cross view clip so what you want to do is cross your eyes slightly so that the two halves of the image become effectively three halves with the center half if you wish being the one with the 3d effect clicking into place and with
this I'll close I hope I've piqued your curiosity for stereoscopic photography or videography maybe just a little bit if you'd like to find out more about stereoscopic photography and videography I can highly recommend to subreddits on reddit namely cross view and parallel view where you find a
lot of stereoscopic images and inspiration also my Twitter handle is obviously all over this talk so feel free to get in touch with me I'd like to say a quick special thanks to Adolfo branches who took the drone footage and graciously gave me permission to use it and with that have fun with all
your new stereo images and videos good luck