03 Jun Giga Panoramas
The Giga is short for Gigapixel, as in millions (mega is thousands), and normally the whole thing is shortened even further to gigapan. They are created by taking a mosaic of photo’s, and then stitching them together to create on massive (and massively detailed) image.
Have you ever gazed upon a breathtakingly beautiful vista, stretching for as far as the eye can see, majestic mountains or the glittering glass sprawl of skyscrapers? Then have you taken a picture, and been disappointed by the result? That’s why.
The limitations of the widest lens means we lose the grandeur of the view, and the aspect that made it so impressive, the all-encompassing vastness.
Humans have a field of view that stretches to roughly 130 degrees. Of course we only focus on a tiny portion of that, but we still take in that full range. A 17mm lens on full-frame camera gives about 104 degrees (about 70 degrees on smaller sensors) 104 degrees sounds fairly wide, but it comes at the effect of some serious optical effects. Distortion is one, but the most important effect is of totally eliminating the “vastness” that you so appreciated when you looked upon the scene with your own eyes. Everything now seems very far away from the viewer, and the wealth of detail that made the original view so interesting are either tiny or completely absent. This is because the camera sensor has a finite resolution, and it just can’t resolve that much detail. Neither can the lens, for that matter.
So what’s the solution?
That’s where the panoramic imaging comes in. Instead of trying to fit the whole amazing view in with an ultra wide lens, we instead break the scene down into multiple pictures. Because we are only now photographing a smaller “chunk” of the scene, we can afford to use a longer lens, which is more zoomed in (We’ll look at lens choice later on). This means we can capture more detail, and because we’re going to stick those multiple pictures together, we’ll still get our wide view (or panorama) It’s almost as if we have a physically larger sensor.
The Gigapixel count is almost a side-effect of stitching together those multiple images, the resolution ramps up pretty quickly as you add images.
The Simplest Panorama
The technique to create a horizontal panorama is very straightforward; take a picture, move the camera to the right, take another picture, and so on until you’ve covered your desired field of view. Then you join them together to create one very wide image. There are actually lots of little complications going on behind the scenes of that simple description, but lots of newer compact digital cameras have functionality built-in to do exactly what I’ve described.
In order to stitch the images together, they need to overlap somewhat, so the stitching software (whether in your camera or on your computer) can join them together. It does this by identifying common points on the overlapping portions. It can’t always do this perfectly, but most software allows you to give it a helping hand by manually picking points to join on. Typically around 25% overlap is recommended so that there is sufficient detail for the software to match on.
Preferably you would take the images with the camera attached to a tripod, so the sequence of images will all be vertically aligned. The current versions of both main stitching applications can stitch together shots you take hand-held, as long as you manage a big enough overlap.
[kml_flashembed publishmethod=”static” fversion=”8.0.0″ movie=”http://blog.cerebis.com/wp-content/uploads/2011/01/pano1.swf” width=”630″ height=”143″ targetclass=”flashmovie” play=”true” loop=”false” scale=”showall”]
If you want a panorama that’s a singe image high, then a tripod and some care is all you need. However, if you want to increase the vertical field of view, as well as the horizontal, then you probably need a specialist camera mount of some kind. The mount is needed to allow you to accurately take the sequence with the required (and consistent) overlap, or registration as its called in the print world. There are a wide variety of mounts available for this purpose, but they come in two main flavours; motorised and manual.
The above matrix shows an example of a panorama made up of 3 columns by 3 rows. In this example, the image-taking sequence started from the top left, then proceeded from left to right, and top to bottom. Each subsequent image overlaps the previous one, both horizontally and vertically.
The manual version requires you to move the camera for each image, which you do in clickable “steps”, where the steps correspond to a set number of degrees. With the more feature-packed (and expensive) products, you get vertical as well as horizontal steps.
An example of a manual tripod heads is the Nodal Ninja, though of course there are many others. Any manual panoramic tripod head requires a degree of discipline to use successfully. You need to be very methodical and consistent in order
The huge advantage to using a motorised camera mount is that it does all the work. And (usually) the machine doesn’t miss out images in a sequence, which is all too easy to do when using a manual panoramic head. They also offer functionality that is not possible with a manual head, such as programmed movement along a path for advanced time-lapses.
Motorised heads generally start their life as mounts for telescopes, and are re-purposed for use in panoramic imaging by adding some sort of programmable control. This control involves connecting the mount to a computer or device running the appropriate software. One such piece of software is Papywizard, which is free and runs on PC, Mac and the Nokia 85 Internet Tablet. It was written specifically to control the Merlin Motorised mount. The Merlin comes in many guises around the world, so there’s a picture so you can identify the mount we’re talking about. It’s also known as the Orion in some markets. You can find lots of panorama and Merlin information at the autopano forums.
The Merlin, and similar mounts like the Giga pan Epic, are capable of 360 degree movement in both vertical and horizontal axes, though that range of motion may be limited by the size of your DSLR and attached lens.
To put it simply, the longer the lens you use, the more images you will need to take in order to achieve a given field of view. For example, if you want to cover a range of 130 degrees across by 30 degrees high, using a 24 mm lens you might need 3 images across by 1 image high, but with a 100 mm lens you will need 12 images across by three high. That’s because the field of view covered by the longer lens is proportionately smaller than a wide lens. The result of this is that the panorama taken with a longer lens will have more detail, and be substantially larger in terms of file size.
The orientation of the camera is also important. With the camera in portrait orientation, there will be less distortion in each image as you sweep the camera across the scene. That’s because you are taking a narrower slice of the scene in each image.
A Stitch in time
Once you’ve taken your shots, whether using a motorised head or a normal tripod, then you need software to “stitch” them all together into one seamless image. For anything more than two images this is extremely difficult to do, but luckily you don’t have to.
There are two main applications available for creating panoramas from your images; Autopanoand ptGui. I’ve used both, and found Gigapano easier to get grips with. Both have their strengths and weaknesses of course, but free trials are available for both, so you can decide for yourself which is best for you.
The software does two main things; it finds the overlapping features in the images in order to determine what image goes where, and secondly they perform deformations in order to create a flat image. The first function (stitching) I have already covered, but what’s this deformation stuff about?
The deformations are necessary because when you swivel the camera around a fixed (pivot) point we are effectively taking shots that create the inside of a cylinder. Left untreated, our panorama would look quite odd, as if the centre of the image was flat but everything to the left and right was streaming towards us like a warp speed special effect from a science fiction movie.
The panorama software carries out deformations (also known as transformations) on your images to correct that effect. In fact, they offer a number of different types of corrections, cylindrical, spherical, planar and more. You can also change the part of the panorama that the software treats as the centre of the image. Normally that centre would be the point where your camera was shooting from, but you can tweak it to achieve your desired effect.
The software is more than good enough to even stitch together images you have taken without benefit of a tripod, just the good old “swivel at the hip” method.
When you generate your panorama, the gigapixel aspect comes in to play. As you would expect, the more images your panorama is made up of, the bigger the size of your resulting panorama file will be Due to the large amount of data involved, creating the panorama can take a very long time. A landscape panorama I took of a local beauty spot contained 220 images and took six hours to render. The resulting file weighed in at almost 2.4 gigabytes.
If you intend to do anything with the image afterwards, like edit or colour correct it in Photoshop, be prepared to work in slow motion. Photoshop is quite good at tiling images in order to handle really large files, but the initial load time will still be substantial. The more RAM you have, the better, and either a RAID hard drive system or an SSD would be advisable.
Sharing the result
Now you’ve got your panorama, what can you do with it? You could get it printed, lots of companies have the facility to create wide prints, though probably not as wide as you can create. The other option is to put your creation online. It’s not as simple as just posting in a web file format, as your file will be too large to be practical. If you resize it drastically to fit within the normal screen resolution, it will probably appear like the image below, wide and short. (Click the image for a slightly larger version)
Instead, you can upload the full sized image to one of the panorama hosting services, such as gigapan.org. These use a specialised image browser to allow users to zoom in to your image, revealing the detail buried in your gigapixel image. On gigapan.org, you can pick points of interest on your panorama, and share them with viewers, to help them discover the details.
- Merlin Motorised camera mount (with tripod)
- Bluetooth module (plugs into the Merlin)
- Nokia N85 or laptop computer with bluetooth
- Lots of AA batteries 🙂 (Kolor.com sells an adaptor that lets you use beefy camera Li-ion batteries)
- Papywizard panoramic control software (free, runs on the N85 or laptop)
- Stitching software (ptGUI and AutoPano are the two main players)
- Potentially HDR compositing software, such as Photomatix
The Merlin head cost around €169 from the UK, and I bought the Nokia N85 on ebay for €60. I was originally using my 13″ laptop, but the weight of that plus the Merlin and the tripod, was just too painful. As well as having built-in bluetooth and running the Papywizard software, the N85 is actually a fantastic little gadget. It’s not a phone as such, it’s a touch-sensitive Internet Tablet with wireless (as well as the previously mentioned bluetooth). It’s also got a proper operating system, and has a large community still developing free apps for it. Very cool indeed, not quite as pretty as my iPhone, but then again it didn’t cost €300 either 🙂