Hello everyone and welcome back to
another 3D Revolution. In this video I'm going to be showing you how you can 3D
scan objects in the real world using an Xbox 360 Kinect There are several different ways of generating a digital 3D model from a real world object each have their own advantages and disadvantages.
You could buy an Xbox 360 Kinect which costs less than 15 pounds online, is easy
to set up and use, and gives pretty good results; you could do photogrammetry
which involves taking photos of an object from every different angle and
using some clever software to generate a model from them, you could use a handheld
3d scanner, and then there's also industrial 3D scanners. Now if you're
watching this video you're probably just doing this for home projects and you're
most likely going to be looking at using a Kinect or photogrammetry. For this
video I'm focusing exclusively on using a Kinect but subscribe to my channel
I'll soon be uploading a tutorial video on how to do photogrammetry as well as a
comparison video looking on the differences and advantages for each
method and what may be most suitable for you. When you're 3d scanning with Xbox
Kinect there's a few options to choose from when deciding which software to use.
I'll be using a scan 3d for this video in this video I'll take you through
every step of 3d scanning an object with an Xbox Kinect
this will include preparing your object for 3d scanning, setting up your Kinect,
looking at the software, the scanning itself, cleaning up your scan, and finally
exporting a model that could be 3d printed or imported into an animation
program. The Kinect can give really good results especially when you're scanning
the right sort of subject. It can struggle a bit with quite small or
detailed objects so to give you the best example of how well it can work for this
video I'll be scanning my Iron Man helmet so let's begin. Place your object in an area you can easily move all around it or place it on a surface you can rotate so you can scan it from every angle, I'm using this lazy susan for my
Iron Man helmet. Light your model as evenly as possible, this is less important if you're not going to be using the photo texture on your model,
but it will still help improve the performance of your 3d scan. Once your Kinect is plugged into the mains as well as a USB port in your computer you should see a live feed of the video on the right hand side of K-scan if you
don't and if the Scan button on the top-left is grayed out click devices and
then reconnect. This forces K-Scan to search for compatible devices connected
to your computer and should bring up a live feed from your Kinect. If it does
you can click back to projects. For a high-quality scan we're going to need to
view our model from as many angles as possible so to begin with set up your
Kinect at a 45 degree angle looking down on your object. Now let's take a look
what the Kinect is actually seeing and adjust it so it's ready for the scan. If
you look at the live preview on the right hand side here you'll see a lot of
black patches, these are areas where the scanner isn't correctly picking up the
information it needs. Ideally your object should be 50 cm away from your
scanner so if you move your Kinect back you can see these black patches
disappear and you get a much cleaner image. Now at the top you have bars which
are just what area your scan is actually looking at.
The top two sliders crop your image reducing the amount of data your
connector's receiving, the bottom two do the same but in depth, so the distance
between your Kinect and the far wall. You can use these sliders to make sure your
Kinect is focusing just on the subject matter you want to scan,
however as we're going to be cleaning up our scans anyway we may as well leave
these all at their full parameters to make sure we capture every bit of detail
we need Now everything set up and ready to go we can get ready to scan. If you make sure you're in the scanning tab in the top left and then go to the meshing
window below this top drop down should have three options: mesh, points and
capture only. Mesh will capture each scan convert it into a mesh and try and align
it with the previous ones, points will do the same thing but instead of converting
it to a mesh it will remain as a points cloud, 'capture only' is the option we're
going to be using. This won't try and align the images as we go and we can
align them after we finished our scan ensuring that nothing accidentally gets
misplaced. When using capture only the alignment
and density modes are grayed out and these aren't features we're going to
need to use today. Underneath this you'll see the scanning window. The first option
here is 'enable batch scanning', this can be really useful if you're doing a scan
of an object which you need to walk physically around as you can set up how
many scans that you want it to do, and how often you want its capture them. This
means that you can walk around the object and allow the Kinect to
automatically take the scans one by one. However when you're rotating your object
and your Kinect is staying in the same spot, I personally find it much easier to
manually click 'scan' each time I wanted to take an image. If I click this now
you'll see a scan appears in the column on the right. To manually click scan
without having to run back to my computer from the object each time I
just use a wireless mouse. Now let's get scanning. This is the easiest step simply
click scan, rotate your model slightly, click scan again and repeat this until
you've done a full rotation of your model.
As you do this you'll notice each of your scans appearing in the list on the
left hand side of the screen. Next, move your Kinect down to the same
height as your object, do another rotation of scanning, and finally move it
down to a lower point than your object and do a final rotation of scanning this
way you've scanned your object from three different heights. Now your scan is complete you'll see a full list of all your captured images on the left hand side. You can reveal & hide these using the tick boxes to the left. We're now going to go through these one by one and delete anything that isn't part of the
object we want to actually capture. You can move around just by clicking and
dragging with the left hand of the mouse. To delete an area simply hold ctrl and
then draw with your left mouse. If you select an area like this and then click
delete in the keyboard it will just remove it from the point cloud. Rotate
the point cloud and delete anything that isn't part of the model you want to
capture once you're happy with an individual image you can click the tick
box again to hide it and click Save. You can then move on to the next one. Now
repeat the same task for every image that you've captured for this scan, this
can take some time but it's definitely worth it as this could make a massive
difference when you're aligning your images ready for your final model Reveal all of your scans at the same
time by selecting one pressing ctrl+A and then clicking one of these tick
boxes you'll reveal the entire point map here but you can see at the moment
they're all misaligned. Navigate to the mesh editor tab at the top and click the
build button they should take each of your scans in turn and try and align
them with a previous one. You should now see your model start to take shape, now
don't be alarmed if your model looks like some sort of weird camouflage this
is absolutely normal as what you're seeing here is every one of your scans
layered one on top of the other. The software's not quite sure which ones
prioritize and as a result you get this weird mottled effect but this won't be
there in the final model. What you can do is use the smooth option on the left
hand side here which will help remove any sharp edges which are currently
showing on your model, once you're happy with your layout click the combine
button at the top this will do a final aligning of all of your scans and merge
them into one final collaborated image This isn't your final model so don't
worry that it still looks a bit bumpy and lumpy all we need to do now is click
the finalize button and this will generate our final model which we can
then export You always want mesh density as high as
possible to ensure a decent quality model but hole filling may depend on if
your model already has holes in it that you don't want to lose, once you're happy
though click OK and it'll generate your final model. And there you have it there
is our 3D scanned Iron Man helmet. I personally think that for a £15
piece of equipment this is an amazing result. Now you can see that at the
bottom there it's still open but that's something that can easily be
fixed in something like mesh mixer which has an easy-to-use hole closing system To export simply go up to the export
button select wherever you'd like to save it to and click OK Well I hope this video has helped you
better understand how to 3D scan objects with an Xbox Kinect remember I'm going
to be uploading a tutorial soon on how to do photogrammetry as well as a
comparison video between the two so make sure you subscribe to my channel and
I'll see you next time Thanks for watching and remember hit
subscribe for more news, tips and livestreams
on everything 3D and tech
