Why would you want to get a 3D model file of your room? Well, there are a couple reasons. If you are an animator, you may want to scan a room to get a realistic setting for animating. Another reason, is that you can use a program to explore your room in 3D. And finally, you could cut out the roof of the 3D model in a 3D modeler pogram and 3D print it! You could even get a color version!
Scanning your room is a bit difficult, and will probably take about an hour to get a really nice model. This is especially fun if you have free time!
Step 1: What Is Required?
There are some things you will need. First, you will need a Kinect (of course!). You can buy one on Amazon for $140 here, or you could probably find one for a lot cheaper on eBay.
Then, you will need a power adapter for the Kinect. Normally, the Xbox would provide the power when you are using it to play games, but it won't get power when you use it on a computer. You can buy one for $8 on Amazon here.
Then, you will need the 3D scanning program. You will also need a mac. It may work on Windows 8, but it hasn't been working for me for some reason. You can get the Pro version of Skanect (the 3D scanning program) for $130, or use the free version. Unfortunately, you cannot really export with the free version, so if you really want to use the 3D scan you will need the full version.
Finally, you will need a messy room. Yes, you read right. The scanning actually works best if there is a lot of texture in the room. It helps it to recognize where it is in the room, relative to where it has already scanned. So an empty room will be hard to scan. It will help if you get our brothers together, add a box of Legos, add a box of toys, then close the door and come back ten minutes later. Unfortunately, you will have to find your brothers in the mix and take them out, as moving objects will mess up the scan. Legos especially will help when scanning the floor!
Step 2: Starting a Scan
First, make sure you have the Kinect plugged into the computer through the power adapter, and the Kinect is plugged into an outlet. Then, start Skanect (after you have installed it, of course). Then, change the settings to look as shown above. Now, press "Start".
You will see three smaller screens, and one large one. The first screen detects colors, the other two are depth sensors. The large screen is all three put together.
Now, you want to start the recording. But before you do, point the Kinect towards a place that has a lot of texture. Bookshelves, unmade bed, messy floor, those will all do. You can see above the starting position I took. You will have a three second countdown (you can adjust this) once you press the record button. When you press it, it should look like the green picture above.
Step 3: How You Should Scan, and How You Shouldn't
Now, you will want to SLOWLY rotate the Kinect around. Try concentrating on one object at first. You can see the GIF above to see how to do this [PIC 1]. Very carefully, you will have to crouch down a bit to get the bottom of objects (in this case the bookshelf), get the sides of an object, etc. One strange feature with the Kinect, is that it will not record shadows. This includes dark/black objects, as you may see in the GIF above [PIC 1]. The top of a suitcase was black, so there will be a hole in that later. But right now, that shouldn't really be a problem. Another awesome awesome feature, is that it can scan dark areas that even a camera could not capture. For example, I was able to scan the inside of a closet, even though it was a bit dark back there.
What You Should Not Do:
Do not start on a blank/low textured object, such as a wall. You can see what happens in the GIF above when I started the scan pointing at a blank wall [PIC 2].
Do not let a blank/low textured object take up more then 50% of the screen at a time, if not it will error. It errors because it cannot tell where it is, or if you are even still moving! You can see what happens in the GIF above [PIC 3].
Do not get too close to an object, even if it has a lot of texture. If you get too close, it will give an error. One reason is that all three sensors on the Kinect aren't put into one exact spot, so they will start seeing different things. Of course, getting too far away will not give you very much detail, especially in the colors. The closest you should get is 3 feet for any object. Getting too close may mess up the scan, too! You can see a visual picture of how close you should get above [PIC 4].
Obviously, don't move too fast. The only way that it knows where things are, is to match things onto things it has already scanned. Moving fast, especially to a place where it has never scanned, will mess it up, and will either error or add objects to wrong places in the model.
Step 4: Fixing the Model and Exporting It
Now, you will want to fix the model. Even a complete scan that took 20 minutes to do has some errors. As you can see in the first picture, the bed has holes in it, the bookshelf has a lot of holes in it, and the bed on the left doesn't look to good.
The first thing you need to do, is skip straight to the process section. Now, on the sidebar, click "Fill Holes", then start. It will take about a minute for a very complete scan, but for the scan I did (only scanned for a bit more then a minute) it will only take about 8 seconds. Now, your model will have mostly filled holes. You can see the difference between picture one and picture two. Now, this part is optional. If you want to 3D print it with color or animate it, you will want to add color. But most of the time, adding color to a 3D print is a bit more expensive, so you not want to do it all the time.
On the sidebar, click Colorize, then start. It may have an error, saying that to inpaint the colors you will need to simplify the model. Inpainting means it will also try painting the holes that were filled. If you have large holes that were filled, then you may want to use the Simplify option on the sidebar. I did not have to do this my scan, as the parts that were filled were not very big.
Now, you want to export it. This is where the Pro version comes in. Each model has a certain amount of sides, or vertices. The model I scanned had 219 thousand vertices. The free version, however, can only export 5 thousand vertices for each model. That is a tiny fraction of the model's vertices, which means that it will remove some of those vertices. A perfect sphere, for example, would have a ton of vertices. A square only has 6 vertices (also called sides). If you tried cutting down the sphere into 4 vertices, then it would trianglish sphere, with only 6 straight sides. This is what will happen to the model. It will become a lot less rounded, and won't look very good, especially if you have a large scan. So if you really want to use this model, you will have to save it as a Skanect file, buy Skanect, then open the Skanect file and export it.
You can see how to export it in the last picture above.
If you think this is a great Instructable, even if you don't actually follow it, please favorite it! It would be great if you could also vote for it in the Coded Contest, too. Thanks!