Engineering student project advises

by Lonavar on 04/07/2014 01:50, 9 messages, last message: 04/09/2014 06:17, 1366 views, last view: 04/22/2024 10:20

Hi,

We are a team of 5 students working on an indoor maping project.
The project main point is to map an all indoor bulding, flat or others indoor places, creating 3D models for video games (at first) but can be used too for many uses.

Using kinect as hardware part for mapping many rooms, and then creating a software which can modelize one big mesh from all the little one created before.

What are you thinking about this project, did you think it can be a good idea ?
Would you like to play in your own flat ?
If you have any tracks that we could follow, please answer this post.

   Board Index   

#1: ..

by chasester on 04/07/2014 05:06

ummm im still lost on exactly what youre doing..

Kinkect as in microsofts xbox one/xbox 360 device

Second by mapping you mean to scan a room to basically form 3d models out the surrounding object? is that even possible?

reply to this message

#2: Re: ..

by Pyccna on 04/07/2014 20:45, refers to #1

I think he means using an Xbox Kinect to map so the Kinect provides the mesh for the map.

Like what game companies do to scan actor's faces into games, except with a building.

reply to this message

#3: Re: ..

by suicizer01 on 04/07/2014 21:07, refers to #2

I wonder why it shouldn't be allowed...

reply to this message

#4: ..

by Pyccna on 04/08/2014 02:09

Probably should make sure Microsoft is okay with this before you start it.

reply to this message

#5: Re: ..

by ImQ009 on 04/08/2014 07:49, refers to #1

Of course it is, I've seen people doing the exact thing for Minecraft

reply to this message

#6: ..

by chasester on 04/08/2014 21:34

You cant scan a panoramic environment with a simple 3d camera 0.0 at best you can get a general idea of the space from one plane but you cant form 3d models from a singles sided 3d dimensional scan ? Plus even if you did develop any way to create models from this scan the amount of polygon faces would be too intense to be used in real-time rendering, you would need to do a polygon face reduction algorithm to make your models less graphically intense. Which as far as i've seen (maya being the best one) it is a very trial and error system, and would be a challenge just to get a passable model.

But good luck any way :)
chasester

reply to this message

#7: Re: ..

by suicizer01 on 04/08/2014 22:48, refers to #6

You certainly can if you would give each dimension a certain colour (like red as front face, blue side face, green up or down face). Perhaps not with a kinect but even a cam does that job.

reply to this message

#8: Re: ..

by Lonavar on 04/09/2014 01:22, refers to #7

We are using the Kinect Fusion : this is différent from the xboxKinnect and only for devellopment so far.

I mean doing basically something like one of this project :
- Matterport
- Kintinuous
Check this if you can.

Right now the biggest difficulties are :
Like you siad it the Kinnect is creating to many polygons when creating models..
I'll check your "maya" algorithm

Plus will he have to make an other algorithm to merge différent recordings and how to make a loop closure (which borders belong together)

The Kinnect Fusion is already doing 3D models from instant record, but for now its only working with objects or very small environment.

reply to this message

#9: ..

by chasester on 04/09/2014 06:17

by "maya" i mean the 3d modeling program made but autodesk, and i doubt that the algorithm is "open-source" :)

But it basically does some lose forms of path finding to determine how much a shape would change if a particular vertex/line/face were to be removed

I take back my other complaints ;p i thought you were talking about the kinect device by microsoft.

chasester

reply to this message

   Board Index