VR application created with point-cloud data.

Hey, first post here, hello all!

I have just started messing around with TD and I already have an idea for a project.
Some time ago I saw this; vimeo.com/89982874. And now I’m wondering if such visuals can be created with TD. I saw websites which publish free laser scans of various areas and objects but there are loadss of file extentions and I’m not sure where I can find some files that TD can read. I think they do come it PLY or E57 format usually.

So, in the end I would be very pleased if I can manage something like this;
youtube.com/watch?v=cdxTsFm58Bc

I do have a background with Processing and Python so I can assume a single csv file would be enough as long as it contains x,y,z and RGB in column/row format but can this be managed in TD and recreated in an aesthetic way?

Also I wonder if can I use my kinect2 to scan streets or where can I download some point-cloud files of simple streets or cities without being too heavy on my computer?

If this can’t be managed all together; I wonder can I upload a 3D model of some sort of city and write a simple patch so one can walk around in it in VR? I think most people would say something like would be easier in Unity but I want TD to be my main piece of software for this as I want to integrate music and sound in later parts as well.

Thank you very much and best wishes.

Yup very doable. The main part is getting it into a readable format. CSV would work fine, the file might be quite big and the initial load might be long, but if you wait it out it should be ok. What you’d want to do is take the data into a DAT and do as little processing on the DAT data as possible, since DATs can become quite slow with a ton of information. You can convert that DAT data into CHOPs and use that for instancing or feeding into a shader to create the scene. We helped a client working on a project exactly like those links, so I guarantee it’s possible :slight_smile:

Thanks for the reply and thumbs up! That is a big relief actually, because I was ready look for other softwares which might be capable of doing this (dunno which ones yet, now I don’t have to look :slight_smile: )

So, may I ask was the data you fed into it was also created by you (I mean did you use your own scans) or found it somewhere? For the first attempt, I think maybe I can find some pre-made patches to check or maybe can download a 3D scan of a small area (a street, or a few trees maybe?)

I checked loads of sites which offer some free lidar scans but none of them were in csv format and was left unsure how can I load them into TD. I also saw there are loads of file formats like PLY and E57, I wonder would they run faster during the loading or creating the environment? I mean compared to a csv file…

Thanks so much!

I tried to load 3 diff formats of point cloud in TD the most common, none worked, so I gave up.
But I’m well curious about the subject. Some websites offering caves pointclouds are cool they offer to let you download a portion of a massive pointcloud, only problem again wrong format.

That photometry stuff in 2nd example is cool, normally we can probably put an object on a spin plate and combine pictures and kinect scans t scan anything with texture (small objects) shame my dog wont stand still.

Not by me personally, the client was making the point clouds themselves though I believe. Have you tried looking for Python libraries that help deal with PLY or E57? I’m sure there are some that exist unless those are proprietary binary formats or something.

Thanks for the answers.

Greenpattern, I wonder what were those formats? Maybe, converting them to .csv files could work, I think I saw some point-cloud data converters somewhere but have to check again. Would probably be a hassle though…

Yea I think the second example would not be very hard for smaller environments and objects as kinect can capture both point-cloud and RGB values.

Elburz, no I haven’t checked actually but not a bad idea! Greenpattern, this could solve our issue of file formats, if there is none maybe I can try to code one as well.

Will write back soon.

My search was a while back, everyone seems most intrested by kinect rather.
I think most time the format was lidar standard *.bin.
In anycase its far too much info for a realtime soft to process.
I was hoping we could find a way of loading in cinema 4d then cut and simplify then to td in a chosen point cloud format.
I’ll search straight away for CSV format.

youtube.com/watch?v=OPCb4x5g8W8
so suited for vr, fly through caves like a bat top speed.

libpointmatcher.readthedocs.io/ … ortExport/

check this!

how is it going with your project :slight_smile: ?

i am also currently working on a point-cloud-based project abd curious, which route you went through and what kind of obstacles you had within that.

if you want to convert/mangle/retouch/edit your point-cloud, i would recommend cloudcompare for this.

if you want to do further animation in C4D or use data from maps for extracting further scandata, then LAZPoint could be an option.

Similar to elburz my way of working with point-clouds so far involved converting CSVs to EXRs in TD (mainly for speed of loading) and used that in GLSL.