Do I need Houdini to do this or can I use another software?

[Shuey1971]

Hi fellow Houdini members,

 

So I've been looking to use Houdini to do something like this. Is she also using other softwares? If so, which ones?

 

https://vimeo.com/121436666

 

The question is whether I need houdini to do it? The person in there is using Houdini and Houdini seems like an awesome product which I want to get involved in. Is there another software that can do this as effectively or nearly but the learning curve isn't as steep?

 

I have no 3D modeling experience but I'm a technologist who uses OpenFrameworks/Cinder and is a use to using C++ not so much of python though. I've read that Houdini has a ridiculous learning curve, very steep according some on the houdini forum as oppose to other 3D softwares. I may not have the time to do get into Houdini as of now.

 

I only have a few days to make the deadline and all, but it doesn't have to be as brilliant as Asphyxia is. I want to make it that brilliant though eventually by the end of May. This will give me more time.

 

I'm use to using Kinect so the only problem i'll have is learning to use Houdini to do something like this. I have all the 3d point cloud and everything down transported as Objs, etc. When I do transport them into Houdini they show up as Mesh. And I tried changing it through materials list but it doesn't look like the point cloud that I imported. Looks like a watered down version. How do I solve this?

 

If I do use Houdini is there something in Houdini i should focus on? Like particles, animation, etc. Its a very vast program and I'm not sure where to start learning what I need to do for the project. Recommend some learning sources please

 

I do have access to Maya/Cinema4D/Blender if those can do this as well.

 

Thanks guys!

Edited March 28, 2015 by Shuey1971

[edward]

Use Houdini. For someone who already understands CG theory, it should be much easier.

In the video, I see: Some program that connects with the Kinect and outputs the points(?), Houdini for the point effects, Maya (for the environment?), and some compositor being used. But then I just skipped through the video. Probably others as well. If you can get out the point positions to Houdini, then I think Houdini is all you need.

[Shuey1971]

Use Houdini. For someone who already understands CG theory, it should be much easier.

In the video, I see: Some program that connects with the Kinect and outputs the points(?), Houdini for the point effects, Maya (for the environment?), and some compositor being used. But then I just skipped through the video. Probably others as well. If you can get out the point positions to Houdini, then I think Houdini is all you need.

Hey thanks for getting back.

 

So I was able to get the points out to Houdini but it looks nothing like my original.

 

It looks like that.

 

While my original looks like the second image

 

Basically it looked like a mesh at first in full geometry settings until I went into the rendering settings and had it display at point cloud. Render>point cloud. 

 

Also, 

Is there a way I can edit each one of those point clouds to be bigger in pixel size or change it's geometry of the individual point?

 

Is there a way I can separate the points?

 

This is project is done in a sequence. It's basically a sequence of .obj.

Edited March 28, 2015 by Shuey1971

[georgeivan]

Something like this?

 

AddSop its your friend,that and the copy node

[edward]

So I was able to get the points out to Houdini but it looks nothing like my original.

What do you mean? The point positions look correct from the images you posted to me.

 

Basically it looked like a mesh at first in full geometry settings until I went into the rendering settings and had it display at point cloud. Render>point cloud.

Ok, so it looks like a bunch of points and not a mesh because you deleted the mesh!?

 

Is there a way I can edit each one of those point clouds to be bigger in pixel size or change it's geometry of the individual point?

Search for "Render as Points" here. So you just set the "width" attribute to a value larger than the default. If you want to render them as spheres, change the "Render as Points" property.

Is there a way I can separate the points?

Define "separated". A point cloud is by definition unconnected.

[dpap]

So I was able to get the points out to Houdini 

 

Could you please elaborate on how you achieved to import Kinect data in Houdini? Did you use another software to do that ?

[Shuey1971]

@georgeivan

Hey thanks,

Will look into addSop.

 

@edward

Thanks for getting back,

 

Yes, the points may be correct but it doesn't look like what I imported. The colors are off, unless I have to change them. I can't see the vertical lines in my houdini import where I could see them before like this in the image posted.

 

No I didn't delete the mesh I just changed it from full geometry to point cloud. I can change it back and it would look like a solid mesh like object thats filled in. 

 

Thanks will look into that "render as point" function and get back with progress.

 

By separation, is there a way I can separate each point from the point cloud to move to a different position in space? Like the cloud is  a series of points. Can I move an individual point away, basically subtracting that point? In the first Image I posted. There is a background of point cloud. It's in the movie format. How would I get rid of it? It's a video of point clouds so I'm not sure if I can.

 

@dpap

 

So what I did was use cinder/OpenFrameworks from a program I made to capture the kinect data and then transferred it as an obj file. It was a video and it changed to a png sequence. from there i changed it through my program settings to import the sequence as an obj sequence. Imported then "1-482" obj files as a sequence into Houdini. It became a point cloud 3d object video which is what I wanted.

Edited March 28, 2015 by Shuey1971

[Shuey1971]

Use Houdini. For someone who already understands CG theory, it should be much easier.

In the video, I see: Some program that connects with the Kinect and outputs the points(?), Houdini for the point effects, Maya (for the environment?), and some compositor being used. But then I just skipped through the video. Probably others as well. If you can get out the point positions to Houdini, then I think Houdini is all you need.

 

Also do you know anything that can output the Kinect into Houdini as well? For an OS X user.

[edward]

Yes, the points may be correct but it doesn't look like what I imported. The colors are off, unless I have to change them.

The .obj file format doesn't support point colors. So you're trying to do a texture look up for the point colors? Offhand, I'm not sure about the support level for texture look ups when rendering points.

If you use the .ply file format instead, then you easily add the point colors by naming the "R G B" components as "Cd1 Cd2 Cd3" in the .ply file. Then you can see them in the viewport.

I can't see the vertical lines in my houdini import where I could see them before like this in the image posted.

The vertical lines? Your Houdini screenshot has *NO* lines.

By separation, is there a way I can separate each point from the point cloud to move to a different position in space? Like the cloud is  a series of points. Can I move an individual point away, basically subtracting that point? In the first Image I posted. There is a background of point cloud. It's in the movie format. How would I get rid of it? It's a video of point clouds so I'm not sure if I can.

Apply any operation that changes the point position, deleting them if you want as well.

Also do you know anything that can output the Kinect into Houdini as well? For an OS X user.

I've never tried to use the Kinect.

[Shuey1971]

The .obj file format doesn't support point colors. So you're trying to do a texture look up for the point colors? Offhand, I'm not sure about the support level for texture look ups when rendering points.

If you use the .ply file format instead, then you easily add the point colors by naming the "R G B" components as "Cd1 Cd2 Cd3" in the .ply file. Then you can see them in the viewport.

The vertical lines? Your Houdini screenshot has *NO* lines.

Apply any operation that changes the point position, deleting them if you want as well.

I've never tried to use the Kinect.

Hey Edward,

 

Thanks for getting back. Everyone has been a big help. I recently figured out how to accomplish this project with Houdini perfectly. Thanks everyone for the help again. I will come back and post progress and more questions with this project in this specific thread so expect something soon.

 

Currently rather than focusing on the Kinect, I'm  trying to learn Houdini thoroughly and experiment with the program. It's quite amazing.

 

@Edward. The program that guy uses is called Brekel my professor told me. It's a paid program (low grade free version as well) that transport kinect data as an obj into 3d softwares like Max/Maya/Houdini. Incase you were wondering as well. It's actually one of the select few products that do this and I don't know why no one has developed a free version of this.

 

Cheers,

Shu

[Shuey1971]

Hey. So I used Brekel to transfer my data into an Alembic file. The abc file when I tried to import it into Houdini nothing showed up. I've only screenshot half the screen to show the other window.

Edited April 5, 2015 by Shuey1971

[Shuey1971]

Imports perfectly in Maya, but for my purposes I need it to work in Houdini.

Edited April 5, 2015 by Shuey1971