@ajeromin I wonder if you were to combine the slam with photogrammetry solutions like pmvs2, the point cloud could be converted to a dense point cloud. I am working on my own solutions for better results
Hey, I use 3DS max and I just got a kinect. Would you be so kind as to post your modified software? I would love to do this, but I don't know much about software development. You can email me at robinsmangat@hotmail.com
@Metapharsical , that was my thought when i saw it. I've yet to try it, but I'm not entirely sure what will happen. The Nyko lens is actually widening the field of view (in order to see more in smaller spaces), so I'd have to move the scanner closer to the object to fill the same amount of screen space (sensor space) as I had without the lens right?
I'm thinking that retro-fitting the Kinect with a zoom lens set over the cam and IR projector would achieve more of what we're after.
@ajeromin That would be pretty interesting if you did that. So by narrowing the field of view with the zoom lens, you are dedicating the camera's sensor & IR projector specifically to the object of interest. There should be a tighter mesh pattern seen which would allow for more data gathered and "higher resolution" where you want it. Creative thinking by you!
@ajeromin yes, I was mistaken. it's a laser point distribution, widening would lower your dpi. It would only be useful in this arrangement when the object had such large concave surfaces that they would be occluded by the current angle of view. That said, it would be rare, and would only happen on the vertical axis because you're rotating the rest. You're right, I dunno, take the nyko and flip the lenses 180° ?? Do you need specific IR filtering lenses or is that redundant?
The markers are just giving the software something to track visually instead of a plain white box. RGBDemo's reconstruction code works by tracking pixel info from the color camera. I didn't change any code for that. Nicolas' code will work fine as is if you build this.
Nice work! You mention some modifications to RGBDemo. Did you actually change Burrus' code or are you just using the RGBDemo library to build this application? What are you actually doing with those markers? Would love to see the code, is it online anywhere?
Not much into to give really. I use RGBDemo 0.6.0 by Nicolas Burrus (Googlable), and I built a spinning foamboard box with pieces of tape on it so the program can visually track it's rotation. Easy breezy! :)
correction, sorry u use the rgb!!!
torontobro 5 days ago
The image is IR so what's the led lighting for exactly?
torontobro 5 days ago
Why the board need to be posted with tapes?
tintinbaby7681 2 months ago
Tried the Nyko.
It does what I suspected. Makes the field of view wider. So even if I mice the Kinect 40% closer, it sees the same as if I had no lens.
Boo.
ajeromin 2 months ago
@ajeromin I wonder if you were to combine the slam with photogrammetry solutions like pmvs2, the point cloud could be converted to a dense point cloud. I am working on my own solutions for better results
goomee3d 1 month ago
Ball pivoting. Yes. :)
ajeromin 2 months ago
By "rolling ball" did you mean "Ball pivoting" surface reconstruction in meshlab?
chytun 2 months ago
This has been flagged as spam show
Hey, I use 3DS max and I just got a kinect. Would you be so kind as to post your modified software? I would love to do this, but I don't know much about software development. You can email me at robinsmangat@hotmail.com
squigglewacks 4 months ago
Could you increase the resolution by using the 'nyko zoom' attachment to the kinect?
Metapharsical 5 months ago
@Metapharsical , that was my thought when i saw it. I've yet to try it, but I'm not entirely sure what will happen. The Nyko lens is actually widening the field of view (in order to see more in smaller spaces), so I'd have to move the scanner closer to the object to fill the same amount of screen space (sensor space) as I had without the lens right?
I'm thinking that retro-fitting the Kinect with a zoom lens set over the cam and IR projector would achieve more of what we're after.
Thoughts?
ajeromin 5 months ago
@ajeromin That would be pretty interesting if you did that. So by narrowing the field of view with the zoom lens, you are dedicating the camera's sensor & IR projector specifically to the object of interest. There should be a tighter mesh pattern seen which would allow for more data gathered and "higher resolution" where you want it. Creative thinking by you!
mados123 5 months ago
Comment removed
Metapharsical 4 months ago
@ajeromin yes, I was mistaken. it's a laser point distribution, widening would lower your dpi. It would only be useful in this arrangement when the object had such large concave surfaces that they would be occluded by the current angle of view. That said, it would be rare, and would only happen on the vertical axis because you're rotating the rest. You're right, I dunno, take the nyko and flip the lenses 180° ?? Do you need specific IR filtering lenses or is that redundant?
Metapharsical 4 months ago
@ajeromin Nice work! have you tried the Nyko lens or any other one ? It would be very interesting to know how it worked.
hansolduron 2 months ago
This sort of research will change the face of independent game development
xxYetterxx 6 months ago
Thank you. I've done both.
This video shows me rotating manually.
I now have a 1 rpm motorized lazy Susan I made from a surplus motorized window curtain motor.
Good luck! Thank Nicolas for his software when you get a chance! :)
ajeromin 6 months ago
Thanks for the info. Do you have a motor in the box or does it work manually? I'm off to build my own now ->
Loiske4real 6 months ago
very cool demo, thanks
funkboxing 6 months ago
The markers are just giving the software something to track visually instead of a plain white box. RGBDemo's reconstruction code works by tracking pixel info from the color camera. I didn't change any code for that. Nicolas' code will work fine as is if you build this.
ajeromin 6 months ago
Love 3DCoat. :)
ajeromin 6 months ago
That's pretty cool. Try 3D-Coat for your retopo, I think you'll be glad you did.
philnolan3d 6 months ago
Nice work! You mention some modifications to RGBDemo. Did you actually change Burrus' code or are you just using the RGBDemo library to build this application? What are you actually doing with those markers? Would love to see the code, is it online anywhere?
Thanks!
atduskgreg 6 months ago
haha... the other side of the blender model was broken... bad^^
ChihabMansour 6 months ago
'info'
ajeromin 6 months ago
Not much into to give really. I use RGBDemo 0.6.0 by Nicolas Burrus (Googlable), and I built a spinning foamboard box with pieces of tape on it so the program can visually track it's rotation. Easy breezy! :)
ajeromin 6 months ago
nice work. Could you provide some more info about this so i can build my own?
Loiske4real 6 months ago
Damn man, very impressive! Just subscribed, can't wait to see a system with multiple Kinects.
DoctaProfessah 6 months ago
whoa... bad ass!
jcpapova 6 months ago
awesome work
claan1 6 months ago
Nice!!! Great work AJ!
nickzucc 6 months ago
I threw up in a good way. Awesome tech!
baue021 6 months ago