Oh yeah, that'd be just great for your city when your high-land-value dense commercial district gets dug up by a giant catzilla needing a dump! How you gonna deal with *that*, Mr Mayor?
Yes, the one-second delay you can see in the videos is caused by the foreground detection filter. The idea is to keep a fixed-length history for each pixel (1s by default), and only update a pixel's output value if its history buffer has a variance smaller than a given threshold. On the upside, it works really well; on the downside, it causes a (short) delay.
Throughout the whole thing my heart kept pounding because I was afraid he'd overfill it with water and it would seep out.. I guess this is what it's like to feel outdated.
That's where the 1-second delay comes from. Anything that doesn't move for 1 second is assumed to be sand, and added to the topography model. It works very well, unless a user holds their hand or arm very still. You can see that a few times in the video. The 1-second delay is not really annoying because when you work with the sand, you see it changing immediately. It's just the virtual topography that's updated a second later.
Love this!! Kids would love it, great sensory activity! And learning about Topographic Maps maps sweet!! Soothing.. so many neat uses... especially for the Sensory kid.
This can certainly be used for more than just that. The amount of rapid prototyping from something like this could help a lot of designers. I really want to use this thing.
Pretty awesome water simulation by using actual sand to scan a heightmap. Looks very realistic and scanning the height of the sand is very cool. Just the scanning part is awesome enough, the water simulation being applied to the heightmap is icing on the cake.
I'm a preschool teacher, and I'm fairly certain you have invented the sandbox of the future. Years from now, all of my kids will expect augmented reality in the classroom!
It's a Linux box with a Core i7 CPU and a Geforce 580. Most of the computation is done on the Geforce. The water simulation is very computationally intensive, and pushes the 580 to its limits. You can see it chugging a bit at several points in the video.
I'm not sure if I understand the question correctly. The Kinect only scans inside the bounds of the sandbox, but any non-sand objects you place on the sand will be picked up, and affect the water flow. During testing I used Lego bricks to build structures or levees, for example.
The 1 second delay is to reduce jitter and, more importantly, filter out hands and tools so they don't interfere with the terrain. If it bugs you so much, you can turn if off by passing "-nas 1" to the AR sandbox executable. The delay would then be between 20 and 25 milliseconds. You're welcome.
This video has been selected by the channel as one of the Most Amazing videos on TH-cam and was added to the channel's playlist accordingly. Thanks for sharing. Regards, bestofYTchannel
The "rain maker" is based on depth as well. It detects non-moving objects inside a configured elevation range. I'm still working on improving that so it's harder to make it rain accidentally. Color-based detection does generally not work because everyting takes on color from the projection.
Fair enough. As far as the simulation goes, it's pretty much top of the pops, as it's a second-order accurate time discretization of the mathematical equations governing shallow water flow. If there's anything better out there, someone needs to tell me. The *rendering* of the resulting water surface, however, is a different story. It's non-photorealistic. I chose this model because it highlights the motion of the water at the scale viewers see the box.
It's a grid-based explicit simulation, details are in the Kurganov&Petrova paper I cite in the description. It's the same method used by civil and hydrological engineers and scientists to simulate water flow in real environments. It's not double-precision (as in floating point) because it turns out the method is stable under numerical round-off, but it's second-order accurate in the numerical sense.
I am research faculty at the University of California, Davis. My job is developing scientific visualization software for virtual reality environments, particularly for the Earth sciences. This is a side project.
little kids playing in the sandbox see this way anyway, it's us adults that need the high tech stuff to get enthused about sand castles again. Great work, Next time I take a trip I'm coming over to your back yard.
What everybody else said. The virtual sandbox is not a simulation of a real sandbox, but of a real landscape. To be precise, the water simulation runs at a 1:100 scale.
It's not done on a CPU, but on a graphics card exploiting its massively parallel processing power. The actual rig is a PC with a Core i7 CPU and a Geforce 580.
It was a lot of work! The hardest part was that I was trying to make it seem as if there were lag -- you know, to make it more like what people expect of Kinect -- and therefore had to anticipate when the projection was going to change one second before it actually did. Took a LOT of trial and error to get that right. In retrospect, it would have been easier to just built the damn thing for real instead of faking it.
The point of tech demos is to demonstrate certain functions of a creation. Some may help further research into a particular subject. Creation is a base of research. So ya, there's a point to this.
That's what I'm trying to figure out... the open boundary conditions I implemented may induce a very large outflow in the outermost cell layer, which would limit the simulation time step. Or it could simply be a bug.
12 ปีที่แล้ว
It's fun how if you leave your hand still for a moment, the topographic map includes it too.
The idea was that you can make water flow out by diggind a trench to the edge of the box. After all, we're trying to simulate landscapes, not sand boxes. :) Without the height indicators, it looks a bit boring. Everything can be disabled independently via command line switches, if one wanted to do that.
How-tos and software will start materializing on my web page in the near future. After all, this is just a prototype, and it is intended to be cloned by science museums. Hardware cost (gaming PC, projector, Kinect, sandbox material) totals between $2000 and $2500.
I think as far as the system is concerned, anything sitting still inside the sandbox relatively near the surface is "sand". If you put a ball on the landscape, it would be treated as a hill. The narrowing contours below the widest part of the ball would not be seen by the system and can't be rendered.
Your Video is a GREAT Inspiration to help ADULTS to REMEMBER TO PLAY. They say Creativity is found in Play.... The POWER in PLAY is deeply Misunderstood. If People KNEW what PLAYING will DO they would ALL be doing MORE of it.
You, sir, are VERY fucking clever ;D It's amazing how well you've manipulated the Kinect System and somehow programmed it to work the way it did. Pure Genious!
This is really really facinating. Reminds me a lot of the holograms in star wars.. I wonder if you could connect VR to it and have a top down 3d aspect to it without the sand.
Sand is real, water and topographical colors are not. Kinect has the ability to see 3D shapes which is what this person is making in the sand. Very cool stuff.
This is one of the coolest applications of AR i've seen so far, considering that it's right in my career field. Have you made any improvements to this since or aware of any evolved products? I'm trying to build one for my work. Really appreciate this. How much did it cost???
We've been adding additional simulation and visualization features over time, such as guided terrain building, subsurface structure visualization, localized rain, absorption, and surface roughness, and a VR client. But it's still the same idea. This particular AR Sandbox cost about $2,700 to build from scratch, everything included, in 2012.
Awesome, wow! I cant wait for the software package release. I want to build one for my daughter, she'd love it. I have all the pieces including a nice DLP projector and lots of 8020. :)
Classic Battletech table top board game. Serious loads of fun with light, medium and heavy mechs running around shooting each other over this. Damn, my husband and I would have so much fun.
Hardware cost is about $2200, less if you already have a gaming PC or a projector or a Kinect. The software will be available as soon as I get it packaged.
This might be a silly suggestion, but wouldn't it be better to use an array of weight sensors to create the topographic overlay? The computer seems to get confused by the movement of hands and placement of tools, so I think it would be more efficient to do it based off of weight rather than distance to the projector. However, I have no idea how much that costs or how difficult it is to implement. Anyway, great idea. I know it would keep me busy for hours
Try playing in a an actual sandbox. sandbox =/= beach. In a sand box the water does stay in holes for a while because there is dirt underneath it or in some cases people have sandboxes with plastic bottoms.
The issue is that the display/camera technology just isnt quite there yet to be able to make transparen acceptable resolution screens or small cameras with a good view angle. We will get there eventually, but its going to take a couple of years more.
Maybe. I personally like the "cartoon water" look, because it highlights small-scale water movement in a way realistic rendering wouldn't. But I appreciate that it's a matter of taste.
Hard to say, but it's probably a matter of resolution and physical accuracy. I've never played From Dust, but from videos I've seen the terrain and water are rather low-res.
plus realistic shaders would also require a lot of computing... you would need HDRI mapping of some kind, plus transparency shading, plus foam generation... water is a lot of work. but this is a cool example of where thing are headed!
I kind of am, it had a tint of purple when it should actually have a tint of green for the realism of actual water. Is it strait purple, no. But it is consisted of 20-30% red which gives it a slight int of red to it.
I read it but assumed that it's specialized simulation completly running on the GPU. The examples for PhysX(CUDA) can also run on the CPU (thats why I wrote accelerated :) ) and the system is visually more appealing. I thought that could be interesting to have a system that runs on a broader variatey of hardware.
Please try version 2.4-004 or newer of Vrui; it has an updated visual selection algorithm that fixes this particular problem. If it doesn't, please provide details on what operating system version and what graphics card and driver you are using.
Augmented reality (AR) is a live, direct or indirect, view of a physical, real-world environment whose elements are augmented by computer-generated sensory input such as sound, video, graphics or GPS data... I'm pretty sure it is AR, Nawmsayin'
Have you considered using a GPU accelerated water simulation? The CUDA development kit has severeal awesome free to use examples for realistic water simulation. Not that it would look bad in the current state but it could look even better :)
You do know that what isn't soaking up the water, and looks green is actual land right? Not sand, which wouldn't soak up the water that quick. The sand from which he's spawning the water does, in fact, absorb the sand, that's why it doesn't stay on it for long. It does flow like that because it's a high-scale topographic map. What you're seeing is not real scale, but a shitload of water being thrown in those valleys.
would be a cool way to model terrains for video games- you could probably import the heightmap into 3ds max and generate terrains based on what you created in the sand box....
I have no idea what im watching,but i like it.
This is 10 years old. Incredible. I wonder what it looks like now or has it been improved upon?
The only difference Ive seen since was some version with "creatures" (really just images moving around on the sand) 5 years ago.
Imagine playing Simcity or The Sims 4 in this. We'll be gods.
Oh yeah, that'd be just great for your city when your high-land-value dense commercial district gets dug up by a giant catzilla needing a dump!
How you gonna deal with *that*, Mr Mayor?
Age of empires XD
Yes, the one-second delay you can see in the videos is caused by the foreground detection filter. The idea is to keep a fixed-length history for each pixel (1s by default), and only update a pixel's output value if its history buffer has a variance smaller than a given threshold. On the upside, it works really well; on the downside, it causes a (short) delay.
Throughout the whole thing my heart kept pounding because I was afraid he'd overfill it with water and it would seep out..
I guess this is what it's like to feel outdated.
That's where the 1-second delay comes from. Anything that doesn't move for 1 second is assumed to be sand, and added to the topography model. It works very well, unless a user holds their hand or arm very still. You can see that a few times in the video.
The 1-second delay is not really annoying because when you work with the sand, you see it changing immediately. It's just the virtual topography that's updated a second later.
Love this!! Kids would love it, great sensory activity! And learning about Topographic Maps maps sweet!! Soothing.. so many neat uses... especially for the Sensory kid.
This can certainly be used for more than just that. The amount of rapid prototyping from something like this could help a lot of designers. I really want to use this thing.
Lol sandbox mode
Pretty awesome water simulation by using actual sand to scan a heightmap. Looks very realistic and scanning the height of the sand is very cool. Just the scanning part is awesome enough, the water simulation being applied to the heightmap is icing on the cake.
That's how god did it. Checkmate ATHEISTS!
So we are gods now?
...I guess so
@@markoftheland3115 water ai
I'm a preschool teacher, and I'm fairly certain you have invented the sandbox of the future. Years from now, all of my kids will expect augmented reality in the classroom!
Could you port this into a minecraft map editor ? That would be epic
It's a Linux box with a Core i7 CPU and a Geforce 580. Most of the computation is done on the Geforce. The water simulation is very computationally intensive, and pushes the 580 to its limits. You can see it chugging a bit at several points in the video.
WANT
I'm not sure if I understand the question correctly. The Kinect only scans inside the bounds of the sandbox, but any non-sand objects you place on the sand will be picked up, and affect the water flow. During testing I used Lego bricks to build structures or levees, for example.
The 1 second delay is to reduce jitter and, more importantly, filter out hands and tools so they don't interfere with the terrain. If it bugs you so much, you can turn if off by passing "-nas 1" to the AR sandbox executable. The delay would then be between 20 and 25 milliseconds. You're welcome.
magical simulation! Enhancing current demostrative methods. i think it can do things more than demostration.
You inspire me that programming with kinect can be interesting and even artistic!
This video has been selected by the channel as one of the Most Amazing videos on TH-cam and was added to the channel's playlist accordingly. Thanks for sharing.
Regards,
bestofYTchannel
The "rain maker" is based on depth as well. It detects non-moving objects inside a configured elevation range. I'm still working on improving that so it's harder to make it rain accidentally.
Color-based detection does generally not work because everyting takes on color from the projection.
This may just be the coolest thing I've ever seen.
Fair enough. As far as the simulation goes, it's pretty much top of the pops, as it's a second-order accurate time discretization of the mathematical equations governing shallow water flow. If there's anything better out there, someone needs to tell me.
The *rendering* of the resulting water surface, however, is a different story. It's non-photorealistic. I chose this model because it highlights the motion of the water at the scale viewers see the box.
Really is a great invention. The learning benefits are endless. Thank you for bringing this to be.
It's a grid-based explicit simulation, details are in the Kurganov&Petrova paper I cite in the description. It's the same method used by civil and hydrological engineers and scientists to simulate water flow in real environments.
It's not double-precision (as in floating point) because it turns out the method is stable under numerical round-off, but it's second-order accurate in the numerical sense.
I am research faculty at the University of California, Davis. My job is developing scientific visualization software for virtual reality environments, particularly for the Earth sciences. This is a side project.
Awesome...this is what software suppose to do, bring cheap solutions to hard problems. Great job sir.
little kids playing in the sandbox see this way anyway, it's us adults that need the high tech stuff to get enthused about sand castles again. Great work, Next time I take a trip I'm coming over to your back yard.
What everybody else said. The virtual sandbox is not a simulation of a real sandbox, but of a real landscape. To be precise, the water simulation runs at a 1:100 scale.
That was an astoundingly good call. Thank you.
Good luck floating your boat with virtual water. In all seriousness though, this a really cool. Nice work!
This is a fantastic application of augmented reality!
It's not done on a CPU, but on a graphics card exploiting its massively parallel processing power. The actual rig is a PC with a Core i7 CPU and a Geforce 580.
Pretty crazy things that people are making with Kinect. Pretty cool.
I like how consumers are making better use of the Kinect than Microsoft ever did.
It was a lot of work! The hardest part was that I was trying to make it seem as if there were lag -- you know, to make it more like what people expect of Kinect -- and therefore had to anticipate when the projection was going to change one second before it actually did. Took a LOT of trial and error to get that right. In retrospect, it would have been easier to just built the damn thing for real instead of faking it.
The point of tech demos is to demonstrate certain functions of a creation. Some may help further research into a particular subject. Creation is a base of research. So ya, there's a point to this.
That's what I'm trying to figure out... the open boundary conditions I implemented may induce a very large outflow in the outermost cell layer, which would limit the simulation time step. Or it could simply be a bug.
It's fun how if you leave your hand still for a moment, the topographic map includes it too.
The idea was that you can make water flow out by diggind a trench to the edge of the box. After all, we're trying to simulate landscapes, not sand boxes. :)
Without the height indicators, it looks a bit boring. Everything can be disabled independently via command line switches, if one wanted to do that.
This brings a whole new meaning to the term sandbox game
bloody brilliant....I mean really... I am deeply impressed
How-tos and software will start materializing on my web page in the near future. After all, this is just a prototype, and it is intended to be cloned by science museums.
Hardware cost (gaming PC, projector, Kinect, sandbox material) totals between $2000 and $2500.
I think as far as the system is concerned, anything sitting still inside the sandbox relatively near the surface is "sand". If you put a ball on the landscape, it would be treated as a hill. The narrowing contours below the widest part of the ball would not be seen by the system and can't be rendered.
Absolutely LOVE this idea! Looks great fun :P
This is really pretty cool. I can only imagine the other possible uses of something like this.
Best video I've seen of this.
Your Video is a GREAT Inspiration to help ADULTS to REMEMBER TO PLAY. They say Creativity is found in Play.... The POWER in PLAY is deeply Misunderstood. If People KNEW what PLAYING will DO they would ALL be doing MORE of it.
You, sir, are VERY fucking clever ;D It's amazing how well you've manipulated the Kinect System and somehow programmed it to work the way it did.
Pure Genious!
This is really really facinating. Reminds me a lot of the holograms in star wars.. I wonder if you could connect VR to it and have a top down 3d aspect to it without the sand.
AR/VR Sandbox: th-cam.com/video/NXD7Dfo7Ngo/w-d-xo.html
not gonna lie, this is pretty cool bro.
Wow that looks so real.
Sand is real, water and topographical colors are not. Kinect has the ability to see 3D shapes which is what this person is making in the sand. Very cool stuff.
This is one of the coolest applications of AR i've seen so far, considering that it's right in my career field. Have you made any improvements to this since or aware of any evolved products? I'm trying to build one for my work. Really appreciate this. How much did it cost???
We've been adding additional simulation and visualization features over time, such as guided terrain building, subsurface structure visualization, localized rain, absorption, and surface roughness, and a VR client. But it's still the same idea.
This particular AR Sandbox cost about $2,700 to build from scratch, everything included, in 2012.
Awesome, wow! I cant wait for the software package release. I want to build one for my daughter, she'd love it. I have all the pieces including a nice DLP projector and lots of 8020. :)
Classic Battletech table top board game. Serious loads of fun with light, medium and heavy mechs running around shooting each other over this. Damn, my husband and I would have so much fun.
Sorry, for the late reply, but okay got it. Thanks, Oliver. Still find this super impressive even now.
Hardware cost is about $2200, less if you already have a gaming PC or a projector or a Kinect. The software will be available as soon as I get it packaged.
This is just TOO COOL... awesome work!!!
This needs to be sold.
the future is going to be awesome.
This might be a silly suggestion, but wouldn't it be better to use an array of weight sensors to create the topographic overlay? The computer seems to get confused by the movement of hands and placement of tools, so I think it would be more efficient to do it based off of weight rather than distance to the projector. However, I have no idea how much that costs or how difficult it is to implement. Anyway, great idea. I know it would keep me busy for hours
This just blew my mind. Wow.
Try playing in a an actual sandbox. sandbox =/= beach. In a sand box the water does stay in holes for a while because there is dirt underneath it or in some cases people have sandboxes with plastic bottoms.
Absolutely, the most significant use of a "gaming" tool!!! :)
The issue is that the display/camera technology just isnt quite there yet to be able to make transparen acceptable resolution screens or small cameras with a good view angle.
We will get there eventually, but its going to take a couple of years more.
Maybe. I personally like the "cartoon water" look, because it highlights small-scale water movement in a way realistic rendering wouldn't. But I appreciate that it's a matter of taste.
Hard to say, but it's probably a matter of resolution and physical accuracy. I've never played From Dust, but from videos I've seen the terrain and water are rather low-res.
plus realistic shaders would also require a lot of computing... you would need HDRI mapping of some kind, plus transparency shading, plus foam generation... water is a lot of work.
but this is a cool example of where thing are headed!
This video would look amazing in 3D
They are playing in a real sandbox and using an XBOX kinect as well as a projector to project virtual water onto the sandbox
I need one of these! it would be totally awesome if it also simulated vegetation growth based off of where water sources were
Holy crap I had no idea that it was a projector on top! Still very cool
I always loved sandboxes, and now I can play with actual sand boxes again! \o/
Easily best song on TH-cam
I kind of am, it had a tint of purple when it should actually have a tint of green for the realism of actual water. Is it strait purple, no. But it is consisted of 20-30% red which gives it a slight int of red to it.
I read it but assumed that it's specialized simulation completly running on the GPU. The examples for PhysX(CUDA) can also run on the CPU (thats why I wrote accelerated :) ) and the system is visually more appealing. I thought that could be interesting to have a system that runs on a broader variatey of hardware.
couldn't have finished the video without your comment :)
Please try version 2.4-004 or newer of Vrui; it has an updated visual selection algorithm that fixes this particular problem. If it doesn't, please provide details on what operating system version and what graphics card and driver you are using.
Augmented reality (AR) is a live, direct or indirect, view of a physical, real-world environment whose elements are augmented by computer-generated sensory input such as sound, video, graphics or GPS data... I'm pretty sure it is AR, Nawmsayin'
this. is. AWESOME.
Best. Sandbox. EVER.
that made the video allot more interesting
this is incredible
AR is so much more interesting than VR. More useful applications as well.
Could we get a how-to in the future? I would love to set this up for myself. Probably pretty expensive, but worth it.
Thanks! I worked hard on it!
Looks like a great new tool for construction
@Saul Villanueva It's a projector
This is quite amazing.
I usually hate sandbox games, but this looks pretty good
This is amazing, you should market this. "The ultimate Techie Zen Garden".
Put me down for two.
Have you considered using a GPU accelerated water simulation? The CUDA development kit has severeal awesome free to use examples for realistic water simulation. Not that it would look bad in the current state but it could look even better :)
I can watch this for days....
You know when technology is getting crazy when you are no longer surprised by something like this lol
Grats! Original and smartly done.
absolutely ! and fabulous elephants as well !
You do know that what isn't soaking up the water, and looks green is actual land right? Not sand, which wouldn't soak up the water that quick. The sand from which he's spawning the water does, in fact, absorb the sand, that's why it doesn't stay on it for long. It does flow like that because it's a high-scale topographic map. What you're seeing is not real scale, but a shitload of water being thrown in those valleys.
would be a cool way to model terrains for video games- you could probably import the heightmap into 3ds max and generate terrains based on what you created in the sand box....
this looks so cool!