A Superb presenter and concise explanations that make sense to a newbie! You have just answered so many questions I had about a project I am pitching...
That is music to my ears!!! Glad it was helpful and useful info. A lot of things in TouchDesigner can often take long if you take roundabout ways to solve problems, but if you know the little tricks, even hard things become very straight forward and intuitive :)
A good option for that would be to use the data from the connected blobtrack1_blobs DAT for instancing geometry, which you could then composite over top of the original input texture. You can see an example of how this might function in the Blob Track TOP OP Snippet called "Track Warping Blobs" (right click on the Blob Track TOP -> OP Snippets to check it out). Hope that helps!
Thanks! You can't adjust the thickness via the Blob Track TOP itself, but it does provide the UV coordinates for the boxes in the blobtrack1_blobs which you could use with an instancing setup. By assigning a Line material to the geometry COMP, you could control the thickness of the lines rendered. Hope that helps!
Thank you! Great tutorial! I was wondering if it's possible to change the width of the blob tracking squares? Like the lines of the squares themselves? I want to achieve a look of CCTV tracking people. So the output of the blob tracking is enough for me :)
Unfortunately you can't adjust the thickness of the blob outlines via the Blob Track TOP itself, but it does provide the UV coordinates for the boxes in the blobtrack1_blobs DAT which you could use with an instancing setup. By assigning a Line material to the geometry COMP, you could control the thickness of the lines rendered. Hope that helps!
Great idea! You could definitely use a Cache TOP, it really depends on how stable your background is. If you're finding you need to program in more regular background snapshots that'd be an easy operator to control for that.
Love the video you are a great presenter!!!! Im trying to use this but instead of the amount of blobs as information I need to use the U and V coordinates. Do I just need to change the script in the constant or am I totally off base lol.
Sure, you're definitely not off base! In the table we pull data from there is also UV co-ordinates, so if you change which cell you're targeting with your expression, you could make a bunch more channels to grab the UV co-ords of the different blobs.
Haven't tried this one, but it seems possible (although might be a bit difficult)! It would probably be highly dependent on the moon phase and the amount of clouds. That said, processing the webcam image with a threshold or similar filter before using the Blob Track TOP might help balance out some of the impact of the clouds, since you could be filtering out all parts of the image below a certain brightness level.
Hey thanks! You could take the detection CHOP channel that we create and make a CHOP reference to the Play parameter of a Movie File In TOP - that way, when a person is detected, the movie will play, and when no person is detected, it will stop. Hope that helps!
@@TheInteractiveImmersiveHQ Great tut! Thanks! What if I want another video to loop when in idle mode (no detection) but a detection video to override the loop when triggered?
Hi, i have a problem, ndiIN has no camera as Source Name, i can't choose anything,why this is happenin? any idea i can solve this? any help wpuld be great! thanks a lot
Hi Raffaele, do you have NDI sources available on your network? If not, you can replace that with a Video Device In TOP and that will let you plug in your webcam instead to the same process.
thanks for your useful information, which kind of sensors do you prefer for making interactive wall(video mapping)?, my area has 5m width and 3m height. i would be thankful if you help me.
There are a lot of options for this, but it really depends on what kind of interaction you're looking to achieve! If you're looking for full body tracking, you might have a look at the Kinect Azure or the Femto Bolt www.orbbec.com/products/tof-camera/femto-bolt/ (fully compatible with Kinect Azure TOP/CHOP). You could also try a software-based approach using some of the built-in Nvidia body tracking operators like the Body Track CHOP, although you'd need 20, 30, or 40 series Nvidia RTX GPU.
I don't need full body tracking, just making touch screen on the wall or floor. I mean making piano on the floor and when stand on a key it makes a sound. It is clear example of sensor that i looking for.
thank you for your help, however currently having an issue where as the NDI In does not show any sources, even including my built-in webcam. Do you possibly know how to fix this?
Hi! Great tutorial! Thanks! I wonder if there is a way to detect with a webcam someone coming from far to close? A basic set up; you have an image. Someone appears and comes close to the webcam and the image zooms in.If he goes backwards or comes out, the image goes back to full size screen.
It depends on how clean the environment and camera feed is. Like I did in the video, one thing you could try to use is the size of the blob as a form of detecting how close someone is to the camera and use that to control zooming into the camera and zooming back out. Have you tried something like that?
@@TheInteractiveImmersiveHQ Hi Elburz! I was wondering if I have to buy a proximity sensor or with the blobs can do it¿? I'll try! Thanks a lot! Your tutorials are great Man!
@@jdiezastronomy Thanks for the kind words! Proximity sensor or depth sensor (like Kinect or Realsense) would definitely be much easier and give more accurate results, but to try to save money you can try using blob size and see if that works to an acceptable level.
@@TheInteractiveImmersiveHQ I have an Art installation this summer in a gallery. For that I'll look for a proximity sensor. But right now, the blob seems good to try. What's the best to control zoom in/out in a hi res still -or video-;just a TOP transform? I work with stills and video only, not 3D stuff.Thanks again!
@@jdiezastronomy It sounds like Transform TOP should work great. Alternative would be to have a Constant TOP going into an Over TOP, and then you can connect your video feed into the Over TOP as well and composite it onto the constant that you set to your canvas resolution. Then on the Over TOP you can use the Scale parameter and transforms for move it around and scale it.
The box is representative of a blob in this case but you can turn off the bounding box drawing, then use the data outputs from the Blob Track TOP to track whatever you like along it.
Our pleasure! I'm mostly working on an Origin Eon 15 Slim laptop from about 2015. It has a Geforce 970M and an i7-6700HQ. I'm due for an upgrade in 2021, mostly just waiting to see Geforce 30xx series roll out to laptops.
@@TheInteractiveImmersiveHQ Thank you Taylor for the question, and Elburz for the answer!! That is wonderful for us noobs to hear - TD is not out of reach!! thank you so much for the software & tuts.
@@borjonx yup, I remember when I started learning TouchDesigner I was using a 15" Macbook with a 330M geforce card, and I would run full live shows on it (although slightly laggy), so you can definitely get a lot of experimenting and learning done on weaker hardware before needing to invest too much money :)
Can someone point me to a video of how to use touch design to use a shadow to projection map my goal is to put a camera underneath a table with a very thin whiteboard and so it will cast a shadow into a camera and then that camera will use that shadow to give live projection mapping around that shadow if you guys can understand what I’m trying to do
Great project! It's a bit too much to dive into in TH-cam comments but what I'd recommend is making a bullet point list of all the features, and then make more sub bullet points that describe every single step needed for each of the processes, then if you can even make further subbullets, you'll naturally end up low-level enough that it will be a lot easier to convert those bullet points into a TouchDesigner network of small operations that add up into the larger functionality.
Assuming you're wanting the number to appear alongside each blob, yes, it is possible, but it's not the simplest thing to set up. You could approach this in two ways, via using the Table DAT option for the Text TOP to draw text at certain positions (which you'd have to calculate via the UV positions of each blob) or using instancing, and specifically texture instancing to draw each number at different positions on the screen. It's worth checking out the OP Snippets for the Blob Track TOP (which you can access by right clicking on the TOP and then clicking OP Snippets), specifically the example named "track warping blobs". This example shows you how to use instancing to overlay geometry onto the tracked blobs. Hope that helps!
Sure! There's lots of ways you could approach object tracking. Some folks find the easiest way is to use some kind of infrared reflective paint on an object, then if you shine an IR light in the environment and use an IR camera to watch the room, you'll only see the objects you've painted light up and most other things will just be black. That way you can then feed into a similar blob tracking setup.
It's possible, but it would take some work to make happen. It would require some calculation based on the blobtrack#_blobs DAT table that gives the position/scale of the blobs. You could calculate the position, scale, and rotation of each line, and then use instancing to generate the geometry. Once rendered, the texture could be composited with the output of the blobtrack TOP. Hope that helps!
I am trying to do a blobtracking in a church. I managed to set up two infrared cameras in the curch an stiched them together. Also I setuped 4 infrared lights, all at 8 m height from the top. But my tracking results are not very good. What am I doing wrong? Or is it the light setup, that is not good? I also tried using blur and threshold like you explained in your youtube-video, but it stills doesnt make it better. Here is an exemple video of the tracking: th-cam.com/users/shortsLXBLTACzUJ8
It doesn't look too bad! I think what you need is more contrast to make a difference between the ground/environment and the person moving in the environment. You could also try doing some background subtraction as well, taking a still image of the floor/environment without anyone in the scene and then subtract that from the live video signal feeding into the blob tracking setup.
@@TheInteractiveImmersiveHQ Thanks for your ideas. Indeed in the meantime I did exactly this, used the backgroundsubraction and then blurred it out. Makes it much better. My big new task is now th light from outside. Beeing in the church yesterday I noticed, that there is a time in the evening, when the sunnlight falls in. A real killer for my background image. :( Thinking of looking at the difference image an as soon it goes over a certain threshold it resets the background image. Lets see. Or do you have another idea? Another point is, how can I keep the position of blobs in the the blob table. E.G. Blob 1 comes in (position is in tablerow 1). Blob2 comes in ( (position is in tablerow 2). All good. But as soon as Blob 1 dissapears, Blob two goes to tablerow 1 and I have some problem for my further processing.
![ประสบการณ์ จ้าง Developer ที่ผมจะไม่มีวันลืม [ภาค2]](http://i.ytimg.com/vi/pZbsSDRrmXA/mqdefault.jpg)
A Superb presenter and concise explanations that make sense to a newbie! You have just answered so many questions I had about a project I am pitching...
Thanks for the kind words! Glad it worked.
Great tutorial, easy, clear, seems low tech but super practical for many situations.
Thanks! Always good to have a few easy tricks like this in your toolkit in case the fancier tools don't work :)
Man you just save me TONS of hours...
That is music to my ears!!! Glad it was helpful and useful info. A lot of things in TouchDesigner can often take long if you take roundabout ways to solve problems, but if you know the little tricks, even hard things become very straight forward and intuitive :)
Excellent and concise tutorial, thank you as always, Elburz!
Thanks, Kelly! Glad you enjoyed it!
Really helpful tutorial Elburz thanks, would love to know how to increase the resolution of the tracking overlay without losing image quality
A good option for that would be to use the data from the connected blobtrack1_blobs DAT for instancing geometry, which you could then composite over top of the original input texture. You can see an example of how this might function in the Blob Track TOP OP Snippet called "Track Warping Blobs" (right click on the Blob Track TOP -> OP Snippets to check it out). Hope that helps!
This tutorial was perfect but could you just please explain how to do have a timer chop for light detection ?
Amazing video! Is there a way to increase the thickness of the squares?
Thanks! You can't adjust the thickness via the Blob Track TOP itself, but it does provide the UV coordinates for the boxes in the blobtrack1_blobs which you could use with an instancing setup. By assigning a Line material to the geometry COMP, you could control the thickness of the lines rendered. Hope that helps!
Thank you! Great tutorial! I was wondering if it's possible to change the width of the blob tracking squares? Like the lines of the squares themselves? I want to achieve a look of CCTV tracking people. So the output of the blob tracking is enough for me :)
Unfortunately you can't adjust the thickness of the blob outlines via the Blob Track TOP itself, but it does provide the UV coordinates for the boxes in the blobtrack1_blobs DAT which you could use with an instancing setup. By assigning a Line material to the geometry COMP, you could control the thickness of the lines rendered. Hope that helps!
That’s clever. Thank you!
Edit: Wonder if using the cache TOP would make a better background filter for the video feed?
Great idea! You could definitely use a Cache TOP, it really depends on how stable your background is. If you're finding you need to program in more regular background snapshots that'd be an easy operator to control for that.
@@TheInteractiveImmersiveHQ Thanks!
Love the video you are a great presenter!!!! Im trying to use this but instead of the amount of blobs as information I need to use the U and V coordinates. Do I just need to change the script in the constant or am I totally off base lol.
Sure, you're definitely not off base! In the table we pull data from there is also UV co-ordinates, so if you change which cell you're targeting with your expression, you could make a bunch more channels to grab the UV co-ords of the different blobs.
really helpful! thank you
Our pleasure!
Hi thank you for your tutorial. Is it possible to use blob tracking to track the moon in the sky? I'm kinda worried about the clouds
Haven't tried this one, but it seems possible (although might be a bit difficult)! It would probably be highly dependent on the moon phase and the amount of clouds.
That said, processing the webcam image with a threshold or similar filter before using the Blob Track TOP might help balance out some of the impact of the clouds, since you could be filtering out all parts of the image below a certain brightness level.
great Idea, Bravo man. what if i want to play a video when a person detects by webcam? thanks.
Hey thanks! You could take the detection CHOP channel that we create and make a CHOP reference to the Play parameter of a Movie File In TOP - that way, when a person is detected, the movie will play, and when no person is detected, it will stop. Hope that helps!
@@TheInteractiveImmersiveHQ Great tut! Thanks! What if I want another video to loop when in idle mode (no detection) but a detection video to override the loop when triggered?
Hi, i have a problem, ndiIN has no camera as Source Name, i can't choose anything,why this is happenin? any idea i can solve this? any help wpuld be great! thanks a lot
Hi Raffaele, do you have NDI sources available on your network? If not, you can replace that with a Video Device In TOP and that will let you plug in your webcam instead to the same process.
thanks for your useful information, which kind of sensors do you prefer for making interactive wall(video mapping)?, my area has 5m width and 3m height. i would be thankful if you help me.
There are a lot of options for this, but it really depends on what kind of interaction you're looking to achieve! If you're looking for full body tracking, you might have a look at the Kinect Azure or the Femto Bolt www.orbbec.com/products/tof-camera/femto-bolt/ (fully compatible with Kinect Azure TOP/CHOP). You could also try a software-based approach using some of the built-in Nvidia body tracking operators like the Body Track CHOP, although you'd need 20, 30, or 40 series Nvidia RTX GPU.
I don't need full body tracking, just making touch screen on the wall or floor. I mean making piano on the floor and when stand on a key it makes a sound. It is clear example of sensor that i looking for.
thank you for your help, however currently having an issue where as the NDI In does not show any sources, even including my built-in webcam. Do you possibly know how to fix this?
To access the webcam directly in TouchDesigner, you can replace the NDI In TOP with a Video Device In TOP. Hope that helps!
and thankyou so much for your video
Our pleasure! Thanks for watching :)
Hi! Great tutorial! Thanks! I wonder if there is a way to detect with a webcam someone coming from far to close? A basic set up; you have an image. Someone appears and comes close to the webcam and the image zooms in.If he goes backwards or comes out, the image goes back to full size screen.
It depends on how clean the environment and camera feed is. Like I did in the video, one thing you could try to use is the size of the blob as a form of detecting how close someone is to the camera and use that to control zooming into the camera and zooming back out. Have you tried something like that?
@@TheInteractiveImmersiveHQ Hi Elburz! I was wondering if I have to buy a proximity sensor or with the blobs can do it¿? I'll try! Thanks a lot! Your tutorials are great Man!
@@jdiezastronomy Thanks for the kind words! Proximity sensor or depth sensor (like Kinect or Realsense) would definitely be much easier and give more accurate results, but to try to save money you can try using blob size and see if that works to an acceptable level.
@@TheInteractiveImmersiveHQ I have an Art installation this summer in a gallery. For that I'll look for a proximity sensor. But right now, the blob seems good to try. What's the best to control zoom in/out in a hi res still -or video-;just a TOP transform? I work with stills and video only, not 3D stuff.Thanks again!
@@jdiezastronomy It sounds like Transform TOP should work great. Alternative would be to have a Constant TOP going into an Over TOP, and then you can connect your video feed into the Over TOP as well and composite it onto the constant that you set to your canvas resolution. Then on the Over TOP you can use the Scale parameter and transforms for move it around and scale it.
Thank you for the tutorial! Is there a way to track it with text instead of boxes?
The box is representative of a blob in this case but you can turn off the bounding box drawing, then use the data outputs from the Blob Track TOP to track whatever you like along it.
Thank you for another awesome video Elburz! I am curious, what PC are you using - more specifically CPU/GPU?
Our pleasure! I'm mostly working on an Origin Eon 15 Slim laptop from about 2015. It has a Geforce 970M and an i7-6700HQ. I'm due for an upgrade in 2021, mostly just waiting to see Geforce 30xx series roll out to laptops.
@@TheInteractiveImmersiveHQ thank you so much ✌🏾
@@TheInteractiveImmersiveHQ Thank you Taylor for the question, and Elburz for the answer!! That is wonderful for us noobs to hear - TD is not out of reach!! thank you so much for the software & tuts.
@@borjonx yup, I remember when I started learning TouchDesigner I was using a 15" Macbook with a 330M geforce card, and I would run full live shows on it (although slightly laggy), so you can definitely get a lot of experimenting and learning done on weaker hardware before needing to invest too much money :)
i have a macbook pro m1, is there anyway to do this on mac? using a different device (not kinect) for tracking.
Can someone point me to a video of how to use touch design to use a shadow to projection map my goal is to put a camera underneath a table with a very thin whiteboard and so it will cast a shadow into a camera and then that camera will use that shadow to give live projection mapping around that shadow if you guys can understand what I’m trying to do
Great project! It's a bit too much to dive into in TH-cam comments but what I'd recommend is making a bullet point list of all the features, and then make more sub bullet points that describe every single step needed for each of the processes, then if you can even make further subbullets, you'll naturally end up low-level enough that it will be a lot easier to convert those bullet points into a TouchDesigner network of small operations that add up into the larger functionality.
Great 🎉
Cheers :)
there is a way to make visible the info numbers of the blop on the screen?
Assuming you're wanting the number to appear alongside each blob, yes, it is possible, but it's not the simplest thing to set up. You could approach this in two ways, via using the Table DAT option for the Text TOP to draw text at certain positions (which you'd have to calculate via the UV positions of each blob) or using instancing, and specifically texture instancing to draw each number at different positions on the screen.
It's worth checking out the OP Snippets for the Blob Track TOP (which you can access by right clicking on the TOP and then clicking OP Snippets), specifically the example named "track warping blobs". This example shows you how to use instancing to overlay geometry onto the tracked blobs. Hope that helps!
Hello! any pointers of the same effect with object tracking?
Sure! There's lots of ways you could approach object tracking. Some folks find the easiest way is to use some kind of infrared reflective paint on an object, then if you shine an IR light in the environment and use an IR camera to watch the room, you'll only see the objects you've painted light up and most other things will just be black. That way you can then feed into a similar blob tracking setup.
Exist a way to join the blops whit lines?
It's possible, but it would take some work to make happen. It would require some calculation based on the blobtrack#_blobs DAT table that gives the position/scale of the blobs. You could calculate the position, scale, and rotation of each line, and then use instancing to generate the geometry. Once rendered, the texture could be composited with the output of the blobtrack TOP. Hope that helps!
clean
Lol good iconic controls Monaco’s comics
I am trying to do a blobtracking in a church. I managed to set up two infrared cameras in the curch an stiched them together. Also I setuped 4 infrared lights, all at 8 m height from the top. But my tracking results are not very good. What am I doing wrong? Or is it the light setup, that is not good? I also tried using blur and threshold like you explained in your youtube-video, but it stills doesnt make it better.
Here is an exemple video of the tracking: th-cam.com/users/shortsLXBLTACzUJ8
It doesn't look too bad! I think what you need is more contrast to make a difference between the ground/environment and the person moving in the environment. You could also try doing some background subtraction as well, taking a still image of the floor/environment without anyone in the scene and then subtract that from the live video signal feeding into the blob tracking setup.
@@TheInteractiveImmersiveHQ Thanks for your ideas. Indeed in the meantime I did exactly this, used the backgroundsubraction and then blurred it out. Makes it much better. My big new task is now th light from outside. Beeing in the church yesterday I noticed, that there is a time in the evening, when the sunnlight falls in. A real killer for my background image. :( Thinking of looking at the difference image an as soon it goes over a certain threshold it resets the background image. Lets see. Or do you have another idea?
Another point is, how can I keep the position of blobs in the the blob table. E.G. Blob 1 comes in (position is in tablerow 1). Blob2 comes in ( (position is in tablerow 2). All good. But as soon as Blob 1 dissapears, Blob two goes to tablerow 1 and I have some problem for my further processing.