RENCI Vis Group Multi-Touch Blog

Developing multitouch hardware and applications for research and experimentation.

Entries Comments


About

Welcome to the Multi-Touch blog for the Renaissance Computing Institute (RENCI) Visualization and Collaborative Environments Group. Our group explores how technology can make scientific problem solving, long-distance collaboration, and even artistic expression more meaningful. Through our development of hardware and software for a multi-touch interface, we hope to empower researchers and scientists with new tools for collaboration and interaction with data in their work.

RENCI brings together a diverse group of professionals from academia and private industry and the Multi-Touch Team is no exception. Here are the core participants in this program:

  • Ray Idazak, Director, Collaborative Environments
  • Xunlei Wu, Ph D, Senior Visualization Researcher for the Duke RENCI Center
  • Jason Coposky, Senior Visualization Engineer
  • Warren Ginn, Senior Research Industrial Designer

For more information about our work on Multi-Touch displays, contact Warren Ginn (warreg at renci dot org). For more information about RENCI, please visit www.renci.org.

Comments

Comment from Nodel
Time: November 21, 2008, 8:09 am

To:Xunlei Wu

I come from Beijing, China, you Chinese?
Your work is too great! Great!
“Tek Satin coating” What? Where to buy?

Comment from admin
Time: November 22, 2008, 4:06 pm

The Tek Satin coating is available from RP Visuals. They supplied out screen for both the Duke Wall and our table. They do both coatings (the projection surface emulsion on the back and the Tek Satin coating on the front).

Comment from admin
Time: November 22, 2008, 4:07 pm

RP Visuals web site is http://www.rpvisuals.com.

Comment from sudhanshu gautam
Time: December 3, 2008, 1:44 am

hey I saw your blog carefully and I found that you guys performs all the work very professionally ,I really like your all the table cad designs.
Hey I am also working on the MT -TABLE so want to you people help can you tell me that how I can modify the LCD SCREEN FOR THE MT-TABLE

Comment from admin
Time: December 3, 2008, 2:20 am

Sorry, our current design does not use a LCD display. It’s rear projected.

Comment from sudhanshu gautam
Time: December 6, 2008, 7:36 am

can you tell me that why you people used six projectors for the wall ,and how much coast of it

Comment from admin
Time: December 6, 2008, 3:06 pm

More projectors = higher resolution for a wall that’s 13-feet across and 5.5-feet high… The projectors currently cost about $2500 - $2700 each.

Comment from sudhanshu gautam
Time: December 13, 2008, 4:11 pm

hey Now we our team working on the MT-wall .I got some Idea from the nui forums ,can you tell me that here we used general purpose touch lib that is open source or you worked on some thing different software .I saw that video that is posted latest on the Nui group they made a Mt wall at the dubai and they used the Four projector .

Can you give me some hints so I can also start work on my MTwall

Comment from admin
Time: December 13, 2008, 4:41 pm

We’re using TouchLib from NUI…

Comment from sudhanshu gautam
Time: December 14, 2008, 1:40 am

Hey first to thanks to give me valuable information ,can you tell me that what kind of the hardware requirement if I want to make a 10′ length mt -wall
mean how much PC,Projector,camera ,which kind of the camera ,type of LLP
can you provide ,me any hardware detail .

Thanks in advance

Comment from admin
Time: December 14, 2008, 2:37 am

I’m sorry. We’re not distributing the complete schematics for our wall or table at this time. But there is a fair amount of information (i.e., the projectors, cameras, illuminators, etc.) within the blog posts.

Thanks again for your interest.

Comment from sudhanshu gautam
Time: December 14, 2008, 3:25 am

ok do you know any community rather than then the nui group that is open source and provide information about the mt table

Comment from admin
Time: December 14, 2008, 10:51 am

I think NUI is really it. But I’m not sure anyone is distributing exact plans on every last detail of the hardware and software… Everyone’s table is different and there is no one exact way to build it… I think everyone has to do a certain amount of experimentation for themselves to figure out what works best for them. In our case, there’s some things that work better than others and it’s by no means perfect - we still have a l0ot of work to do and we’ll be sharing what we can as we make improvements. But to just make a carbon copy of anyone’s setup won’t do you as much good as doing some of your own development.

Good luck!

Comment from admin
Time: December 16, 2008, 3:28 pm

By the way, all questions and comments need to be submitted in English.

Thanks.

Comment from tuxboy
Time: February 25, 2009, 4:56 pm

Kudos !
i agree that multitple projectors give a better resolution.. but ..how do you guys work with it. i mean - for one single computer’s display how can I divide it on say 6-7 screens equally ? are you guys sharing the source code. and anyway, the apps u’ve done are really nice, where’s the source code ? i am really keen on checking it out and improving it :)

Comment from jason coposky
Time: February 25, 2009, 5:59 pm

under either windows or linux you can tile a group of outputs from video cards into a single functioning
desktop. we use 3 nvidia 8 series video cards in the multi-touch wall configuration. each vertical pair
are set as a vertical span in the nvidia control panel. this would be a TwinView in xorg.conf with xinerama
enabled.

if you need to split a single video signal i would recommend the matrix triple head 2 go which we have used
quite well in our 16 tile single desktop vizwalls. 8 outputs on a nvidia model II quadroplex are split into 16
video signals all grouped similarly into a single desktop.

Comment from tuxboy
Time: February 25, 2009, 6:42 pm

thank you Jason :) you guys rock , i am on your blog

Comment from tuxboy
Time: March 4, 2009, 2:24 am

what about the apps that are running on ur table. is it open source ? i would like to peep into the code, and maybe develop something further :)

Comment from Tri Wahyudi
Time: March 26, 2009, 10:55 am

Hi Jason, I understand your explanation on tile a group of outputs from the video cards. But how about the Camera, do you write your own touchlib program that can stiching several camera?
I already build my MT Table with single projector and camera, working on Tbeta. Now I am really keen to go further with multi camera and multi projector stiching together like the MT Wall.

Comment from K Armstead
Time: August 13, 2009, 2:33 pm

What if I need a display only…no touching required…that’s at least 6′ high by 24′ wide?? Could I use your product?? What is the largest display area that you could project…in case we need to upgrade the display at some point to make it larger?? Please help.

Comment from admin
Time: August 13, 2009, 11:24 pm

Hi Kimberly,

If you want a large multi-projector display, the sky’s the limit… provided you had enough graphics cards to drive it… Big is one thing… big AND high-resolution is another… Upgrades can take the form of just zooming the image larger, or adding additional display, or swapping out higher-resolution displays… It has more to do with what you’re trying to do, and how much money you have to spend…

w

Comment from Elisa
Time: August 19, 2009, 6:46 am

Hi,
I also want to know how do you handle the multiple cameras input. What software do you use to stitch all cameras into a single big image and feed into Tbeta? I saw your architecture diagram which indicate a server is used to handle the multiple cameras’ input. May I know how it works?
Thank you very much.
Elisa.

Comment from Jason Coposky
Time: August 19, 2009, 10:27 am

for the current incarnation at the duke facility we have a touchlib thread running for each camera. we then take the touches from each camera and sort out the duplicates due to overlap in adjacent cameras. from there we send the touches to our gesture engine. stitching the images together and then processing a huge single image is considerably less scalable as it bottle necks at the filter-graph and blob detection.

Comment from john
Time: September 12, 2010, 2:32 pm

Hi,

Do you sell the tables or the software?
OR you can help me to build a table┬┐? and create contents┬┐
it is possible that you write to me to my email please

Regards

Comment from admin
Time: September 12, 2010, 9:04 pm

Sorry, this work is for academic research only and is not for sale.

Thanks for your interest.

Write a comment