Forum Replies Created

Viewing 15 posts - 1 through 15 (of 16 total)
  • Author
    Posts

  • GrantWells
    Participant

    Hi Kasper, it’s likely that the error is due to the measured point being below the existing surface plane. Try hitting F2 to recalculate the surface plane after each F1 point capture until it figures out the actual texture surface.

    Also if you’re just using the sand surface then make sure it’s reasonably flat and level. Preferable for the first capture is a flat board or card. You may have to alter height manually in (? Sorry can’t remember the the filename) to raise or lower the bed height.

    OF bigger concern…. It looks like you’re using a short throw projector so you will need to make sure the Kinect and projector have the same ‘viewpoint’ ie centered above the bed otherwise the image has a very different geometry to the Kinects camera and will never align well.(yes my hard learning to here).

    Good luck
    Rgds
    Grant

    in reply to: Wiki Please – so we can jointly edit and contribute #103714

    GrantWells
    Participant

    Its generally about ten clicks to add a wiki to a wordpress site..

    in reply to: A wiki for distilled sandbox wisdom? #103713

    GrantWells
    Participant

    Crikey a wiki would be good. Then the answers would be collaboratively improved not buried in the sands of spam.

    😉

    in reply to: Motion sensor for the projector turning off/on? #103712

    GrantWells
    Participant

    Hi, a while ago I roughed up an arduino with IR motion sensor mounted near the projector (epson). I used the arduino to capture the IR codes, Then had a routine that would send the power down code when a timer timed out, unless it had recently received a IR sensor trigger event. Note that one issue is to ensure there are separate power on, and power off, codes otherwise they will inevitably be accidentally swapped.

    From memory the workaround was to use the input selector code as the wake up / proxy for power on command. If the projector was off then sending the AV1 input command would wake it up and select the right input, whereas if it was already on, then sending that command will cause it to only reselect the correct input, so one command covers both power conditions correctly. The routine doesn’t check for triggering again until the timer expires to avoid constantly sending wakeup commands.

    I also included hardwired buttons for wake and sleep on the button panel that initiated sending the IR code to the projector – just to cover all options.

    This wasn’t on a Benq projector and sorry it was too long ago for the specifics off the top of my head, however I’m about to climb back into this so will make a post when its reliable…

    rgds
    Grant

    in reply to: AR sandbox help #103533

    GrantWells
    Participant

    Good luck and at least try not to make it look too easy !!

    I found malamakers script seriously handy https://arsandbox.ucdavis.edu/forums/topic/great-and-easy-startup-script/

    and I assume you found projectioncentrals throw distance calculator for sizing sandbox design.

    http://www.projectorcentral.com/BenQ-MW632ST-projection-calculator-pro.htm

    in reply to: AR sandbox help #103524

    GrantWells
    Participant

    Hi Olly, as David said – with a short time frame then prior Linux knowledge is probably the biggest hurdle.

    On a good day… (and it’s your 3rd rebuild..) you simply follow the install instructions and it bursts into life.

    In the real world, the install instructions are seriously cryptic until you’ve been around the loop a few times.

    If you have some Linux knowledge – or can get a tame geek – then the only real hurdle to having a go at it, is the cost of a specific type of projector. As far as I understand it – the quirk with the projector is that it needs an uncommon throw ratio of around 1:1 or less – when it is 1 metre (3ft) from the sand it basically needs to throw an image 1 metre(3ft) or more wide, and this is an unusual configuration.

    Its neither the common short throw projector – like often used in a classroom, nor is it normal/long throw which is the common style. I have just got an epson powerlite 1761w Ebay US$400 to test.

    For testing you can leave the water simulation turned off and the software/xbox sensor combination will run on any(?) (most?) gaming grade machines, so you can get started pretty easily, then grab a proper graphics card when you are confident.

    The physical construction is pretty straight forward and can be pretty cheap/temporary. Its only really navigating Linux thats the big headache for us noobs.

    FWIW – you should have a quick try – its easy to find out how much of a learning curve it is for you – just grab/borrow a decent gaming PC and a kinect camera, drop in a spare drive with a new linux install as per the instructions. Don’t worry about a projector for a start – get it working on two screens – if you can get it starting/stopping/ and beginning to calibrate the alignment process then all good and it’s time to think about a projector. Ignore the sand until everything is working on a table top or wall, then start thinking about the hardware build.

    Good luck.

    rgds
    Grant

    in reply to: Sharing my modifications #103500

    GrantWells
    Participant

    Brilliant, thanks for sharing the code – I’ll have a go at this next week and report back. Much appreciated.

    rgds
    Grant

    in reply to: Glitch / freeze on Startup #103438

    GrantWells
    Participant

    Hey Daniel, its just a suggestion – based on my complete lack of skill in scripts / linux in general – and simple tests. I’ve struck network timeouts causing a long delay in startups on windows machines so personally I’d beg/borrow/steal a wifi dongle for a few days and see if the problem changed. I’ve had a machine that hangs regularly when particular USB devices were left plugged in – sorry I don’t recall the resolution to that, maybe a different slot or bios default.

    Also FWIW – I use a Logitech wireless keyboard trackpad combo that gets switched off and put away in the supervisors cupboard as the simple first line of defence, no cables and no mouse surface needed. Logitech because it uses the unified receiver which can be easily replaced.

    rgds
    Grant

    in reply to: Glitch / freeze on Startup #103436

    GrantWells
    Participant

    It could be revealing to find out if the same issue occurs when the system has ( / has not) internet access – maybe an app checking for updates or some other unrequested ‘feature’ is keeping the system busy.

    in reply to: Projector Setup #103373

    GrantWells
    Participant

    Hi, almost certainly the issue is that the projector is an ultra short throw style, meaning it’s designed to go very close to it’s screen and project a very wide image. So for example on a coffee table say only 4ft from the wall – perfect for home gaming or movies.

    So at 40inches it’s simply projecting an image that is way too big for your sandbox. So you have three options.
    A) it’s possible that the projector has a digital zoom or clamp function, where you can set a menu item in the projector to shrink the image.
    B) move the projector closer to the sand surface, by mounting it on a vertical pole at the rear centre of the sandbox, rather than overhead. The the downside of this is that the image alignment​ will be compromised, in practice it will still look great and work fine for a learning project. The effect is the contour lines will be effectively tilted to the projection lens rather than truly horizontal. This is the practical solution and will still be amazing.
    C) change projectors.

    The Kinect does need to stay about 40inches and vertically centered on the sandbox.

    The rawimage viewer should just see the edges of the sandbox and then these are eliminated during the calibration stage.

    I’d recommend using the startup script from malamaker as it makes the calibration etc much easier.

    Good luck, the calibration stage can be challenging. Hang in there..

    in reply to: Great and easy startup script #103316

    GrantWells
    Participant

    I’ve just banged into the same problem and taken a stab at correcting the paths for the sandbox 2.3 / kinect 2.8 / vrui 4.2 combination and new locations.

    I’ve not run the complete physical setup yet so not certain its 100% correct but seems to be working – can someone with actual skill check and validate please.

    Replace the contents of startAugmentedSandbox.sh in ~/UC-Davis-AR-Sandbox-Configuration-Menu folder with the following:

    
    #!/bin/sh
    # 
    # Author: Malamaker (J.Malavolti) Created: 2016-03-23 Last Updated: 2016-03-24
    # fixed paths GW 2017-05-24 to suit sandbox 2.3 / kinect 2.8 / vrui 4.2
    
    echo "Welcome to the Augmented Reality Sandbox Configuration and Start Menu"
    
    while true; do
    	echo "Make a selection from the following List"
    	echo "  1) Align your kinect sensor"
    	echo "  2) Align your projector"
    	echo "  3) Calibrate sensor and projector"
    	echo "  4) Start AR Sandbox"
    	echo "  5) Exit"
    	read -p "Selection: " userSelection
    
    	case "$userSelection" in
    		1) 	echo "Starting Kinect Sensor Alignment tool"
    			cd /usr/local
    			./bin/RawKinectViewer -compress 0
    			;;
    			
    		2)	echo "Starting Projector Alignment tool"
    			cd /usr/local
    			./bin/XBackground
    			;;
    			
    		3)	echo "Starting calibration tool"
    			read -p "Enter projector witdth: " calWidth
    			read -p "Enter projector height: " calHeight
    			cd ~/src/SARndbox-2.3
    			./bin/CalibrateProjector -s $calWidth $calHeight
    			;;
    
    		4)	echo "Starting AR Sandbox"
    			cd ~/src/SARndbox-2.3
    			./bin/SARndbox -uhm -fpv
    			;;
    
    		5)	echo "Exiting Augmented Reality Sandbox Menu"
    			exit
    			;;
    	esac
    	
    	sleep 2
    	echo "\n\n"
    done
    in reply to: Calibration crosshairs #102374

    GrantWells
    Participant

    Hey Rob,

    I’ve had this before also – three suggestions.

    1) A flakey kinect sensor – try redoing the intrinsic calibration step – one of mine seems a lot less reliable than the other and the intrinsic calibration seems to help a bit..
    2) During sandbox calibration hit the second key that you set up to force it to recapture and recalculate with the new surface data
    3) try a different surface level – the video uses a flat card, which puts the targets well above the sand surface

    And finally Malamakers script makes it easy. Malamakers startup script link

    rgds
    Grant

    in reply to: Sandbox "in a well" — need help! #102286

    GrantWells
    Participant

    😉 good news.

    Hmm.. my projector is a promethean PRM-30, and my kinect must be pushing 5′ also. Maybe thats the source of my calibration issues.. Glad it worked out.

    in reply to: Welcome to the AR Sandbox Forum! #102282

    GrantWells
    Participant

    Hi All, I’m in Christchurch New Zealand and have built a box to find out if its useful for public engagement work around earthquake recovery and water management issues we are facing locally.

    I’ve built mine with a glass bottom so that we can place maps and diagrams under the table to guide the users. Not sure how well this will work but we’ll find out soon enough.

    rgds
    Grant

    in reply to: Common Issues (READ THIS FIRST!) #102281

    GrantWells
    Participant

    HI, this is most likely because of the wrong corner point collection sequence, there was an inconsistency between the documented instructions and the video. Its a backwards Z pattern not a clockwise process.

    I won’t repeat it here at risk of yet another source of confusion – its elsewhere on the forum and correct on the video.

    rgds
    Grant

Viewing 15 posts - 1 through 15 (of 16 total)