GearHeads Corner

User Video Links => WebCam experience => Topic started by: ArtF on April 17, 2015, 11:47:58 AM



Title: WebCam Data
Post by: ArtF on April 17, 2015, 11:47:58 AM
Hi All:

  This is a dropzone for any webcam data ( like your confidenc e levels) , experienc es or comments / suggestio ns
related to the Stereo Photometr ics module. All data helps to make the module strionger in future.

Thx
Art


Title: Re: WebCam Data
Post by: KKirk on April 19, 2015, 08:48:19 PM
I am getting confidenc e levels like 23.48, 44.25, 36.148 and etc.  Picture looks good but every time I select "add st'" the program immediate ly goes into an infinite loop and a "3 finger salute" is the only way out.  I have yet to get even 1 good output.

Karl


Title: Re: WebCam Data
Post by: ArtF on April 19, 2015, 08:58:14 PM
Karl:

 Excellent numbers. OK, so you have generated the normals, and created the 3d? Did you decimate the stl before using AddSTL? Decimate it down to around 75,000 or less vertices. Does it lock up when you hit addSTL now?

Art


Title: Re: WebCam Data
Post by: DanL on April 25, 2015, 07:44:19 PM
hi art tried the webcam in my scanner box the picks came out ok with me turning lights on and off by hand.

had settings at 176 x 144 YUY2, confidenc es at 270.9, gradient scale 0.01 I put the image out at x 176, y 144 and z 14.81831, vertices 6331. see attached pic

it would have worked better if i turned lights on and of at the correct time got it close a couple of times pic`s came out quite clear so if you wont to add out put light control through a arduino that would be cool this year next year when ever.

if you did I think it would be as good or better than the scanner program I use.


Title: Re: WebCam Data
Post by: ArtF on April 25, 2015, 08:30:11 PM
Hi Dan:

 Thx for the numbers. If the photos are not perfectly timed it will screw it up..no question. :)

Art


Title: Re: WebCam Data
Post by: DanL on April 25, 2015, 09:07:45 PM
yep I will try sticking my spear off next week screen in there to get the correct timing it will make it better, useing twin screens makes it easy


Title: Re: WebCam Data
Post by: DanL on June 09, 2015, 12:31:53 AM
Hi art had a good play today I have the screen in my scanner box works good forgot to save the last image I did it came out quite well

the confidenc e was 64.899, resolutio n 640 x 480, fire 250ms

couple of thing I found the colour of what you are taking a image of can make or brake a image capture dark colour`s don't work very well at all light colour`s are fine, black pretty much leaves a very white colour, so if the back ground is black it will leave a white back ground.

I will play more tomorrow using a black back ground hopefully I will rember to save the images this time


Title: Re: WebCam Data
Post by: ArtF on June 09, 2015, 06:08:13 AM
Dan:

 Send me a email to support and I can send you the new version with a much different
algorithm .. much faster and pulls much better. You may find it better, but there are no docs
or video for it yet, its a bit more complex as more selection s exist in its operation s.

  But then, the results are much more impressiv e..

Art


Title: Re: WebCam Data
Post by: DanL on June 10, 2015, 07:20:53 PM
hi art did not see your reply did some playing with a black back ground so much better the area around what I did images off is single colour and flat.

I will add images of result from cut 3D

a is same settings as before as a negative, 3 is image


Title: Re: WebCam Data
Post by: ArtF on June 10, 2015, 08:06:04 PM
lol.. Those look cool..

I see your email, Ill send you a fresh version..

Art


Title: Re: WebCam Data
Post by: DanL on June 10, 2015, 10:03:01 PM
I have run in to a big problem the X and Y are the same as the aspect size but Z is coming out as just over 3 meter in depth I change x to 200 it goes down to 1 M


Title: Re: WebCam Data
Post by: ArtF on June 11, 2015, 06:07:22 AM
Dan:

  Wow, those are pulling out nicely.. just lower your scale slider, it should lower back to a reasonabe l pull, I aim for about 150 as a good scale.. but it varies from data to data set..

Art


Title: Re: WebCam Data
Post by: DanL on June 11, 2015, 04:42:19 PM
yer they are better than my laser scanner can do I will try that


Title: Re: WebCam Data
Post by: DanL on June 11, 2015, 10:01:09 PM
Hi art moving the slider down to 1 leaves not a very good image, having it set to 2.5 is a lot better but it has a 80 mm curve ever negative or image it`s all the same.

so the scale of negative to image is good

but the curve is way to much the whole image will be curved cutting the side of helps a little bit.

I have tried a flat black back ground it make`s it worse bigger curve.

A flat white back ground is better, between black and white is 200 mm

it would be good if you can reduce the Gradient scale so if its at 10 it would be 1/4 or less of what it is now but keeping the image just as good or have it how it was in the last version I tried the one on your site had barely any curve to wear there was a flat back ground sorry to be a pain

picks 1 is negative, 1 on gradient scale pick 2 is negative, 2.5 on gradient scale, picks 1a, 2a is image same setting`s 

threshold is 0.1, zero noise gradients is un ticked, ambient correctio n same, fullres resolutio n is 176 x 144 yuy2--format 2


Title: Re: WebCam Data
Post by: DanL on June 12, 2015, 02:09:30 AM
been playing with the web cam on my laptop the differenc es from the webcam in a box to the one on the laptop is massive everythin g comes out fine with the webcam on my laptop.

I thinks it just that there is no back ground noise in the box so the changes you made are to much if the webcam has no back ground noise and is in a box I will try having the web cam sitting on my laptop to see what happens I will post those latter


Title: Re: WebCam Data
Post by: DanL on June 12, 2015, 03:10:09 AM
here is a pick done with my webcam sitting on my laptop I have backgroun d lights on so its not the best but it half the depth of doing it in my scanner box with the Gradient scale at 3


Title: Re: WebCam Data
Post by: ArtF on June 12, 2015, 05:46:58 AM
Dan:

  Also, look closely at the images, any chance the timing is off on the scanner box? The photo is actually taken at the end of the flash cycle, not at the beginning . When the screen flashes the one side, you have
to be pretty carefull about when exactly that photo is taken.
   I guess I really need to put out a trigger just in case for people like yourself that can use external
processes .. But your right, it could just be your images are too good.. Of course, you dont need to use Gearotic at all for taking of the pictures, so long as you save them in alphabeti cal order so the
reconstru ction knows which is which..
   
    In the fall when I revisit the 3dfromweb cam, Ill get some images from you and analyse why
their so strong. Ive moved on to a new investiga tion now that will keep me busy till the fall,
so we'll discuss this in more detail then. The new 3d will be released than as well.

Thx
Art


Title: Re: WebCam Data
Post by: DanL on June 12, 2015, 06:12:37 AM
ok sweet I will play with what is there now and just fart around with it I might try a infinity wall so there is not a backgroun d for the cam to pick up I will do a negative and positive to see what happens, it a bloody good program.

I whacked the slider to the max`s it was very very deep 6 m but what was at the end was really sharp and clean.

I will post picks if the infinity wall works and its just a second screen in the box and its just a cheap webcam with auto focus and zoom.

thank you for another one of your cool products and letting me play with it


Title: Re: WebCam Data
Post by: ArtF on June 12, 2015, 06:55:19 AM
Dan:

  Excellent, let us know how it works out. :)

Art


Title: Re: WebCam Data
Post by: DanL on June 20, 2015, 02:24:16 AM
hi art I have gone back to the dev version of gearotic I could not get anything useful out of the version you sent me with out being 100 mm plus deep I played for quite a long time everythin g I tried something s helped some not.

I think the problem is the webcam I am using compered to what you have mines a Genius FaceCam 1020.

I have it set to 640 x 480 YUY2
250ms
Confidenc e is around 100
Gradient Scale 1.50
pedestal closure ticked (what ever it for)

pick attached I got a new program for the web cam I have and can zoom in and pan with it.

so it is better than any other program for scanning I have now




Title: Re: WebCam Data
Post by: ArtF on June 20, 2015, 05:32:54 AM
Dan:

 If you get a chance, could you zip up a 4set of shots saved from the screen? Id like to
experimen t with your data in the new code. Pulling out too far is an odd one.. Id like to see
if I can make any adjustmen ts to allow a camera such as yours to me a bit more muted
for the release of the new code in the fall..

Thx
Art


Title: Re: WebCam Data
Post by: DanL on June 20, 2015, 08:12:11 PM
will do soon


Title: Re: WebCam Data
Post by: DanL on June 24, 2015, 09:25:17 PM
Hi art lost the new version of the scanner you sent me would the picks taken with the dev version be ok if not can you send me the file again

ta
Daniel


Title: Re: WebCam Data
Post by: ArtF on June 25, 2015, 06:10:19 AM
Hi Dan:

  Any 4 pics will do, that section never changed, I just wanted to see data with such a high confidenc e rating.. ( 100 is pretty
huge, I get 30 or so typically .. ). :)

Art


Title: Re: WebCam Data
Post by: DanL on June 25, 2015, 06:34:13 PM
ok I will do a set, white back ground, black backgroun d, mdf back ground and no back ground the endless backgroun d did not help


Title: Re: WebCam Data
Post by: ArtF on June 25, 2015, 08:06:12 PM
Dan

  Thx,. no rush. I wont touch the 3d code now till the fall , Ill clean it up a bit then before the
first release on the next developme nt year. I really just want to analyse the numbers to see
just how powerfull your gradient fields are , and to see if I can put in code to fix that when a person
is lucky enough to have that good a setup.

Thx
Art


Title: Re: WebCam Data
Post by: DanL on June 25, 2015, 10:30:53 PM
the set up is diy wast materials and bits what where laying around you could do the same with a dead microwave it`s just a box with a door