Hi Fellas,

I have been watching this thread for a while now but honestly I can only follow along with about 1/2 of what you are talking about.  Obviously you both seem to have about 100X more technical knowledge on this subject but, I thought I would jump in anyway.  Excuse me if I sound like a nitwit but all I have is my experience during the past week of 'playing around'.

I totally agree with you that the sharpness and location of the laser and the quality of the camera play a huge role in the acquisiton of data but using the low end stuff I have been able to generate some really good preliminary results.  The clouds I have been able to generate are so rich with data that one of my problems has been that I acquire TOO much data (points) and this makes thing really hard on the meshing software.  Funny thing is, Art's plug in allows for the user to acquire even more data.  I don't know what one would do with all of these points.

Which leads me to my next question...  What do you guys plan on scanning?  Is it bigger than a 'breadbox'? Or are you guys looking to scan really small stuff like coins?  The reason I ask is because I think that the type of subject matter has a bearing on how you need to  up your camera/laser ( i.e. small stuff = low/close camera laser).  I need to start messing around with this myself.  I have already scanned small things and hopefully will try some bigger stuff this week.

The camera laser are obvioulsy important but so far, I have been able to get good looking clouds.  IMHO the next roadblock is what to do with them after they are clouds.  You will soon see that depending upon what you scan that although you get good clouds they often have holes or other areas that need to be edited.

Mach Cloud is pretty good and you can't beat the price but, due to its 'work in progress' nature and the demands on Art's time I think it will be a while before we see some of the features that I think it needs to be really useful.  So, that put me to RHINO.  That program is cool and does WAAAY more than I think most people need.  Plus it's kinda hard to swallow the $ price tag  when it won't generate useful toolpaths unless you 'pony up' more $ for their plugin.

After you generate meshes, then you need to generate toolpath.  Again, there are not too many software options there.  Meshcam has produced some pretty encourgaing results and I think I'll buy it because I can't find much else but it does crash and 'bog down' depending upon the detail of the mesh fill I 'feed it'.  Depending upon how detailed your mesh is, toolpath generation can consume computer processor time like 'carter's got liver pills'.  If you've been thinking about upgrading your processor and memory- now would be a good time.

There are so many variables to this process that it's hard to focus on just which one will improve the end result.  I'm not really sure that it's all in acquisition.  While that is no doubt important, at this point in my experimentation I would say that it's only about 30% of the whole pie.

Regards,
Sid

Hi Sid

First off never apologise for asking questions. I was pretty lost at first too until  the penny finally dropped from Toms explanations and a google about how to focus View cameras. BTW IMHO it is Tom that is the resident expert on this. I'm just a guy "with all the gear and no i dear" . well maybe a little.

Anyway I think you are right in stating that the getting of the data is only half  (or less) of the story and what you do with it can be the bigger task. However remember "********* in, *********t out" so the better the data one can get at the beginning the easier it should be to create a good mesh at the end. I would look at using something like Gmax to manipulate the mesh as its s/w made expressly for this. I have no experience with it yet but its free and sounds like the go to clean up a mesh and reduce the number of vectors before importing it into a CAD program.

I am still at the begining of my CNC router build but I hope to do all kinds of stuff with it once is finished. The types of things I see myself wanting to scan will be anything from small figurines to large foils like rudders. Ideally I would like a setup that can measure depths up to the limit of a cutting tool so at least 100mm. To do this I need to balance the FOV of the camera lens and the pixel resolution of the image that this FOV dictates. Range versus resolution.  I think I just answered my own question,...hmmm if I want say 0.1mm resolution then times this by the pixel resolution of the CCD to give me my max FOV.

I also think that this video scan setup could be used as a non contact zero finder with a bit of work. I think it is because of the scale issue you mention that Tom is planning to have the camera perpendicular to the image/laser plane as this makes changing lenses easier.  A Scheimflug mount complicates things a bit as the change in focal length also effects the image focal plane tilt.

The idea is to get a good setup that doesn't cost the earth and is fairly easy to build. (good, fast, inexpensive pick any two out of the three). I just happen to have some security style cameras and lenses so thats what I'm using. Having said that they are not that expensive of Ebay from what I've seen.

I do think many people may be overlooking the impact that getting the laser plane and the optical plan angles correct as possibly they see it as being too dificult. Get your head around how a View camera works and it will all become clear (pun ) That is the gist of what Tom has been saying all along and I think he's right.

Happy testing

Cheers

Mark

Hello,

Here is a general drawing of my design. I wanted to keep it simple for discussion so I deleted a lot of the details that make it easy to orient and machine in a fixture.

Note that the included angle between the laser and the camera is 90 degrees. That allows the lens to be attached to the camera in the normal way. Then rotate the the threaded lens to image the laser plane onto the ccd plane.

The camera will have a visible signal if the target crosses the laser plane within the camera's field of view.

A shorter focal length lens will have a greater angular field of view so will be less accurate. Also, inexpensive wide angle lenses are more likely to produce an image with barrel distortion. Barrel distortion can be corrected in software but that requires more complicated calibration and calculation methods.

Tom Hubin
thubin@earthlink.net

Hey Fellas,

Yesterday I tried to relocate my camera and laser per Tom's drawing.  I can't totally explain why but, I wasn't able to get any good results.  I understand the principle behind calibrating this way but, for some reason it just doesn't work with my setup.

Here are a couple of observations  when I mounted my camera as shown on the drawing it is upside down.  This results in the my calibration line for my cube showing up lower rather than higher as the table (Zero).  Now, that I am sitting here typing this I am wondering if I could have used a - (neg) value in the cube size setting box?  But that's not what I did    Instead I used a setting in the camera driver to 'flip' the image so it appeared correctly on the screen.

While this fixed the problem with my Y axis, it didin't invert my X axis.  So, when doing a single pass scan in the Y, I did get a cloud that looked OK.   But I have been expanding my horizons as of late and have started to scan objects as long as 6 Feet using the X scan/step.   When I did scans that required multiple passes in the X, the cloud was 'confused' and the stripes could not be aligned.

BTW- here's a little 'Gem' of info... I discovered that Rhino is able to open striped clouds and AUTOMATICALLY align the striped segments almost perfectly.

I have also noticed that there are definite "Sweet Spots" in camera & laser positioning.  At this point I have tried around 10 different positions/configurations and seem to always come back to a certain range where things look good.  I suspect this just has to do with the way this Plug in was written in terms of the calibration on the cube process and the angle/height calculation.

No matter!  I have been getting some REALLY super results.  I don't have time right now to post any pix (I have to go out and try and make some $$$ today to support my CNC "habit"  ) but, I will try later. 

IMHO,  even though the plugin is not 'perfected',  it is really pretty damn good at doing what it supposed to do.  I do have a list of things I have noticed that are 'funky'  but so far, I have been able to find ways to work around most of the issues.  Hopefully others will see this also when they start generating clouds.

I would like to see just a tad more functionality added to this plugin.  I don't know if this is possible but it would be great if you could control the STEP in scan quadrants (like one has the ability to control the Feed Rate at different lines of G-Code).  I've noticed that fo some of the things I scan some areas require a 1mm step to get good resolution and other areas can get away with 10mm because there is not a lot of detail there.  Currently one can not do this so you would have to use the 1mm step for the whole scan.  This results in a HUGE point cloud file which is a little unwieldy.  I suppose if someone was good with RHino, you could peice multiple scans together to accomplish this but, this takes more time.

Gotta run now- I'll try and post some pix later.

Sid





 

Here are a couple of observations  when I mounted my camera as shown on the drawing it is upside down.  This results in the my calibration line for my cube showing up lower rather than higher as the table (Zero).  Now, that I am sitting here typing this I am wondering if I could have used a - (neg) value in the cube size setting box?  But that's not what I did    Instead I used a setting in the camera driver to 'flip' the image so it appeared correctly on the screen.

Hello Sid,

I have mentioned the camera orientation options in another thread.

While you can rotate the camera at any angle from 0 to 360 degrees, it is more likely that the camera will be oriented in one of four positions. Rightside up, upside down, rotated +90 degrees, rotated -90 degrees. In addition, you might end up using a mirror someplace and that would produce a mirror image of one of these four. So there are 8 likely setups.

All of these can be handled with only three operations. Mirror X, Mirror Y, Rotate 90 degrees...in any order. Flipping an image is mach3's term for mirroring an image.

Tom Hubin
thubin@earthlink.net

Hello,

Here are two photos of my laser triangulation gauge installed on my Sherline 5410 CNC mill. It is mounted on a Sherline spindle spacer block and that, in turn, is installed in place of the spindle and motor.

The third photo is the underside of the laser triangulation gauge with the camera on the left end, the laser on the right end, and the Sherline spacer attached near the center of the rear face.

I took some photos of the finished metalwork but the lighting was bad. I will retake those and post them another day.

I took some photos of the video display. They are ok but I would prefer a snapshot of the screen. How do I capture the screen as a gif or a jpg or other acceptable file format?

I also have some photos of focus procedures and experiments with agc methods. The photos need to be organized before posting so that will happen some day after another day.

Tom Hubin
thubin@earthlink.net 

Tom,

To screen capture In windows, I use the <Alt> and <Prt Scr> keys held down together to copy the "active window" into the paste buffer.  Note: just using the <Prt Scr> key will copy the entire display.

Then go to Paint, Paste it and edit away.

Elkaholic

Pretty Nice looking camera mount Tom!

Press and hold the 'Alt' button then press the 'PrtScn' button- this will capture your screen to the clipboard.  You can then usually open a graphic editor (like paint) and paste the clipboard, save the file & upload.

Sid

Hello Sid,

> when I mounted my camera as shown on the drawing it is upside down.
> This results in the my calibration line for my cube showing up lower rather than higher as the table (Zero).
> I used a setting in the camera driver to 'flip' the image so it appeared correctly on the screen.

I think I am having the same problem. My image needs to be inverted too. I can do this with the camera vendors software and I can do this with MS Photo Editor. Where did you find this control within Mach3?

> While this fixed the problem with my Y axis, it didin't invert my X axis.

Some camera software uses ROTATE 180 DEGREES to invert both axes. Others mirror horizontally and mirror vertically. These mirror options might be labeled MIRROR or FLIP or one of each.

Tom Hubin
thubin@earthlink.net

Hey Tom,

I fouund (and used) the flip feature in the camera's driver software settings.  You can find that in Mach by clicking the stop video, then camera source- should bring up the camera settings.  Remember to Start the video again when you exit out of settings.

Good Luck!

Sid