I’ve watched a lot of videos and read numerous articles in the Lightburn library but still have some basic questions on camera setup. My Creality 10 watt laser bed is 400mm x 415mm but I’ve downsized to 400mm x 350mm as a separate laser setup for using the camera (per an earlier question). I’m using the Creality canvas enclosure with the camera mounted at the top so it is about 12" above the work surface. Camera is an inexpensive 170° USB unit from Amazon.
There seems to be inconsistency in the directions on where to place the calibration pattern. Some instructions seem to indicate it should be on the work surface and others seem to show it lifted above. Indeed the calibration steps image look like the calibration pattern is on top of a cardboard box, thus lifting it off the work surface. So which is it: on the work surface or lifted above.
Some things I’ve read indicate the calibration pattern should remain flat and parallel to the work surface while other seem to show it tilted so that its face is always perpendicular to the camera lens when placed around the perimeter locations of the work surface. Here again the suggested image sketches while calibrating seem to show it tilted so as to be perpendicular but most everything else I’ve seen or read seems to say it should always be flat on the work surface. So which is it?
The basic download site for the calibration patterns seems to indicate the AprilTag 6x9 pattern is the preferred pattern. However I can’t get my camera to see the markers unless I use a full 8-1/2 x 11 sheet of paper that fills most of my workspace. The circles pattern works fine though. Curious why I can’t get the AprilTag 6x9 to work. Maybe a non-fully compatible USB camera?
My USB camera is a 170° fisheye lens and as described in the instructions has been difficult to work with in the calibration process. As such I’ve done dozens of calibration attempts as I’ve gotten the hang of this. After doing multiple attempts over a couple hours I start getting the dreaded black screen on CAPTURE attempts. If I continue to repeat these capture attempts though I might actually manage to finally get a visible shot. Removing the pattern for a capture attempt and then putting it back seems to help. I’ve read a couple forum questions where this is discussed but they don’t really address what I’ve experienced. The camera always shows the image in the calibration window but the capture buttons sometimes results in a black image. Very strange.
Question of focusing the camera lens: does the camera focal point affect the calibration results. If I’m supposed to run the calibration with the pattern above the work surface will it matter if I focus the camera on that higher location and then refocus it on the work surface when done? I’ve found nothing on this anywhere.
My most recent calibration attempt has resulted in a relatively straight and square work surface image on screen. However when I draw a square in Lightburn that fills most of the work surface (approx. a 250mm x 250mm square on my 400mm x 300mm work area) the resulting laser created lines are not landing 100% as expected. They are nice and parallel to the drawing lines. However while the top and left hand lines seem right where they should be the right hand and bottom lines, while nice a parallel to the drawing, are about 10mm beyond the drawing lines. That is the laser created square is bigger than the drawing square. From what I’ve read I think I should be able to adjust this using the Width, Height, X Shift, and Y Shift fields in the CAMERA CONTROL window. Am I on the right track here? I’m away from the shop right now so can’t test this theory.
I’ve put a lot together here in one post so apologize for the treatise. Hopefully folks won’t find it just too much to consider at a single reading.
That’s a lot of questions at once…
Let’s start with the camera itself. The camera should (must?) be able to cover your entire work area just right with a little bit of overlap, a few cm. Resolution should be as high as possible, I personally consider 4MP to be the lower limit. Focus should be as precise as possible on the surface of the material. (It can be tricky for machines without raising/lowering machine beds) A compromise must probably be found to be able to give a sharp image over the entire area. In your case, I will focus and calibrate on the material height you use most, or you should save a profile for each material height.
Calibration - calibration is very dependent on the camera’s resolution and your lighting conditions, as well as proper focus. Without it, you can give up April Tags as a calibration tool. Print the old dot map out on a printer, in A5 format or something that fits your machine, and glue it flat on a piece of MDF or stable and similar material. Avoid shiny surfaces of your test pattern map.
I have the best results by laying the map flat on my machine bed and moving it around as instructed. Experiment with your lighting for good results, in the Windows version you can adjust for the camera’s contrast etc.
It is extremely important that you follow the order of the 4 control marks and that you hit it very precisely in the center during the last calibration. You must not move the engraved test pattern, otherwise you will have to start all over again. Only IF you have the opportunity to make a permanent fixture that is very precise to lock the test pattern, you can reuse it.
That is the way I do it to achieve the best results - on my system.
Good advice at the end, be patient and do not settle for half good/bad results otherwise the use of the camera will be useless. Final precision should be sub mm accurate, I am under 0.5mm.
Place it on the work surface. The exact height doesn’t matter; the cardboard is placed to get a better image recognition and remove surfaces like a honeycomb pattern.
It should stay flat. Everything else doesn’t make sense. You want to calculate the image distortion in the image plane.
It’s not the camera itself, but the camera resolution. If it’s too low, the camera will have problems detecting it. If the circles work, choose them.
The camera focus should not change. A fixed focus camera works best. Focus on the desired calibration plane.
Not really. If the result does not fit, do the calibration again. You can adjust some minimal deviations using those functions, but if you need those, the calibration was incorrect before.
You might add some screenshots to your post to better visualize your issues.
Thank you both for the extensive replies. I’ve been on the road all day and just got back to my laptop.
Before heading out this morning I had a few minutes to work on this again. I took yesterday’s calibration and used the shift and width/height buttons to dial it in where it belongs. The triangle was off by 2 to 3 mm’s when I started and was right on an 8" x 8" x 8" triangle when done. I also gave the camera a slight twist as I realized it wasn’t perfectly aligned with the bed. I’ll go with those results for now and see how they hold up over time.
The only thing in your replies that was new to me was the resolution factor of the camera. When I get back to the laser next week I will see where that stands. Cheap camera probably means lower than ideal resolution but as I said I’ll see how it goes for now.
Yes, it’s a stupid error in the calibration process, 0 means no result but is misinterpreted. Sometimes the captured image also turns black at the 0 result.
Thanks. That explains it plus the black screen when it comes up.
I checked the camera and it’s 640x480 so waaayyyyy to weak a resolution. I’ve ordered a 4k 3.6mm camera to try. Just double the $20 price of the first one as it’s on sale. Certainly worth a shot.