Forum

Welcome Guest 

Show/Hide Header

Welcome Guest, posting in this forum requires registration.





Pages: [1]
Author Topic: Test image for troubleshoting
volansky.i
Newbie
Posts: 3
Permalink
Post Test image for troubleshoting
on: July 23, 2016, 11:23
Quote

Is there a test image that I could run on my polar-graph to get an idea of what's wrong with it?
Currently it seems that I have a few problems (the image is skewed, there are no details and the machine draws out of the paper) and after configuration few times using the tutorials I was thinking maybe there's an image (or a set of images) that I could print and easily figure what's going on here (just like the pen width test).

sandy
Administrator
Posts: 1317
Permalink
sandy
Post Re: Test image for troubleshoting
on: July 23, 2016, 14:23
Quote

It'd be great wouldn't it, and I've wanted something like this for ages, but I've never been able to think of a kind of drawing that would reveal something useful.
Any rectilinear shape would reveal miscalibration (ie a square would come out curved), but miscalibration can be caused by a few different things - and the nature of the distortion doesn't tell you which one is the problem.

Roughly speaking, all distortions are symptoms of a mismatch between how the machine is defined in software, and how it is in reality. So, the machine is larger in reality than the software thinks, has different kinds of motors, different sized sprockets or different steps per rev, that kind of thing.

A second class of distortions are about a mismatch in pen position, and these really stem from bad calibration, either an error in the measurement of the home point, forgetting to "set home", or drive-train problems where the gondola has been allowed to move without being told to (weak torque, being pulled or getting stuck).

The problem is that there isn't much different in the symptoms between the two classes of distortions. A curved line that should be straight could mean:
* class 2 distortion, the pen is in a different place to where the controller software, and the machine thinks it is (at least three different potential causes for this, even assuming a perfectly specified machine).
* class 1 distortion, the motors have been defined as 400 steps per rev, when in fact they're 200 steps per rev, or any one of six other key pieces of configuration.

It's very common that the machine is specified perfectly, but the over-excited drawer has simply forgotten to "upload machine spec".

It's pretty hard to diagnose anything except "there's something wrong" with a test image.

What probably _is_ possible, is a kind of interactive process, a sort of "expert system" whereby the system guides you through a series of small steps with an initial assumption of being at home. So it'd draw a straight line across the page, and ask you to measure it, and if it's different from what was expected, then it'd suggest some things that could have caused it.

Until that happens, you've got to do that stuff manually.

One feature that is useful for this is the "draw outline", that will draw a rectangle where you've got your selection area defined.

Another thing I've just thought of (re-reading your message), is are you talking about vector drawing or the pixel styles?

sn

volansky.i
Newbie
Posts: 3
Permalink
Post Re: Test image for troubleshoting
on: July 23, 2016, 22:28
Quote

Wow didn't expect such a comprehensive response in such a short time, thank you! and I'm drawing pixel styles.

Also I must admit I'm still not sure about what the values of the grid size (and sample area) should be and because I got problems with the calibration I can't be sure if they are right or wrong, are there any pointers on how to know that you're on the right range? I tried to figured it out from the tutorial but couldn't really get a decisive answer as it changes with each photo and I don't know how it suppose to look.

sandy
Administrator
Posts: 1317
Permalink
sandy
Post Re: Test image for troubleshoting
on: July 30, 2016, 23:23
Quote

It's quite tricky, partly because of the units that system uses. The grid size is specified in motor steps - so it's different for every machine and there isn't a default setting that suits everyone.

Getting good results is also interdependent on a mix of grid size and pen tip size too.

First, measure the pen tip size. Use a grid size that will let 10 (or so) pixels fit on your page. Do a series of pen width tests to figure out what your pen tip width setting is. If you happen to already know the tip size (or can accurately measure it), then just set it.

Use the "select area" feature to select particular small sections of the picture where there is detail in the very dark and very light areas, and see how they render. See whether you get the range of detail you need. You don't even need to calibrate your system properly for this - I tend to just pull the pen to a blank area of the paper to test this - otherwise you're going over the same area time and again.

Rule of thumb: A larger grid makes a more interesting _drawing_ with a wider dynamic range (more different levels of shading). A smaller grid makes for something more representative of the input image, but it tends to be more shallow in range (only a few different levels of shading).

sn

Pages: [1]
Mingle Forum by cartpauj
Version: 1.0.34 ; Page loaded in: 0.015 seconds.