Forum

Welcome Guest 

Show/Hide Header

Welcome Guest, posting in this forum requires registration.





Pages: [1]
Author Topic: Saw tooth circle and dubious text
Visualbyte
Beginner
Posts: 33
Permalink
Post Saw tooth circle and dubious text
on: March 3, 2015, 12:26
Quote

How it is meant to generate circles?

sandy
Administrator
Posts: 1235
Permalink
sandy
Post Re: Saw tooth circle and dubious text
on: March 3, 2015, 13:19
Quote

How does it comes out on the page?

There is no typical way to generate single-line fonts (sometimes called "engraving typefaces" or fonts), but my esteemed colleague Kongorilla started a thread about this a few days ago - http://www.polargraph.co.uk/forum/polargraphs-group2/general-forum1/single-line-fonts-roundup-thread381/.

I have not come across any software which does this natively, but EvilMadScience has a plugin for inkscape that can do it. It's designed for the eggbot, I think it's part of the eggbot extensions for inkscape. Not used it myself.

sn

Visualbyte
Beginner
Posts: 33
Permalink
Post Re: Saw tooth circle and dubious text
on: March 3, 2015, 15:48
Quote

I found the Inkscape plugin.

sandy
Administrator
Posts: 1235
Permalink
sandy
Post Re: Saw tooth circle and dubious text
on: March 3, 2015, 21:21
Quote

The preview in the controller is only ever that: a preview, so the jaggedyness is really because of a quirk of how the controller quantises the points into it's own grid. It's not accurate when you get to the pixel level. A circle should "draw" the same regardless of what generates it, or how messy it looks in preview.

Visualbyte
Beginner
Posts: 33
Permalink
Post Re: Saw tooth circle and dubious text
on: March 3, 2015, 22:34
Quote

I had been exporting the queue from another machine and it could be there was a discrepancy in the pen size. It ended up pretty jagged on the paper but the other circles I happened to draw which came out ok the queue was compiled on the tethered machine.

DaniK
Beginner
Posts: 33
Permalink
DaniK
Post Re: Saw tooth circle and dubious text
on: January 9, 2016, 16:45
Quote

Hi,
I think I ran into a similar problem ....i tried to produce a similar image to Sandy's Rasterbator moon image (http://www.polargraph.co.uk/forum/polargraphs-group2/drawing-techniques-forum4/rasssssssttterrrrrbaaaator-thread49/) but of course without rasterbator 😎 ... and therefore generated 2D circles in a 3D modeling software exported them into illustrator and saved an svg file from there...
the circles look perfect in iIllustrator (second image) and in the PolargraphController they start to look as if they have small dents ...but the generated vector paths are very strange (first image)...and they get drawn like this as well unfortunately (pen NOT wobbling around)(third image).....
tried several times already (with different files and sizes)...

Does anyone have any suggestions on how to start troubleshooting on this?
Thanks!

Daniela
Image
Image
Image

sandy
Administrator
Posts: 1235
Permalink
sandy
Post Re: Saw tooth circle and dubious text
on: January 9, 2016, 23:40
Quote

Very interesting, I've known about this phenomenon for a while, but haven't ever really looked much closer into it.

There's three things at work here.

1. Polygonizing. Geomerative has a way of converting curves into polygons. The method that the Polargraph Controller app uses is called RG.ADAPTATIVE, but there's an alternative called RG.UNIFORMLENGTH and seems to result in a more regular output. You can change which it uses by modifying the line: https://github.com/euphy/polargraphcontroller/blob/master/polargraphcontroller.pde#L540

See pic at the bottom to see the difference it makes.

There's also a couple of extra functions in Geomerative that you can use to change how curves are polygonized:

    setPolygonizerLength

    public static void setPolygonizerLength(float length)

    Use this to set the segmentator length for the UNIFORMLENGTH segmentator and set the segmentator to UNIFORMLENGTH.

    Parameters:
        length - the length of each resulting segment.

    setPolygonizerStep

    public static void setPolygonizerStep(float step)

    Use this to set the segmentator step for the UNIFORMSTEP segmentator and set the segmentator to UNIFORMSTEP.

    Parameters:
        step - if a float from +0.0 to 1.0 is passed it's considered as the step, else it's considered as the number of steps. When a value of 0.0 is used the steps will be calculated automatically depending on an estimation of the length of the curve. The special value -1 is the same as 0.0 but also turning of the segmentation of lines (faster segmentation).

(from http://www.ricardmarxer.com/geomerative/documentation/geomerative/RG.html#polygonize%28geomerative.RShape%29)

2. Basic resolution. There isn't actually a ton of physical resolution to the machine, but there's a little more to dig out if you need it. The steps per rev is set to 200, and the step multiplier is 8. So that means that there are 8 microsteps in between each full motor step. The machine is addressed using full steps, but moves in microsteps. The reason is that microsteps are unreliable - if you stop on a microstep then the motors get hot and the movement is unreliable. You can never be sure where it picks back up from, and it could be as much as half a full step out - and the errors will compound. As long as you aren't planning on leaving the machine motionless for more than a minute or two, you can use microsteps to address the machine too. Open your properties file and change

machine.motors.stepsPerRev=1600.0
machine.step.multiplier=1

This will give you a bit of extra resolution. It makes me uneasy! See pic at the bottom to see the difference it makes.

3. Last thing - plotting using the wrong coordinates system. Just looking at the shapes from the queue preview (the red lines), it's obvious that they're snapped to a cartesian grid, whereas the commands themselves use a native coodinate system. During the preview phase, the commands are grabbed from the queue, converted back into cartesian, then scaled to the screen. So the jagged edges on the screen don't necessarily correspond to the jagged edges that get drawn. HOWEVER, if they are distorted on-screen, then I think there's probably a pretty good chance they'll be distorted on paper too - just slightly differently. The degree of distortion is probably a good indicator.

I can think of a solution to this, but it doesn't do anything to get extra fidelity onto the page, so I'm not that interested in doing it.

~

Here's a pic showing improvements that can come from changing the polygonizer style, and increasing the steps per rev.
Image

DaniK
Beginner
Posts: 33
Permalink
DaniK
Post Re: Saw tooth circle and dubious text
on: January 10, 2016, 10:39
Quote

Hi Sandy!

Thank you so much for this extensive answer! I will dig into the different possibilities and get back here with the results....

Additionally, maybe that also helps in any way, I tried it with simple rectangles (closed polylines) a little later yesterday, as I read several threads in the forum that it could be just that preview problem and thought that with rectangles there shouldn't be that visual snapping problem or if so, evenly along one line of rectangles...but it wasn't either.....(see images)...although I have to admit I like the output anyways a lot, I have the feeling no one would doubt, that I drew the thousands of pixels by hand if I told them so 😉 .... if I don't find a solution, I just decided that I then added a new drawing style to my machine....

(Got the same visualizing problem with the "SpiralfromImage" sketch from krummrey...it basically didn't follow any of the lines and only matched here and there with the endpoints of one....and basically after I figured that out, with every svg I then imported....but I haven't tested the actual drawing output yet...)

thanks again!
Image
Image
Image

DaniK
Beginner
Posts: 33
Permalink
DaniK
Post Re: Saw tooth circle and dubious text
on: January 10, 2016, 15:45
Quote

after some more testing regarding the rectangles:
it definitely gets better by increasing the steps per rev to 1600 and step multiplier 1...especially for small rectangles (image 2)....also increasing the general size of the rectangles of a maximum edgelength of 6mm to 12mm helps a lot in terms of accuracy 😉 with the additional change to 1600 revs (lower half on each image) they almost have 90° angles...
but I am very pleased with the result of the small rectangles now! (bottom image 2)....running the whole drawing right now! (still A3 in size...maybe I should bigger size and lower in resolution in general with my polargraph...might solve lots of problems....)

the RG.UNIFORMLENGTH did nothing recognizable to the small circles (I have to admit they are very small)...will try some more with them and look into Geomerative....and also the polargraph firmware itself, i guess, to learn how everything gets actually drawn....the vector options seems to run the motors very different to the pixel style settings....
So far I don't really get yet why my machine is "trembling" like that, drawing svg circles...as the spiral pixel settings produce really nice circles even in small scale (succesfully tried with grid size 6)....I probably need to find the right way around the svg import and vector drawing for the circles....(like what the G2 and G3 commands do to the quality of circles in comparison to G1....)

And Sandy, do you - by any chance - still know what the smallest and biggest circle radius of your moon drawing was? Just found your high resolution close up picture on flickr...looks similar to my outcome 😉 ...
Also tried the rasterbator now....same problems with the pdf/svg....maybe the settings were also to small...
Image
Image

sandy
Administrator
Posts: 1235
Permalink
sandy
Post Re: Saw tooth circle and dubious text
on: January 10, 2016, 23:33
Quote

I like your investigation - yep I've seen the same kind of distortion in small shapes.

The reason why the small circles are so jittery (compared to the smooth spirals) is that the circles are made of lots of small single lines, whereas the spirals are a planned curve. That is the only case where the machine knows how to draw curves. The small single lines wreaks havoc with the path planning code / acceleration, and that whole function is actually pretty awful - crude if nothing else.

You can probably mitigate a bit of the messiness by running the machine very slowly - acceleration and max speed.

The odd geometry of your squares is because of the coordinates system. When you draw using the native pixel styles, everything is smooth and regular and parallel (or rather concentric) because it moves in the machine's native coordinates system. When you try to use a coordinates from cartesian system, they get snapped to the native system, and there's almost no chance it'll end up being strictly perpendicular and parallel in a cartesian system.

Personally, that slightly uneven pattern makes me so happy, so very very happy. I love it. That pattern of squares makes my heart sing man, it really does! It's exactly the little return gift that pushed me into keeping developing this kind of machine. Unfortunately for anyone who has bought a kit ignoring my warnings (http://www.polargraph.co.uk/faq/) and expected to get a precision plotter, that is something I see as the essential character of the machine - it'll never be eliminated, and I will never try to do it!

(Truth is, if you want a high precision printer, you pay high-precision dollars...)

That said, the way that vector drawing is handled is frankly pretty embarrassing and I covet the path planning algorithms of gocupi etc pretty badly.

Do you mind if I pinch these pics to write up a blog about this investigtion? I know you might see this as a failure, but I think it's really thrilling, really exciting! A new facet!

sn

DaniK
Beginner
Posts: 33
Permalink
DaniK
Post Re: Saw tooth circle and dubious text
on: January 11, 2016, 15:38
Quote

The opposite Sandy, I don't see it as a failure at all! Intentionally you always have to start somewhere...but 3 excessive years at art school have at least taught me to worship the "accidents, failures and coincidences" during a (design)process and also why 😉
That's exactly why I got a polargraph and not a Cameo, Roland DXY or comparable...way more fun this way.
Just wanted to not get too far off from a square like in the first test....and the "krixikraxi circles" still shout for some straightening up...(next weekend maybe)

I am also way more interested in the machine - human "collaboration"....fiddling around too much with 3dPrinters, Lasercutters and CNC machines for work anyways....
Did some (extensive 😯 ) manual hatching, yesterday night, of the first polargraph fullmoon (not sure how I find it yet and let's see where that leads on the long road)...the whole drawings:
Image
Image
and some close-ups in the end....

I am quite familiar with the Cartesian GCODE commanding....so having to look into the "native coordinate system" of the polargraph is what I meant by saying that I need to figure out how to generate the appropriate C-commands directly...as i have perfect circles (8 control points) I just need to do some brain-exercise and look into (arduino/processing) your code....is there any documentation I could start with (the "sweet spot calculations" I did a while back already caused some headache....)

Sure! Use whatever you need! I would be even a little flattered that this motivates you for a whole blog-entry! 😀

And my grasshopper definition you can find here:
http://www.polargraph.co.uk/forum/polargraphs-group2/drawing-techniques-forum4/rhino-and-grasshoppper-thread407.0/#postid-2856

Image
Image
Image

DaniK
Beginner
Posts: 33
Permalink
DaniK
Post Re: Saw tooth circle and dubious text
on: January 19, 2016, 21:29
Quote

Just a little update I wanted to share:

I decided to try to save some time and relieve some pressure from my poor tiny servo-motor (after thousands of lifts for all the testing) and especially the pen-tip (everything drawn so far with a stabilo 88, the tip has a lifetime of about 3-4 moons (each 3844 lifts), and then you can already really see a difference in linewidth)
I've done that by connecting the little rectangles into almost one continuous polyline.....I split them vertically in half...the right polyline is always drawn upwards, the left one downwards...and then there is only one servo lift to get to the next line 😉

I quite enjoy how one can see now exactly how/the way it seems to snaps the Cartesian grid points! The right polylines are very similar to each other and so are the left ones...and together they form kind of little diamonds strung together on lines.

drawn on the same paper with the same pen as the other tests (stabilo 88, unfortunately at the end of its lifetime with a very flattened tip...)
Image
Image

Pages: [1]
Mingle Forum by cartpauj
Version: 1.0.34 ; Page loaded in: 0.025 seconds.