How do I create a random pattern of vectors?

[Brady Watson]

I need to make a random Swiss cheese pattern of holes .75, 1 and 1.25" dia in a piece 5" wide by 18 feet long. Since the piece is so long, the pattern needs to be truly random...or at least random enough that your eye will not be drawn to any noticable pattern.

Can anyone tell me how they would go about making a random pattern like this and what software/techniques they would use? Corel seems like the most likely choice, although I don't know how to do it.

Any help or tips would be appreciated greatly!!!

Thanks!
-Brady

[srwtlc]

Brady,

Since I was working in Corel as I read this I thought it would be faster to do one for you to look at than to tell you how at the moment. Do you have CorelDraw? I can send you a .cdr,.dxf or an .eps to look at.

You have mail ;-)

[Brady Watson]

Looks great! Thanks!

You have mail.

-Brady

[mikejohn]

Scott
How did you do it so they didn't overlap?
Just curious
Mike

[Brady Watson]

I'd like to know how to 'fish' too!


-Brady

[srwtlc]

Mike, here's the email text I just sent Brady. By mirroring as such there where no overlaps.

I first made a rectangle (random), then in the 'Object Size" field give it the parameters that you want (be sure to unlock the fields first or it will maintain aspect ratio). Then I made three circles the size you stated in the same manner. Then select one or all three of them and do a duplicate (ctrl-d) a number of times to get a bunch of the circles. You can then drag and drop them into place where you want until you get a random pattern that you like. Then select all of them (group if you want) and do a mirror copy of them. The best way to do this is to bring up the 'Transformation Docker' with 'Arrange/Transformations/Scale', choose mirror horizontal or vertical, place a check mark in the box for the direction and 'Apply To Duplicate'. You should have the mirrored set selected now and you can just mirror it again and again until you get to or go past the end of your piece. I then just deleted the extras and filled in the corners to finish it out. I seldom worry about where 0,0 is in Corel and just reposition it in my cam program (Vector).


Scott

[srwtlc]

Here's a shorter sample


4290

[gerald_d]

From the picture, the pattern repeats 2.3 times - but it does look random enough.

[mikejohn]

4291
This is truly random.
The x,y position is random, and the selection of .75, 1.0, or 1.25 is Random (Pseudo Random I know, but very close to true random)
In Excell
A column of random numbers, between 1 and 3 was made.
A column of random numbers for the x position between 1 and 10 was made.
A column of random numbers for the y position between 1 and 10 was made.
This was saved as a .csv file
this was opened as a text file, and odd commas removed
This was saved, the file extension was changed to .scr, and the file opened in AutoCad as a script.
The whole thing took about 5 minutes.
To make it 5"X 18 feet would be a moments work.
To ensure the circles don't touch, or overlap, requires another 30 minutes of mathematics, to ensure staying truly random.
Unfortunately, I leave to trek across Europe in a few hours, wont be home for a week.
Should a .dxf file still be needed at that time, I am willing to make it.
If, as was asked, it needs to be truly random,then it can have no human input.
If it just needs to look OK, then the Corel drawing may well be just the thing
-Mike

[srwtlc]

Obviously the more random and the farther out you place the first group of holes before mirroring, the less noticable the repeat will be. For making up a quick sample, I just went out only so far before mirroring. You could make two different groups and mirror them randomly also.

Scott

[mikejohn]

Scott,
I believe your approach would be similar to my own if I needed to produce a visually random display of holes.
I was only trying to show that, if you needed truly random(Bradys words), possibly for a scientific reason, then it is possible to do this.
Or almost! Computer generated random numbers are not truly random, and the mathematical model to remove holes touching or overlapping each other, or the edge, would also reduce the true 'Randomness'
What interests me is the possiblity to do what you want with a shopbot, providing you can make the software work for you
-Mike

[Brady Watson]

That works well enough for me. I can add/delete areas where the pattern seems noticable. It's just hard to see small circles over an 18' span on my 19" monitor. It would be nice if one of the software packages had a random pattern generator (or at least random enough, Mike) I think that is a major problem in making things on the cnc look real when a random pattern is involved. Fibonacci sequences help, but...I don't want to spend the time reinventing the wheel if there is something out there already.

Thanks for the help!
-Brady

[paco]

I too looking for something similar; some of my customer ask me to mimic a caring like aspect at the cleared area of signs using ball-end bits of different size but the first result I tested gives a robotic like carving finish... I would need to spend long time to create a finish aspect that "could" look similar to what they get from sandblasting the cleared area background. On the other hand, I could be lower price than masking/sandblasting though??... They say they don't like the process of masking and sandblasting their signs but it's the only way to get what they are looking for... any idea?! Anybody already worked on this type of thing?
This kind of "strategie" could be called texturing...

[artisan]

4292

[artisan]

Somebody say Swiss Cheese? The above 12 ft "sandwich" was carved in foam on the Bot and was part of a larger sign....D

[Brady Watson]

Paco,
If you had to charge for your time to make a blasted-look sign, you would be a LOT higher in cost than a sandblasted sign.

Darrell,
Outstanding work as always!!! Is that a softshell cat in there?


-Brady

[paco]

Really?! Hummmm... maybe it would depends on the needed resulting aspect; sure a "same" finish as a sandblast wooden sign would be costly ad would require a 3D file (to look like erode wood grain)... but I plan to test something in 2 1/2D with many random "gouging"... Anyway thanks for your view Brady, I appreciate!

[Brady Watson]

The only reason I say that sandblasting is cheaper is because I was approched by a sign guy to try carving the texture instead of blasting it. What normally would take 30 min all together to blast, takes hours and hours to carve 2.5D or 3D.

-Brady

[gerald_d]

Talking of random patterns, we have been approached to do "reproduction" Dogon doors (http://www.nigerbend.com/catdir/catalog_24.php3). Snag is, how does one reproduce that crude geometry? Wondering if there is a "morphing" program that allows one to "disturb" a vectorised drawing?

Notice the hinges on the doors. This one (http://www.nigerbend.com/catalog/mini_Akandoor.jpg) ($2500 for 28"x55"!) already seems to be fake because there is no wear on the hinge pins. There is also fake wear - the top hinge must wear on the door side and the bottom on the "frame" side to be correct

[Brady Watson]

Gerald, not sure I am following which style of door you are referring to. (distorting vectors on)

1st thing that comes to mind for distorting vectors in that way is to have your grandson/nephew/niece try to reproduce one from a line drawing....or...Part Wizard has a shear function that lets you distort vectors in the X and/or Y.

Those prices are insane...carve your design in some timber, sandblast it and run a torch over a few places and it'll look old...

-Brady

[Brady Watson]

Here's the 10'6" one in 1/4" mild steel...18 footer is next. All cut on SB plasma cutter.




4293
4294

[paco]

Gerald-
You sure have GREAT product that get my attention in your country... the one I prefer is "Oldest genuine door"... but I would need to fit it with some wheater strip if I installed this on my house up here in Canada!!!


Brady-
Do you happen to have a zoom view of the right pic; I'm curious about the cut from an ShopBot plasma... do you need much grinding/cleaning?
I may get you the feedback that I will get from my customers when they see what I can provide 'em for what they are paying for as for sandblast signs... is 30 minutes inclued making and installing the mask?... still thanks for your input!

[artisan]

I tend to agree with Brady on the sandblasting issue. It really just depends on time though....and how busy you are. If you have plenty of work, then routing a sandblasted look is not cost effective. However....having said that, here is a job I just installed Friday. The requirements were for a natural woodgrain look, but all materials had to be weatherproof as the sign is mounted on an aluminum frame in front of a waterfall. I chose aluminum and HDU, sealed with epoxy. The woodgrain is not visible in this wide shot, but it is VERY pronounced in person. I'll be shooting better pics later in the week. In this instance, routing was the way to go....12' x 4' panels...6 hours each (I want an Alpha)....D

4295

[artisan]

Here is a barely better shot of the sign itself

4296....D

[Brady Watson]

Paco,
I'll try to post pics that show more detail. If you setup the plasma parameters correctly, the cuts are very clean. It took me less than 1 minute to run the orbial sander down the entire 9' piece to get a perfect dross free surface.

-Brady

[paco]

Even at the "stop & go" at nodes of the CAD drawing??! I'll be watching... I'm very curious about this...

[Brady Watson]

Yes...even at the lead-ins and lead-outs. I'll post pics tomorrow when I get back in the shop. On inside cuts, the lead-ins are towards the center of the cut (scrap...like the waste portion of a hole cut-out) and outside profiles lead in from outside of the profile extents.

I have to hand it to Bill Young, Gordon, Ted and Chris Burns for getting me a Windows control software version that is designed for the Plasma Bot. I cannot believe the incredible difference that the software has made on cut quality. I use all the same settings as the DOS version, and even though they were good...the Win version is unreal. I cut a bracket out of 3/8" mild tonight for a hot-rodder that had next to no bevel on the cut. He spent a fortune having parts cut out on a water jet and he told me that the plasma cuts were tighter and smoother than the WJ. (to me...that is hard to believe...but just because you have a tool doesn't mean you know how to use it...)

-Brady

[paco]

In fact Brady, I meant the deceleration and "re-acceleration" at some points within a drawing; often when I cut curves with my PRT Alpha, the tool slow down at nodes then get back to it's full set speed... Do you observe that too or not on both of your SB tools? Here a pic with thoses slow "spots" on a particular drawing...

4297

I just test cutting white insulating foam yesterday that another guy failled to cut properly; at first test it cut FAST and VERY clean to me?!?! Though, I don't have any difficulty believe that many CNC owner don't get all they can out of their tool... Many CNC owners/users see thoses tool as some kind of "magic" machinery that will trow out finish parts without any efforts of their own... Anyway, I too make mistake but I just don't stop testing further to improve....

[Brady Watson]

I don't have any problems in cut quality on the router or plasma because of curves or ramping...You may want to set your ramp percentage up a little higher if it is stalling on sweeping curves.

Here's 2 pics I took today. You can see some remanents of the piercing points where the tool moves from the center to the inside of the vector coordinates. This is pretty much standard stuff. The angle at which I took the pics really focuses on them and makes them more apparent than it would look from the top. The little bit of dross that you see on the top of the 2nd piece was removed in less than a minute with an orbital sander (entire 10' length)...you can take it off with your fingernail if you had to.


4298
4299

-Brady

[paco]

Hummm... I will take a look at this setting tomorrow and do some more testing... Thanks!

The plasma cut look just fine!! Not much to remove... I'm suprised! Is this torch use only a cutting gas or an assist/shield too?

[Brady Watson]

Just air from the compressor. You don't need a shielding gas unless you are cutting exotic materials. SS may benefit from Argon or Nitrogen, but that gets very expensive.

It may be that I can completely eliminate the dross on the top side by fine tuning, but as it is I am happy with the results.

-Brady

[paco]

Top dross can be (sometime) reduce by lowering the cutting height (voltage)... since it will narrow more cutting gas pressure through the kerf instead of loosing some at the surface of the material... but, as you say Brady, the most important thing is to be happy with the result!
By the way, shielding gas is normaly a gas that flow around (using a secondary "nozzle" which is often called a shield cap) the cutting gas to improve piercing thick material and to reduce top dross and/or improve cut edge quality by introducing a secondary gas to "clean" the cut edge... cutting gas are air, nitrogen or oxygen. Shield gas can be the same as cutting gas and one can use cardon dioxyde (cheap), propane mixture, hydrogen mixture, argon for non ferrous metal... steel is really best dealed with oxygen as cutting gas with some air as shieling gas if the torch allow it... this use the principle of the oxy-fuel cutting process... the same is true with laser... though laser don't use shielding gas as far as I know!?

[daniel_carr]

VS3D has tools for creating a random texture. It also has tools for performing polynomial warping (which is more sophisticated than simple skewing).

Below is an image showing a random backgound with raised letters. The orange vectors (a little hard to see) were drawn by hand in VS3D (just click and drag). The box around the letters was originally perfectly straight, but the vectors were used to warp (stretch) the surface.

If you are making a sign that has raised letters, and you are using an area clear to reveal them, it won't make much difference in the total cutting time if you use a flat background or a textured background.




4300