gerberMUTATOR opinions please

[hagster]

So i've been mucking around with gerber files over the xmas break. My general idea was to create a tool that would make my PCBs stand out and look a little different.

The tool I have created can perform various transforms on standard gerber output of most PCB packages.

Current transforms are:-
Random Noise
Sine Wave
Moving Average
Fade In/Out (to make tracks come out of the pads square).

The amplitude/settings of each effect can be manually adjusted, and they can be applied in any order.

There's still a lot of bugs and I only support a small subset of the full gerber spec so far. I really like the effect on the Silk Screen layer and makes it look 'hand-drawn'.

I have been doing this as a bit of fun really (my alternative to doing crosswords or Sudoku's), but I wanted to post some results to see what people think.

There is loads more I could do, such as changing the size/shape of the pads etc, or allowing different effects on different tracks. The intention is not to improve the circuit performance at all and I expect in most cases it to have the opposite effect. It is also likely to cause havoc with any design rules for PCB manufacture.

[Dago]

That is quite neat!

Might not use it for work projects but maybe something else...

[hagster]

Here's a screenshot with a bit more of an over the top effect added.

Moving Ave-> Sine -> Fade In/Out.

There's loads of shorts etc in this image. I would ideally like to make it so I can draw boxes round critical areas to reduce the amount of distortion in these.

Green areas are Filled Polygons and the yellow and purple spots just mark the start and end of tracks.

[free_electron]

just what we need. The internet is already full of crap layouts . we really don't need a tool to make em even crappier...

[hagster]

Thought it might not be everyone's cup of tea. I don't think there are any good technical reasons to do this(as it currently stands).

A good layout meets all the electrical, emc, mechanical, safety and mechanical requirements. After that you are free to do whatever you want (if it's for a hobby) be that a snazzy logo or Morse code hidden in the pin spacings.

[Mechatrommer]

just what we need. The internet is already full of crap layouts . we really don't need a tool to make em even crappier...

export the "standard non transformed" gerber into bitmap, open it in photoshop, you pick... eye candy? alien skin? ripple crayon or whatnot you name it. when we say people love to waste their time they hate us. when we said you are wasting your time they said no i'm experimenting, yeah they have more time than i have just to know that... sooner or later. go to gaming forum you'll get more applause, this is not helping.

[AlfBaz]

Don't be deterred by some of these boorish statements. Experimentation can lead you to places you never thought of. Even though they usually lead to dead ends they can still arm you with knowledge that may be useful elsewhere. Who's to say that in further studies you may be able to use what you have learnt to calculate coupled regions for example.

Even Andrew Wiles was aware of the stifled attitudes in the mathematical community when he spent six years secretly trying to solve Fermat's last theorem

[hagster]

I'm open to all opinions for and against.

[ConKbot]

Have it so you can enter design rules, and maybe even open eagle design rule files, so you can have it go wild, and then go though perform a DRC validation, and smooth out/fix problem areas.  Would be useful as a DRC checker gerber viewer on its own, with the mutator/warper/obfuscator a bonus.

[hagster]

I fear that good DRC checking might be too difficult. The Gerber has no knowledge of net lists etc so it's hard to work out what can and can't connect. It gets even harder with power planes etc.

I think the best way forward would be to integrate with KiCAD. That way you can leverage all the built in tools and the whole think becomes seamless. They are starting to do python scripting support so it should be fairly easy. I'd need to look onto it more to see if it's feasable.

[ConKbot]

I fear that good DRC checking might be too difficult. The Gerber has no knowledge of net lists etc so it's hard to work out what can and can't connect. It gets even harder with power planes etc.

I think the best way forward would be to integrate with KiCAD. That way you can leverage all the built in tools and the whole think becomes seamless. They are starting to do python scripting support so it should be fairly easy. I'd need to look onto it more to see if it's feasable.

Not a full DRC I guess, more of a DFM check. Most PCB manufacturers have a DFM check when you upload gerbers, so it can be done, just not sure how hard it would be to code.  At least checking to see if youre maintaining traces/spaces would be a start.

[Mechatrommer]

Most PCB manufacturers have a DFM check when you upload gerbers, so it can be done, just not sure how hard it would be to code.  At least checking to see if youre maintaining traces/spaces would be a start.

pathetically speaking, from my 1 minute meditation.... assuming your input is a m x n dimension bitmap, traces colored black, spaces colored white. scan for each white pixel (radius centre), set radius (min tolerans), look for black pixel within radius, if found we have nonDFM pcb... an O(m.n.a) operation (with a = pi.r^2 = circle area in pixels) would be a start. looking from what the OP has achieved, he could and should do alot better than that.

[free_electron]

If you want to make something useful : write a program that can create a gerber polygon from a jpeg image. So we can import nice graphics on silkscreen or copper layers.
You should be able to set minimum feature size and

Bonus points if you let us import SVG and specify exactly how large the output should be.
Like take this picture, minimum trace or width is 6 mils and render it into a 25mm by 10mm area as a geber polygon.

[amyk]

Maybe you could use it subtly to embed a sort of watermark in the design?

But frankly, having my PCBs automatically look like a 5-year-old kid drew them is a fun idea

[hagster]

If you want to make something useful : write a program that can create a gerber polygon from a jpeg image. So we can import nice graphics on silkscreen or copper layers.
You should be able to set minimum feature size and

Bonus points if you let us import SVG and specify exactly how large the output should be.
Like take this picture, minimum trace or width is 6 mils and render it into a 25mm by 10mm area as a geber polygon.

Do you mean like this. http://dangerousprototypes.com/forum/viewtopic.php?f=56&t=4389

http://dangerousprototypes.com/forum/viewtopic.php?t=4452

The method they used seems to be tied to Eagle. I haven't found a more cross EDA software solution. Maybe via DXF would be the way to go.

[scientist]

just what we need. The internet is already full of crap layouts . we really don't need a tool to make em even crappier...

It's for artistic purposes, you insensitive clod!

[Mechatrommer]

It's for artistic purposes, you insensitive clod!

artistic purpose from raw gerber engineering data, why the hell i've not think about that? excellent!

[HackedFridgeMagnet]

Love the concept, but it does look a bit ugly, maybe sweeping multi-lane racetrack curves or crazy angles would look nicer.

Maybe not every where too.

[c4757p]

Ugly can be OK. There's something I really like about the utilitarian aesthetic apathy in these PCBs, though I doubt this tool is anything more than a curiosity.

[Mechatrommer]

making it "open source" is definitely a "good" thing to do, by doing this, they can invite "supporters" to improve the "algorithm" esp person like me

[hagster]

making it "open source" is definitely a "good" thing to do, by doing this, they can invite "supporters" to improve the "algorithm" esp person like me

I've never opensourced anything before. This is written in VB.NET using Visual Studio 2010. Any recommendations on where to host it?

[T3sl4co1l]

just what we need. The internet is already full of crap layouts . we really don't need a tool to make em even crappier...

It's for artistic purposes, you insensitive clod!

Now... be that as it may... consider this.

Real Artists do not use Comic Sans.

Real Typographers do not use Comic Sans.

(Typography is relevant, because a PCB is printed.  A good PCB reads like a good page layout: a good bit more abstract, but still as much functional as aesthetic.)

So why would you want to make your board look like Comic Sans?

Now... I'm not judging.  You are more than welcome to explore artistic interests, in fact I encourage it!  Just beware that: 1. what you think is a neat trick might not be so well interpreted by others, and 2. as you get better at things, you'll probably realize that art is inseparable from its interpretation, and adjust your expectations and goals accordingly.  Which is just another way of saying, as you make more art, you make better art.

I will add one important caveat though: that logo in your example raises a good point.  In general, companies don't take well to people screwing with their trademarks; this process might be considered defacing or "tarnishing", and probably wouldn't be looked on too kindly, say, if a careless employee applied the process before ordering boards, or a board house did the same.  You'll definitely want to have a rule to, say, exclude a region on a layer, or something like that, to preserve explicit features.

Tim

[Mechatrommer]

making it "open source" is definitely a "good" thing to do, by doing this, they can invite "supporters" to improve the "algorithm" esp person like me

I've never opensourced anything before. This is written in VB.NET using Visual Studio 2010. Any recommendations on where to host it?

i mean the TopoPCB software linked just before my post, it looks so artificially advanced. well, you can put your code here if you want to but dont expect a support from me since i'm a NET.blind guy