MIME-Version: 1.0
In-Reply-To: <51BBA5E7.4040006@prochac.sk>
References: <CAK=z9GWkMwS3NiLEgpvWuPgBVrGB+sHqb7DLCgYZTC3C1OyLSg AT mail DOT gmail DOT com>
	<51B6CDB6 DOT 1010909 AT estechnical DOT co DOT uk>
	<51B6DB0E DOT 8000108 AT prochac DOT sk>
	<CAK=z9GUWzhidWQsvxhWqYpSnyvUnmsut1+gatrMVvJqMM6L_UA AT mail DOT gmail DOT com>
	<kpfolc$gma$1 AT ger DOT gmane DOT org>
	<51BBA5E7 DOT 4040006 AT prochac DOT sk>
Date: Sat, 15 Jun 2013 02:03:26 -0400
Message-ID: <CAM2RGhQD-xY9dB37+Z-pkDwRQBrqv1aOew9a2W-J1n2g00Gdew@mail.gmail.com>
Subject: Re: [geda-user] 3D modelling and gEDA
From: Evan Foss <evanfoss AT gmail DOT com>
To: geda-user AT delorie DOT com
Content-Type: text/plain; charset=ISO-8859-1
Reply-To: geda-user AT delorie DOT com
Errors-To: nobody AT delorie DOT com
Precedence: bulk

I would like to point out that many if not most 3D CAD formats are BOT
(Bags Of Triangles). Yes it stinks and the mechanical people hate it
just as much as we do. STL is in use basically everywhere and it has
the same problem.

IGES/STEP would have been better because they uses cubic splines. This
same functionality is available in Eagle right now using google
Sketchup. I know that is based on COLLADA which I think is also not
based on BOT.

Going from 3D part model to 2D footprint is not a good plan for two
reasons. First it ignores what ever added wisdom the manufacture put
into their recommended footprint. Second it ignores details like the
way drill hole sizes shrink based on plating thickness. This is an off
topic comment but I wish I could have footprint hole sizes based on
desired open size diameter so that I could have PCB adjust the drill
size based on the copper thickness I will be using.

Ether way Matthew please do not take the above comments from me as a
criticism. The code clearly adds some nice functionality. Nice work.


On Fri, Jun 14, 2013 at 7:23 PM, Milan Prochac <milan AT prochac DOT sk> wrote:
> On 14. 6. 2013 20:50, Kai-Martin Knaak wrote:
>>
>> I am not a particular fan of VRML.
>
>
> Me too. VRML is really horrible format by design. It tries to be friendly to
> hardware renderers (I guess) and it lacks lot of features.
>
> My plan was to visualize the PCB design. The VRML was chosen due its
> simplicity and possibility to examine nearly-realistic 3D model in real
> time. Other options - like HQ  renderers (povray) or 3D modeller software -
> were not so straightforward, as they required offline rendering for each
> view or required additional software (and some extra knowledge to get some
> results from it).
>
> Additionally, the simplicity of VRML made it good candidate for my first 3D
> export filter, as it was quite easy implement it.
>
>
>>   More often than not, when I import VRML
>> models into 3D CAD applications, the result was less than perfect (Autocad
>> Inventor, Varicad, solid works). Round surfaces are pieced together by
>> literally thousends of triangles. The object is represented by its
>> combined outer surface rather than by a set of basic object that combine
>> with boolean operations. My ad hoc explanation is, that this format is
>> specifically geared toward "virtual reality".
>
>
> The problem is that most of VRML exporters produce VRML models created from
> triangles. VRML format defines basic primitives like cube, cylinder, sphere
> - but they are not used in that case. As the complete CSG is missing in
> VRML, there is no other way to export composite object from CSG capable 3D
> modeller.
>
> VRML should be considered as final format (like PDF) meant for viewing or
> printing. It is not suitable for further processing, importing or
> modifications.
>
>
>> I'd be more than happy to see a lossless path from pcb to a open sourced
>> 3D CAD application and back. This would raise the level of awesomeness of
>> geda by an order of magnitude!
>
>
> Generally,  the PCB design and generic 3D modelling are very different
> worlds. The path PCB -> 3D modeller is not so hard, but the opposite way is
> almost impossible, I guess.
>
> Well, the VRML export filter is real piece of working code (I will offer
> patch on  Launchpad next days). It is capable to export the board design,
> populate it with components, it has some fancy features - it can create the
> the models on fly, adjust description or color, change position, combine IC
> with socket or transistor with cooler.
>
> It can be used as inspiration or base for another 3D export filters. I'd be
> happy to assist.
>
> The real pain is component library. I am not very satisfied with my current
> solution, as it is format-specific (each format requires its own models or
> generators). I am playing with the idea to define models in some high level
> (but simple) meta-language, which would be easily translated to target
> format. And as next step ... the whole board could be exported by generic
> "3D export filter" in that meta-language and converted into format of your
> choice afterwards...
>
> But it is all one-way road.
>
> Milan
>
>> ---<)kaimartin(>---
>
>


-- 
Home
http://evanfoss.googlepages.com/
Work
http://forge.abcd.harvard.edu/gf/project/epl_engineering/wiki/