2552p06

[Damien Roux]

Hi there, it has been a long time...

I use to author a few parts some years ago, I quit because of lack of time but I found back motivation and decided to complete a huge patterned part I was working on.




Before submitting it to the part tracker I'd like to ask for some help: DATHeader is giving me some collinear Verticles. How can I easily solve that?
Also, when I open some subfiles with LDForge it gives me the following: some kind of conditionnal lines that I don't know why they are there. I guess there are some gaps due to projection with SlicerPro. Any idea on how to correct that?





I attached one of the subfiles that have this problem. Thanks for your help!

[Philippe Hurbain]

Hi Damien,
Glad to see you back! I'll have a look - but it's not simple

[Philippe Hurbain]

Here is the method I used to fix your subfile. I guess that most problems were introduced by Slicerpro, in the two areas where the pattern was split (needlessly) because of the structure of the initial surface (one big quad in the middle, two small ones on sides). I guess that the unwanted condlines were introduced by Datheader: because of rounding errors, some quads were warped, and Datheader split them back into triangles with a condline in the middle.
- the first step was to come back to a completely flat surface, using MLCad. Since the surface mainly projects in x direction, it will be flattened this way (file 2552p06bflat.dat). To do this, in MLCad I selected everything, then in the "enter position rotation" tool, enter the transformation matrix with all x coefficients set to 0.

Code:

0 0 0 0 1 0 0 0 1

- next remove all unnecessary conditional lines: select one -> select all same type -> delete. Save file.
- Then have a look at the pattern, using LPC (import .dat in x direction - I probably used the wrong side since in the end the resulting file was BFC backwards, easy to fix with LDDP). Overlap analysis (Edit -> detect triangle overlaps) shows that there is a problem in the left area. I deleted all triangles there and retriangulated the area. Another change I made was to simplify the mesh where it was cut by slicerpro. Select nearby vertices and merge them (Merge to average, ctrl-W). Take care not to average on a too large area, as this could result in new overlaps!
- Export the result as .dat (file: 2552p06b-cor.dat). A check shows that all colinear vertices are now gone.
- We need to project again the surface. We do that with a slant transformation. In the initial file, we see that for a 144 ldu height we need to tilt in y direction by 40 ldu. This gives the slant ratio: 40/144=0.2777778. The transformation matrix is thus:

Code:

1 0.2777778 0 0 1 0 0 0 1

And finally we translate the result to its original position, x=290. After running through Datheader to round values and merge triangles into quad (since this is a big file, instruct Datheader to use Rectifier to do that, it's much faster!), we get final result: 2552p06b-cor-final.dat

[Damien Roux]

Hi Philo, thanks for the help. You description was quite detailled. I should make it out

I'll try to correct all my subparts like that, and then we will meet on the tracker I guess!

[Damien Roux]

It's on the tracker : http://www.ldraw.org/cgi-bin/ptdetail.cg...552p06.dat.

[Philippe Hurbain]

I checked the part using Edger2, it shows a few problems in the ramp area patterns (attached image -testing new image attachment!- and LDraw file). I haven't thouroughly looked at all problems, but in the middle of ramp it's a triangle overlap.
Otherwise it looks very good!
   

[Damien Roux]

I've made the corrections.