30485pb01 Windscreen 8 x 6 x 3 Curved Top Angled Canopy Pattern, how to finish?

[Daniel Goerner]

This is what I want:
.

The problem is, most of the printed area shown above uses primitives, and is also bent in a way that slicerpro can't handle (in a satifying way), or am I wrong? Any ideas how I could start (except rebuliding the part from scratch)?

Oh, and if we find a way, I would do this one also:

[Philippe Hurbain]

and is also bent in a way that slicerpro can't handle (in a satifying way)

Perfectly right... you can't properly do that in one step, since some surfaces are parallel to projection direction!. You have to do a partial pattern viewed and projected from top, and another viewed and projected from side.
In the attached file, the pink area will get the top projected pattern, yellow area will get left projected pattern, and green area will complete the half windscreen pattern subpart (or, since here the pattern is simple, put directly the pattern and its x-mirror in the patterned part). The only tricky part is to have both patterns meet properly (here it should be easy).

most of the printed area shown above uses primitives

Either inline everything in the former files, or check "expand subfiles in former (-sf)" option. In this case, don't forget to specify your LDraw folder in search path.

[Daniel Goerner]

Um, okay, that sounds not too difficult. I'll try it.

[Daniel Goerner]

Well, I learned something today. Primitives are beyond me. I don't get how they work, bend etc. I just tried to mirror your pattern to complete it. The triangles worked fine, but the primitives went nuts.

I tried to make a pattern for a tile round 2x2 and don't get how to put it on the part since the surface is a primitive. I tried every import method, and when I exported it, the pattern was inside the surface. Well, all that was before I found out, that the pattern I'm working on for the whole day already exists. 4150ps8. Although it could use improvement with a gradient.

I'm done for today.

[Philippe Hurbain]

I just tried to mirror your pattern to complete it. The triangles worked fine, but the primitives went nuts.

Could you describe roughly what you did and post the result?

I tried every import method, and when I exported it, the pattern was inside the surface.

For the next round tile you try...
- put the top primitive in a new file
- inline it with LDDP (Tools > Inline part, shortcut ctrl-i), save the result
- Import that in LPC (project on yz plane, top)
- Create the pattern on this template
- Export the result ((project on yz plane, top)
...done.

[Daniel Goerner]

Could you describe roughly what you did and post the result?

Well, mirroring for dummies: I just threw out the studs and subparts, copied everything else in your file, pasted it, and clicked Movement->Mirror->X-Axis. Which was wrong, of course, because eventually it didn't work. ;-)

[Daniel Goerner]

Okay, now I inlined every primitive, deleted all the headers (very confusing) and mirrored the rest. Now it worked. So, the next pattern will be done. ;-)

[Philippe Hurbain]

Yeah, symmetry in LDDP doesn't work great for primitives. I suggest to use MLCad for that.
First set your rotation point on world origin (settings > rotation point). Select everything you want to create mirror, copy/paste, then Edit > Move > Keyboard entry (or keboard icon in toolbar). There check "Use rotation matrix value". Transformation matrix is displayed as "1 0 0 0 1 0 0 0 1". Change the 1st "1" to -1 for a x mirror, 2nd for a y mirror, 3rd for a z mirror.
Note that if a single primitive is selected, you may see the absolute transformation matrix of the primitive instead of identity matrix "1 0 0 0 1 0 0 0 1". In this case just select another element that will be mirrored too... but it doesn't matter much if you select a comment line

[Daniel Goerner]

I'll keep that in mind for the future.

One technical question: No that I'm using triangles insted of primitives, does it line up properly with the subpart, when I'm done?

And one other question: The pattern on the actual part seems to be white. But it's not completely opaque, so it seems more like a light gray. Which color should I use, since white seems to bright.

Attached is the first attempt. A watermelon with cockpit. ;-)

[Philippe Hurbain]

No that I'm using triangles insted of primitives, does it line up properly with the subpart, when I'm done?

Yes, it should!
Always a delicate problem to match partially transparent colors! Maybe Very_Light_Bluish_Gray (151)?

Attached is the first attempt. A watermelon with cockpit. ;-)

Looks great! - despite watermelon is one of the very few things I really dislike ;D

[Daniel Goerner]

Looks great!

Thank you, but no, it doesn't. I got the proportions completely wrong. I'm working on the second attempt right now.

And someone should establish semi-opaque colors for ldraw. ;-)

[Daniel Goerner]

I decided to make the pattern light gray, since it's darker than the space shuttle part, which is definitely white.

BTW: can i post comments in the review section without another upload of the part?

[Tim Gould]

Yes. Just novote.

Tim

[Daniel Goerner]

Yes. Just novote.

Tim

Oh, I didn't know I was allowed to do that.

[Daniel Goerner]

Another question: When I inline the primitives, the headers of those primitives are put inside my file. Do I have to keep these? It's a lot of data.

[Philippe Hurbain]

No, you shouldn't keep anything but code!

[Daniel Goerner]

I ran a file with subparts through Edger2. Where the subparts meet are some unmatched edges. Is that okay or should I work on that?

[Philippe Hurbain]

There may be _some_ mismatches because of limited precision. You can try to increase a bit precision value in Edger2 to see if the unmatched lines disappear. Pattern may also have created some T-junctions... Attach the file so that we have a look... (tomorrow for me

[Daniel Goerner]

Okay, if I set the precision from 0,001 to 0,0011 the line dissapears. I wanted to merge the two subfiles but there is an unmatched edge between these two. I can't get rid of it (except with the precision thing).

BTW: Chris, would 30485ps0 be okay for this?

[Chris Dee]

Yes - please use 30485ps0.dat

[Philippe Hurbain]

Excellent! Just one thing, one triangle on top of up going middle bar is main color instead of gray.

I wanted to merge the two subfiles but there is an unmatched edge between these two. I can't get rid of it (except with the precision thing).

This is indeed a slight mismatch of coordinates (different projection -> different rounding!). Running Unificator with default parameters on the merged file cures the problem (attached).

[Daniel Goerner]

Wow. I have to say, you're a genius, Philo. Writing all these little gadget programs to make a partauthor's life easier. Thank you.

[Daniel Goerner]

Library Question: I have created 3 subfiles for this. Two only to be used with the Falcon canopy and one that I would use on the Mars Mission part, too. How do I name this 3rd part, since it's not compatible with the original 30485.

[Daniel Goerner]

Ready for final check before submitting.

[Philippe Hurbain]

Library Question: I have created 3 subfiles for this. Two only to be used with the Falcon canopy and one that I would use on the Mars Mission part, too. How do I name this 3rd part, since it's not compatible with the original 30485.

I wouldn't create the 3rd subpart, but use the primitive-based elements directly in main part. This can be discussed, as it introduces some T-junctions, but here it's imho a small price to pay.

Attached what I would do (that doesn't mean I am right .
- moved condlines to their matching subparts.
- corrected the wrongly colored triangle in s01.
- s02 also had the lower right part that was BFC inverted (lines 155..223).
- The edge lines between conical front and cylindrical back was forgotten (it was not present in my "template"...). Restored it. As it was not present when you recreated condlines with edger2, Edger2 did create condlines there that were not needed. Deleted these condlines.

Name is wrong too, M. F. doesn't mean anything to the non-initiated. Unfortunately I was not able to coin a better name, submit as-is and ask for suggestion.

[Daniel Goerner]

I wouldn't create the 3rd subpart, but use the primitive-based elements directly in main part. This can be discussed, as it introduces some T-junctions, but here it's imho a small price to pay.

Attached what I would do (that doesn't mean I am right .
- moved condlines to their matching subparts.
- corrected the wrongly colored triangle in s01.
- s02 also had the lower right part that was BFC inverted (lines 155..223).
- The edge lines between conical front and cylindrical back was forgotten (it was not present in my "template"...). Restored it. As it was not present when you recreated condlines with edger2, Edger2 did create condlines there that were not needed. Deleted these condlines.

Name is wrong too, M. F. doesn't mean anything to the non-initiated. Unfortunately I was not able to coin a better name, submit as-is and ask for suggestion.

I'm the noob around here, so I'll leave that to the experts. ;-)

[Steffen]

I don't want to interrupt this creative process here,
but the question came to my mind if this forum currently evolves as a second parts tracker.
I think that we should avoid that for the long run, because people will start
downloading unofficial parts no longer from the PT, but also from here,
and files improvement takes place here, not on the PT.
Thus, this kind of creates a parallel universe.
I think you could just do this parts improvement iteration on the normal parts tracker.
To me, the parts tracker simply is a cooperative workbench.
What do you think?

[Philippe Hurbain]

I initially feared that too, but it does have a documentary/educational side that is interesting. This discussion and intermediate files remain here, this can be helpful to other parts authors in the future.

[Daniel Goerner]

You both got a point there. But as Steffen said, this is a creative and also learning process as I'm trying to learn how to do it right. And the more I learn the lesser parts I will upload here ;-).

Back to topic:

- moved condlines to their matching subparts.
- The edge lines between conical front and cylindrical back was forgotten (it was not present in my "template"...). Restored it. As it was not present when you recreated condlines with edger2, Edger2 did create condlines there that were not needed. Deleted these condlines.

I didn't use Edger2 to create the conlines. The condlines were created when I inlined the Primitives.
I'm still trying to find out what conditional lines are needed for. How do I know which condlines are needed and which not? And what condline belongs to what? Where's the difference between conlines and edgelines? I tried to create edgelines around my slopepattern, but Edger2 only said 4 unmatched edges and did nothing else. This all is still very confusing.

[Magnus Forsberg]

Over at the Part Tracker I wrote this, here:

Ok, I'll try to explain the concept of the conditional line.
(Mainly to myself, but it might help someone else too)
Take two triangles, like this:
3 16 0 0 0 1 0 0 0.5 0 1
3 16 0 0 0 0.5 0.2 -1 1 0 0
Around them are four edgelines.
2 24 0 0 0 0.5 0 1
2 24 0.5 0 1 1 0 0
2 24 1 0 0 0.5 0.2 -1
2 24 0.5 0.2 -1 0 0 0

If there is an angle between these surfaces,
there must also be a conditional/optional line, along their common edge,
like this:
5 4 0 0 0 1 0 0 0.5 0 1 0.5 0.2 -1

not like this:
5 14 1 0 0 0 0 0 0.5 0 1 0.5 0.1 -1
The red line is the only correct one, since the controlpoints
are on the same plane as the farthest points of each of the two tringles.
This conditional line is only visible, if the controlpoints are
below the "horizon" created by the conditional line.
The yellow line has a control point above the end of one of the triangles.
It therefor "disappears" too early.

There is one exception to the rule, cond-lines in primitives always have tangential controlpoints, at their end surfaces.
Cond-lines in a cyli-primitive are made to meet a flat surface next to it.
However, if that surface is placed in an other angle, you must add an extra cond-line.

Edit:
I've attached a file to study. Turn on/off "All conditional lines" and "Show control points" in LDView.

[Daniel Goerner]

Thank you for the answer, but it's like I'm trying to read chinese.

[Tim Gould]

Conditional lines exist so that parts with curved surfaces, that would otherwise have no edge, can have an edge that shows only when the angle is right.

Think of a cylindercone made of 16 triangles with one point at the top, and the other two on a 16-agon at the bottom. At the side of each triangle you'd like an edge, but only if that edge will be at the side of the cylindercone. So how do you know if it's at the side?

Well... each edge is shared by two triangles. If the base of one of those triangles goes from left to right (eg. out the front), and the base of the other goes from right to left (eg. out the back) then the edge is at the side of the cylindercone. Otherwise it is not. SO we can use this to set up control points.

See the attached file. I've marked the cond-line so that it, and its control points are tan normally, but the conditional line shows up black when it's activated. Play around with rotating this and see what happens to the triangles and lines when the black line appears and disappears.

Hope that helps.

[Steffen]

you meant cone instead of cylinder, right?

[Tim Gould]

Yes. I blame the early morning

[Magnus Forsberg]

I've added some pictures to my post. I hope they help in understanding.

[Tim Gould]

Cool. Actually... a wiki page on conditional lines would be a very good thing.

[Daniel Goerner]

Play around with rotating this and see what happens to the triangles and lines when the black line appears and disappears.

Either my LDView is broken or absolutely nothing happens. Well, except a black line appearing and disappearing. I don't see controlpoints when I turn them on. I even deleted the condline and I don't see any difference. What should I see?

If there is an angle between these surfaces,
there must also be a conditional/optional line, along their common edge,

Why? what happens if it's not there?

I'm sorry, but I still don't get it. I feel like a painter who's trying to wrap his head around a crash course in architecture.

[Tim Gould]

Does that help?

[Philippe Hurbain]

Not sure it will help, but I'll try to tell it differently...
The initial purpose of conditional is to make sure to have an outline edge around the shape of the part (think "Ligne claire" cartoons characters), so that it can be seen clearly on the background. The edges of a cube are always visible so they are treated as permanent lines. But what to do for rounded shapes? if we put permanent lines, the facets forming the shape will be highlighted - something we don't want. What we need is a mechanism where the line is visible only when it appears on the border of the shape. The magics comes from control points. When projected on screen with current viewing direction, if the two control points are on each side of the conditional line, this mean that the conditional line is in the middle of two visible surfaces, so it is not on the border of the shape and must not be visible. If on the other hand the two control points are projected on the same side of the lines, this means that the conditional line is on the border between shape and background and must be visible. With this verbose explanation and Tim's image, I hope you'll get the hang of it!

[Daniel Goerner]

It dawns.

But I still don't get the control points visible in LDView. Activating "Show Conditional Control Points" does nothing. Should they look like in Tim's image?

[Tim Gould]

I added it all artificially. They should flash on when you rotate the model. I've highlighted an example on 4-4cyli.dat

You can put 'show all' but I don't find it so helpful since it's hard to tell what is a real line, a control point, or an optional line.



Tim

[Daniel Goerner]

The unmatched edges around my part, should I add edgelines there? And if so, how do I do that? Edger2 seems only be able to create umatched edges in red...

Wow, and then I used Edger2 to create conditional lines. It did that. About 150. And I only need about 20. How do I find out which condlines are needed and which not, without going through every line of code? Because, I think, when you advanced authors look at the code you can see lines, triangles, etc, I only see a bunch of numbers and have to put a 0 in front of every line to see, which line disappears. Really, condlines are starting to p*** me off.

[Tim Gould]

In the case of this part I think you'd be best copying the edge lines from the unprinted part and sticking with just those. You'll too easily end up with too many using the print since it has many more quads.

Tim