Complete rework of parts and further questions

[Manfred Schaefer]

Hello,

I have uploaded some official parts to the Parts Tracker. I had added references and set numbers to these parts. Some more will follow.
However, errors were discovered in some parts during the release. Or for other parts I changed BFC from CW to CCW and additionally these parts are certified to CC 2.0.
The point now is that I want to revise these parts so that they can leave the Parts Tracker again and I am at the beginning of this process. 
It will be a long way for me, but probably also a long thread, as I have many questions and many more will come.
If one or the other question seems too trivial or is clear to you, then I ask for your understanding. Sometimes it can also be a comprehension problem on my part, as English is not my native language.

Now to the first point. If a part is also licensed under CC 2.0, then this part must be rewritten from scratch. I will call this part description code. Willy wrote me that no line of code may be taken over.
This is very difficult or even impossible. If, for example, primitives were used or subfiles were used, then it is necessary to use them again.
Can a statement be made here about how big the difference must be, or how much code may be the same, preferably as a percentage, so that the new part counts as a new part with a new author?
The primitives and subfiles could be dispensed with by writing them as individual lines of code and then releasing the part. This makes it a new part with a new author, after which the primitives and subfiles are reinserted and released again. Would such a way be conceivable?
At the pattern, the values for the triangles and squares may vary slightly in the decimal places.

Further questions will follow.
Thank you.

Greetings

Manfred

[Willy Tschager]

This is very difficult or even impossible. If, for example, primitives were used or subfiles were used, then it is necessary to use them again.
Can a statement be made here about how big the difference must be, or how much code may be the same, preferably as a percentage, so that the new part counts as a new part with a new author?

For the pattern it's easy. Taking a pic into LDPC and redoing the pattern will almost certainly be different from the current pattern.

As for the prims/subs (especially the tori) just copy them from existing heads. All combinations have been used at nausea.

w.

[Manfred Schaefer]

For the pattern it's easy. Taking a pic into LDPC and redoing the pattern will almost certainly be different from the current pattern.

As for the prims/subs (especially the tori) just copy them from existing heads. All combinations have been used at nausea.

w.

Hello Willy,

thank you, I have understood.

Greetings

Manfred

[Manfred Schaefer]

Hello,

the next question. Five quadrilaterals are shown below.

   

Are all variants approved according to the specification?

Greetings

Manfred

[Orion Pobursky]

Are all variants approved according to the specification?

4 is a concave quad is is not allowed per the file format spec. 

From the spec:

Quads must not be concave (two or more other polygons must be used instead).

[Manfred Schaefer]

4 is a concave quad is is not allowed per the file format spec. 

From the spec:

Hello Orion,

Thank you.
I had only thought of lenses when I heard the term “concave” and didn't see any connection to quadrilaterals.
I found a good description on the Internet and now know when a quadrilateral is concave, namely when both diagonals lie within the quadrilateral.

Best regards

Manfred

[Philippe Hurbain]

I found a good description on the Internet and now know when a quadrilateral is concave, namely when both diagonals lie within the quadrilateral.

To make it clear...
For a convex quadrilateral, both diagonals lie within the quadrilateral. For a concave quadrilateral (NOT ALLOWED), one diagonal lies outside the quadrilateral.

[Manfred Schaefer]

To make it clear...
For a convex quadrilateral, both diagonals lie within the quadrilateral. For a concave quadrilateral (NOT ALLOWED), one diagonal lies outside the quadrilateral.

Hello Philippe,

thank you for your hint. I made an error in my answer, correct is convex and not concave.
I meant convex.

Greetings

Manfred

[Manfred Schaefer]

Hello,

to better understand the primitives, I took a look at Minfig Head 3626.
As I understand it, the primitives are a kind of container, which in turn contain further primitives or character instructions.
The primitives are available in three resolutions P is the standard resolution, 48 is the high or 3x resolution and 8 is the low or half resolution.
Normally the standard resolution is used, this resolution can be changed by the program for rendering, for example, if the primitive with the same name is available in a higher resolution.
I hope I have understood this correctly so far.
In the following picture I have compared a part of the minifigure head in standard resolution and high resolution. I used LDView for the images.
The high resolution was forced by the fact that I wrote 48\ in front of the primitives. I made a mistake with Stud2, I should have made this change on the lowest level.
I noticed that there are small gaps or overlaps here, because the torus t04o6250 is not available in high resolution. See the two ellipses.

   

Would it make sense to create them in high resolution?
I know with printed minifig heads there would have to be one or more variants, as there is no high resolution for the printed parts and the problem mentioned above would exist in the opposite direction.

Greetings

Manfred

[Willy Tschager]

Would it make sense to create them in high resolution?

No. There is something I call "Too much detail". You have to take into consideration that while making a pattern you're working on it isolated from the rest, where you get the feeling that more detail would render nicely.

How many time you're lookin' at a close-up of your head in a BI or a render of a set and how much would you gain compared to the bloat of a file 4 times bigger in size? The strenght of LDraw was also that the polycount was/is reasonable.

w.

[Manfred Schaefer]

No. There is something I call "Too much detail". You have to take into consideration that while making a pattern you're working on it isolated from the rest, where you get the feeling that more detail would render nicely.

How many time you're lookin' at a close-up of your head in a BI or a render of a set and how much would you gain compared to the bloat of a file 4 times bigger in size? The strenght of LDraw was also that the polycount was/is reasonable.

w.

Hello Willy,

I understand that there should be a good compromise in the level of detail and the number of polygons. The number of polygons also stands for the time needed for rendering.
Perhaps I expressed myself incorrectly in my question.
I don't want to replace the torus, but rather offer an additional torus with a higher resolution to avoid the above-mentioned gaps and overlaps when rendering in high resolution.
With the standard resolution, everything would be the same.

Regards

Manfred

[Travis Cobbs]

I don't want to replace the torus, but rather offer an additional torus with a higher resolution to avoid the above-mentioned gaps and overlaps when rendering in high resolution.

While LDView does have an option to use high-resolution primitives when available, I don't feel that the existence of this option should encourage people to produce 48 versions of primitives that aren't needed for any other reason. I think most people who want higher quality in LDView will instead increase the curve quality setting, which applies to all curved primitives supported by LDView.

[Magnus Forsberg]

While LDView does have an option to use high-resolution primitives when available, I don't feel that the existence of this option should encourage people to produce 48 versions of primitives that aren't needed for any other reason. I think most people who want higher quality in LDView will instead increase the curve quality setting, which applies to all curved primitives supported by LDView.

Manfred.

There is a difference here you need to understand. The primitives are not "replaced" by another high resolution primitive.
What you see in LDView is created by the "math magic" performed by the program. The "high resolution" is adjustable, and not dependant of a hi-res primitive.

[Manfred Schaefer]

Manfred.

There is a difference here you need to understand. The primitives are not "replaced" by another high resolution primitive.
What you see in LDView is created by the "math magic" performed by the program. The "high resolution" is adjustable, and not dependant of a hi-res primitive.

Okay, I understand.
I also noticed that LDView has more than 48 elements for a circular ring at a higher resolution.
The high resolutions (48) are necessary when large patterns are involved and the segmentation would be visible or disturbing with circles.
Greetings
Manfred

[Manfred Schaefer]

Hello,
I have now made my first attempts to create a digital pattern.
Most of the patterns used almost only triangles. However, I prefer quadrilateral.
Below I have shown a part of a pattern (Chinese character) and wanted to know whether my approach is allowed or permissible.
In reality, the sign is approx. 4 x 4 mm² in size.
I don't see any problem with the T-junction.
At what angle, other than 90°, is it no longer referred to as a T-junction?

   

Greetings

Manfred

[Magnus Forsberg]

I see no T-junctions in your image.  
T-junction is not an angle issue. It's a question about surfaces next to eachother.

Have you found the LDraw Wiki pages?
https://wiki.ldraw.org/wiki/T-Junction

[Manfred Schaefer]

I see no T-junctions in your image.  
T-junction is not an angle issue. It's a question about surfaces next to eachother.

Have you found the LDraw Wiki pages?
https://wiki.ldraw.org/wiki/T-Junction

Hello Magnus,
Thank you for your reply. I was mainly concerned with the use of quadrilaterals instead of triangles. As I have seen some patterns that have used triangles although quadrilaterals would have been possible.
I thought that using quadrilaterals was uncommon or considered bad style.
I was aware of the site you mentioned. I try to avoid T-junctions. In the picture with the triangle and quadrilateral, there could be a problem because point D is not part of the triangle and the shapes are not connected at that point. There could then be a gap here due to rounding errors.
Greetings

Manfred

[Orion Pobursky]

I thought that using quadrilaterals was uncommon or considered bad style.

Nope. In fact, when I do patterns I usually do all triangles and then run the file through DAT Header's optimizer to combine as many triangles into quads as possible.

[Manfred Schaefer]

Nope. In fact, when I do patterns I usually do all triangles and then run the file through DAT Header's optimizer to combine as many triangles into quads as possible.

Hello Orion,
thanks for the tip. I have looked at the documentation for the program.
Using the program to convert triangles into quadrilaterals is a good idea. I am currently doing it manually in my trials. But in future the conversation of the polygons into triangles should be done automatic. The subsequent conversion to quadrilaterals could then be realized with this program.

Greetings

Manfred

[Manfred Schaefer]

Hello,
here are my next questions. I have created a printed part and uploaded it to the Parts Tracker.
While working on the pattern, the following questions arose.
How many decimal places are useful for a flat pattern?
In one part on the Parts Tracker, two was mentioned as a comment. Which I would also consider sufficient. Is two a rule, or just a recommendation, or should the value be greater?
I used the prim “2-4ring6.dat” and was surprised that all the individual triangles are drawn. Wouldn't the following solution also be possible?

1 4 0 0 0 0 1 0 0 0 0 1 0 0 0 1 1-4ring6.dat
1 4 0 0 0 0 0 0 -1 0 1 0 1 0 1 0 0 1-4ring6.dat

Greetings

Manfred

[Orion Pobursky]

How many decimal places are useful for a flat pattern?
In one part on the Parts Tracker, two was mentioned as a comment. Which I would also consider sufficient. Is two a rule, or just a recommendation, or should the value be greater?

There's no hard rule. Typically 2 or 3 depending on size of pattern and how it looks. Anything more than 3 you should probably justify with a comment. 

Personally, I'd just go with 3.

[Gerald Lasser]

Two is usually fine. 

More decimaly only if the vertex need to snap to a prim, use the Unificator for that. The two decimals also fit most logos or curved patterns, even if you inline a circle, round it, not visible.

And if there are very fine structures, well, but LEGO does not print that accurate...

Remember:
1 LDU = 0,4 mm
0,1 LDU = 0,04 mm
0,01 LDU = 4 micrometers -> 6350 dpi!

[Manfred Schaefer]

Thank you for the answers to the question about decimal places.