Acceptable vertex mismatch error

[Paul Griffin]

For parts 2997 and 2998, I have noticed many edges that simply don't match up. In one case, two "identical" vertices overlap by .09 u; in another, there is about a .07 u gap. Without the ability to identify edges, I am unable to smooth normals, and I'm certain I will encounter other side effects.

Should I consider such a mismatch to be /too large/ and request/submit a part update, or should I instead implement some kind of vertex snapping tool? Or should I do both?

It would be useful information if there is some standard on identifying identical vertices, or, at the least, on the minimum acceptable feature size (minimum length of any polygon edge).

[Philippe Hurbain]

Clearly these parts are not precise enough. This is especially clear on the tire, we can easily see gaps (gaps are easier to spot visually than overlaps!).

Coordinates in LDraw files are generally rounded to 3 dp. Add to that some rounding in transformation matrix of subparts, and you can reach mismatches as high as 0.01. Higher mismatches are generally visible, or can be spotted with Edger2 (standard "vertices unification0" threshold of Edger2 is 0.001).

[Philippe Hurbain]

2997 BFCed, precision improved and sent to part admin...

[Paul Griffin]

Aww you beat me to it...but that's okay. I haven't edited any parts as of yet.

I don't have access to Edger2 (I use the eeevil Mac platform). But as you've started editing them, the following pictures may help you identify the overlap in 2998.




Just look for where it isn't smooth.

Maybe it would help if Edger2 implemented a feature that flagged any two points were >0.01 u and <(minimum part edge length) from each other. Though I 'spose I could write up my own parser that could identify such an issue if it would help.

[Philippe Hurbain]

I have not yet considered improving the rim (significantly more complex that the tire...).

I don't have access to Edger2 (I use the eeevil Mac platform).

The source code is available and it's a simple C++ console application, so I guess you could compile it on Mac?

Maybe it would help if Edger2 implemented a feature that flagged any two points were >0.01 u

It doesn't do that directly, but it can flag mismatched edges between facets. This either means a T-junction, or unmatched vertices.
I also have another tool (Unificator) that can automatically fuse vertices close enough.