What role does vertices play in detail portrayal in an stl file?

Hi.
I have received a scan of a figure that is resin and 5.7cm tall , in rendered view (default settings Rhino V5) Scan1 shows defined pocket seams, stitching marks, press studs, buckles etc.
There was a flaw on part of it.
The file received (Scan2) less the flaw was a little softer on the detail definition.
It was then sent again (Scan 3), it looks exactly the same as Scan2.

Do the figures from Rhino show why ?

Scan1 was:-
70,687kb
Closed double precision
polygon mesh: 983,014 vertices,
1,444,850 polygons with normals
(145% more vertices than Scan2)
(117% more polys than Scan2)

Scan2 was :-
60,206kb
Closed double precision
polygon mesh: 676,249 vertices,
1,230,718 polygons with normal
(68% vertices of Scan1 )
(85% polys of Scan1)
(85% of Megabytes of Scan1 )

Scan3 :-
76,588kb
Closed double precision
polygon mesh: 848,511 vertices,
1,565,592 polygons with normals
(86% of the vertices that Scan1 has)
(108% more polys than Scan1)
(108% more Megabytes than Scan 1)
[ 125% more vertices than Scan2 ]
[ 127% more polys than Scan2 ]

I am told vertices are the lines between polys, so how become the Scan3 has 85% vertices and 108% polys compared to Scan 1 ? i.e. polys increase vertices decreased.

I think if the vertices equal or increase to those of Scan1 then it will be as good as Scan 1. Vertices hold the key. Am I right ?

How become, comparing Scan 3 to Scan 2 they look exactly the same, yet Scan3 is 125% more vertices than Scan2 and 127% more polys than Scan2 ?

I think the figure was re-scanned for Scan2 then this ‘large’ scan resampled, or whatever the term is, to create a file of workable megabyte size, scan3 is taking the ‘large’ Scan2 and re-sampling it again.

I know that an initial scan was 400Mb and hell to download let alone get into Rhino, and yet when I did manage to, it had a surface as if made of chards of glass ! Since then a 19Mb file also had the same effect. Its not the size, its what you do with it ! :grin:

Just what I wonder will a scan matching Scan1 have ?
Can one extract from a master scan an .stl file and dictate the number of vertices ?

Cheers

Steve

Just to keep the nomenclature straight:

Polygons are made up of straight lines called EDGES.

There are, to my knowledge, no 0, 1, or 2 edged polygons. Also no polygons with a negative number of edges. So the world of polygons consists of 3 or more edges joined at their ends.
A 3 edged polygon (triangle) by definition is planar. With more edges it can but need not be.

The POINTS where EDGES meet are called VERTICES. (Singular: VERTEX)

Vertices not only define the edge connections of a single polygon, but are also the places where adjoining polygons are attached via a vertex or two that are common to both polygons.

Here is where the math wizards at Rhino can take over if they desire and correct my mis-statements and amplify what I’ve said.

1 Like

Hi Steve - if I understand the question, which I may not:If scan 3 is a resampling upwards of a reduced scan 2, it will not be any more detailed than scan 2 - any original detail lost in down-sampling from the original for 2, is gone. Do get a mesh with better detail than mesh 2 you’d need to start with the original and down-sample it less.

-Pascal

Hi,
Understood on terminology. Thanks.

Pascal, I see what you are saying, if they redid the scan but to a lesser detail, then fiddling with that will never get back the detail that Scan 1 had.
I will check on that.
Maybe we need to take scan 1 and edit the frazzle from the boots using a free mesh editor. (any recommendations welcome on what to get).

Any further figure scans must be on settings used for Scan 1, but to make sure they dont cause any granular looking anomalies.

Steve

Given what vertices are, still not sure how increasing polys doesnt increase vertices but reduced them.

Steve

Hi Steve,

An STL file is like a digital photo. You can run filters on it but its not going to make it any better.

Usually some type of noise reduction is the only thing that might improve a result.
Upsampling (increasing # of triangles) is rarely useful as it dos not really increase the resolution, and typically will require smoothing, which may actually reduce the resolution.