HLR algorithm gives empty shapes for GLTF model.

 Prajwal Shelar wrote:

Does it use some other algorithms ?

It is a visualization-only HLR feature basically drawing a silhouette of closed parts (Graphic3d_Aspects::ToDrawSilhouette()).

 Prajwal Shelar wrote

I tried building solid shape by using Poly_Triangulations obtained from the faces of shapes of gltf and from these Poly_Triangulations I build wires,faces,shells and finally solid shape by using (BRepBuilderAPI_MakeSolid)

From attached screenshot it looks like some individual triangles being put into dedicated TopoDS_Face. Maybe sharing result .brep shape could be useful.

The first problem is geometry continuity of Edges shared between Faces. HLR is expected to draw edges on hard connections, and in your case it is almost every edge as planar faces doesn't connect smoothly: HLRBRep_Algo produces the following result on your shape:

vs. HLRAlgo_PolyAlgo:

I think that HLRAlgo_PolyAlgo does some simpler check to detect hard edges here - by comparing surface normals at connected edges or something like that, so that we see redundant edges being dropped on a planar topping. So that to get rid of redundant edges at round corners you need to detect which triangles are smoothly connected (with some angle tolerance) and encode this information somehow into B-Rep.

However, in your case a simple angle smoothing check wouldn't give desired result. If we will consider 20 degrees as a smooth angle we will end up with almost no hard edges in your case (I'm drawing Face Boundaries here as their drawing is based on BRep_Tool::MaxContinuity() check that could be easily overridden in Draw with help of BRepLib::EncodeRegularity()):

pload MODELING VISUALIZATION
restore HLRInputShape.brep b
encoderegularity b 20
vinit View1
vdisplay -dispMode 1 b
vaspects b -faceBoundaryDraw 1 -mostContinuity c0
vfit

This is because this shape has smoothing fillets, so that almost all connections are expected to be smooth. Image from your expectations cannot be achieved like that:

To detect edge lines on your attached image you really need an algorithm that would detect features like fillets, holes, round corners and other things (depending on complexity of your real models), classify them and perform segmentation of your input STL file into domains.

Only after that processing (usually called "defeaturing") it would make sense to think about generating a HLR image out of that...