esri / i3s-lib Goto Github PK

It's hard for me to read the code.Is there a description for this library？

In function i3slib::i3s::convert_geomery(), I notice that src_pos return a relative position of mesh, which should be a std::vector<Vec3d>, and the src_pos.size() should return the size of vertices. But when I use a vertices with triangluar indices, the size() function of Mesh_attrb return an index size like this:

so, the next step code (in row 213)

will always touch an assert: ERROR: mesh->size() != rel_pos.size();
I want to know, did I miss something? or the indices should not calculate like this?

How to add additional information to a simple node data, it seems that it lacks interfaces about it.

There is no way how to provide bounding volume via the API since the the library generates bounding volumes (both OBB and MBS) for each node on its own.

Bounding volume of a non-empty node is computed only from node's mesh (vertices) and ignores bounding volumes/geometries of node's subtree. If a more detailed nodes are larger than the their parent nodes then they are not properly visualized (at least in AGOL) because the renderer does not descend to subtree of a node that is outside of viewport.

Also, the scene's extent in 3dSceneLayer.json does not cover whole scene.

When we process a library-generated SLPK file by the ArcGIS Pro Upgrade Scene Layer tool the upgraded SLPK file does not suffer with these problems. Node's bounding volume covers node and whole its subtree. Also, the scene's extent is correct.

So, current layer export via i3s-lib is broken and deviates from the functionality provided by the ArcGIS Pro.

We have generated a sample SLPK dataset via the i3s-lib library, where:

• Level 1 has a simple mesh of a cube (12 faces), with 10 meter half-axis.
• At each subsequent level the cube's half-axis grows by 10 meters.
• Bottom of the tree ends up at level 41 with cube with 410 meter half-axis.
• All cubes share common center.

Library generated dataset suffers from the above described problems: root node has OBB with halfSize = [10, 10, 10] and MBS radius = 17.32, i.e. covering only the top-level 20x20x20 m cube.

ArcGIS Pro re-generated (upgraded) dataset has correct node bounding volumes: root node has OBB with halfSize = [410, 410, 410] and MBS radius = 710.14, i.e. covering whole subtree including the bottom 820x820x820 m cube.

Demonstrate exact process for building on at least Ubuntu 18.04/20.04 using Github Actions as CI. Various pre-requisites include.

• New cmake required for draco
• GCC 8 or Clang being installed

Currently building this on Ubuntu 18.04 on WSL will make a CI PR when the steps are worked out.

I set the program to generate the slpk data of the ktx2 format map, but in the web side call display is 0.jpg, I extracted the 0_0_2.bin.ktx.gz data into the ktx2 map loader prompt format error，What's wrong with this, please

I am working on an example using this library to convert SFM derived urban models to SLPK format. I have the attached model tile to donate under creative commons (CC-BY-SA) licence to this library and use as a focus for development for this functionality.

I can supply the model as a PR as well once we discuss appropriate folder structure under which testdata should sit in the repository.

The obj2slpk function proposed will follow the raster2slpk pattern:

obj2slpk input.obj input_lod1.obj input_lod2.obj ... output.slpk

To avoid extra dependencies being added to this repository it is recommended that the LOD's for the given OBJ are generated separately using tools such as.

• VCGLib
• Blender
• Simplygon
• And any other implementations of the triangle-edge collapse and similar mesh simplification algorithms.
  Tile_+007_+008.zip

测试数据无法下载，能否提供一个网络环境比较好的地址，位置：**

Testdata sharing
I am working and testing your example application,here is a test data download url:
https://www.cc.gatech.edu/projects/large_models/ps.html
if you are chinese netizen,you can download here:
链接：https://pan.baidu.com/s/1B6E6T2GPso0xmmPxlzQ01A
提取码：xzbd

We would like to request support for shared resources (textures), as we have many geotypical (shared) textures. Not sharing textures leads to very large slpk.

We would like to request support for textures with alpha channel. Currently, these are not handled correctly.

From a conversation with Juan Ruiz/ESRI:

I did see some issues with rgba in i3slib::i3s::create_texture_set. The current implementation is encoding rgba as jpg, not png. I3slib::i3s::raw_to_jpg doesn’t support 4 channel textures. Support is “to do”. After updating to encode rgba as png, then there’s an issue with i3slib::i3s::raw_to_png. The implementation is also “to do.” This causes create_node to fail.

Hello, @rvargasESRI @whatnick @vsautin1

We tried changingthe default tile size 32 to 64, as follows:
std::vector<std::vector> grids;
grids.emplace_back(std::move(elevation_data));
build_downsampled_grids(size, grids.front().data(), 64, grids);
std::reverse(std::begin(grids), std::end(grids));

//
auto writer = create_writer(slpk_file_path);
if (!writer)
return 1;

ENU_to_WGS_transformation transformation({ -123.4583943, 47.6204856 });
i3slib::i3s::Node_id node_id = 0;

if(!process(*writer, size, cell_size, grids, textures, 64, 128, 0, 64, { 0, 0 }, 128, { 0, 0 }, node_id, &transformation))
return 1;

Then run the app and get the new slpk , and published it to be a scene service. And, it shows that the scene service display normally in the browser.

Why the exception? Did we miss any other parameter?

Looking forward to your response.

测试数据无法下载，能否提供一个网络环境比较好的地址，位置：**