writeXKTModelToArrayBuffer throws invalid size error

[intrida-dev]

writeXKTModelToArrayBuffer throws invalid size error

[xeolabs]

We have a solution for this, I'll describe below.

This happens when NodeJS cannot allocate all the necessary memory it needs to convert large files. NodeJS always has a hard 1GB limit on how much memory it can allocate for text strings in memory. When we try to process large glTF and JSON files, users often encounter the error you've shown.

In xeokit-sdk 2.4.0 we can now avoid this error by using the model splitting feature described in this tutorial.

For splitting, we extended ifc2gltfcxconverter and convert2xkt, to automatically split models into parts, and we also extended xeokit-sdk and xeokit-bim-viewer to seamlessly load split models. Note that ifc2gltfcxconverter is our closed source tool provided by Creoox with a commercial license.

As a result, we are now able to convert and load models like this 2GB Baku stadium IFC, so it seems to solve this problem pretty nicely:

[intrida-dev]

Thanks for your reply. We will follow this when want to create split model out of IFC dataset.

Currently we have a JSON (not IFC) with graphical + metadata information. You will find XktCreator.js in XktCreator.zip attached earlier. In that file, createModel() is constructing XKTModel from input json file.

So we want to know how can we create multiple xkts here. Is it by creating multiple XKTModels? If yes, then what should be the condition to split the model into different xkts?

[xeolabs]

Unfortunately we can't help with the development of user's application code outside of a commercial support agreement, but hopefully this helps:

• Splitting has been only tested and documented by us for IFC files processed via ifc2gltfcxconverter
• You could however still use the classes in xeokit-convert to build multiple XKTModels and serialize each of those to a separate XKT file, and generate a manifest file that lists those XKT files, so they can be loaded as a set into a xeokit viewer. We don't demonstrate that anywhere, but the classes are certainly usable like that.
• Splitting criteria - you could use the same criteria as ifc2gltfcxconverter, which is basically to output objects into a glTF until the glTF hits some specified size, eg 10MB, then save that glTF, start the next glTF, when that's got 10MB objects, save that, and so on.

[intrida-dev]

Thanks for you reply. This will help us in splitting large models.