Ara 3D

I'm speaking next week at the NXT BLD / NXT DEV in London on May 14th: https://lnkd.in/eYAzznaT. I'll be giving two separate talks. The first one "Fast, Open, Queryable BIM Data on Your Own Terms" will be about setting performance expectations for what you should be able to achieve when querying data across thousands of projects using modern tools and approaches (hint: it's probably an order of magnitude faster than you expected). The second talk "Portable AEC Workflows with the Open Ara 3D SDK" will reveal why spending two years building a geometry kernel and 3D authoring tool from scratch was perfectly timed for the new era of AI driven coding. I'll be in London all week, and I hope to reconnect with as many of my friends and connections as possible.

AI Rendering almost feels like cheating! While there are often small discrepancies between from the original mesh, you can get some pretty amazing results with a single click of the button without having to worry about UVs, Materials, and lighting set-ups. This is a new beta feature being developed for the pro version of Ara 3D studio.

At Ara 3D we are working hard building BIMLakehouse.com. A system that can search across million of parameters across thousands of real-world BIM projects in msec, and provide access to any geometry in seconds. To achieve this goal, we needed to develop a lot of technology. First we needed to be able to convert IFC files and Revit files into an efficient normalized format for loading into, and extracting out of the data store. This technology is BIM Open Schema (https://lnkd.in/eyCUGG9w). An open binary format based on Parquet files. As you see in the picture, a sample file can be less than 10% of the original size, and loads 5x faster. Second we need to create a test-bed for the format and the libraries for querying, transforming, loading, and extracting the data. This is one of the primary purposes of #Ara3DStudio. The same libraries used in the desktop tool are used in the Lakehouse. It allows us to quickly validate and test our APIs and technology stack, and make sure that we can keep costs down and responsiveness up. It also plays a key role, in preventing vendor lock-in. Third, we needed to create an open-source WebGL viewer (https://lnkd.in/eYQU9qHB). This is a table stakes for any web-service, and we wanted to build a flexible and efficient viewer for and with the community. It helped prove to us that transmitting BIM data and geometry stored in Parquet files would be a good path forward for the web. --- A big thank you to Yskert Schindel from Vyssuals.com who has made many contributions to the web-viewer and been using it in his online services. Also a big thank you to Tomo Sugeta who has helped validate and build tools for the BIM Open Schema format.

I recently saw the claim that desktop BIM software is dead, but can desktop BIM software do things differently and stay relevant? Most of the software still in use today for designing and engineering the built environment was designed 30 years ago for the desktop. As we know, a lot has changed in this 30 years, in terms of hardware, connectivity, and software. We saw the rise of multi-core machines, GPUs, smart-phones, internet, SIMD instructions, and of-course AI. As data-sets have grown, as well as the complexity of the software, the performance and user-experience have either degraded or at the very least not kept up to user's expectations. People sometimes mistakenly blame desktop software or the vendors, but as software grows in complexity it becomes harder to adapt and change. I'm a bit reluctant to say that the software used is poorly engineered. It is all quite marvelous when we consider it in context, but collectively professional developers today know a lot more about how to engineer software that is less fragile and that performs better at scale. If we want to improve workflows and capabilities for AEC professional, putting everything on the cloud and accessing it via web-browsers only, solves some problems but creates new ones. When used from a desktop or laptop computer, the browser is a sandbox and abstraction layer that prevents software from accessing the true capabilities of the hardware. Some of the limitations imposed are: 1. Performance - You don't have as much RAM, you can't access the full CPU instruction set, you can't access the hard-drive, you can't control the threading, you don't have the full GPU instruction set. This all adds up. 2. Costs - you have to pay additional costs to store and access your data, and that other provider (whether it is AWS, Microsoft Azure, Google, or whatever) is charging you a steep mark-up. 3. Latency - synchronizing and accessing the data is slow, especially if you can only do it via a browser, and only when connected explicitly. Having the central source of truth for a project on the cloud makes perfect sense, but architects and engineers need to be able to work with and view as much of the project locally as possible when they want. So the issue is not desktop versus cloud. That is a false dichotomy. The solution is hybrid. Desktop software connected to the cloud. Using native capabilities of the operating system and hardware where it makes sense, and synchronizing and connecting to data in the cloud, without getting in the way of the user. Plus web-based applications for providing additional supporting workflows using the same central data. I'd love to hear from you. Do you agree or disagree there is potential for significant improvements to desktop BIM software for architects and engineers? Any other thoughts to contribute?