Chaos’ Post

That’s is actually a huge improvement. All scenes in AEC are huge, that speed up is a game changer ! Not only we use scatter but we use it a lot 😄 That is important for Vantage pipelines in AEC

Yes, you have a good way to transmit/process data by using advanced technologies. But we need metadata to show the power of data. Can your product, or any other AI products you know, interpret and answer the following Chinese-English multilingual questions of BI? "Who, in the Ontario province of Canada, has new US patents granted on the nearest Tuesday, when the USPTO releases the newly granted US patents on a weekly basis?" "Who, in the "江蘇‘’ province of China, has new US patents granted on the nearest Tuesday, when the USPTO releases the newly granted US patents on a weekly basis?" With our intellectual property (IP), we can answer. Do you or any of your contacts need our expertise and our IP? We're selling, NOT just talking. Thanks. Here is our pitch deck: "Experiment results showed that with our intellectual property (IP), a copyrighted multilingual metadata, we are doing what AI, like ChatGPT, can't do in data analytics. Without metadata, NO data can be found/retrieved, even by the most advanced technologies, like AI, supercomputers, etc. https://lnkd.in/g-aJFnXR Our IP can also make your information service UNIQUE in the world." P.S. We wrote this post manually, without using any AI tool.

⏩ Export .vrscenes up to 20x faster! ⏩ With V-Ray 6 for 3ds Max, update 2—speed up your workflow with up to 20x faster exports and far smaller file sizes on scenes containing lots of scattering plug-in data. V-Ray GPU's time to the first pixel has also been accelerated. 🔗 See what else is new: https://bit.ly/3Gnzkno

I have been using llama 3 8b (Q8 - 8.54gb) for almost two weeks now, previously I used phi 2 and 3 and mistral 7b, gemma models but needless to say llama 3 8b is the best sota model, thanks Meta for open sourcing it, ui that I am currently using for text generation is lmstudio, I have also tried oobabooga and ollama, on my rtx 3060 12gb card (offloading 32 layers completely on gpu) first token takes around 1 sec, rest of the output rate is about 15 tok's per sec. Let me know what other models should i try. Seems like llama 3 70b needs around 24gb vram (3090/3090ti/4090) and 64 gb ram.

🌟 New Research Alert! 🌟 Excited to share a new blog post on real-time instance segmentation with disk covering. The post discusses a method that achieves state-of-the art results on challenging datasets while maintaining an impressive real-time inference time of 0.040 s on a single RTX 3090 GPU. Check it out for insights into achieving both speed and accuracy in instance segmentation. Read more here: https://bit.ly/48E8Eus

Explore the current state of the NVIDIA RTX Branch of #UnrealEngine and discover the latest optimizations, features, and enhancements for developers. Learn more in our upcoming webinar: nvda.ws/3RNJ4g6

Get up to speed on the current state of #RTX ray tracing in Unreal Engine and learn what’s on the horizon for the NVIDIA RTX Branch of #UnrealEngine. Learn more in our upcoming webinar. #GameDev #UnrealEngine5

Learn how you can optimize the lighting in your scenes with the latest advancements in #RTX Direct Illumination (RTXDI). Join us tomorrow and see how to leverage the NVIDIA RTX branch of Unreal Engine. #GameDev #UnrealEngine5

Setting up a next-generation LED production stage? Unreal Engine 5.3 adds Experimental Support to nDisplay for SMPTE ST 2110, utilizing NVIDIA hardware and Rivermax SDK. Laying the groundwork for a range of hardware configurations that open up new possibilities for LED stages - the most exciting configuration uses a dedicated machine for each camera frustum, maximizing the potential rendering resolution, increasing frame rate, and allowing for a more complex scene geometry and lighting than previously possible. This solution offers the ability to tackle challenges like wider angle lenses that require greater resolution and multi-camera shoots that stress the current system. It also implies lower latency in the system, due to simplification of the signal chain.

Continuing on with a few updates as I get time Added a scroll driven central column for the About page with self assembling hexagonal geometry. This uses instanced mesh for a single drawcall for the lot, though I am still animating per instance on the CPU (this could be moved to GPU and I may on next refactor) I've also added speed dependent chromatic aberration and lensing (most visible on the floating pillars top and bottom). You can play wth the hexagons on hover Minor changes are also a few shaders for the LED effect on the arrow, and smoothing. This runs fine on mobile, but not quite buttery smooth yet #threejs #webdevelopment