Kalakshar US

Valve may not have touched the Half-Life franchise in over a decade apart from releasing its VR-only game Alyx, but that isn’t stopping enthusiasts from giving the game a visual overhaul. NVIDIA has unveiled a community-led Half-Life 2 RTX: An RTX Remix Project that, as the name implies, will remaster the classic shooter for PCs with GeForce RTX graphics. The team isn’t just adding ray tracing, though — this is an attempt to modernize the overall look and feel of the game.

The ray-traced lights are the star attraction, of course, but the modders are also using an early version of RTX Remix to add extra model detail (through Valve’s Hammer editor) and rework materials with physical-based rendering properties. The result is what you’d expect. Where the original Half-Life 2 graphics look flat and otherwise dated, the RTX port is moodier and far more detailed. You might want to spend extra time inspecting Dr. Kleiner’s desk or the pet headcrab Lamarr. Not surprisingly, the refresh makes use of additional NVIDIA tech like DLSS 3 upscaling, Reflex anti-lag and RTX IO GPU-accelerated storage.

The project is only just getting started, and there’s no tentative release date. Right now, this is more of a marketing showcase than a practical release. It comes alongside news of DLSS 3.5, which uses AI to improve ray-traced light quality by generating pixels between sampled rays. Titles like Alan Wake 2 and Cyberpunk 2077: Phantom Liberty will support the feature on launch.

The unofficial port is notable all the same. Existing RTX conversions like those for Portal and Quake II are pretty, but limited by either the age of a game or its relative scale. Half-Life 2 set a new standard for modern first-person shooters between its tightly integrated story, expansive (and seamless) world and physics-driven gameplay. Now, it’s getting an upgrade that could keep it relevant on modern PCs.

This article originally appeared on Engadget at https://www.engadget.com/half-life-2-is-getting-an-unofficial-rtx-remaster-130006917.html?src=rss 

Last year, NVIDIA unveiled DLSS 3 with frame interpolation, which used its AI-driven rendering accelerator to add extra frames to games. Now at Gamescom it’s introducing DLSS 3.5, which adds Ray Reconstruction, a new feature that will use the company’s neural network to improve the quality of ray traced images. It’ll be available for all RTX GPUs—unlike DLSS 3’s frame interpolation, which only works with RTX 40-series cards.

NVIDIA says Ray Reconstruction will replace “hand-tuned denoisers with an NVIDIA supercomputer-trained AI network that generates higher-quality pixels in between sampled rays.” That’s similar to NVIDIA’s original pitch for DLSS — making low-res textures look better thanks to AI — and it could potentially lead to better ray tracing performance as well. In images shown to media, Ray Reconstruction appears to deliver sharper reflections and textures in supported titles. (See comparisons below.)

According to the company, Cyberpunk 2077 in Overdrive Mode (its most powerful ray tracing offering) hit 108 fps with DLSS 3.5 and Ray Reconstruction, while the same system reached 100fps with DLSS 3 alone, 63fps with DLSS 2 (which lacks Frame Generation) and 20fps without any DLSS help.

Just like previous DLSS releases, developers will have to manually implement support for Ray Reconstruction. Cyberpunk 2077 (and its expansion Phantom Liberty) will be the first DLSS 3.5 title in September, followed by Portal RTX and Alan Wake 2. NVIDIA will be showing off Ray Reconstruction at Gamescom this week, and hopefully we’ll get a look ourselves sometime soon.

This article originally appeared on Engadget at https://www.engadget.com/nvidias-dlss-35-makes-ray-traced-games-look-better-with-ai-130012143.html?src=rss 

Though it’s not quite ready to usher in the Doolittle future we’ve all been waiting for, modern AI translation methods are proving more than sufficient in accurately transforming humanity’s roughly 6,500 spoken and written communication systems between one another. The problem is that each of these models tends to only do one or two tasks really well — translate and convert text to speech, speech to text or between either of the two sets — so you end up having to smash a bunch of models on top of each other to create the generalized performance seen in the likes of Google Translate or Facebook’s myriad language services. 

That’s a computationally intensive process, so Meta developed a single model that can do it all. SeamlessM4T is “a foundational multilingual and multitask model that seamlessly translates and transcribes across speech and text,” Meta’s blog from Tuesday reads. It can translate between any of nearly 100 languages for speech-to-text and text-to-text functions, speech-to-speech and text-to-speech supports those same languages as inputs and outputs them in any of 36 others tongues, including English. 

In their blog post, Meta’s research team notes that SeamlessM4T “significantly improve[s] performance for the low and mid-resource languages we support,” while maintaining “strong performance on high-resource languages, such as English, Spanish, and German.” Meta built SeamlessM4T from its existing PyTorch-based multitask UnitY model architecture, which already natively performs the various modal translations as well as automatic speech recognition. It utilizes the BERT 2.0 system for audio encoding, breaking down inputs into their component tokens for analysis, and a HiFi-GAN unit vocoder to generate spoken responses. 

Meta has also curated a massive open-source speech-to-speech and speech-to-text parallel corpus, dubbed SeamlessAlign. The company mined “tens of billions of sentences” and “four million hours” of speech from publicly available repositories to “automatically align more than 443,000 hours of speech with texts, and create about 29,000 hours of speech-to-speech alignments,” per the blog. When tested for robustness, SeamlessM4T reportedly outperformed its (current state-of-the-art) predecessor against background noises and speaker style variations by 37 percent and 48 percent, respectively.

As with most all of its previous machine translation efforts — whether that’s Llama 2, Massively Multilingual Speech (MMS), Universal Speech Translator (UST), or the ambitious No Language Left Behind (NLLB) project — SeamlessM4T is being open-sourced. “we believe SeamlessM4T is an important breakthrough in the AI community’s quest toward creating universal multitask systems,” the team wrote. “Keeping with our approach to open science, we are excited to share our model publicly to allow researchers and developers to build on this technology.” If you’re interested in working with SeamlessM4T for yourself, head over to GitHub to download the model, training data and documentation.

This article originally appeared on Engadget at https://www.engadget.com/metas-new-multimodal-translator-uses-a-single-model-to-speak-100-languages-133040214.html?src=rss 

Now’s a good moment to get a smartwatch that can easily handle your end-of-summer hikes. Amazon is selling the Apple Watch Ultra with a green Alpine Loop at a new all-time low price of $700, or $100 off, after a checkout voucher. That’s the same price as a 45mm Series 8 in steel, making it the obvious choice if you want more rugged Apple wristwear.

The Apple Watch Ultra remains the company’s most powerful smartwatch, and it’s the clear pick if you’re an outdoor adventurer. The large, extra-bright screen makes it easy to read even in direct sunlight, and the added water resistance is helpful for recreational dives. The action button also comes in handy for marking hike waypoints or starting the next leg of a run. And it’s hard to ignore the extra battery life — this watch can last an entire weekend without a charge, depending on how you use it.

You’ll need an iPhone to even consider the Apple Watch Ultra, of course. Its size may also be off-putting if you have thin wrists or simply prefer sleeker timepieces. There’s also the question of timing — Apple might introduce a refreshed Ultra at an event that could be just weeks away. If you’re more interested in value than having the absolute latest model, though, this discount is hard to top.

Follow @EngadgetDeals on Twitter and subscribe to the Engadget Deals newsletter for the latest tech deals and buying advice.

This article originally appeared on Engadget at https://www.engadget.com/the-apple-watch-ultra-falls-to-a-new-low-of-700-133522809.html?src=rss