GeForce RTX 50 Sequence Desktop and Laptop computer GPUs, unveiled at the moment on the CES commerce present, are poised to energy the subsequent period of generative and agentic AI content material creation — providing new instruments and capabilities for video, livestreaming, 3D and extra.
Constructed on the NVIDIA Blackwell structure, GeForce RTX 50 Sequence GPUs can run artistic generative AI fashions as much as 2x sooner in a smaller reminiscence footprint, in contrast with the earlier era. They characteristic ninth-generation NVIDIA encoders for superior video modifying and livestreaming, and include NVIDIA DLSS 4 and as much as 32GB of VRAM to sort out large 3D tasks.
These GPUs include varied software program updates, together with two new AI-powered NVIDIA Broadcast results, updates to RTX Video and RTX Remix, and NVIDIA NIM microservices — prepackaged and optimized fashions constructed to jumpstart AI content material creation workflows on RTX AI PCs.
Constructed for the Generative AI Period
Generative AI can create sensational outcomes for creators, however with fashions rising in each complexity and scale, generative AI will be tough to run even on the newest {hardware}.
The GeForce RTX 50 Sequence provides FP4 help to assist handle this concern. FP4 is a decrease quantization technique, just like file compression, that decreases mannequin sizes. In contrast with FP16 — the default technique that the majority fashions characteristic — FP4 makes use of lower than half of the reminiscence and 50 Sequence GPUs present over 2x efficiency in comparison with the earlier era. This may be completed with just about no loss in high quality with superior quantization strategies supplied by NVIDIA TensorRT Mannequin Optimizer.
For instance, Black Forest Labs’ FLUX.1 [dev] mannequin at FP16 requires over 23GB of VRAM, that means it might solely be supported by the GeForce RTX 4090 {and professional} GPUs. With FP4, FLUX.1 [dev] requires lower than 10GB, so it might run regionally on extra GeForce RTX GPUs.
With a GeForce RTX 4090 with FP16, the FLUX.1 [dev] mannequin can generate pictures in 15 seconds with 30 steps. With a GeForce RTX 5090 with FP4, pictures will be generated in simply over 5 seconds.
A brand new NVIDIA AI Blueprint for 3D-guided generative AI based mostly on FLUX.1 [dev], which will likely be supplied as an NVIDIA NIM microservice, gives artists higher management over text-based picture era. With this blueprint, creators can use easy 3D objects — created by hand or generated with AI — and lay them out in a 3D renderer like Blender to information AI picture era.
A prepackaged workflow powered by the FLUX NIM microservice and ComfyUI can then generate high-quality pictures that match the 3D scene’s composition.
The NVIDIA Blueprint for 3D-guided generative AI is predicted to be accessible via GitHub utilizing a one-click installer in February.
Stability AI introduced that its Secure Level Conscious 3D, or SPAR3D, mannequin will likely be accessible this month on RTX AI PCs. Because of RTX acceleration, the brand new mannequin from Stability AI will assist remodel 3D design, delivering distinctive management over 3D content material creation by enabling real-time modifying and the power to generate an object in lower than a second from a single picture.
Skilled-Grade Video for All
GeForce RTX 50 Sequence GPUs ship a generational leap in NVIDIA encoders and decoders with help for the 4:2:2 pro-grade shade format, multiview-HEVC (MV-HEVC) for 3D and digital actuality (VR) video, and the brand new AV1 Extremely Excessive High quality mode.
Most client cameras are confined to 4:2:0 shade compression, which reduces the quantity of shade info. 4:2:0 is often enough for video playback on browsers, however it might’t present the colour depth wanted for superior video editors to paint grade movies. The 4:2:2 format offers double the colour info with only a 1.3x improve in RAW file measurement — providing a really perfect stability for video modifying workflows.
Decoding 4:2:2 video will be difficult as a result of elevated file sizes. GeForce RTX 50 Sequence GPUs embrace 4:2:2 {hardware} help that may decode as much as eight occasions the 4K 60 frames per second (fps) video sources per decoder, enabling clean multi-camera video modifying.
The GeForce RTX 5090 GPU is provided with three encoders and two decoders, the GeForce RTX 5080 GPU consists of two encoders and two decoders, the 5070 Ti GPUs has two encoders with a single decoder, and the GeForce RTX 5070 GPU features a single encoder and decoder. These multi-encoder and decoder setups, paired with sooner GPUs, allow the GeForce RTX 5090 to export video 60% sooner than the GeForce RTX 4090 and at 4x velocity in contrast with the GeForce RTX 3090.
GeForce RTX 50 Sequence GPUs additionally characteristic the ninth-generation NVIDIA video encoder, NVENC, that provides a 5% enchancment in video high quality on HEVC and AV1 encoding (BD-BR), in addition to a brand new AV1 Extremely High quality mode that achieves 5% extra compression on the similar high quality. Additionally they embrace the sixth-generation NVIDIA decoder, with 2x the decode velocity for H.264 video.
NVIDIA is collaborating with Adobe Premiere Professional, Blackmagic Design’s DaVinci Resolve, Capcut and Wondershare Filmora to combine these applied sciences, beginning in February.
3D video is beginning to catch on due to the expansion of VR, AR and combined actuality headsets. The brand new RTX 50 Sequence GPUs additionally include help for MV-HEVC codecs to unlock such codecs within the close to future.
Livestreaming Enhanced
Livestreaming is a juggling act, the place the streamer has to entertain the viewers, produce a present and play a online game — all on the similar time. High streamers can afford to rent producers and moderators to share the workload, however most need to handle these duties on their very own and sometimes in lengthy shifts — till now.
Streamlabs, a Logitech model and main supplier of broadcasting software program and instruments for content material creators, is collaborating with NVIDIA and Inworld AI to create the Streamlabs Clever Streaming Assistant.
Streamlabs Clever Streaming Assistant is an AI agent that may act as a sidekick, producer and technical help. The sidekick that may be part of streams as a 3D avatar to reply questions, touch upon gameplay or chats, or assist provoke conversations throughout quiet durations. It will possibly assist produce streams, switching to probably the most related scenes and taking part in audio and video cues throughout fascinating gameplay moments. It will possibly even function an IT assistant that helps configure streams and troubleshoot points.
Streamlabs Clever Streaming Assistant is powered by NVIDIA ACE applied sciences for creating digital people and Inworld AI, an AI framework for agentic AI experiences. The assistant will likely be accessible later this 12 months.
Hundreds of thousands have used the NVIDIA Broadcast app to show places of work and dorm rooms into residence studios utilizing AI-powered options that enhance audio and video high quality — with no need costly, specialised tools.
Two new AI-powered beta results are being added to the NVIDIA Broadcast app.
The primary, Studio Voice, enhances the sound of a consumer’s microphone to match that of a high-quality microphone. The opposite, Digital Key Mild, can relight a topic’s face to ship even protection as if it had been well-lit by two lights.
As a result of they harness demanding AI fashions, these beta options are advisable for video conferencing or non-gaming livestreams utilizing a GeForce RTX 5080 GPU or increased. NVIDIA is working to broaden these options to extra GeForce RTX GPUs in future updates.
The NVIDIA Broadcast improve additionally consists of an up to date consumer interface that enables customers to use extra results concurrently, in addition to enhancements to the background noise removing, digital background and eye contact results.
The up to date NVIDIA Broadcast app will likely be accessible in February.
Livestreamers may also profit from NVENC — 5% BD-BR video high quality enchancment for HEVC and AV1 — within the newest beta of Twitch’s Enhanced Broadcast characteristic in OBS, and the improved AV1 encoder for streaming in Discord or YouTube.
RTX Video — an AI characteristic that enhances video playback on in style web browsers like Google Chrome and Microsoft Edge, and regionally with Video Tremendous Decision and HDR — is getting an replace to lower GPU utilization by 30%, increasing the lineup of GeForce RTX GPUs that may run Video Tremendous Decision with increased high quality.
The RTX Video replace is slated for a future NVIDIA App launch.
Unprecedented 3D Render Efficiency
The GeForce RTX 5090 GPU gives 32GB of GPU reminiscence — the biggest of any GeForce RTX GPU ever, marking a 33% improve over the GeForce RTX 4090 GPU. This lets 3D artists construct bigger, richer worlds whereas utilizing a number of purposes concurrently. Plus, new RTX 50 Sequence fourth-generation RT Cores can run 3D purposes 40% sooner.
DLSS 4 debuts Multi Body Technology to spice up body charges through the use of AI to generate as much as three frames per rendered body. This allows animators to easily navigate a scene with 4x as many frames, or render 3D content material at 60 fps or extra.
D5 Render and Chaos Vantage, two in style professional-grade 3D apps for architects and designers, will add help for DLSS 4 in February.
3D artists have adopted generative AI to spice up productiveness in producing draft 3D meshes, HDRi maps and even animations to prototype a scene. At CES, Stability AI introduced SPAR3D, its new 3D mannequin that may generate 3D meshes from pictures in seconds with RTX acceleration.
NVIDIA RTX Remix — a modding platform that lets modders seize recreation property, robotically improve supplies with generative AI instruments and create beautiful RTX remasters with full ray tracing — helps DLSS 4, rising graphical constancy and body charges to maximise realism and immersion throughout gameplay.
RTX Remix quickly plans to help Neural Radiance Cache, a neural shader that makes use of AI to coach on dwell recreation knowledge and estimate per-pixel correct oblique lighting. RTX Remix creators may also anticipate entry to RTX Pores and skin of their mods, the primary ray-traced sub-surface scattering implementation in video games. With RTX Pores and skin, RTX Remix mods anticipate to characteristic characters with new ranges of realism, as gentle will mirror and propagate via their pores and skin, grounding them within the worlds they inhabit.
GeForce RTX 5090 and 5080 GPUs will likely be accessible for buy beginning Jan. 30 — adopted by GeForce RTX 5070 Ti and 5070 GPUs in February and RTX 50 Sequence laptops in March.
All programs geared up with GeForce RTX GPUs embrace the NVIDIA Studio platform optimizations, with over 130 GPU-accelerated content material creation apps, in addition to NVIDIA Studio Drivers, examined extensively and launched month-to-month to reinforce efficiency and maximize stability in in style artistic purposes.
Keep tuned for extra updates on the GeForce RTX 50 Sequence. Study extra about how the GeForce RTX 50 Sequence supercharges gaming, and take a look at all of NVIDIA’s bulletins at CES.
Each month brings new artistic app updates and optimizations powered by the NVIDIA Studio
Comply with NVIDIA Studio on Instagram, X and Fb. Entry tutorials on the Studio YouTube channel and get updates instantly in your inbox by subscribing to the Studio publication.
See discover relating to software program product info.