NVIDIA held a series of private sessions to show the media some of the news and the most important advances recently achieved in the field of artificial intelligence, and how could it be otherwise, I had the opportunity to participate in these demonstrations. I had very high expectations and I can say that she did not disappoint me at all.

The first session focused on AI image generation, but in a unique way. In this demo, NVIDIA gave us the option to “become” a superherousing a series of clipart images, a camera that captured our face, ComfyUI ControlNet as an image creation platform, and a GeForce RTX 4090 as a hardware-level AI control engine.

The result was spectacular. AI could not only do it apply our face into these images and integrate them in a natural and realistic way, but he could also introduce numerous changes to adapt the design of the superhero and the background in a very simple way. The entire process was completed within seconds. According to NVIDIA GeForce RTX 4090 speed up to 10 times image generation times.

Below we could see one of the most important novelties that were introduced in RTX Remix, and this platform becomes open source. Starting this June, the following will be available:

• Open source Remix toolsetwhich expands the number of available tools dedicated to AI and modding.
• Remix Runtime Development Kit, which allows Remix Runtime to be integrated into applications and games that do not support DirectX 8 and DirectX 9.
• Remix REST API development and toolkitwhich expands the availability of tools focused on generative AI, DCC and mods.

AI powered remastering of classic games with RTX Remix much simpler, faster and more intuitive. In the demo, we could see a demo focused on creating and integrating textures, which was done in seconds with a simple command.

To conclude this first sample session we find “an old acquaintance”, ChatRTXa platform that we’ve had the chance to test at the time and will continue to improve with the launch new development kit. If you want to know more, don’t miss our ChatRTX analysis.

NVIDIA doesn’t forget about gaming either

The green giant bets on graphics cards SFF form factor compatible, which allows you to build a much more compact gaming PC with a very interesting design, without having to give up a good level of performance and without cooling problems. Right now, the top of the line that we can find compatible with this format is a GeForce RTX 4080 SUPER.

NVIDIA DLSS 3.5 technology was also present in this session and Alan Wake was used to demonstrate the benefits it offers. Performance in this game uses DLSS Super Resolution, Frame Generation and Ray Reconstruction from 29 FPS on average to 99 FPS on average with a GeForce RTX 4090. If you want to know more, I recommend checking out our technical analysis of Alan Wake 2.

NVDIA also showed improvements that AV1 encoding offers compared to H.264 running at 120 FPS in 4K resolution, all with a bitrate of 10 Mbps The difference is huge, as we can see in the attached images. Last but not least, the company showed us some of the news introduced in the new version of the NVIDIA application. The most interesting was without a doubt automatic overclocking functionwhich can increase the working speed of the GPU completely safely and without loss of warranty.

NVIDIA ACE, the technology it represents, could not be missing a huge leap in NPC interactivity and realism, and it promises to open a new stage in the world of video games. We already had a chance to dive deeper into it last April, and I can confirm that seeing it in motion is spectacular.

In conclusion, we have The G-Assist project, a project that is currently in its early stages but has all the makings of a breakthrough in the gaming world. This is an LLM trained to use data from a variety of sources that you can offers very useful information about games.

Want examples? Imagine getting lost in a game and needing a guide to progress, with Project G-Assist you would have direct access to the guide you so desperately need. You will also be able to bid technical information on optimal configuration games to push them to 60 FPS with our hardware, among other things.

AI in professional applications: qualitative and quantitative improvements

NVIDIA Studio has become a sure bet for professionals in the world of editing, design and content creation. AI has been key in this sensebecause NVIDIA knows how to play its cards very well and has implemented valuable functions and features that make a big difference.

Among the most important innovations that the green giant showed us is implementation DLSS Super resolution in D5 Render, a very popular rendering tool in architecture and interior design. This technology allows rendering at lower than native resolution and upscaling to the target resolution of AI applications and combining different images, all to achieve the perfect image.

The difference this technology made in the technical demo was so great that the project moved with absolute fluidity using DLSSand it worked in fits and starts once said technology was disabled. I don’t need to tell you what this means for a professional who has to work with heavy projects in real time.

Another interesting feature is simultaneous scene encoding, which was implemented in the well-known CapCut application, a platform also using generative AI, as many of our readers already know. This new feature allows you to encode two images at once, which cuts export time in half when editing video.

Topaz Photo AI is another of the platforms most enhanced by accelerated AI through NVIDIA GeForce RTX graphics cards as it integrates AI-enhanced scaling, noise reduction and object removal features. The technical demonstration that I had the opportunity to see clearly showed me that these functions were well integrated, as they are applied in a very short time and achieve very good results.