Nvidia DLSS 4 vs AMD FSR 4: Which is better?

Nvidia's DLSS upscaling technology is widely regarded as the best in gaming. The introduction of DLSS 4, featuring the Transformer model, has further solidified the company's dominance in the industry. While AMD has a similar technology called FSR, it has been far from a competitive equal. However, Team Red's new FSR 4, powered by AI, aims to change that.

While Nvidia's DLSS is still the superior option, the gap is closer than ever, with AMD's FSR 4 having a few advantages over Team Green's technology. In this article, we analyze both to determine to what degree one is better than the other.

Note: Some aspects of this article are subjective and reflect the writer's opinions.

---

Nvidia DLSS 4 vs AMD FSR 4: Detailed comparison

FSR 4 vs FSR 3.1 and Native 4K (Image via AMD)

Image quality of upscaling

Nvidia DLSS uses AI to deliver exceptional image quality, with each new version further improving the visuals. As the technology relies on AI, it is only compatible with newer Nvidia RTX graphics cards.

There are two AI models for DLSS 4: CNN and Transformer. The CNN model is ideal for those who prioritize performance and decent image quality. Meanwhile, the Transformer model provides significantly better visuals but with a slightly smaller performance boost.

Traditionally, AMD relied on compute-based algorithms for its FSR technology up to version FSR 3. This allowed a wide range of AMD Radeon GPUs, including integrated ones, to use FSR for enhanced performance.

Even Nvidia graphics cards, including the older GTX cards, could use FSR to achieve higher frame rates in gaming.

However, AMD's FSR faced criticism for its subpar image quality, which fell short of DLSS' older CNN model and was even further behind the new Transformer.

Fans had long urged AMD to adopt AI, and the company finally responded. With the introduction of FSR 4, AMD revamped its algorithms to incorporate AI. This shift has improved the upscaling technology, resolving many previous issues.

FSR 4 now surpasses the DLSS CNN model in image quality, but it remains slightly behind the Transformer model in still scenes. However, in motion, FSR 4 is not only better than DLSS CNN but also DLSS Transformer in many instances.

The downside is that FSR 4 upscaling is only compatible with the newer Radeon 9000 series GPUs. In contrast, DLSS 4 upscaling is compatible with all Nvidia RTX GPUs, including older RTX 2000 series cards.

---

Image quality of Frame Generation and Multi-Frame Generation

Upscaling technologies like DLSS increase performance (Image via Nvidia)

Frame Generation (FG) technology works by inserting a previously rendered frame between two real frames to theoretically double a game's FPS. Nvidia uses AI to minimize artifacts, ghosting, and other visual issues to some degree.

AMD's FG technology, on the other hand, uses compute cores for processing. As a result, it produces cleaner images with fewer ghosting and artifact issues compared to Nvidia's CNN model. However, Nvidia's Transformer model for FG still offers slightly better results than AMD's FSR 4 FG.

When it comes to Multi-Frame Generation, Nvidia remains the clear winner since there's no such technology from Team Red yet.

---

Nvidia DLSS 4 vs AMD FSR 4: Which one is better?

When it comes to upscaling, the DLSS 4 Transformer model retains its edge in pure image quality. However, FSR 4 can occasionally surpass it in terms of motion.

The story is the same in Frame Generation. AMD's FSR 4 is better than DLSS FG with the CNN model, producing fewer ghosting and artifacts. However, DLSS FG with the Transformer model beats FSR 4 FG.

While Nvidia still holds the lead, the company no longer has a significant upper hand with the DLSS technology. It will be intriguing to see how both technologies evolve with future updates.

---