Dynamic Lighting Was Better Nine Years Ago | A Warning About 9TH Gen's Neglect.

Summary

This video critiques the performance regressions in Unreal Engine 5.5, particularly focusing on lighting and graphical effects that strain modern consoles like the PS5 and Xbox Series X. The speaker highlights inefficiencies in Unreal's cloud shading, global illumination, and ambient occlusion techniques, arguing that lighter, optimized methods used in earlier games (e.g., *Quantum Break*, *The Division*) could improve performance without sacrificing visual fidelity. They advocate for a shift away from pushing higher resolutions or overly dynamic lighting in favor of more efficient, realistic systems that maximize hardware potential while maintaining quality in both static and dynamic environments.

Takeaways

• 😀 Unreal Engine 5.5 suffers significant performance regression, with cloud shading running 800% slower than in 5.4, and still relying on temporal anti-aliasing to hide jittering.
• 😀 Developers should focus on optimizing for ninth-generation consoles like PS5 and Xbox Series X, which offer the best balance of performance and image quality.
• 😀 The Series S should not dictate visual and performance standards due to its underperformance and the lack of meaningful hardware enhancements compared to mid-gen consoles.
• 😀 Resolution upgrades (e.g., 4K) should not be prioritized over performance and visual quality, as higher resolution requires a disproportionate increase in computational power.
• 😀 60fps should be the target for modern gaming, as it offers a clear improvement over previous-gen experiences without overstretching hardware capabilities.
• 😀 Developers should focus on improving anti-aliasing methods rather than aiming for higher resolutions, as this significantly improves image quality on current hardware.
• 😀 Light maps are unsuitable for modern games due to their baking process, memory consumption, and inability to work with dynamic elements, requiring alternative approaches like probe-based lighting.
• 😀 Traditional ambient occlusion (AO) methods like SSAO cause issues such as halo artifacts, and should be replaced by more advanced solutions like MXAO or new AO methods that better handle lighting awareness.
• 😀 New techniques like Radiance Cascades offer a potential solution for optimizing global illumination, allowing for realistic bounce light and shadowing with minimal computational cost.
• 😀 Despite modern advancements like Lumen, current techniques like dynamic GI and SSAO still suffer from issues like smearing, noise, and splotchiness, which significantly degrade visual quality in dynamic scenes.
• 😀 The gaming industry is overly focused on high-end hardware features at the expense of optimizing for existing hardware, leading to performance regressions and lower image quality for the majority of gamers.

Q & A

What is the primary issue discussed with Unreal Engine 5.5 in the script?

The primary issue is that Unreal Engine 5.5's cloud shading effect runs significantly slower compared to Unreal Engine 5.4, causing performance regressions. The author emphasizes that despite prior warnings to Epic Games, these issues were not addressed in the release.

How does the author describe the performance issues with Unreal Engine 5.5?

The author explains that cloud shading in Unreal Engine 5.5 is over 800% slower than in version 5.4. Additionally, it still relies on temporal smear techniques to hide jittering, which is not an acceptable long-term solution.

What is the target hardware discussed for the game development mentioned in the video?

The target hardware is the PS5 and Xbox Series X, as they provide well-established performance enhancements over their base-gen and mid-gen predecessors, making them ideal for modern game development.

Why does the author criticize the Series S in terms of image quality standards?

The author criticizes the Series S because it does not uphold the established hardware improvements found in the Series X and PS5. This results in poor image quality and excessive upscaling, and the author believes the Series S should not dictate the progress of visual standards for higher-end consoles.

What is the problem with the current approach to resolution and frame rate in modern games?

The author argues that focusing on performance or resolution modes leads to inefficient use of hardware. Instead, they suggest that 60 FPS should be prioritized as a clear improvement over last-gen consoles without exceeding hardware capabilities. They also emphasize the need to focus on improving image quality, particularly through better anti-aliasing, rather than simply raising resolution.

What issues does the author identify with lightmaps in game development?

The author points out several drawbacks to using lightmaps, including the time-consuming baking process, memory challenges in large open worlds, and the incompatibility with dynamic elements like destructible objects and moving light sources.

How does the author propose to improve global illumination (GI) techniques?

The author suggests several alternative GI methods that are more optimized and efficient than current approaches, such as using probe-based lighting systems, advanced ambient occlusion techniques, and methods from games like 'Quantum Break' and 'Metal Gear Solid V'. These approaches focus on reducing the computational cost while maintaining high visual quality.

What are the problems with screen-space ambient occlusion (SSAO) that the author mentions?

The author critiques SSAO for issues such as halo artifacts, lack of lighting awareness, and excessive noise. They highlight how some alternatives, like mxao, avoid these problems by using cheaper dithering techniques to improve performance.

What is the author’s stance on Lumen GI in Unreal Engine?

While Lumen GI in Unreal Engine is an improvement, the author highlights its performance costs (3.8 milliseconds per frame in a large city environment) and the persistent issues such as disclusion errors, splotchy lighting, and smearing, which make it less reliable in large-scale scenes compared to other methods.

How does the author describe the potential for future improvements in graphics optimization?

The author believes that there is a significant opportunity to explore more efficient GI techniques, such as Radiance Cascades and probe-based methods, that can lead to better performance and quality in modern games. They emphasize that industry focus on these techniques could lead to better optimization and a more sustainable approach to rendering.