Where computing power will go: Metaverse + decentralized rendering

After sorting out all the public chain projects that received financing in 2022, we further explored the direction of the segmented Metaverse public chain.

In the new public chain track, there are many public chains trying to find new uses for PoW computing power. This use is no longer used to maintain network security, but to generate proofs, execute specific business, etc.

One business direction that is gradually increasing in number is decentralized rendering based on Metaverse application scenarios. This type of public chain is positioned in vertical application scenarios related to the Metaverse, providing 2D/3D rendering services based on computing power and in line with Web3 logic.

In addition to the new public chains Caduceus and Portalverse Network that received financing last year, there is also the Metaverse public chain iPolloverse announced last year, as well as the early old project Render Network that first involved this concept. We have sorted out the above public chains:

Traditional rendering services rely on high-performance CPUs or GPUs and are known as one of the “mainstays of computing power.”

"Rendering" can be simply understood as "graphics + computing power + infrastructure". Traditional computing power rendering is one of the mature markets at the Web2 level. In the GPU rendering market, rendering refers to the process of "generating a certain visual style image through a series of calculations after setting many parameters for the modeled 3D graphics" when making special effects. The rendered objects have different requirements for texture, material, texture, lighting, global illumination, shader, lens blur, depth, etc., and require different visual styles, but no matter what form of graphics rendering, it requires a lot of computing power.

Traditional industries for rendering services include film, television, animation, games, architecture/interior design, advertising/film/packaging, academic research, medical treatment, exhibitions, etc.

Different application scenarios have very different requirements for rendering quality, accuracy, and speed, and require different CPU/GPU solutions: for example, film and television products require realistic effects, and often adopt pre-rendering/offline rendering to ensure image quality, but the speed is extremely slow; while games and AR displays require interaction and real-time rendering, with high frame rate requirements but low requirements for image quality details.

Offline rendering: more used in film and television special effects, 3D renderings, CG animations, home decoration design and other scenes.

Real-time rendering: Mainly used in cloud games, AR/VR, live video special effects, cloud creation and other scenarios with high requirements for low latency and interactivity.

In the ultimate pursuit of real scenes, it is difficult to render full-scale models in real time for scenes such as AI, virtual reality, and 3D games.

Interestingly, some scene rendering is based on different technical solutions, and different hardware devices can implement rendering solutions with different characteristics. For example, the traditional OctaneRender (which has been acquired by Hollywood partner rendering company OTOY) is a GPU-based physical rendering engine that uses NVIDIA's RTX ray tracing GPU for hardware acceleration.

With the intervention of Web3 infrastructure, decentralized distributed networks can allow more participants to join such an ecosystem, such as CPU/GPU miners and individual creators who need rendering services, providing new opportunities for Metaverse/VR/AR scenarios.

To a certain extent, decentralized rendering has the potential to break through traditional rendering: existing Web2 games require extensive use of interactive images. When the PC runs the game, the graphics card will render a large number of images in the background to form a continuous visual effect (game animation). When the quality of the rendered images is too high or the animation transitions are too frequent, the number of images increases significantly, and the graphics card computing power cannot keep up, which will cause freezes and delays, especially for real-time rendering. Web2 cloud games have always found it difficult to solve the problems of high latency and high cost. Web3 rendering infrastructure has advantages in high-precision simulation and real-time rendering.

For example: RNDR rendering network, an early leading project in the track, mentioned in the latest released iPad pro that when the local computing power of iPad is insufficient, it will use the computing power of Render Network on the cloud to assist in rendering.

In the existing Web3 rendering infrastructure, Tokenomics is mainly used to incentivize ecosystem users to contribute idle graphics card computing power, and to form an edge computing cloud rendering network closer to each user with more distributed personal GPU nodes. It also balances the rendering task initiator and the computing power provider, and gives benefits to GPU miners while ensuring the cost-effectiveness of rendering services, so as to maintain the sustainability of the ecosystem.

At the same time, such platforms also need to provide services closer to the application layer and better introduce creators, a key role in Web3, into the ecosystem. Taking the early RNDR rendering network as an example, it proposed two roles: "creator" (3D image creators who need additional GPU computing power) and "node provider" (users with idle GPU computing power). In pursuit of low-cost rendering and towards the masses.

As the infrastructure of Web3, the Metaverse public chain can better support the entire chain ecosystem.

For example, taking the public chain network architecture iPolloverse as an example, the network architecture has four layers: from low to high, they are Meta computing layer, network layer, rendering layer and ecological layer. According to public information, its test network has reached 1 GPU supporting 500 users.

Judging from the early leading projects in the track, Render, and then to the new public chain with the concept of metaverse rendering, the supply side of this track is sufficient, that is, the GPU computing power end has been occupied by many infrastructures. This new type of mining involves many roles, including GPU cloud service providers.

Therefore, the next thing to observe is the demand side of Metaverse rendering:

Refer to the three basic parameters that make up the Metaverse: Avatar (number of people/roles), NFT assets, and infrastructure. Decentralized rendering network services can touch all three of the above parameters.

From the perspective of shared GPU computing power, this concept based on edge computing is not new, but the rise of the Metaverse concept in 2022, as well as the implementation of application scenarios such as GameFi and Chain Games, provide new application scenarios and industry scale expectations for decentralized rendering. As an infrastructure with practical application scenarios, the Metaverse rendering public chain will be the focus of attention in the development of rendering demand in the future.

In addition, from the perspective of complex computing, outside of the metaverse scenario, some concepts and scenarios that still use complex computing are still increasing. For example, RNDR is testing the ChatGPT API to provide computing power. These new concepts include AI computing power, protein simulation, weather computing, etc. These frameworks that require a lot of computing power have opened up new imaginations.