Low-delay game picture optimization method based on cloud image rendering

Abstract

The invention discloses a low-delay game picture optimization method based on cloud image rendering, and the method achieves the optimization through the following three core steps: 1, constructing a cloud rendering engine cluster based on a containerization technology, and packaging a complete game rendering environment by each containerization rendering unit, an intelligent task scheduling system and an automatic capacity expansion and contraction function are matched to realize optimal utilization of rendering resources; 2, optimizing a picture rendering and transmission link, analyzing game scene data, constructing an exclusive rendering assembly line for key elements, generating frame pictures in advance for dynamic elements by adopting a prediction rendering technology, and guaranteeing transmission stability in combination with adaptive code rate transmission and network coding; and 3, client adaptation is enhanced, when the client is started, hardware configuration is automatically detected and reported to the cloud, the cloud customizes an exclusive receiving decoding scheme, and the client reduces display delay by using a hardware acceleration module. The cloud rendering resource utilization rate is effectively improved, the game picture transmission fluency is guaranteed, and the picture display delay is reduced.

Description

Technical Field

[0001] This invention relates to the field of cloud image rendering technology, specifically to a low-latency game screen optimization method based on cloud image rendering. Background Technology

[0002] With the rapid development of the gaming industry, players' demands for game graphics quality and smoothness are constantly increasing. Traditional local game rendering methods are limited by the performance of terminal hardware, making it difficult to achieve high resolution and high frame rate rendering when handling complex game scenes. This often leads to stuttering, frame drops, and other problems, affecting the gaming experience.

[0003] With the rise of cloud computing technology, cloud rendering has become an effective way to solve local hardware bottlenecks. Existing cloud rendering technologies mostly adopt a centralized architecture, concentrating rendering tasks on large server clusters and allocating tasks on a first-come, first-served basis, without finely encapsulating the rendering environment. Task scheduling relies solely on the overall server load, failing to fully consider factors such as client geographical location, dynamic network changes, and task complexity, leading to unreasonable resource allocation, with some servers idle while some tasks are processed inefficiently.

[0004] In the image transmission stage, existing technologies mostly use fixed bitrate transmission, which cannot adjust parameters according to real-time network conditions. When the network is unstable, stuttering or transmission interruption is likely to occur. At the same time, the client adaptation is not perfect. It simply receives and displays the image without combining local hardware configuration optimization and lacks functions such as dynamic resolution switching. It cannot make full use of hardware acceleration capabilities, which further increases the image display latency. Summary of the Invention

[0005] To address the problems of low resource utilization, unstable image transmission, and poor client compatibility in existing technologies, this invention provides a low-latency game screen optimization method based on cloud image rendering. By constructing a containerized cluster, intelligent task scheduling, adaptive transmission optimization, and client adaptation enhancement, it improves resource utilization, ensures smooth image transmission, reduces display latency, and enhances the gaming experience.

[0006] To address the aforementioned technical problems, this invention provides a low-latency game graphics optimization method based on cloud-based image rendering, comprising the following three core steps:

[0007] 1. Cloud rendering engine construction:

[0008] A cloud-based rendering engine cluster based on containerization technology is built. Each containerized rendering unit runs independently and is encapsulated with a complete game rendering environment, including an operating system, rendering software, and drivers. The Kubernetes container management platform is used to coordinate the start, stop, scaling and resizing operations of each containerized rendering unit. The image corresponding to each containerized rendering unit contains a complete operating system, graphics rendering library, graphics driver, and rendering software configuration.

[0009] Deploy an intelligent task scheduling system in the cloud rendering engine cluster. The intelligent task scheduling system takes into account the geographical distribution of game clients, the current network load, and the complexity of rendering tasks, and dynamically allocates rendering tasks to the most suitable containerized rendering units.

[0010] The intelligent task scheduling system calculates the load index of each containerized rendering unit using a load assessment formula. The formula for calculating the load index is as follows: ;

[0011] in, This represents the load value of container i. For the CPU utilization of container i, For the memory usage of container i, For the network bandwidth utilization of container i, This is the maximum capacity of the container resources. Using this formula, the resource utilization of each container can be calculated, its load assessed to determine if it has reached a threshold, and thus whether it is necessary to expand or reclaim containers for the cluster. The adaptability formula evaluates the adaptability of containerized rendering units to rendering tasks; the adaptability formula is... ,in, Let represent the adaptability of container i to the task, and T represent the resource requirements of the rendering task. For the load of the container. According to this formula, if the container's fit... A higher value indicates that the container has a large amount of remaining computing resources and can prioritize processing this rendering task.

[0012] Containerized rendering units have automatic scaling capabilities, calculating the current rendering pressure of the cluster using a rendering pressure value formula. The rendering pressure value formula is as follows: Where P represents the current rendering pressure value of the cluster. It is the CPU utilization of container i. It is the memory utilization of container i. P represents the network bandwidth utilization of container i, and N is the total number of containers in the cluster. This formula combines the CPU, memory, and bandwidth usage of containers, reflecting the overall load of the current rendering task. If the load value P exceeds a set threshold, a scaling-up operation is triggered; if it falls below a certain lower limit, a scaling-down operation is triggered.

[0013] 2. Optimization of image rendering and transmission links:

[0014] After receiving the rendering request from the game client, the cloud parses the game scene data in the request, extracts key scene elements, including characters, props, and special effects, and builds a dedicated rendering pipeline for key scene elements to improve rendering efficiency.

[0015] For dynamic scene elements in the game, predictive rendering technology is used to calculate the future positions of the dynamic scene elements using a position prediction formula. The position prediction formula is as follows: Where P(t) is the current position of the element at time t, V(t) is the velocity of the element at time t, A(t) is the acceleration of the element at time t, and Δt is the prediction time step. Based on the predicted future position, the subsequent frames are rendered in advance and stored in the cloud cache so that the client can quickly obtain them and reduce the waiting time for real-time rendering.

[0016] Adaptive bitrate transmission technology based on streaming media protocols is adopted to monitor parameters such as packet loss rate and latency jitter of the network link in real time, and dynamically adjust the transmission bitrate and frame rate to ensure smooth transmission of the image. Network coding technology is used to perform redundant encoding on the transmitted image data, and the original data is restored by decoding algorithm at the receiving end, which enhances the anti-packet loss capability of the image transmission.

[0017] 3. Client-side adaptation and enhancement:

[0018] When the game client starts, it automatically detects the local hardware configuration and reports it to the cloud. The local hardware configuration includes CPU frequency, memory size, and storage capacity. Based on the local hardware configuration information, the cloud customizes a dedicated screen reception and decoding scheme for the client. The dedicated screen reception and decoding scheme includes selecting an appropriate decoder and adjusting the size of the receive buffer.

[0019] The client utilizes a local hardware acceleration module, which is the GPU's hardware acceleration function, to accelerate the rendering of the received optimized game screen, further reducing screen display latency and improving the frame rate.

[0020] The client supports dynamic resolution switching, triggering resolution adjustment via an adaptive resolution scaling formula. The adaptive resolution scaling formula is as follows: Where R is the adaptive resolution ratio, B is the current network bandwidth, L is the real-time latency between the client and the cloud, and S is the current frame size of the rendered screen. When R is lower than the preset resolution threshold, the screen resolution is reduced to ensure basic game smoothness. When R is higher than the preset resolution threshold, the screen resolution is increased to improve screen quality.

[0021] In summary, due to the adoption of the above-mentioned technologies, the beneficial effects of this invention are:

[0022] 1. Significantly improved resource utilization: Through containerized cluster construction and intelligent task scheduling system, the rendering tasks and server resources are accurately matched, avoiding the problems of resource idleness or overload under centralized architecture. At the same time, the automatic scaling function further improves the flexibility of resource utilization and reduces resource waste.

[0023] 2. Enhanced screen transmission stability: Adaptive bitrate transmission technology can dynamically adapt to changes in network conditions, and network coding technology improves the ability to resist packet loss, effectively reducing the risk of screen stuttering and transmission interruption, and ensuring the continuous and smooth transmission of game screens.

[0024] 3. Client compatibility optimization: The client hardware detection and customized solution are combined to make full use of local hardware acceleration capabilities and reduce display latency; the dynamic resolution switching function enables the client to adapt to complex network environments, balance screen quality and smoothness when the network fluctuates, and significantly improve the user gaming experience under different devices and network conditions. Attached Figure Description

[0025] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention, making other features, objects, and advantages of the invention more apparent. The illustrative embodiments of the invention illustrated in the drawings and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0026] Figure 1 This is a flowchart of the steps of the present invention;

[0027] Figure 2 This is a signaling interaction diagram of the present invention;

[0028] Figure 3 This is a system framework diagram of the present invention. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to represent selected embodiments of the invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] In the description of this invention, it should be understood that the terms indicating orientation or positional relationship are based on the orientation or positional relationship shown in the drawings and are only for the convenience of describing the invention and simplifying the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention.

[0031] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific context of the specification.

[0032] This invention provides a low-latency game screen optimization method based on cloud image rendering, including the following embodiments:

[0033] Example 1

[0034] This embodiment applies to a small to medium-sized cloud gaming platform that serves 5,000 to 10,000 users and covers various game types such as role-playing and action games. The terminal devices are mainly PCs and smartphones.

[0035] Step 1) Building a cloud rendering engine cluster based on containerization technology. Kubernetes is used as the container management platform. The cluster initially deploys 20 containerized rendering units. Each containerized rendering unit's image contains a Linux operating system, OpenGL rendering software, and the corresponding version of the graphics card driver, ensuring that each container runs independently and the rendering environment is consistent.

[0036] Deploy an intelligent task scheduling system, setting the CPU utilization weight coefficient to 0.4, the memory utilization weight coefficient to 0.3, and the network bandwidth utilization weight coefficient to 0.3, with a preset rendering pressure threshold of 70 and a preset lower limit of 30.

[0037] When a PC client initiates a rendering request for a large-scale role-playing game, the intelligent task scheduling system first obtains the client's latitude and longitude coordinates as 120.12°E, 30.25°N. It then traverses the latitude and longitude coordinates of each containerized rendering unit in the cluster and calculates, using the geographical distance calculation formula, that container A's latitude and longitude coordinates of 120.10°E, 30.23°N are the closest to the client's geographical distance.

[0038] Meanwhile, the CPU utilization of container A is 45%, the memory utilization is 35%, and the network bandwidth utilization is 40%. The load index of container A is calculated as 45×0.4+35×0.3+40×0.3=41.5 using the load assessment formula. The resource requirement T of this rendering task is 50. The adaptability of container A to this task is calculated as 50÷41.5≈1.20 using the adaptability formula, which is a relatively high adaptability.

[0039] At this point, the overall rendering pressure value P of the cluster is calculated by the formula as the sum of the average values ​​of CPU utilization, memory utilization, and network bandwidth utilization of each container divided by 3. The calculated value of P is 55, which is lower than the threshold of 70. Therefore, no expansion is required, and the intelligent task scheduling system allocates the rendering task to container A.

[0040] When the number of platform users reaches a peak of 10,000, the overall rendering pressure value P of the cluster is calculated to be 75, which exceeds the threshold of 70. The system automatically starts 10 idle container instances to join the cluster. After expansion, the cluster rendering pressure value drops to 58, ensuring efficient execution of rendering tasks. When the number of users drops to 5,000, the overall rendering pressure value P of the cluster is 28, which is lower than the lower limit of 30. The system reclaims 5 redundant containers to avoid resource waste.

[0041] Step 2) Optimization of screen rendering and transmission link: After receiving the rendering request from the PC client, the cloud analyzes the game scene data, extracts key scene elements such as characters, props, and skill effects, builds a dedicated rendering pipeline, prioritizes the rendering tasks of these key elements, and improves rendering efficiency.

[0042] The system employs adaptive bitrate transmission technology based on the HLS streaming media protocol, which monitors the packet loss rate and latency jitter of the network link in real time. The initial transmission bitrate is set to 8 Mbps and the frame rate to 60 fps. When the network packet loss rate is detected to rise from 0.5 to 3, the system automatically adjusts the transmission bitrate to 5 Mbps and maintains the frame rate at 60 fps; when the packet loss rate drops below 1, the transmission bitrate returns to 8 Mbps.

[0043] By using RS network coding technology to perform redundant encoding on the transmitted image data, with the encoding redundancy set to 0.2, even if the network packet loss rate does not exceed 20%, the receiving end can completely recover the original data through the decoding algorithm, ensuring the integrity of the image transmission.

[0044] Step 3) Client Adaptation and Enhancement: When the PC client starts, it automatically detects the local hardware configuration, which is 3.6GHz CPU, 16GB memory, 512GB storage, and NVIDIA GTX1660 GPU, and reports this information to the cloud.

[0045] Based on the hardware configuration, the cloud platform selects an H.265 decoder and adjusts the receive buffer size to 512KB to adapt to the GPU's hardware acceleration function.

[0046] The client utilizes the hardware acceleration capabilities of the NVIDIA GTX 1660 to accelerate the rendering of received game footage, reducing the display latency from 80ms to 45ms and stabilizing the frame rate at 60fps.

[0047] When the network is stable, the client receives a 1080p resolution screen. When the network bandwidth drops from 10Mbps to 4Mbps and the real-time latency increases from 40ms to 80ms, and the current frame size is 1.2MB, the adaptive resolution ratio is calculated as R=4÷(80×1.2)≈0.42, which is lower than the preset resolution threshold of 0.5. The client automatically reduces the screen resolution to 720p to ensure smooth gameplay. When the network bandwidth recovers to 10Mbps and the latency drops to 40ms, the calculated R=10÷(40×1.2)≈2.08, which is higher than the threshold of 0.5. The resolution is then restored to 1080p.

[0048] In this embodiment, the resource utilization rate of the cloud gaming platform increased from 65% to 88%, the screen stuttering rate decreased from 12% to 2.5%, and user satisfaction was significantly improved.

[0049] Example 2

[0050] This embodiment is applied to a large-scale cloud gaming platform, serving more than 50,000 users, mainly competitive games. These games have extremely high requirements for screen latency and smoothness, and the terminal devices include PCs, game consoles, smartphones, etc.

[0051] Step 1) Building a cloud rendering engine cluster based on containerization technology. The cluster uses the Kubernetes container management platform and initially deploys 100 containerized rendering units. Each containerized rendering unit encapsulates the Windows Server operating system, DirectX 12 rendering software, and high-performance graphics card drivers. The image also integrates game-specific optimization plugins to improve rendering efficiency.

[0052] Deploy an intelligent task scheduling system, setting the CPU utilization weight coefficient to 0.5, the memory utilization weight coefficient to 0.2, and the network bandwidth utilization weight coefficient to 0.3, with a preset rendering pressure threshold of 75 and a preset lower limit of 25.

[0053] When a game console client initiates a rendering request for a competitive game, the rendering task is highly complex, with a resource requirement T of 80. The intelligent task scheduling system obtains the client's latitude and longitude coordinates as 116.4°E and 39.9°N, calculates the geographical distance between each container and the client, and selects the 5 closest containers.

[0054] Obtain the load data for these 5 containers. Container B has a CPU utilization of 50%, a memory utilization of 30%, and a network bandwidth utilization of 35%. The load index calculated using the load assessment formula is 50×0.5+30×0.2+35×0.3=41.5. The adaptability of container B to this task is calculated using the adaptability formula as 80÷41.5≈1.93, which is the highest adaptability.

[0055] At this point, the overall rendering pressure value P of the cluster is calculated to be 68, which is lower than the threshold of 75. The intelligent task scheduling system then assigns the rendering task to container B.

[0056] When a competitive game hosts a large-scale event and the number of users surges to 80,000, the overall rendering pressure value P of the cluster rises to 82, exceeding the threshold of 75. The system automatically starts 50 idle container instances to expand the capacity, and after the expansion, P drops to 65. After the event ends, the number of users drops to 30,000, and P is 22, which is lower than the lower limit of 25. The system reclaims 30 redundant containers.

[0057] Step 2) Optimize the rendering and transmission links: The cloud analyzes the scene data of the competitive game, extracts key scene elements such as moving characters, flying bullets, and explosion effects, and builds a dedicated rendering pipeline.

[0058] For the dynamic scene element of a bullet, its future position is calculated using a position prediction formula. Given that the bullet's current position P(t) at time t is x1, y1, z1, its velocity V(t) is vx, vy, vz, and its acceleration A(t) is ax, ay, az, and the prediction time step Δt is 16.7ms, which is the time interval between each frame, the positions of the bullet at times t+Δt, t+2Δt, and t+3Δt are calculated, and these three frames are rendered in advance and stored in the cloud cache.

[0059] When a client requests a frame, the cloud simultaneously returns the current frame and three pre-rendered predicted frames, reducing real-time rendering wait time and lowering the frame rendering latency from 60ms to 25ms.

[0060] Adaptive bitrate transmission technology based on the DASH streaming media protocol is adopted to monitor network packet loss rate and latency jitter in real time. The initial transmission bitrate is 15Mbps and the frame rate is 120fps. When the network latency jitter is detected to increase from 10ms to 30ms and the packet loss rate increases to 4, the system adjusts the transmission bitrate to 10Mbps and maintains the frame rate at 120fps. When the network conditions improve, the latency jitter drops below 15ms, the packet loss rate drops below 1, and the transmission bitrate returns to 15Mbps.

[0061] LDPC network coding technology is used to perform redundant coding on the transmitted image data, with the coding redundancy set to 0.3, which enhances the data transmission's ability to resist packet loss and ensures that the receiving end can completely recover the original data when the network packet loss rate does not exceed 30%.

[0062] Step 3) Client Adaptation and Enhancement: When the game console client starts, it detects the local hardware configuration, which is 4.2GHz CPU, 32GB memory, 1TB storage capacity, and AMD Radeon RX6800XT GPU. The configuration information is then reported to the cloud.

[0063] A customized solution was developed for it in the cloud, selecting the AV1 decoder and adjusting the receive buffer size to 1024KB to fully adapt to the GPU's hardware acceleration capabilities.

[0064] The client utilizes the hardware acceleration capabilities of the AMD Radeon RX 6800 XT to accelerate the rendering of the received images, further reducing the display latency to 18ms and stabilizing the frame rate at 120fps.

[0065] When the network bandwidth drops from 20Mbps to 8Mbps and the real-time latency increases from 30ms to 70ms, with the current frame size being 1.8MB, the adaptive resolution ratio is calculated as R=8÷(70×1.8)≈0.63, which is higher than the preset threshold of 0.5, thus maintaining the 1080p resolution. When the network bandwidth drops to 5Mbps and the latency increases to 100ms, the calculated R=5÷(100×1.8)≈0.28, which is lower than the threshold of 0.5, and the client automatically reduces the resolution to 900p to ensure smooth gameplay. After the network recovers, the resolution returns to 1080p.

[0066] In this embodiment, the screen latency of competitive games is stabilized within 20ms, the screen stutter rate is less than 1, and the resource utilization rate is improved from 62 before optimization to 90, which meets the requirements of large-scale competitive games for low latency and high smoothness.

Claims

1. A low-latency game screen optimization method based on cloud image rendering, characterized in that: Includes the following steps: Cloud rendering engine construction: Build a cloud rendering engine cluster based on containerization technology. Each containerized rendering unit runs independently and is encapsulated with a complete game rendering environment, including an operating system, rendering software, and drivers. Deploy an intelligent task scheduling system in the cloud rendering engine cluster. The intelligent task scheduling system dynamically allocates rendering tasks to the most suitable containerized rendering unit based on the geographical distribution of game clients, the current network load, and the complexity of the rendering tasks. Rendering and transmission link optimization: After receiving the rendering request from the game client, the cloud parses the game scene data in the request, extracts key scene elements, and builds a dedicated rendering pipeline for the key scene elements. Adaptive bitrate transmission technology based on streaming media protocols is adopted to monitor the packet loss rate and latency jitter parameters of the network link in real time and dynamically adjust the transmission bitrate and frame rate of the picture; network coding technology is used to perform redundant encoding on the transmitted picture data, and the original data is restored at the receiving end through decoding algorithms; Client adaptation and enhancement: When the game client starts, it automatically detects the local hardware configuration and reports it to the cloud. The local hardware configuration includes CPU frequency, memory size, and storage capacity. Based on the local hardware configuration information, the cloud platform customizes a unique screen reception and decoding solution for the client. The client utilizes a local hardware acceleration module to accelerate the rendering of the received optimized game graphics; the client supports dynamic resolution switching, adjusting the screen resolution according to network conditions.

2. The low-latency game screen optimization method based on cloud image rendering according to claim 1, characterized in that, The cloud rendering engine cluster uses the Kubernetes container management platform to coordinate the start, stop, scaling and shrinking operations of each containerized rendering unit. The image corresponding to each containerized rendering unit contains a complete operating system, graphics rendering library, graphics driver and rendering software configuration.

3. The low-latency game screen optimization method based on cloud image rendering according to claim 1, characterized in that, The intelligent task scheduling system calculates the load index of each containerized rendering unit using a load evaluation formula, the formula for which is: ,in For the CPU utilization of container i, For the memory usage of container i, For the network bandwidth utilization of container i, This is the weighting coefficient for CPU utilization.

4. The low-latency game screen optimization method based on cloud image rendering according to claim 1, characterized in that, The intelligent task scheduling system evaluates the adaptability of containerized rendering units to rendering tasks using an adaptability formula, which is: ,in Let represent the adaptability of container i to the task, and T represent the resource requirements of the rendering task. This represents the load value of container i.

5. The low-latency game screen optimization method based on cloud image rendering according to claim 1, characterized in that, The intelligent task scheduling system calculates the geographical distance between the game client and the containerized rendering unit using Euclidean distance. The formula for calculating the geographical distance is as follows: ,in Geographical distance, and The latitude and longitude coordinates of the containerized rendering unit. and These are the latitude and longitude coordinates of the game client.

6. The low-latency game screen optimization method based on cloud image rendering according to claim 1, characterized in that, The containerized rendering unit has automatic scaling capabilities and calculates the current rendering pressure of the cluster using a rendering pressure value formula, which is: Where P is the rendering pressure value. For the CPU utilization of container i, For the memory usage of container i, P represents the network bandwidth utilization of container i, and N represents the total number of containerized rendering units in the cluster. When P exceeds a preset threshold, expansion is triggered, and when P falls below a preset lower limit, shrinkage is triggered.

7. The low-latency game screen optimization method based on cloud image rendering according to claim 1, characterized in that, In the image rendering and transmission link optimization steps, predictive rendering technology is used for dynamic scene elements in the game. The future positions of the dynamic scene elements are calculated using a position prediction formula, which is as follows: Where P(t) is the current position of the element at time t, V(t) is the velocity of the element at time t, A(t) is the acceleration of the element at time t, and Δt is the predicted time step. The subsequent frames are rendered in advance and stored in the cloud cache.

8. The low-latency game screen optimization method based on cloud image rendering according to claim 1, characterized in that, The dynamic resolution switching is triggered by an adaptive resolution scaling formula, which is as follows: Where R is the adaptive resolution ratio, B is the current network bandwidth, L is the real-time latency between the client and the cloud, and S is the frame size of the current screen rendering. When R is lower than the preset resolution threshold, the screen resolution is reduced; when R is higher than the preset resolution threshold, the screen resolution is increased.

9. The low-latency game screen optimization method based on cloud image rendering according to claim 1, characterized in that, The dedicated image reception and decoding scheme includes selecting a suitable decoder and adjusting the reception buffer size.

10. The low-latency game screen optimization method based on cloud image rendering according to claim 1, characterized in that, The local hardware acceleration module is the hardware acceleration function of the GPU.

Images