WO2013097210A1

WO2013097210A1 - Online rendering method and offline rendering method and relevant device based on cloud application

Info

Publication number: WO2013097210A1
Application number: PCT/CN2011/085117
Authority: WO
Inventors: 朱旭琪
Original assignee: 华为技术有限公司
Priority date: 2011-12-31
Filing date: 2011-12-31
Publication date: 2013-07-04
Also published as: CN103299347B; CN103299347A

Abstract

Disclosed are an online rendering method and offline rendering method and relevant device based on cloud application, for improving rendering efficiency. The embodiments of the present invention include: receiving a rendering task request; determining a model file to be rendered according to the rendering task request; dividing the model file into N₁ display units according to the play order of each frame in the model file, the display unit being the minimum unit for compression and transmission, wherein each display unit contains several frames, the play orders of which are adjacent, with at least N₁-1 display units including W frames; composing the frames with the same intra-unit play order in each display unit into a rendering unit so as to obtain W rendering units, wherein the intra-unit play order of each frame in the display unit is the play order of the frame relative to other frames in the display unit to which it belongs; and respectively allocating W rendering units to W different rendering nodes for rendering.

Description

Online application method and offline rendering method based on cloud application and related device

The present invention relates to the field of rendering cloud applications, and more particularly to an online rendering method and an off-line rendering method and related apparatus based on a cloud application.

Background technique

In recent years, the emergence of cloud computing coincides with the needs of rendering service applications, and the rendering cloud has been born. Cloud computing is highly virtualized with hardware resources and network storage, with efficient resource utilization and dynamic features, making it easy to perform the high-performance computing required for rendering. Rendering a cloud-accelerated rendering task is handled internally in a similar way to rendering a cluster. It also breaks down the rendering tasks into small units and then uses the cloud's powerful computing power to perform these task units.

There are two main types of application methods for rendering clouds: First, self-service rendering services, users can submit and control rendering tasks themselves. The purpose is to quickly complete a large number of rendering tasks by rendering the cloud's powerful computing power. It can also be called offline rendering ( Offline Rendering), commonly used in 3D animation, film effects, etc., can consume a lot of computing resources and time for the pursuit of realism; Second, support a variety of online rendering applications, called Real-time Rendering (Real-time Rendering) or Online rendering, due to the rich business form of online rendering applications, and the increasing pursuit of visual effects such as special effects in various fields, its application range will rapidly expand. For example, online 3D (3-Dmensions) design, 3D navigation, 3D display, and even real-time 3D games can be supported as the network improves.

However, current cloud-based renderings have low rendering efficiency and poor user experience.

Summary of the invention

Embodiments of the present invention provide an online rendering method and an offline rendering method and related devices based on a cloud application, which are used to improve rendering efficiency.

In view of this, the embodiments of the present invention provide:

An online rendering method based on cloud applications, including:

Receiving a rendering task request;

Determining the model file to be rendered according to the rendering task request;

According to the playing order of each frame in the model file, the model file is divided into two display units, and the display unit is a minimum unit of compression and transmission, wherein each display unit includes several playing units. Arranging adjacent frames, wherein at least N _r l display units include W frames;

Forming a rendering unit with the same playing order in each unit of the display unit to obtain W rendering units, wherein the playing order of each frame in the display unit is the playing of the frame relative to other frames in the display unit to which it belongs Order

W rendering units are assigned to W different rendering nodes for rendering.

An online rendering device based on a cloud application, comprising:

a first receiving module, configured to receive a rendering task request;

a first determining module, configured to determine, according to the rendering task request, a model file that needs to be rendered; a first decomposition module, configured to divide the model file into two display units according to a playing order of each frame in the model file, where the display unit is compressed And a minimum unit of transmission; wherein each display unit includes a plurality of frames adjacent to each other in play order, wherein at least N _r l display units include W frames; and a first combination module is used for each display unit The frames in the same playing order are composed of the rendering units to obtain the W rendering units. The playing order in the unit of each frame in the display unit is the playing order of the frame relative to other frames in the display unit to which it belongs.

The first allocation module is configured to allocate W rendering units to W different rendering nodes for rendering.

An offline rendering method based on a cloud application, comprising:

Receiving a rendering task request;

According to the playing order of each frame in the model file, the model file is divided into N ₂ groups, wherein at least N ₂ -1 groups include Z frames, Z is greater than 0 and less than the number of available rendering nodes;

The frames in the same playing order are grouped into rendering units to obtain Z rendering units, wherein the playing order of each frame in each group is the playing order of the frame relative to other frames in the group to which it belongs; Z rendering units are assigned to Z different rendering nodes for rendering.

An offline rendering device based on a cloud application, comprising:

a second receiving module, configured to receive a rendering task request;

a third determining module, configured to determine, according to the rendering task request, a model file that needs to be rendered; a second decomposition module, configured to divide the model file into N ₂ groups according to a playing order of each frame in the model file, where at least N ₂ -1 groups contain Z frames, Z is greater than 0 and less than available rendering Number of nodes;

a second combination module, configured to group the frames with the same playing order in each group into a rendering unit, to obtain Z rendering units, where the playing order of each frame in each group is the frame relative to the group to which it belongs The order in which other frames are played;

The second allocation module is configured to respectively allocate Z rendering units to Z different rendering nodes for rendering.

As can be seen from the foregoing technical solutions, an embodiment of the present invention provides an online rendering method and an offline rendering method and related devices based on a cloud application. In online rendering, according to a playback order of frames in a model file that needs to be rendered, The model file is divided into a plurality of display units, and then frames with the same playing order in the display unit are combined into a rendering unit, wherein each rendering unit has different priorities regarding the playing order, and then each rendering unit is respectively assigned to a different rendering. The node realizes frame parallel rendering of each display unit, and each display unit can be rendered in the order of playing time; in offline rendering, in order to save rendering time, rationally utilize resources, and perform task decomposition before rendering, effectively Improve the efficiency of rendering.

DRAWINGS

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following description of the embodiments will be briefly described. It is obvious that the drawings in the following description are only some embodiments of the present invention. One of ordinary skill in the art can also obtain other drawings based on these drawings without undue creative effort.

FIG. 1 is a flowchart of a cloud application-based online rendering method according to an embodiment of the present invention; FIG. 2 is a flowchart of a cloud application-based online rendering method according to an embodiment of the present invention; A schematic diagram of the process of the online rendering method provided by the example;

4 is a schematic diagram of another process of an online rendering method according to an embodiment of the present invention;

FIG. 5 is a schematic structural diagram of an online rendering device based on a cloud application according to an embodiment of the present invention; FIG.

FIG. 6 is a schematic structural diagram of another online rendering apparatus according to an embodiment of the present invention; FIG. 7 is a flowchart of an offline rendering method based on a cloud application according to an embodiment of the present invention; A schematic diagram of the process of the offline rendering method;

9 is a schematic diagram of a conventional offline rendering method; FIG. 10 is a schematic structural diagram of an offline rendering device based on a cloud application according to an embodiment of the present disclosure;

FIG. 11 is a schematic structural diagram of another offline rendering apparatus according to an embodiment of the present invention.

detailed description

BRIEF DESCRIPTION OF THE DRAWINGS The technical solutions in the embodiments of the present invention will be described in detail below with reference to the accompanying drawings. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present invention without creative work are within the scope of the present invention.

The details are described below separately.

The embodiment of the invention provides an online rendering method based on a cloud application. For the method flow, please refer to FIG. 1 , the method includes:

S101. Receive a rendering task request.

S102. Determine, according to the rendering task request, a model file that needs to be rendered;

In some embodiments, step S101 and step S102 may be: the user triggers an online 3D application, which is equivalent to submitting an online rendering task request to the background, and determining a model file, or scene, to be rendered according to the online rendering task request. Document

The model file may be obtained in advance, or may be obtained by real-time calculation of the frame information in the model file according to the trigger condition; it is not specifically limited herein.

S103. Divide the model file into a display unit according to a play order of each frame in the model file, where the display unit is a minimum unit of compression and transmission, where each display unit includes a plurality of frames adjacent to each other in a play order. At least N _r l display units include W frames;

It can be known from steps S101 and S102 that if the model file to be rendered contains a frame, M in the model file is divided into display units according to the playing order of each frame in the model file; the display unit is the smallest unit of compression and transmission; The display unit includes a plurality of frames adjacent to each other in the playing order, that is, the frames in each display unit are a set of frames arranged in the playing order thereof, and at least N _r l display units in the plurality of display units Contains W frames.

In some embodiments, the total number of frames in the rendered model file can be set to be a frame, The display unit is the smallest unit of compression and transmission. If it is known that the number of frames in the display unit required for smooth display at this time is W, the number of display units N _{1 =} ― ;

W where, the expression indicates that when it can be divisible by W, the display unit is one, and each display unit contains W frames; when it cannot be divisible by W, the fractional part of the result is rounded off, and the integer part is + 1 It is found that the number of display units is N _{l ,} wherein the first N _r l display units contain W frames, and the last display unit has (M Ni- x W ) frames.

5104. The frame of the same playback order in each unit of the display unit is composed into a rendering unit, to obtain W rendering units, where the playing order of each frame in the display unit is the frame relative to other frames in the display unit to which the frame belongs. Play order

For each display unit, the first frame played in the unit is composed into a rendering unit. Similarly, the second frame played in the unit is composed into a rendering unit, and so on, and the W frames played in the unit are composed. A rendering unit, so that to get W rendering units.

It should be noted that the playing order in the unit of each frame in the display unit is the playing order of the frame relative to other frames in the display unit to which it belongs;

For example: According to the playing order of each frame in the model file, the first frame played in the first display unit is the first frame played by all the frames of the model file, and if the model file includes a display unit, Frame, the last frame played in the unit is the Wth frame of all frames of the model file; if the first frame played in the last display unit is the first frame of the model file ( ( Nrl ) x W+1 ) frames played, the last frame played in the unit is the XWth frame of all frames of the model file.

5105. Assign W rendering units to W different rendering nodes for rendering; it can be understood that each rendering unit needs to be assigned to a rendering node for rendering, and at the same time, to ensure that all frames of one display unit are simultaneously generated. Therefore, the number of nodes participating in rendering is also W.

It should be noted that the cloud application-based online rendering method provided in this embodiment is applicable not only to a 3D animation model file that needs to be rendered in advance, but also to real-time calculation of each frame scene by an artificial intelligence or the like. a model file of content and parameters, that is, the model file is obtained by real-time calculation of the information of the frame according to the trigger condition; For the model file in the first case, according to the total frame number M _l and the size W of the display unit, according to the play order of the frame, the frame may be first divided into several display units, and then decomposed into W rendering units. , assigned to W different rendering nodes for rendering; if for the model file in the second case, according to the total number of frames M _l and the size of the display unit W, each frame is calculated in real time according to the playing order of the frame, And sequentially assigned to W rendering units frame by frame, and assigned to W different rendering nodes for rendering. In essence, they are all rendered by a frame that is composed of the same playback order in each unit of the display unit.

The cloud application-based online rendering method provided by the above embodiment divides the model file into a plurality of display units according to the playing order of the frames in the model file to be rendered, and then renders the frames with the same playing order in the display unit. a unit, wherein each rendering unit has a different priority with respect to the playing order, and then each rendering unit is respectively assigned to a different rendering node, that is, the task is decomposed before rendering, and the frame parallel of each display unit is realized. Rendering, try to ensure that they complete rendering at the same time; and each display unit can be rendered in the order of playback time, which effectively improves the rendering efficiency.

The embodiment of the invention further provides an online rendering method based on a cloud application, which performs evaluation of the rendering capability of the allocated rendering nodes to undertake rendering tasks of rendering units of different priorities. As shown in Figure 2, the method includes:

5201. Receive a rendering task request.

5202. Determine, according to the rendering task request, a model file that needs to be rendered;

It can be understood that the content of step S201 is the same as that of step S101, and the content of step S202 is the same as that of step S102, and the related description may be referred to, and will not be specifically described herein.

5203. Divide the model file into a display unit according to a play order of each frame in the model file, where the display unit is a minimum unit of compression and transmission, where each display unit includes a plurality of frames adjacent to the play order, where At least N _r l display units include W frames;

It can be known from steps S201 and S202 that, in the online rendering, the model file to be rendered may be a 3D animation sequence containing frames; according to the playing order of each frame in the model file, the M in the model file is divided into display units. The display unit is a minimum unit of compression and transmission; each display unit includes a plurality of frames adjacent to each other in play order, that is, each display unit is a set of frames arranged in the play order of the frames therein, and At least N _r l display orders in the display unit The element includes W frames, wherein the first N _r l display units include W frames, and the N ₁ display unit includes (M ( N l ) W ) frames.

In some embodiments, the total number of frames in the rendered model file can be set to be a frame. Since the display unit is the smallest unit of compression and transmission, if it is known that the number of frames required for smooth display at this time is W, the display is performed. The number of units _{Nl =} "i];

W where, the expression indicates that when Mi can be divisible by W, the display unit is one, and each display unit contains W frames; when it cannot be divisible by W, the fractional part of the result is rounded off, the integer part +1 indicates that the number of display units is N. Among them, there may be W frames in one display unit, and (M Ni-x W ) frames in the last display unit.

It should be noted that the number of frames required to ensure smooth display W can be derived according to the following formula:

W≥ {[max(ti(res), t ₂ (res), · . - ti(res), · . - tw(res)) + TcT(res, B, W, R) + Ttoi] xf} ( 1) , where t es) is the time required to render a frame for the assigned render node i, i is an integer from 1 to W; mainly related to resolution res; max ^res), t ₂ (res), ...ti(res), ... t _w (res)) is the maximum single frame rendering time; T _CT is the total time required for the difference compression (Compression) and transmission (Transmission) W frame, and resolution Res, network speed B, display unit size W and the difference encoding compression ratio R used; T _tol is the tolerance time, that is, the sum of the system waiting time and the frameless compensation time after the last frame rendering in one display unit is completed. ; f is the frame rate, the unit is s (frames per second), which is the number of frames played per second.

Generally, the rendering result of the 3D animation model file is an image sequence, and the image sequence obtained after the rendering is completed needs to be subjected to difference encoding and network transmission difference encoding, wherein the difference encoding refers to a series of images rendered (equivalent In the process of compressing one frame and one frame in the video, when the compression of the first image is completed, the other images in the back can be compared with the first image, and then the difference image, that is, the two images. Different parts are compressed, which can get better compression effect. In addition, the resolution res (resolution) can also be called resolution, which is the precision of the screen image, which refers to the number of pixels that the display can display;

The lack of frame compensation means that the frame that has not been completed within the waiting time cannot be processed, and the image frame method such as motion prediction and interpolation needs to be used to estimate the frame image by using the obtained frame before and after the playback order. For example, if the 10th frame in a display unit is not completed within the waiting time, the ninth and eleventh frames can be used to estimate the 10th frame, so that all the frames of one display unit are completed. The frame number of the forward and backward reference may be changed, and may not be one frame, and may be two frames or three frames, which is not specifically limited herein.

Generally, when a user submits a rendering task request and enters the system, its total frame number M _l resolution res, the frame rate f can be considered as given. The average network speed B, the difference code compression ratio R and the tolerance time T _{tol of the} current or time period can be collected inside the system. The maximum single frame rendering time max(t!(res), t ₂ (res), ...ti(res), ...t _w (res)) is not strongly correlated with W, so the system can be used. The maximum rendering time of the empirical value estimate is represented; the local configuration and rendering time can also be provided by the user as a reference, and combined with the configuration of the current rendering node, the maximum single frame rendering time is comprehensively estimated.

The minimum unit of compression and transmission, that is, the size of the display unit W, can generally be obtained by the following steps: First, according to the number of rendering nodes available in the system, given an initial value of W Wo, Wo The value is greater than zero and less than the integer between the number of render nodes available in the system; wherein, the available render node can be understood as a render node whose idle or render node has a capability score higher than a preset threshold; _{Q is} substituted into the right side of the formula (1), and a value W _est(1 ) is calculated. If Wo ≥ West(l), W=W is obtained. At this point, the calculation of W is completed; if WQ < West(l), it is necessary to reduce _W by a certain step size _ΔW . Value, get WfW. -△ _W and record = West(l) - Wo, then repeat the calculation of the second step to get W _est(2 ). If Wi≥ W _est (, let w=W _l complete the calculation of W; if Wl < West (2), then continue to decrease in step size Δ _{w to} obtain W ₂ = W _r Δ _w , record ^W2 = West(2) - Wl ₀ and so on, until the value of W is obtained.

If a suitable value of W cannot be obtained in a positive integer greater than or equal to 1 and less than W _Q , the smooth display effect of the user end cannot be guaranteed, and the maximum value that minimizes the value is selected as W, where j=l , 2, 3, ... ^ .

S204. The frame of the same playback order in each unit of the display unit is formed into a rendering unit, to obtain W rendering units, where the playing order of each frame in the display unit is the frame relative to other frames in the display unit to which the frame belongs. Play order

In some embodiments, it may be considered that each frame is played in the order of playing in each display unit. The frame in the display unit is given a frame number;

For example: For the first frame in each display unit, the frame number "1" can be given. Correspondingly, for the second frame in each display unit, the frame number "2" can be given, and so on, for the display unit. In the i-th frame, i=l, 2, 3, ..., W, the frame number "i" can be given, and the frame number "W" can be given to the last frame in the display unit, that is, the Wth frame. " , , so that the same frame number can be given to the frames in the same playback order in the units in each display unit.

Wherein, the frames having the same playing order in the units in each display unit are combined into a rendering unit, and the frames in which the same frame number is given in each display unit may be composed into one rendering unit;

For example: For each display unit, a frame with a frame number "1" is assigned to form a rendering unit. For ease of understanding, a rendering unit composed of a frame with a frame number "1" can be recorded as a rendering unit 1. Similarly, The rendering unit composed of the frame number given with the frame number "2" is referred to as the rendering unit 2, and so on, and the rendering unit composed of the frame to which the frame number "i" is given can be referred to as the rendering unit i, and the frame number is given. The rendering unit composed of the frame of W" can be recorded as the rendering unit W, so that the frames with the same playing order in the units in each display unit can be composed into the rendering unit to obtain W rendering units.

For example: According to the playback order of each frame in the model file, the frame to which the frame number "1" is given in the first display unit is the first frame of the frame of all the model files, which is given the frame number "W". The frame is the Wth played frame of all the frames of the model file, but the frame given the frame number "1" in the second display unit is the first frame played in the second display unit, but the frame is the model. The W+1th frame of all the frames of the file, the frame given the frame number "W" is the 2 x W play frames of all frames of the model file; therefore, for the second and subsequent kth display units , k is an integer greater than 1 and less than ^, and the frame given the frame number "1" in the unit is the ((k-1) X W+1) played frame of all frames of the model file, which is given the frame The frame of the serial number "W" is the kxth frame of the frame of all the model files. As shown in FIG. 3, the frame number of each frame in the display unit can be given in the order of play, as shown in the frame in the figure. The order of these frames in the entire model file is the number shown on the horizontal axis below. 3, it is considered that the model file preclude the use of cross-frame is decomposed to achieve the rendering.

S205. Evaluating a rendering node rendering by rendering hardware information of the node and real-time load statistics The strength of the dyeing ability, determine that the rendering nodes with the strongest rendering ability participate in the rendering of w rendering units;

It can be seen from step S203 and step S204 that the frames decomposed into the same rendering unit can be assigned the same frame number according to the playing order of the frames; it can be understood that the decomposed rendering units have different meanings regarding the playing order. Priority, since the priority of each rendering unit is determined by the playback time sequence of the frames it contains, there are no two frames played at the same time, so the playback time must be sequential, so each rendering unit must have different priorities. level.

In addition, since each frame has a smaller frame number, the frame needs to be played earlier, and is also a reference object of the subsequently played frame. Here, a related description about the difference image compression process may be contacted to indicate the frame to be played later. Compression is done with the previous frame as the reference object.

Therefore, it should be ensured that frames with smaller frame numbers are rendered first. Based on this, a rendering unit including a frame having a small frame number can be considered to have a high priority, and a rendering unit including a frame having a large frame number can be considered to have a low priority.

In terms of task scheduling, the high-priority rendering unit needs to be handed over to the rendering node with strong rendering capability, and the low-priority rendering unit needs to be handed over to the rendering node with weak rendering capability. In this embodiment, the rendering nodes are sequentially assigned to the rendering nodes from strong to weak according to the priority of the rendering unit from high to low.

The rendering ability of the rendering node can be evaluated and determined by hardware configuration information and real-time load statistics. In the art, the hardware configuration information of the rendering node may mainly include the following parameters: memory size, swap area SWAP size, CPU (Central Processing Unit) core and its processing speed, file system size, network bandwidth, etc. The real-time load statistics can include the following parameters: memory, swap area, CPU and file system occupancy, process, number of threads, network transfer rate, and so on. These parameters are comprehensively scored, and the ability scores of each render node are calculated. For example, the strength of the marked rendering capability is 0 to 100 points, and the rendering node with the highest score indicates that the rendering node has the strongest rendering capability.

In this embodiment, preferably, the W rendering nodes with the strongest rendering ability in the system are selected to participate in the rendering of the W rendering units; if the W rendering nodes participating in the rendering are marked according to the strength of the rendering capability, the rendering capability is the most A strong render node is denoted as render node 1, and so on, and the render node with the weakest rendering ability can be recorded as the render node W. S206. The W rendering units are respectively allocated to the W rendering nodes that have the strongest rendering capability for rendering, and the rendering units composed of the frames with the earlier playing order in the display unit are sequentially allocated to the rendering nodes with stronger rendering capabilities. Rendering, and assigning a rendering unit consisting of a frame having a lower playing order in the display unit to a rendering node having a weak rendering capability for rendering;

It can be understood that the rendering unit composed of the frame with the earlier playing order in the display unit is allocated to the rendering node with stronger rendering ability for rendering; as shown in step S204, the frame given the frame number "1" can be rendered. The unit is denoted as rendering unit 1, and the frame given the frame number "1" can be represented as the first played frame in the display unit to which it belongs, so it can be considered that the rendering unit 1 is assigned to the rendering with the strongest rendering ability. Node 1 is rendered;

Similarly, the rendering unit 2 is assigned to the rendering node 2 for rendering; and so on, the rendering unit composed of the frames with the later playing order in the display unit is assigned to the rendering node with weak rendering ability for rendering, and the frame is to be given The frame of the serial number "W" can be regarded as the last played frame in the display unit to which it belongs, and the rendering unit W is assigned to the rendering node W having the weakest rendering ability for rendering.

Preferably, after the last frame in each display unit is completed, the rendering capability of the rendering node allocated to the rendering unit to which the unfinished frame belongs in the display unit is scored again within the waiting time, and the waiting time is The waiting time in the tolerance time in step S203, wherein the tolerance time is the sum of the waiting time of the system and the time of the frameless compensation after the last frame rendering in one display unit is completed.

If there is a rendering node that does not participate in the task and the score is higher than the rendering node of the currently unfinished rendering task, then the rendering node is replaced, and the new rendering node completes the subsequent frame of the rendering unit originally rendered by the replaced rendering node. Rendering task. If there are multiple rendering nodes that need to be replaced, the rendering nodes of the corresponding rendering capabilities are assigned according to the priority of the rendering unit.

It should be noted that the cloud application-based online rendering method provided in this embodiment is applicable not only to a 3D animation model file that needs to be rendered in advance, but also to real-time calculation of each frame scene by an artificial intelligence or the like. a model file of content and parameters, that is, the model file is obtained by real-time calculation of the information of the frame according to the trigger condition;

For the model file in the first case, according to the total frame number M _l and the size W of the display unit, according to the play order of the frame, the frame may be first divided into several display units, and then decomposed into W rendering units. , assigned to W different rendering nodes for rendering, please refer to Figure 3; For the model file in the second case, according to the total frame number M _l and the size W of the display unit, each frame is calculated in real time according to the play order of the frame, and is sequentially allocated to W rendering units frame by frame, and allocated. Rendering for W different render nodes, see Figure 4. In essence, they are all rendered by a frame that is composed of the same playback order in each unit of the display unit.

The online rendering method based on the cloud application provided by the embodiment, according to the playing sequence of the frame in the model file to be rendered, the model file is divided into several display units, and then the frames with the same playing order in the display unit are rendered. a unit, wherein each rendering unit has a different priority with respect to the playing order, and then is sequentially assigned to a rendering node with a rendering capability from strong to weak according to the priority of the rendering unit from high to low. Not only the task is decomposed before rendering, but also the frame parallel rendering of each display unit is realized, which effectively improves the rendering efficiency, and also performs task scheduling for the decomposition result, and renders the rendering node before assigning the rendering node. Perform strength and weakness assessments to make rendering more reasonable.

The following embodiment uses an online 3D animation design rendering cloud application as an example application scenario, and specifically describes the online rendering method proposed by the present invention.

This embodiment can be a network working situation of a professional animator or a self-oriented self.

3D creative entertainment. After completing a certain animation design, the user needs to preview his own creation effect on the line. At this time, the user sends a request to render the specified animation to the rendering cloud server, and the server can analyze and obtain according to the user's model file/scene file and other materials. The total number of frames, frame rate, resolution, etc. that need to be rendered, and then rendered using the method proposed by the present invention.

Suppose there are 50 available render nodes in the system and the hardware configuration is the same. According to the rendering ability of each rendering node, it is scored and named according to the rendering ability from strong to weak, that is, the rendering node 1 has the highest score, indicating that the rendering capability is the strongest, and the rendering capability of the rendering node 50 is the weakest.

The maximum time to render a frame is max res t^res tiOes), ••• wOes)) is 500ms. It is known that the number of rendering nodes available in the system is 50, and the frame rate f is 24φ _δ . According to the calculation method of W above, 4 It is estimated that after several attempts, the size of the display unit is 24, and the time T _CT and the tolerance time T _{tol of the} difference compression and transmission are 100 ms in total.

For the convenience of calculation and understanding, assuming that the total number of rendered frames in the embodiment is 2400 frames, there are 100 display units, and each display unit is 24 frames, thereby being able to be launched, and the rendering unit is 100, which is required at the same time. There are 24 rendering nodes involved in rendering, and a rendering unit is also 100 frames in size. According to the cloud application-based online rendering method provided by the present invention, reference may be made to the method process diagram of FIG. 3, and the application scenario is applied to the embodiment.

Table I

As can be seen from the data in Table 1, the first, second, second, second, second, and second frames are the first rendering unit; the second, the second, the second, the second, the second and second The first to the 2400th frames are cross-decomposed into 24 rendering units, and are divided into 100 display units of a group of 24 frames in the order in which the frames are played.

By the strength of the rendering capabilities of the aforementioned 50 render nodes, the rendering unit 1 can be assigned to the rendering node 1 for rendering, and so on, and the rendering unit 24 can be assigned to the rendering node 24 for rendering.

If the last frame in the first display unit, that is, after the 24th frame is completed, it is found that the 20th frame has not finished rendering within the waiting time in the tolerance time, the rendering node 20 is re-scored, and the other is not The current scores of the participating render nodes 25-50 are compared. If it is found that the score of the rendering node 25 is higher than the rendering node 20 at this time, then the rendering node 25 is used instead of the rendering node 20, and the subsequent rendering task of the rendering unit 20 is performed, that is, the 44th, 68, 2396 frames are rendered.

In this example, the cloud application-based online rendering method proposed by the present invention divides 2400 frames in the model file into 100 display units according to the playback order of the frames in the model file to be rendered, and then displays the display unit. Frames with the same playback order form a rendering unit to get 24 rendering units. Wherein each rendering unit has a different priority with respect to the playing order, and then according to the rendering unit The priority levels are assigned from high to low, and are sequentially assigned to rendering nodes with strong rendering capabilities, that is, rendering units 1-24 are sequentially assigned to rendering nodes 1-24. In this way, after the user issues a preview request, it can wait up to 600ms to see the continuous and smooth rendered animation.

This embodiment uses the cloud application-based online rendering method proposed by the present invention, which not only decomposes the task before rendering according to the playing order of the frame, but also realizes frame parallel rendering of each display unit, effectively improving the rendering. The efficiency of the task is also scheduled for the result of the decomposition, that is, the higher the priority of the rendering unit, the stronger the rendering ability of the rendering node, making the rendering more reasonable; in line with the real-time requirements of online rendering, high timeliness, users Experience is good.

In addition, the method of the present invention can also be applied to an online 3D game scene under the premise that the speed of network transmission, I/O storage, and rendering can meet the real-time interaction requirements of the game.

When the user triggers a level or a plot, if the 3D animation model file that needs to be rendered can be obtained in advance by the game engine, the method provided by the embodiment of the present invention can be used for task decomposition and scheduling to implement rendering. 3, the specific process can refer to the content of the relevant steps of the method embodiment, which will not be described in detail herein;

If the logic of the game engine needs to perform real-time calculation of each frame scene by the operation of artificial intelligence, as shown in FIG. 4, the specific process can refer to the content of the relevant steps of the method embodiment, which will not be described in detail herein.

Similarly, the embodiment uses the online rendering method proposed by the present invention, which effectively improves the rendering efficiency, meets the requirements of online rendering real-time, has high timeliness, and has a good user experience.

In order to facilitate the implementation of the technical solutions of the embodiments of the present invention, the embodiments of the present invention further provide related devices for implementing the foregoing solutions.

An embodiment of the present invention provides an online rendering device based on a cloud application, and a schematic structural diagram thereof is shown in FIG. 5. The device includes:

The first receiving module 101 is configured to receive a rendering task request.

a first determining module 102, configured to determine, according to a rendering task request, a model file that needs to be rendered; a first decomposition module 103, configured to divide the model file into a display unit according to a playing order of each frame in the model file, where the display unit is a minimum unit of compression and transmission; wherein each display unit includes a plurality of frames adjacent to each other in play order, wherein at least N _r l display units include W Frames

The first combination module 104 is configured to form a frame with the same playing order in each unit of the display unit to obtain a rendering unit, where the playing order of each frame in the display unit is the frame relative to the frame. Display the playing order of other frames in the unit;

The first allocation module 105 is configured to allocate W rendering units to W different rendering nodes for rendering.

The cloud application-based online rendering device provided by the above embodiment divides the model file into a plurality of display units according to the playing order of the frames in the model file to be rendered, and then renders the frames with the same playing order in the display unit. a unit, wherein each rendering unit has a different priority with respect to the playing order, and then each rendering unit is respectively assigned to a different rendering node, that is, the task is decomposed before rendering, and the frame parallel of each display unit is realized. Rendering, try to ensure that they complete rendering at the same time; and each display unit can be rendered in the order of playback time, which effectively improves the rendering efficiency.

Referring to FIG. 6, an embodiment of the present invention further provides an online rendering device based on a cloud application, including: a first receiving module 101, a first determining module 102, a first splitting module 103, a first combining module 104, and an evaluating module 106. The second determining module 107, the first allocating module 105, wherein the first receiving module 101, the first determining module 102, and the first splitting module 103 can be specifically implemented according to the function module function description in the foregoing device embodiment, where Explain in detail.

The first combination module 104 in this embodiment may include:

The first frame numbering sub-module 1041 is configured to assign a frame number of the same frame to the same playing sequence in the unit in each display unit according to the playing sequence of the frames in each display unit;

The first combination sub-module 1042 is configured to group the frames in the respective display units with the same frame number into one rendering unit.

The evaluation module 106 in the embodiment of the present invention is configured to evaluate the strength of the rendering capability of the rendering node by rendering the hardware configuration information of the node and the real-time load statistics.

The second determining module 107 is configured to determine that the W rendering nodes with the strongest rendering capability participate in the rendering of the W rendering units obtained by the first combining module 104.

After the operation of the evaluation module 106 and the second determination module 107 is completed, the first distribution module 105 is instructed to perform an action. At this time, the first distribution module 105 may include: a first allocation sub-module 1051, configured to allocate W rendering units to the W rendering nodes determined by the second determining module 206 to perform rendering;

The second allocation sub-module 1052 is configured to sequentially allocate a rendering unit consisting of a frame with a higher playing order in the display unit to a rendering node with a stronger rendering capability, and perform a rendering of a frame having a lower playing order in the belonging display unit. The rendering unit is assigned to render nodes with weak rendering capabilities for rendering.

It can be understood that the cloud application-based online rendering device provided in this embodiment may be implemented by using the method in the foregoing method embodiment, and the specific implementation process may refer to the related description of the foregoing method embodiment. I will not repeat them here.

The online rendering device based on the cloud application provided by the above embodiment performs the decomposition of the task before rendering, realizes the parallel rendering of each display unit, and ensures that they are simultaneously rendered at the same time; and each display unit can follow The playback time sequence is rendered, which effectively improves the rendering efficiency.

An embodiment of the present invention provides an offline rendering method based on a cloud application. For the method flow, please refer to FIG. 7. The method includes:

S301. Receive a rendering task request.

5302. Determine, according to the rendering task request, a model file that needs to be rendered;

In some embodiments, step S301 and step S302 may be: the user triggers a self-service rendering service, which is equivalent to submitting an offline rendering task request to the background, and determining a model file to be rendered according to the offline rendering task request, or Scene file

The model file may be a self-service rendering or rendering corresponding model file such as a farm, which is not specifically limited herein.

5303. Divide the model file into N ₂ groups according to a play order of each frame in the model file, where at least N ₂ -1 groups include Z frames, where Z is greater than 0 and less than the number of available render nodes; S301 and S302 can know that if the model file to be rendered includes an M ₂ frame; according to the playing order of each frame in the model file, the M ₂ frame in the model file is divided into N ₂ groups; each group includes several playing sequences. adjacent frames, i.e., each group is a collection of frames arranged in order of playback within its frame, and at least N ₂ -l groups contained in Z frames in the N ₂ groups. In some embodiments, the total number of frames in the rendered model file may be set to M ₂ frames and grouped, if the set group contains Z frames, where Z is greater than 0 and less than the number of available rendering nodes, then the group Number of N ₂ =

Wherein, the expression indicates that when M ₂ can be divisible by Z, it is divided into N ₂ groups, and each group contains Z frames; when M ₂ cannot be divisible by Z, the fractional part of the result is discarded. The integer part +1 gives the number of packets as N ₂ , where it can be that the first N ₂ - l groups contain Z frames, and the last group has (M ₂ -( N ₂ - l) x Z ) frame.

S304. The frames in the same playing order in each group are combined into a rendering unit to obtain Z rendering units, where the playing order of each frame in each group is the playing order of the frame relative to other frames in the group to which it belongs. ;

For each group, the first frame played in the group is composed into a rendering unit. Similarly, the second frame played in the group is composed into a rendering unit, and so on, and the Z frames in the group are composed into one frame. Render the unit so that you get Z rendering units.

It should be noted that the intra-group play order of each frame in each group is the play order of the frame relative to other frames in the group to which it belongs;

For example: According to the playing order of each frame in the model file, the first frame played in the first group is the first frame played in the frame of the model file, if the model file N contains ₂ Z groups. Frame, then the last frame played in the group is the Zth frame of all frames of the model file; if the first frame played in the last group is the first frame of the model file ((N ₂ - l) χ Ζ +1) played frames, the last frame played in the group is the N ₂ x Z play frames of all frames of the model file.

Preferably, in some embodiments, it may be considered that a frame number is assigned to a frame in each group according to a play sequence of frames in each group;

For example: For the first frame in each group, the frame number "1" can be given. Correspondingly, for the second frame in each group, the frame number "2" can be given, and so on, for the first in the group. X frames, x=l, 2, 3,... _: Z, can give the frame number "X", for the last frame in each group, that is, the Zth frame, the frame number "Z" can be given, It is possible to obtain a frame number with the same frame sequence in the same playback order in each group. Wherein, the frame having the same playing order in each group is composed of a frame, and the frame with the same frame number in each group is composed into one rendering unit;

For example: For each group, the frame with the frame number "1" is assigned to form a rendering unit. For the sake of understanding, the rendering unit composed of the frame with the frame number "1" can be recorded as the rendering unit 1. For the same reason, The rendering unit composed of the frame number given with the frame number "2" is denoted as the rendering unit 2, and so on, and the rendering unit composed of the frame given the frame number "X" can be recorded as the rendering unit X, and the frame number is given. The rendering unit of the frame can be recorded as a rendering unit, so that the frames with the same playing order in each group can be composed into rendering units to obtain one rendering unit.

For example: According to the playback order of each frame in the model file, the frame given the frame number "1" in the first group is the first frame of the frame of all the model files, which is given the frame number "Ζ". The frame is the next frame played by all frames of the model file, but for the frame in the second group that is given the frame number "Γ" is the first frame played in the second group, but the frame is the entire frame of the model file. The Z+1th frame, the frame to which the frame number "Ζ" is given is the second frame of the frame of all the model files; therefore, for the second and subsequent yth groups, y is greater than 1 and less than An integer of N ₂ whose frame is given the frame number "1" is the ((y-1) xZ+1)-played frame of all the frames of the model file, and the frame to which the frame number "Z" is given is The yxZ-played frames of all frames of the model file, as shown in FIG. 8, frame numbers can be assigned to the frames in each group according to the play order, as shown in the frame, and the order of the frames in the entire model file. Is the number shown on the horizontal axis below; as shown in Figure 8, it can be considered that the frame in the model file is By decomposition to achieve cross-rendering.

S305. Assign Z rendering units to Z different rendering nodes for rendering. It can be understood that each rendering unit needs to be assigned to a rendering node for rendering, so the number of nodes participating in rendering is also Z.

Preferably, it is known from step S304 that a rendering unit composed of frames of the frame number "1" in each group can be recorded as the rendering unit 1, and similarly, a rendering unit composed of frames with the frame number "2" can be composed. It can be recorded as rendering unit 2, and so on, the rendering unit composed of the frame given the frame number "Z" can be recorded as the rendering unit Z; for ease of understanding, the rendering node rendering the rendering unit 1 can be recorded as a rendering node 1. Similarly, the rendering node that renders the rendering unit 2 can be recorded as the rendering node 2, and so on, and the rendering node that renders the rendering unit Z can be recorded as the rendering node Z.

The cloud application-based offline rendering method proposed by the present invention, according to a frame in a model file that needs to be rendered The playing sequence divides the model file into several groups, and then frames the same playing sequence in the group into a rendering unit, and then assigns each rendering unit to a different rendering node, which can make the rendering task decomposition more reasonable. , improve rendering efficiency, thus reducing the rendering time to complete the entire task.

In the following embodiment, an application scenario of a 3D animation model file rendering cloud application is taken as an example, and the offline rendering method based on the cloud application is specifically described.

For model files, there may be problems with the complexity and rendering effects of individual frames, which may result in different rendering times for each frame in an animation. An offline rendering method in the prior art is illustrated. FIG. 9 is a schematic diagram of an existing offline rendering task decomposition manner. If there are a total of M ₂ frames to be rendered, one frame per N ₂ frames is used as a rendering unit. If M ₂ can be divisible by N ₂ , it can be divided into Z rendering units, which are respectively participated by Z rendering nodes. The task, that is, M ₂ = Z x N ₂ , let l ≤ x ≤ Z, where x represents the xth rendering unit or the xth rendering node. When all the shards, ie all rendering units, have been rendered, these sequence images can be packaged for download. Here, the rendering unit (sharding) is generated step by piece according to the time sequence of the frame, and each fragment is randomly selected by the same probability to render the rendering node.

It should be noted that, in this embodiment, the number of rendering nodes participating in the rendering is mainly in the embodiment, and the number of rendering nodes is set to .

If it is assumed that the total number of frames M ₂ to be rendered is 2400, the number of nodes participating in rendering Z is 24, and the size of a rendering unit N ₂ is 100; the rendering node 1 can be set to render rendering unit 1, ie, 1st to 100th Frame, render node 2 renders render unit 2, ie 101st to 200th frames, and so on. Since the complexity and setting of each frame of the animated model file are different, it can be assumed that most of the frames need to be completed every 20s per frame, but the 505th to 600th frames need to be completed for 50s per frame. It can be inferred that the rendering node 1 can complete the rendering task of the allocated rendering unit 1 in 2000s, the rendering node 2 can also complete the rendering task of the allocated rendering unit 2 in 2000s, and so on; however, since the rendering node 6 needs Rendering the 505th to 600th frames requires 20x4+50x96=4880s. Thus, the time required to complete the rendering of the entire animated model file is the maximum rendering time of the rendered rendering unit for each rendering node, ie 4880s. It can be found that at this time, the other 23 rendering nodes complete the task in 2000s, and then all wait for the rendering node 6, so the completion of the entire task is delayed. If the cloud application-based offline rendering method proposed by the present invention is used, reference may be made to FIG. 8. The same 2400 frames are rendered, the number of participating rendering nodes is 24, and the rendering unit size is 100. The decomposition results of its tasks can be referred to Table 2:

Table II

According to the data in Table 2: 1 25 49 73 2353 2377 frame is the first rendering unit, assigned to rendering node 1; 2 26 50 74 2354 2378 frame is the second rendering unit, assigned to rendering node 2; analogy. The 506th to 600th frames, which are thus rendered for a long time, are distributed to all 24 rendering nodes for processing. At this time, the rendering unit 1 rendered by the rendering node 1 includes the four complex frames 505, 529, 553, and 577, so the total time to complete the rendering unit 1 is 20x96+50x4=2120s; likewise, the rendering node The rendering unit 2 of the rendering 2 includes the four complex frames 506, 530, 554, and 578, and the total time is also 2120s, and so on. In this way, all render nodes complete the assigned render unit with 2120s, and the total task time is 2120s. Compared to the 4880s used in the prior art method, the rendering time is greatly reduced.

The offline rendering method based on the cloud application proposed by the embodiment of the invention can make the rendering task decomposition more reasonable, improve the rendering efficiency, and shorten the rendering time of completing the entire task.

In practical applications, online rendering cloud applications also face different rendering time of each frame, but it is less likely to be due to large differences in complexity between adjacent consecutive frames; and tolerance time during rendering It is also set to take into account the effects of small differences that may exist between successive frames of complexity. In addition, in the application scenario of offline rendering, the assumption of rendering time is only to simulate the actual situation in a simple and clear way, and in actual cases, the rendering time of each frame is continuously floating and changing, which is not so simple. split into two. However, this assumption is similar to the actual situation and can fully explain the problems faced.

An embodiment of the present invention provides an offline rendering device based on a cloud application. For a schematic structural diagram, see FIG. 10, the device includes:

a second receiving module 301, configured to receive a rendering task request;

The third determining module 302 is configured to determine, according to the rendering task request, the model file that needs to be rendered; the second decomposition module 303 is configured to divide the model file into N ₂ groups according to a playing order of each frame in the model file, where at least There are Z frames in N ₂ -1 groups, Z is greater than 0 and less than the number of available rendering nodes;

The second combining module 304 is configured to group the frames with the same playing order in each group into the rendering unit to obtain Z rendering units, wherein the playing order of each frame in each group is the group relative to the group to which the frame belongs. The order in which other frames are played;

The second allocation module 305 is configured to respectively allocate Z rendering units to Z different rendering nodes for rendering.

Preferably, referring to FIG. 11, the second combining module 304 in the offline rendering apparatus provided in this embodiment may include:

The second frame numbering sub-module 3041 is configured to assign a frame number of the same frame to the same playing sequence in each group according to the playing sequence of the frames in each group;

The second combination sub-module 3042 is configured to form frames of the same frame number in each group into one rendering unit.

It can be understood that the cloud application-based offline rendering device provided in this embodiment may be implemented in accordance with the method in the foregoing method embodiment, and the specific implementation process may refer to the related description of the foregoing method embodiment. I will not repeat them here.

The offline rendering device based on the cloud application proposed by the embodiment of the invention can make the rendering task decomposition more reasonable, improve the rendering efficiency, and shorten the rendering time of completing the entire task. A person skilled in the art can understand that all or part of the steps of implementing the above method embodiments may be completed by using hardware related to the program instructions. The foregoing program may be stored in a computer readable storage medium, and the program is executed when executed. The foregoing steps include the steps of the foregoing method embodiments; and the foregoing storage medium includes: a medium that can store program codes, such as a ROM, a RAM, a magnetic disk, or an optical disk.

The cloud application-based online rendering method and offline rendering method and related devices provided by the present invention are described in detail above. For those skilled in the art, according to the embodiments of the present invention, in the specific implementation manner and application scope There is a change in the above, and the contents of the present specification should not be construed as limiting the invention.

Claims

Rights request

1. An online rendering method based on a cloud application, comprising:

Receiving a rendering task request;

Determining a model file that needs to be rendered according to the rendering task request;

Dividing the model file into a display unit according to a play order of each frame in the model file, where the display unit is a minimum unit of compression and transmission, wherein each display unit includes a plurality of play sequences adjacent to each other. a frame, wherein at least N _r l display units include W frames;

The W rendering units are respectively assigned to W different rendering nodes for rendering.

The method according to claim 1, wherein the distributing the W rendering units to the W different rendering nodes for rendering before further comprises:

The rendering node's rendering capability is evaluated by rendering the node's hardware configuration information and real-time load statistics. It is determined that the W rendering nodes with the strongest rendering capabilities participate in the rendering of the W rendering units.

The method according to claim 2, wherein the assigning the W rendering units to W different rendering nodes for rendering comprises:

The W rendering units are respectively allocated to the determined W rendering nodes with the strongest rendering capability for rendering, and the rendering units composed of the frames with the earlier playing order in the display unit are sequentially assigned to the rendering capability. The rendering node performs rendering, and the rendering unit composed of the frames with the later playing order in the display unit is allocated to the rendering node with weak rendering ability for rendering.

The method according to claim 1, wherein the frame having the same playing order in the units in each display unit is composed of a rendering unit, and includes:

According to the playing sequence of the frames in each display unit, the same frame number is given to the same frame in the unit in each display unit;

Frames given the same frame number in each display unit constitute one rendering unit.

The method according to any one of claims 1 to 4, wherein the W satisfies: W ≥ |[max(ti(res), t2(res), · .. ti(res), .tw(res)) + TcT(res, B, W, R) + Tt. i] xf| , Where t es) is the time required to render a frame for the assigned render node i, i is an integer from 1 to W; T _CT is the total time for the difference compression and transmission of the W frame; T _tol is the tolerance time, The tolerance time is the sum of the waiting time and the frameless compensation time after the last frame in one display unit is completed; f is the frame rate, the frame rate is the number of frames played per second; res is the resolution, B is the network Speed, R is the difference encoding compression ratio.

6. An online rendering device based on a cloud application, comprising:

a first receiving module, configured to receive a rendering task request;

a first determining module, configured to determine, according to the rendering task request, a model file that needs to be rendered; a first decomposition module, configured to divide the model file into a display unit according to a playing order of each frame in the model file, where The display unit is a minimum unit of compression and transmission; wherein each display unit includes a plurality of frames adjacent to each other in a play order, wherein at least N _r l display units include W frames;

a first combination module, configured to form a rendering unit with the same playback order in each unit of the display unit to obtain a rendering unit; wherein, the playing order of each frame in the display unit is the display relative to the frame to which the frame belongs. The order in which other frames in the unit are played;

a first allocation module, configured to allocate the W rendering units to W different rendering nodes for rendering.

7. The device according to claim 6, wherein the device further comprises:

An evaluation module, configured to evaluate the rendering capability of the rendering node by rendering hardware information of the node and real-time load statistics;

The second determining module is configured to determine that the W rendering nodes with the strongest rendering capability participate in the rendering of the W rendering units obtained by the first combining module.

The device according to claim 7, wherein the first allocating module comprises: a first allocating submodule, configured to respectively allocate W rendering units to the rendering capability determined by the second determining module The strongest W rendering nodes are rendered;

a second allocation sub-module, configured to sequentially allocate a rendering unit consisting of a frame with a higher playing order in the display unit to a rendering node with a stronger rendering capability, and render the frame with a later playing sequence in the belonging display unit. Units are allocated for rendering nodes that have weak rendering capabilities.

The device according to claim 6, wherein the first combination module comprises: a first frame number sub-module, configured to assign a frame number of the same frame to the same playing sequence in each unit of the display unit according to the playing sequence of the frames in each display unit;

The first combination sub-module is configured to form a frame by assigning frames of the same frame number in each display unit.

10. Apparatus according to any one of claims 6 to 9, wherein said W satisfies:

W ≥ |[max(ti(res), t2(res), · .. ti(res), · . . tw(res)) + TcT(res, B, W, R) + Tt. i] xf| , where t es) is the time required to render a frame for the assigned render node i, i is an integer from 1 to W; T _CT is the total time for difference compression and transmission of W frames; T _tol To tolerate time, the tolerance time is the sum of the waiting time and the frameless compensation time after the last frame is rendered in one display unit; f is the frame rate, the frame rate is the number of frames played per second; res is the resolution Rate, B is the network speed, and R is the difference encoding compression ratio.

11. An offline rendering method based on a cloud application, comprising:

Receiving a rendering task request;

According to the playing order of each frame in the model file, the model file is divided into N ₂ groups, wherein at least N ₂ -1 groups include Z frames, and the Z is greater than 0 and smaller than the number of available rendering nodes; The frames in the same playing order are grouped into rendering units to obtain Z rendering units, wherein the playing order of each frame in each group is the playing order of the frame relative to other frames in the group to which it belongs; The Z rendering units are respectively assigned to Z different rendering nodes for rendering.

The method according to claim 11, wherein the frame having the same playing order in each group is composed of a rendering unit, including:

According to the playing sequence of the frames in each group, the same frame number is given to the frames in the same playing sequence in each group;

Frames assigned the same frame number within each group constitute one rendering unit.

13. An offline rendering device based on a cloud application, comprising:

a second receiving module, configured to receive a rendering task request;

Third determining module, based on the rendering model file for a task request determined to be rendered; second decomposition module configured in accordance with the model file playback order of each frame, the model file is divided into N ₂ groups, At least N ₂ -1 groups include Z frames, and the Z is greater than 0 and less than Use the number of render nodes;

And a second allocation module, configured to respectively allocate the Z rendering units to Z different rendering nodes for rendering.

The device according to claim 13, wherein the second combination module comprises: a second frame numbering sub-module, configured to play the same sequence in each group according to a play sequence of frames in each group Frames are given the same frame number;

The second combination sub-module is configured to form a rendering unit by assigning frames of the same frame number within each group.