Search

CN-121658262-B - Multi-application-oriented non-integrated single-engine background rendering method and system

CN121658262BCN 121658262 BCN121658262 BCN 121658262BCN-121658262-B

Abstract

The invention discloses a multi-application-oriented non-integrated single-engine background rendering method and a multi-application-oriented non-integrated single-engine background rendering system, which relate to the technical field of computer graphics background rendering and multi-application graphics resource scheduling and are used for solving the problems of pseudo-change amplification and rendering competition caused by insufficient session isolation, video memory and dynamic change of window caliber under the multi-application non-integrated access condition, establishing a session by an application, submitting an output picture specification, a content description and a resource list, checking and normalizing an engine, generating a stable node identifier and a node fingerprint table, and creating an off-screen frame buffer. The engine forms resource quota according to the video memory and identifies caliber jump pseudo-change, and the submission is gated. And the rendering period is redrawn according to the increment of a change area or a full frame, multi-view multiplexing is supported, and the result is locally synthesized or remotely encoded and output and is updated in an interactive closed loop.

Inventors

  • ZHU ZHENSHUI

Assignees

  • 上海正项信息科技有限公司

Dates

Publication Date
20260512
Application Date
20260206

Claims (10)

  1. 1. The non-integrated single engine background rendering method for the multiple applications is characterized by comprising the following steps of: firstly, receiving a session establishment request to generate a rendering session, checking and then accessing an initial content description and a resource list, creating an off-screen frame buffer and returning a ready mark; Recording an aperture state vector, calculating aperture jump intensity, decomposing normalized content description into an aperture sensitive attribute set and an aperture insensitive attribute set, calculating aperture sensitive node fingerprints and aperture insensitive node fingerprints to obtain a pseudo-change ratio, forming a three-source dislocation coupling abnormal index by combining resource rolling pulse intensity, determining a submitted share, a working thread quota and a pseudo-change suppression rule according to a gating quota and slow-release lifting, and suppressing aperture change amplification; step three, receiving content description increment to obtain a change node set, selecting a region increment redrawing path or a full-frame redrawing path, and generating and submitting a drawing command according to a submitting share; and step four, outputting a frame result after setting the synchronous fence, processing the interaction event record to update the normalized content description, and recalculating the change node set for redrawing in the next rendering period.
  2. 2. The method of claim 1, wherein in the first step, the application submits a session establishment request on the premise that the non-integrated background rendering engine processes itself, generates a rendering session and returns a session identification and a session key after the application identification, the output picture specification and the authentication credentials are checked and authenticated by the background rendering engine, and then the application submits an initial content description and a resource list, the background rendering engine performs structure check and resource reference check on the content description, calculates a content check value and resource list consistency comparison on resource entity byte data, loads resources through the post, and registers session attribution.
  3. 3. The method of claim 2, wherein in the step one, deterministic normalization processing is performed on the content description based on verification passing to fix the representation, then a stable node identifier is derived from the normalized hierarchical relationship and node attributes, and then a session-level baseline fingerprint is written and a node fingerprint table is established to support the determination of a subsequent change node set, and finally an off-screen frame buffer bound with the rendering session is created according to the output picture specification and is marked as ready and returned to a ready mark after the deterministic normalization processing is completed, the stable node identifier derivation is completed, the session-level baseline fingerprint writing is completed, the node fingerprint table is established and the off-screen frame buffer creation is successful.
  4. 4. The method of claim 1, wherein in the second step, the background rendering engine establishes a session rendering context for the rendering session, forcibly limits the access range of the session rendering context to normalized content description under the rendering session name, loaded resource entities and bound off-screen frame buffering, simultaneously creates a command buffer allocation pool and a resource binding table according to a session identification partition, registers an additional session identification for the resource objects, and executes session consistency check before each resource handle analysis and binding and drawing command submission, and refuses cross-session resource use by taking the session identification as a pre-condition.
  5. 5. The method of claim 4, wherein in the second step, after isolation is established, the background rendering engine calculates the basic occupation of off-screen frame buffer and adds up the actual occupation of loaded texture objects, font objects and model objects to form the account book of the video memory of the rendering session based on the normalized representation of the output frame specification, the preset mapping from the pixel format to the number of bytes per pixel, the fixed configuration of the off-screen frame buffer number and the input of the available video memory capacity obtained by the inquiry of the graphic driving interface in the first step, then sums up all the account books of the rendering session with the available video memory capacity, if not, writes the account book into the resource quota table and keeps the output frame specification unchanged, and if not, corrects the effective output frame specification according to the deterministic resolution scaling factor and recalculates the off-screen frame buffer occupation and writes back the resource quota table.
  6. 6. The method of claim 1, wherein in the second step, the background rendering engine records the caliber state vector formed by window width, window height, pixel density and effective output picture specification width and calculates caliber jump intensity in each statistic period, meanwhile, the normalized content description is decomposed into caliber sensitive attribute set and caliber insensitive attribute set according to whether the caliber participation unit conversion and layout normalization are relied on, caliber sensitive node fingerprint and caliber insensitive node fingerprint are calculated respectively, two kinds of change node sets are obtained, and the degree of caliber change regarded as content change is further described by pseudo change ratio.
  7. 7. The method for rendering the non-integrated single engine background facing multiple applications according to claim 6, wherein in the second step, the background rendering engine builds the resource rolling pulse intensity based on the version number or the number of changes of the content check value in the increment of the resource list and the number of checks disagreement or loading refusal, on the basis, a three-source dislocation coupling abnormal index is built by using the caliber jump intensity, the pseudo-change ratio and the resource rolling pulse intensity instead of adopting a conventional weight strategy of pixel scale or fixed weight summation, the burst property of main capture dislocation superposition is obtained through multiplication amplification and extremum, and the conversation is switched to a gating limit or a slow release rise interval by preset threshold and recovery conditions; When the coupling abnormal state is established and the pseudo-change ratio exceeds a threshold value, a pseudo-change suppression rule is further started, only command recording and submission associated with the caliber insensitive change node set is allowed, a large-area request triggered by caliber insensitive change is marked as delay processing, caliber jump is avoided from the source to be amplified into content change competition, the same submission share mechanism is simultaneously mapped to a working thread quota of the central processing unit in a consistent mode, the same submission share mechanism is adjusted to be not lower than a minimum value in the abnormal state, and the released thread quota is preferentially used for content check value comparison, resource loading registration and caliber state vector updating.
  8. 8. The multi-application-oriented non-integrated single-engine background rendering method according to claim 1, wherein in the third step, the background rendering engine receives content description increment with session identification according to rendering period and performs deterministic standardization, extracts a change node set based on node fingerprint difference and generates a change region set by geometric layout thereof so as to cover a route between region increment redrawing and full-frame redrawing, commands generation and resource binding are strictly limited in session rendering context and are subjected to resource binding list and session consistency check, and meanwhile, a graphics processor submits submitted share obtained according to the second step to perform throttling and time budget control, and when multi-session only viewing angles are different and geometric references are highly consistent, the multi-view set is integrated into the multi-view set to generate multi-view output by single drawing multiplexing and is respectively written into each off-screen frame buffer.
  9. 9. The method of claim 1, wherein in the fourth step, the background rendering engine only takes frames from the session off-screen frame buffer and outputs the frames after the synchronization fence is set, the frames are submitted to the operating system synthesizer by the shared surface in the local background mode and are subjected to session consistency check before being submitted, format conversion and low-delay coding are performed in the remote streaming mode, code rate post-attached session identification is determined by combining link feedback, and the feedback interaction event record is applied, the engine maps event coordinates to the effective output picture specification and judges the positioning node by using the stable node identification or hit, and recalculation node fingerprints are generated to change node sets after normalized content description is updated to drive the next period redrawing.
  10. 10. The multi-application-oriented non-integrated single-engine background rendering system is used for realizing the multi-application-oriented non-integrated single-engine background rendering method according to any one of claims 1-9, and is characterized by comprising a rendering session access module, a three-source dislocation coupling gating module, a parallel incremental multiplexing rendering module, an image output interaction closed-loop module and signal connection among the modules; The rendering session access module is used for receiving the session establishment request to generate a rendering session, accessing the initial content description and the resource list after verification, creating an off-screen frame buffer and returning a ready mark; The three-source dislocation coupling gating module is used for recording the caliber state vector and calculating caliber jump intensity, decomposing normalized content description into caliber sensitive attribute set and caliber insensitive attribute set, calculating caliber sensitive node fingerprint and caliber insensitive node fingerprint to obtain pseudo-change ratio, combining resource rolling pulse intensity to form three-source dislocation coupling abnormal index, determining submitted share, working thread quota and pseudo-change suppression rule according to gating quota and slow-release lift, and suppressing caliber change amplification; The parallel increment multiplexing rendering module is used for receiving the content description increment to obtain a change node set, selecting a region increment redrawing path or a full-frame redrawing path, and generating and submitting a drawing command according to the submitting share; The image output interactive closed loop module is used for outputting a frame result after the synchronous fence is set, processing the interactive event record to update the normalized content description, recalculating the change node set for redrawing in the next rendering period.

Description

Multi-application-oriented non-integrated single-engine background rendering method and system Technical Field The invention relates to the technical field of computer graphics background rendering and multi-application graphics resource scheduling, in particular to a non-integrated single-engine background rendering method and system for multiple applications. Background The prior multi-application background rendering adopts an embedded rendering engine or a multi-process multi-engine parallel mode of each application, which is easy to cause process volume expansion, inconsistent version updating and rising of interface adaptation cost, when a multi-session shared graphic processor is arranged on the same terminal, the display memory occupation fluctuation, the step change of window size and pixel density can cause frequent adjustment of resolution, further, the layout normalization aperture change is misjudged as content change, abnormal amplification of a change node set, increment redrawing failure and aggravation of submission competition are caused, and meanwhile, when the resource byte data lacks unified check and session attribution constraint, cross-session resource mixed use, handle error binding and safety boundary break easily occur. The remote streaming scene is also affected by link throughput jitter, the coding rate is difficult to stabilize, and overlay resource rolling update can cause frame rate jitter and picture tearing. The present invention proposes a solution to the above-mentioned problems. Disclosure of Invention In order to overcome the above-mentioned drawbacks of the prior art, embodiments of the present invention provide a multi-application-oriented non-integrated single-engine background rendering method and system, so as to solve the problems set forth in the above-mentioned background art. In order to achieve the above purpose, the present invention provides the following technical solutions: In a preferred embodiment, the method comprises: firstly, receiving a session establishment request to generate a rendering session, checking and then accessing an initial content description and a resource list, creating an off-screen frame buffer and returning a ready mark; Recording an aperture state vector, calculating aperture jump intensity, decomposing normalized content description into an aperture sensitive attribute set and an aperture insensitive attribute set, calculating aperture sensitive node fingerprints and aperture insensitive node fingerprints to obtain a pseudo-change ratio, forming a three-source dislocation coupling abnormal index by combining resource rolling pulse intensity, determining a submitted share, a working thread quota and a pseudo-change suppression rule according to a gating quota and slow-release lifting, and suppressing aperture change amplification; step three, receiving content description increment to obtain a change node set, selecting a region increment redrawing path or a full-frame redrawing path, and generating and submitting a drawing command according to a submitting share; and step four, outputting a frame result after setting the synchronous fence, processing the interaction event record to update the normalized content description, and recalculating the change node set for redrawing in the next rendering period. In a preferred embodiment, in step one, an application submits a session establishment request on the premise that a background rendering engine is not integrated with the application process, completes access by using an application identifier, an output picture specification and an authentication certificate, and generates a rendering session and returns a session identifier and a session key after verification and authentication by the background rendering engine. In a preferred embodiment, in step one, deterministic normalization processing is performed on the content description on the basis of verification to fix representation, then a stable node identifier is derived from the normalized hierarchical relationship and node attributes, and then a session-level baseline fingerprint is written in and a node fingerprint table is established to support determination of a subsequent change node set, and finally an off-screen frame buffer bound with the rendering session is created according to the output frame specification and the rendering session is marked as ready and returned to a ready mark after completion of the deterministic normalization processing, completion of the stable node identifier derivation, completion of session-level baseline fingerprint writing, completion of node fingerprint table establishment and successful creation of the off-screen frame buffer. In a preferred embodiment, in step two, the background rendering engine establishes a session rendering context for the rendering session, forcibly limits the access scope of the session rendering context to the normalized content description under the renderin