A dynamic parallax rendering and optimization method

Abstract

The application provides a dynamic parallax rendering and optimization method, and relates to the technical field of parallax rendering optimization, which comprises the following steps: loading a visual response prediction model based on psychophysical experiment calibration; a rendering engine submits a predicted rendering completion time and a scene dynamic degree in real time; a visual perception synchronous controller calculates an optimal target display trigger time and a waveform modulation parameter set of a current frame according to the model, the scene dynamic degree and a feedback actual display time of a last frame; a display driving circuit drives a pixel to emit light at an accurate time according to a modulated voltage waveform, and dynamically compensates for waveform deviation through high-frequency monitoring during the period; the system measures and feeds back an actual photoelectric delay of each frame, which is used for online calibration of model parameters; and the application synchronizes the final light signal sequence with the human eye visual processing rhythm by prospectively coordinating the rendering end time and accurately modulating the light pulse characteristics, so that the stereoscopic realism, dynamic fluency and visual fatigue are significantly improved at the perception level.

Description

Technical Field

[0001] This invention relates to the field of parallax rendering optimization technology, and in particular to a dynamic parallax rendering and optimization method. Background Technology

[0002] In the fields of visual computing and graphics rendering, achieving a realistic 3D immersive experience has always been a core pursuit. Dynamic parallax rendering technology, which simulates the changes in depth and motion visuals caused by binocular parallax and head movements, is the cornerstone of building this immersive experience. From early geometry-based parallax mapping to today's mainstream screen-space rendering technology, its evolution has always revolved around a core contradiction: the conflict between the infinitely increasing scene complexity, display resolution, and limited real-time computing power.

[0003] Traditional optimization paths mainly fall into two categories. The first is a geometry-based simplification path, which uses techniques like parallax mapping and Level of Detail (LOD) models to sacrifice some geometric accuracy during preprocessing or runtime in exchange for rendering speed. The second is a picture-space-based efficiency path, such as screen-space ambient occlusion and reflections. Its advantage lies in its ability to quickly generate visual effects using current frame buffer information, but it is powerless against off-screen information and is prone to distortion and artifacts in complex occlusion, transparency, or reflection scenes. Both paths aim to compress computational overhead in their respective dimensions to maintain high frame rates. However, as hardware performance approaches its physical limits, a more fundamental bottleneck gradually emerges: a systemic mismatch between rendering, display, and perception. To compensate for unavoidable rendering latency, the industry has introduced post-processing techniques such as asynchronous time warp; display hardware, on the other hand, continuously pursues higher refresh rates and faster pixel response times. These efforts implicitly share a common premise: that it is feasible to treat the rendering pipeline, signal modulation, and display response as a segmentable, linearly optimized pipeline. This premise overlooks the fact that the human visual system is not a passive, linear receiver, but an active perceptual system with complex physiological feedback. The "ideal" parallax information calculated by the rendering engine, during its conversion into physical light signals via the display driver circuit, undergoes unpredictable micro-temporal jitter and waveform distortion. This distorted light signal then becomes frequency-dissonant with the unique electrochemical processing rhythm of the human visual nervous system for motion depth information. This disharmony is a novel systemic incoordination resulting from the coupling of dynamic characteristics across three different domains: rendering, display, and perception. The resulting experiential defects are subtle—even with top-tier hardware specifications such as frame rate and refresh rate, users will subconsciously perceive a stiff or artificial sense of depth in fast-moving scenes and are more prone to visual fatigue. Traditional methods repeatedly optimize within each domain but struggle to address this problem rooted in the domain's interactive interface. Therefore, a dynamic parallax rendering and optimization method is urgently needed to at least address the aforementioned shortcomings. Summary of the Invention

[0004] The purpose of this invention is to provide a dynamic parallax rendering and optimization method to solve problems such as visual defects, stiff experience, and visual fatigue caused by dynamic mismatch between rendering, display, and visual perception systems in existing technologies. The essence of these problems is that existing technologies employ divide-and-conquer and linear superposition optimization methods, which cannot adapt to the collaborative rhythm of the human visual system as a nonlinear active perception system. The specific technical solution of this invention is as follows: This invention provides a dynamic parallax rendering and optimization method, comprising: S10, Obtain and load the pre-built visual response prediction model; S20: When the dynamic parallax rendering engine starts rendering the current frame, it submits the predicted rendering completion timestamp of the current frame to the visual perception synchronization controller and calculates the dynamic parameters of the current scene. S30, the visual perception synchronization controller receives the prediction rendering completion timestamp and dynamic parameters, receives the actual line scan end timestamp of the previous frame, and calculates the optimal target display trigger time and waveform modulation parameter set for the current frame based on the visual response prediction model. S40 receives pixel data from the final frame buffer, waits for the clock to reach the optimal target display trigger time, and drives the pixel units of the display panel to emit light according to the voltage timing and waveform defined by the waveform modulation parameter set. S50 captures the actual waveform of the driving voltage in real time during the display driving process and compares it with the preset waveform in the waveform modulation parameter set. If the deviation between the real-time voltage and the preset curve is detected to exceed 1.8% of the calibrated voltage during the voltage ramp-up stage, dynamic compensation is performed by adjusting the input code value of the digital-to-analog converter. S60 measures the end-to-end photoelectric delay from the issuance of the drive command to the actual generation of photons, and feeds the measured value back to the visual perception synchronization controller for online calibration of the expected photoelectric delay parameters of the display hardware in the visual response prediction model.

[0005] Furthermore, the visual response prediction model in S10 defines a transfer function from the display light pulse to the visual cortex to generate stable motion perception, with parameters including the visual integration time window, the absolute perception delay reference, and the expected photoelectric delay constant of the display hardware.

[0006] Furthermore, the pre-establishment method for the visual response prediction model includes: presenting a standard test pattern at a standard observation distance of 0.7 meters and an ambient illuminance of 200 lux, recording the threshold, collecting eye-tracking data and motor cortical potentials in electroencephalogram signals, and establishing an original dataset; Based on the original dataset, the functional expression of the visual response prediction model is obtained through nonlinear regression fitting. The data structure, which includes a lookup table of basic delay functions, dynamically adjusted function coefficients, and initial hardware parameters, is compiled. During system initialization, the visual perception synchronization controller loads the configuration file from non-volatile memory to complete the model establishment and readiness.

[0007] Furthermore, the method for calculating the predicted rendering completion timestamp includes: at the start of frame rendering, the dynamic parallax rendering engine first analyzes the command list and resource status of the current frame, calculates the rendering time of this frame, and the predicted rendering completion timestamp is obtained by adding the rendering time of this frame to the current system clock; the value of the scene dynamics parameter is the ratio of the maximum displacement of pixels in the scene between two consecutive frames to the length of the screen diagonal; when calculating the scene dynamics parameter, the average acceleration estimate of moving objects in the scene is introduced to correct the scene dynamics parameter; when submitting the predicted rendering completion timestamp, the dynamic parallax rendering engine submits the rendering complexity estimate of the current frame, which is obtained by normalizing the weighted sum of the number of triangles in the scene, the amount of lighting calculation, and the number of parallax texture samplings.

[0008] Furthermore, the S30 also includes: The controller reads the actual line scan end timestamp of the previous frame and the target trigger time set in the previous frame, calculates the drift error, and uses the error value for feedback correction; Based on the dynamic parameters of the current frame, query the basic delay mapping table to obtain the basic delay compensation value and obtain the basic target time; Feedback and dynamic adjustments are made to calculate the final optimal target display trigger time: the base target time is corrected by a proportional-integral controller to eliminate systematic timing drift; The controller queries the waveform parameter mapping table based on the scene dynamic parameters and the user fatigue estimate to determine the waveform modulation parameter set; the waveform modulation parameter set is a structure that includes the initial voltage ratio, the target stable voltage ratio, the voltage ramp-up time, and the overshoot voltage ratio.

[0009] Furthermore, the S40 also includes: The waveform modulation parameter set is received, a matching voltage-time reference curve is selected, and after discretization, it is loaded into the waveform buffer of the high-speed digital-to-analog converter. Before the optimal target display trigger time arrives, the output voltage of the digital-to-analog converter is preset to the starting voltage level. When the system clock reaches the optimal target display trigger time, a high-resolution timer triggers the waveform generator to drive the digital-to-analog converter to output voltage, thereby controlling the charging current of the pixel capacitor and generating a drive voltage waveform. During the voltage stabilization phase, the junction temperature is monitored in real time by a temperature sensor inside the driver chip. If the junction temperature exceeds 85 degrees Celsius, the target stable voltage is automatically lowered, and the adjustment information is fed back to the controller in real time.

[0010] Furthermore, the S50 also includes: During the voltage ramp-up phase, the high-frequency monitoring circuit compares the real-time voltage with the preset voltage. When the deviation between the real-time voltage and the preset voltage exceeds 1.8% of the calibrated voltage, the integral value of the voltage error is calculated; Based on the integral value of the voltage error and the current differential error, the gate voltage control word of the subsequent driving transistor is adjusted by looking up the compensation table to perform real-time waveform shaping.

[0011] Furthermore, the S60 also includes: At the start of the vertical blanking period for each frame or every set number of frames, the test signal generation module in the display driver circuit sends a drive pulse to a dedicated test light-emitting diode integrated in the non-display area of ​​the display panel, and the driver circuit records the precise timestamp of the pulse transmission. The light pulses emitted by the test LED are received by a photoelectric sensor, converted and shaped to obtain digital pulses; Record the precise timestamp of the arrival of the rising edge of the digital pulse, calculate the measured value of the end-to-end photoelectric delay, send the measured value to the visual perception synchronization controller, and update the expected photoelectric delay parameters of the display hardware in the visual response prediction model through the calibration formula.

[0012] Furthermore, if the corrected scene dynamics parameter is greater than 0.7 and there are objects moving laterally in the scene, a high-motion-enhanced waveform is used. The specific steps include: Set the starting voltage to 10% of the rated voltage and the target voltage to 100% of the rated voltage; Set the voltage ramp-up time to 0.9 microseconds; The overshoot voltage ratio is set to 8.5%, the overshoot duration is 0.3 microseconds, and the voltage decays to the target stable voltage within 0.5 microseconds. The waveform parameter set is dynamically applied to the display area where the moving object is located.

[0013] Furthermore, if the estimated user fatigue level is greater than 0.65, a low-fatigue-relieving waveform is used; if the corrected scene dynamics parameter is less than 0.3 and the estimated rendering complexity is greater than 0.6, a high-quality energy-saving waveform is used. The specific steps for creating a low-fatigue-relieving waveform include: setting the initial voltage to 15% of the rated voltage; setting the voltage rise time to 2.2 microseconds; setting the overshoot voltage ratio to -2%, meaning that before the voltage rises to the target stable voltage, it first reaches 102% of the rated voltage and then falls back to 100%; and applying the waveform globally to the entire display screen. The specific steps for achieving a high-quality energy-saving waveform include: setting the initial voltage to 20% of the rated voltage; setting the voltage ramp-up time to 2.8 microseconds; setting the overshoot voltage ratio to 0%; and during the voltage stabilization phase, automatically reducing the target stable voltage to 95% of the rated voltage after the duration exceeds 2 milliseconds.

[0014] Furthermore, an electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method.

[0015] Furthermore, a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method.

[0016] The beneficial effects of this invention are as follows: This invention, through a visual perception synchronization controller, incorporates the response characteristics of the human visual system, graphics rendering latency, and the photoelectric conversion characteristics of display hardware into a unified closed-loop control framework. It coordinates the rendering completion timing in a look-ahead manner and precisely modulates the drive signal, achieving dynamic synchronization among "perception-rendering-display," fundamentally solving the problem of cross-domain dynamic detuning. Its effect lies in ensuring that the light pulse sequence entering the human eye is synchronized with the visual processing rhythm from the signal source, thereby significantly improving the sense of stereoscopic realism and motion smoothness at the perception level, and reducing visual fatigue caused by prolonged use. Furthermore, the method enhances the system's adaptability to different scene dynamics, rendering complexity, and user states through online calibration and adaptive waveform modulation.

[0017] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of a dynamic parallax rendering and optimization method in an embodiment of the present invention. Detailed Implementation

[0019] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0020] This embodiment provides a dynamic parallax rendering and optimization method, such as Figure 1 As shown, it includes: S10, Obtain and load the pre-built visual response prediction model; S20, when the dynamic parallax rendering engine starts rendering the current frame, it submits the predicted rendering completion timestamp of the current frame to the visual perception synchronization controller and calculates the dynamic parameters of the current scene. S30, the visual perception synchronization controller receives the prediction rendering completion timestamp and dynamic parameters from S20, and simultaneously receives the actual line scan end timestamp of the previous frame from the display driving circuit. Based on the visual response prediction model, it calculates the optimal target display trigger time and the corresponding waveform modulation parameter set for the current frame. S40, the display driving circuit receives the pixel data of the final frame buffer, waits for the system clock to reach the optimal target display trigger time determined in S30, and drives the pixel units of the display panel to emit light according to the voltage timing and waveform defined by the waveform modulation parameter set. S50, during the display driving process, the high-frequency monitoring circuit built into the driver chip captures the actual waveform of the driving voltage in real time at a sampling rate of 1 MHz and compares it with the preset waveform in the waveform modulation parameter set. If the deviation between the real-time voltage and the preset curve is detected to exceed 1.8% of the calibrated voltage during the voltage ramp-up stage, dynamic compensation is performed by adjusting the input code value of the digital-to-analog converter. S60, after each frame display process is completed, the system measures the end-to-end photoelectric delay from the issuance of the driving command to the actual generation of photons through an integrated photoelectric sensor located at the edge of the display panel, and feeds back this measured value to the visual perception synchronization controller for online calibration of the display hardware expected photoelectric delay parameter T_photon_expected in the visual response prediction model.

[0021] The principle and effects of the above technical solution are as follows: This invention introduces a visual perception synchronization controller as the core hub, and for the first time incorporates the bioelectrochemical response characteristics of the human visual system, the logical latency of graphics rendering, and the electro-optical conversion characteristics of display hardware into a unified closed-loop control framework. This method proactively uses the predicted neural response rhythm as a benchmark to proactively coordinate the timing of rendering completion and precisely modulate the waveform and timing of the driving electrical signal, achieving dynamic synchronization of the three domains of "perception-rendering-display," fundamentally solving the deep-seated pain point of "cross-domain dynamic detuning" in the industry. Its effect is that it ensures from the signal source that the light pulse sequence ultimately entering the human eye is synchronized with the visual processing rhythm, thereby producing an unexpected leap in stereoscopic realism and smoothness at the perception level, and reducing visual fatigue caused by prolonged use.

[0022] In one specific embodiment, the visual response prediction model defines a transfer function from the display light pulse to the visual cortex to generate stable motion perception. Its parameters include a visual integration time window T_window (e.g., 15.2 ms), an absolute perception delay baseline T_delay_base calibrated through psychophysical experiments (e.g., 35.5 ms), and a display hardware expected photoelectric delay constant T_photon_expected. The display hardware expected photoelectric delay constant represents the current estimate of the system hardware delay from the issuance of the driving command to the actual emission of the pixel. Its initial value is based on a hardware calibration preset (e.g., 3.2 ms) and will be continuously calibrated through online measurements during system operation.

[0023] In one specific embodiment, the acquisition and loading of the pre-established visual response prediction model in step S10 is implemented as follows: S10, Establish and load the visual response prediction model. This model is a mathematical framework for predicting the overall system delay and perceptual quality from the emission of a display light signal to the generation of stable motion perception in the visual cortex. Its core consists of a perceptual transfer function library with adjustable parameters and a delay prediction table based on scene feature queries. The pre-establishment method for the model includes: S11, Model Parameter Calibration Experiment: At a standard viewing distance of 0.7 meters and an ambient illuminance of 200 lux, a series of standard test patterns were presented to 20 subjects with normal or corrected-to-normal vision using a reference display with a response time of less than 0.1 milliseconds. These patterns included sinusoidal gratings moving at different spatial frequencies (0.1 to 5.0 cycles per degree) and motion speeds (0 to 30 degrees per second). For each test pattern, the timing parameters of its displayed light pulses were systematically varied, including: pulse onset phase, minute jitter in the inter-frame interval (range -3.0 to +3.0 milliseconds), and the steepness of the pulse rise edge (10% to 90% rise time varying between 0.5 and 3.0 microseconds). A raw dataset was established by recording the threshold of "clear, stable, and jitter-free motion perception" reported by the subjects, and simultaneously acquiring their eye-tracking data and motion-related cortical potentials in EEG signals.

[0024] S12, Model Function Construction: Based on the original dataset in S11, the functional expression of the visual response prediction model is obtained through nonlinear regression fitting. The model output is the predicted optimal perceptual display time offset. The calculation formula is: ,in The base latency function is obtained by querying a two-dimensional lookup table, which takes the predicted rendering completion timestamp T_render_pred and the scene dynamics D_scene as input. For example, when T_render_pred is stable at 12.5 milliseconds and D_scene is 0.8, the base latency value output by the lookup table is 8.2 milliseconds. This table is generated by summarizing the calibration experimental data. The function is dynamically adjusted and is a linear combination of the head angular velocity ω_head and the rendering complexity C_complex. Where α is the head motion compensation coefficient, with an empirical value of 0.05 milliseconds / (degrees / second); β is the complexity delay penalty coefficient, with an empirical value of 0.5 milliseconds (when the normalized value of C_complex increases by 0.1). T_photon_expected is the expected photoelectric delay constant of the display hardware, with an initial preset value of 3.2 milliseconds, and is updated online during system operation via step S60. This is a personalized offset, learned based on the user's initial calibration or long-term usage habits, with an initial value of 0.

[0025] S13, Model Integration and Loading: This includes the above-mentioned... Function lookup table The data structures for the function coefficients (α, β) and initial hardware parameters are compiled into a configuration file or embedded firmware. During system initialization, the visual perception synchronization controller loads this configuration file from non-volatile memory to complete the model establishment and readiness.

[0026] In one specific embodiment, S20, when the dynamic parallax rendering engine starts rendering the current frame, it submits the predicted rendering completion timestamp T_render_pred of the current frame to the visual perception synchronization controller, and calculates the current scene dynamics D_scene. The value of scene dynamics D_scene is the ratio of the maximum displacement of all pixels in the scene between two consecutive frames to the length of the screen diagonal, which is between 0 and 1. The specific implementation of calculating the scene dynamics D_scene and submitting the prediction rendering completion timestamp T_render_pred in step S20 is as follows: S21, Calculation of the predicted rendering completion timestamp T_render_pred: At the start of frame rendering, the dynamic parallax rendering engine first analyzes the command list and resource status of the current frame, and calculates the rendering time T_render_estimate for this frame using the following formula: Where N_poly is the total number of triangles to be processed in this frame, K1 is the baseline throughput of the rendering pipeline for processing triangles, with an empirical value of 1,200,000 triangles / millisecond; N_pixel is the total number of pixels in this frame, C_shader is the average arithmetic logic complexity of the pixel shader in this frame (ranging from 1.0 to 5.0), K2 is the baseline computation rate of the shader core, with an empirical value of 800,000 pixels / millisecond / unit of complexity; T_fixed is the fixed drive and synchronization overhead, with an empirical value of 0.3 milliseconds. Subsequently, the predicted rendering completion timestamp is obtained by adding the rendering time of this frame to the current system clock, using the formula: .

[0027] S22, Calculation of scene dynamics D_scene: The rendering engine obtains the depth buffers of the previous and current frames and calculates the displacement vector of each pixel in the world coordinate system. The dynamics parameter D_scene is calculated by the following formula: .in, D_diagonal is the maximum displacement of all pixels between two consecutive frames, in meters; D_avg is the physical length of the diagonal of the display screen, in meters; A_avg is the scalar mean of the average acceleration of all moving objects in the scene, in meters per second squared. These calculations are performed in a dedicated hardware motion analysis module, updated once per frame.

[0028] In one specific embodiment, the calculation of the optimal target display trigger time T_target and the corresponding waveform modulation parameter set W_mod based on the visual response prediction model in step S30 is specifically implemented as follows: S31, Calculate the timing drift error E_drift: The controller reads the actual row scan end timestamp T_scan_prev_actual of the previous frame and the target trigger time T_target_prev set in the previous frame, and calculates the drift error: E_drift = T_scan_prev_actual - T_target_prev. This error value is used for feedback correction.

[0029] S32, Calculate the base target time T_target_base: Based on the dynamic parameter D_scene of the current frame, query the preset base delay mapping table to obtain the base delay compensation value T_comp_base. The mapping table is as follows: if D_scene < 0.3, then T_comp_base = 11.5 milliseconds; if 0.3 ≤ D_scene < 0.7, then T_comp_base = 9.0 milliseconds; if D_scene ≥ 0.7, then T_comp_base = 7.0 milliseconds. Therefore, T_target_base = T_render_pred + T_comp_base.

[0030] S33, perform feedback and dynamic adjustments, and calculate the final T_target: The final target display trigger time T_target is calculated using the following formula: Where Kp is the scaling factor, with a value of 0.7; Ki is the integral factor, with a value of 0.1; and ΣE_drift is the cumulative sum of E_drift over the past 5 frames. This step constitutes a scaling-integral controller used to eliminate systematic timing drift.

[0031] S34, Determine the waveform modulation parameter set W_mod: The controller queries the waveform parameter mapping table based on D_scene and the user fatigue estimate U_fatigue to determine W_mod. W_mod is a structure containing the following parameters: starting voltage ratio V_start (%), target stable voltage ratio V_target (%), voltage rise time T_rise (microseconds), and overshoot voltage ratio O_vershoot (%). The mapping is represented as follows: If D_scene ≥ 0.7 and U_fatigue < 0.6, then W_mod = {V_start: 10%, T_rise: 0.9, O_vershoot: 8.5}; if D_scene < 0.3, then W_mod = {V_start: 20%, T_rise: 2.8, O_vershoot: 0}.

[0032] In one specific embodiment, the process of driving the pixel units of the display panel to emit light according to the voltage timing and waveform defined by the W_mod parameter set in step S40 is implemented as follows: S41, Waveform Parameter Loading and Digital-to-Analog Converter Preset: The programmable waveform generator in the display driver circuit receives the W_mod parameter set. The waveform generator internally stores multiple predefined voltage-time reference curves. Based on the T_rise and O_vershoot values ​​in W_mod, the best-matching reference curve is selected and discretized into 512 control points, which are then loaded into the waveform buffer of the high-speed digital-to-analog converter. Two microseconds before the arrival of T_target, the output voltage of the digital-to-analog converter is preset to the starting voltage level defined by V_start.

[0033] S42, Precise Timing Trigger and Waveform Generation: A high-resolution timer with an accuracy higher than 0.1 microseconds within the system sends a trigger signal to the waveform generator when the system clock reaches T_target. The waveform generator immediately begins driving the digital-to-analog converter output voltage according to a pre-loaded control point sequence. This output voltage serves as a reference voltage for the source drive circuit, controlling the charging current of the pixel capacitor, thereby precisely generating a drive voltage waveform with specific rise time and overshoot characteristics as defined by W_mod.

[0034] In one specific embodiment, if the deviation between the real-time voltage and the preset curve is detected to exceed 1.8% of the calibrated voltage during the voltage ramp-up phase, dynamic compensation is performed by adjusting the input code value of the digital-to-analog converter within the next 5 microseconds.

[0035] In one specific embodiment, the online calibration of the display hardware expected photoelectric delay parameter T_photon_expected in the visual response prediction model is performed with a calibration weight coefficient of 0.05. The measurement of the end-to-end photoelectric delay T_photon_actual from the issuance of the driving command to the actual generation of photons in step S60 is specifically implemented as follows: S61, Generate and send a test light pulse: At the beginning of the vertical blanking period of each frame or every N frames, the test signal generation module in the display driver circuit sends a drive pulse to a dedicated test LED integrated in the non-display area of ​​the display panel. The amplitude of the pulse is the same as the data voltage, and the duration is 0.1 milliseconds. The driver circuit records the precise timestamp T_send of the pulse transmission.

[0036] S62, Receiving and Converting Optical Signals: A photoelectric sensor (such as a photodiode) located 5 mm away from the test LED receives the light pulses emitted by the test LED. The weak current signal generated by the photoelectric sensor is converted into a voltage signal by a transimpedance amplifier, and then shaped by a high-speed comparator (hysteresis voltage 10 mV) to obtain a digital pulse with steep edges.

[0037] S63, Capture Timestamp and Calculate Delay: The digital pulse is input to a high-precision time-to-digital converter module, which records the precise timestamp T_receive of the pulse's rising edge arrival. The end-to-end photoelectric delay T_photon_actual is calculated using the following formula: T_photon_actual = T_receive - T_send. The time resolution of the time-to-digital converter is 0.05 microseconds, with a measurement error within ±0.1 microseconds. This measured value is then sent to a visual perception synchronization controller for online calibration of model parameters.

[0038] In one embodiment, S10 further includes: S101 uses psychophysical experiments to calibrate the basic model parameters. The experiment is conducted at a standard observation distance of 0.7 meters. A motion grating with a spatial frequency of 0.5 cycles per degree is presented to the subject. Its brightness is modulated in a variety of combinations with frequencies from 30 Hz to 120 Hz and waveform rise times from 0.5 microseconds to 3 microseconds. The visual evoked potential signals are recorded simultaneously. S102, by analyzing the latency changes of the N75 and P100 components in the VEP signal, empirical formulas for the perception delay and stimulus frequency f and voltage rise time T_rise were obtained through fitting: This formula is then integrated into the visual response prediction model for dynamic fine-tuning of T_delay_base.

[0039] The principle and effect of the above technical solution are as follows: This step ensures that the visual response prediction model is not a fixed theoretical value, but a dynamic model constructed based on actual physiological measurement data and containing nonlinear response laws. By introducing empirical formulas related to stimulus characteristics, the model can more accurately predict the actual processing delay of the human visual system under different display content, providing a scientific and personalized computational basis for subsequent accurate synchronization, enhancing the method's adaptability and the realism of the final synchronization effect.

[0040] In one embodiment, S20 further includes: S201, when the dynamic parallax rendering engine submits T_render_pred, it also submits the rendering complexity estimate C_complex for the current frame. This value is obtained by normalizing the weighted sum of the number of triangles to be processed in the scene, the amount of lighting calculation, and the number of parallax texture samplings. S202, when calculating the scene dynamics D_scene, not only pixel displacement is calculated, but also the average acceleration estimate A_avg of moving objects in the scene is introduced. The final D_scene is corrected as follows: , where A_avg is in meters per second squared.

[0041] The principle and effect of the above technical solution are as follows: This step provides richer and more forward-looking scene context information. The rendering complexity C_complex helps the controller predict the risk of rendering time fluctuations; while the introduction of acceleration correction D_scene' can more sensitively capture the changing trend of motion state (such as sudden acceleration or deceleration), enabling the controller to make more aggressive or conservative synchronization strategy adjustments in advance, thereby maintaining stable perceptual synchronization in dynamic scenes and avoiding synchronization failure caused by prediction lag.

[0042] In one embodiment, S30 further includes: S301, the visual perception synchronization controller queries a preset delay compensation mapping table based on the scene dynamics parameter D_scene' of the current frame to determine the corresponding basic delay compensation value T_comp_base. For example, when D_scene' ∈ [0, 0.3), T_comp_base = 11.5 milliseconds; when D_scene' ∈ [0.3, 0.7), T_comp_base = 9.0 milliseconds; when D_scene' ∈ [0.7, 1.0], T_comp_base = 7.0 milliseconds. Subsequently, the basic target time T_target_base is calculated according to the formula T_target_base = T_render_pred + T_comp_base; S302, based on the feedback T_scan_prev_actual and the previous frame's T_target_prev, calculate the timing drift error E_drift = T_scan_prev_actual - T_target_prev, and use first-order proportional-integral adjustment to correct T_target_base with a proportional coefficient Kp=0.7 and an integral coefficient Ki=0.1 to obtain T_target_adjusted; S303, based on D_scene' and C_complex, selects a waveform modulation parameter set W_mod from the preset waveform library, which includes: starting voltage V_start, target stable voltage V_target, voltage rise time T_rise, and overshoot voltage ratio O_vershoot.

[0043] The principle and effects of the above technical solution are as follows: This step reveals the specific algorithm by which the controller makes intelligent decisions. By combining table lookup with closed-loop feedback control, it ensures rapid response to different scenario modes while eliminating systematic timing deviations through the integral stage, thus ensuring long-term synchronization stability. Directly linking scenario characteristics with waveform parameters provides clear instructions for subsequent precise electrical signal modulation, which is a key step in achieving cross-domain collaboration.

[0044] In one embodiment, S40 further includes: S401, 2 microseconds before the arrival of time T_target, the driving circuit presets the start voltage of the row scanning circuit to V_start; S402, at time T_target, triggers the waveform generator so that its output voltage rises from V_start to V_target according to T_rise defined in W_mod, and strictly follows the overshoot curve defined in W_mod; During the voltage stabilization phase, the S403 monitors the junction temperature T_junction in real time using a temperature sensor within the driver chip. If T_junction exceeds 85 degrees Celsius, it automatically lowers V_target by 0.5% and feeds this adjustment information back to the controller in real time.

[0045] In one embodiment, S50 further includes: S501, the high-frequency monitoring circuit compares the real-time voltage V_actual(t) with the preset voltage V_preset(t) every 0.5 microseconds during the voltage ramp-up phase; S502, when detected Immediately calculate the integral value of the voltage error E_int; S503, in the next 5 microseconds, adjusts the gate voltage control word of the subsequent driving transistor directly by looking up the compensation table based on E_int and the current error derivative, and performs real-time waveform shaping.

[0046] In one embodiment, S60 further includes: S601, at the beginning of the vertical blanking period of each frame, the driving circuit sends a driving pulse with a duration of 0.1 milliseconds to the test LED next to the integrated photoelectric sensor; S602, the photoelectric sensor records the timestamp of the received light signal, and calculates the difference between the timestamp of the drive pulse and the timestamp of the received light signal to obtain the end-to-end photoelectric delay T_photon_actual of the current frame, with a measurement accuracy of 0.05 milliseconds; S603, Online calibration of hardware latency parameters: The controller updates the expected photoelectric latency parameters of the display hardware in the visual response prediction model using the following formula. = Where λ is the calibration weight coefficient (e.g., 0.05), T_photon_expected_new is the updated expected value, and T_photon_expected_old is the original expected value. Through this online learning process, the model's estimate of hardware latency continuously approximates the measured value.

[0047] In one specific embodiment, in S40, if D_scene' > 0.7 and there are objects moving laterally at high speeds in the scene, then a "high-motion enhancement waveform" is used, and the specific steps are as follows: SA41: Set the starting voltage V_start to 10% of the rated voltage and the target voltage V_target to 100%.

[0048] SA42: Set the voltage ramp-up time T_rise to 0.9 microseconds to produce a steep rising edge.

[0049] SA43: Set the overshoot voltage ratio O_vershoot to 8.5%, the overshoot duration to 0.3 microseconds, and then decay to the target stable voltage V_target within 0.5 microseconds.

[0050] SA44: Dynamically apply this waveform parameter set W_mod_high to the display area where the moving object is located.

[0051] The high-motion enhancement waveform generates a light pulse with a sharp leading edge and slight overshoot, which resonates with the response characteristics of M-type ganglion cells in the human visual system that are highly sensitive to the initiation of motion, thereby enhancing the three-dimensional prominence of moving objects and the instantaneous depth perception at the physical signal level.

[0052] In S40, if D_scene' < 0.3 and C_complex > 0.6, then a "high-quality energy-saving waveform" is adopted. The specific steps for the high-quality energy-saving waveform are as follows: SB41: Set the starting voltage V_start to 20% of the rated voltage.

[0053] SB42: Set the voltage ramp-up time T_rise to 2.8 microseconds to achieve a smooth and slow voltage rise.

[0054] SB43: Set the overshoot voltage ratio O_vershoot to 0% to avoid any unnecessary power consumption.

[0055] SB44: During the voltage stabilization phase, if the duration exceeds 2 milliseconds, automatically reduce V_target to 95% of the rated voltage.

[0056] The high-quality energy-saving waveform reduces the electrical power stress of the display unit and crosstalk between pixels through a smooth driving method, making it particularly suitable for rendering complex static or slow scenes. While ensuring high-quality image presentation, it also reduces system power consumption and extends the lifespan of display components.

[0057] In S40, if the estimated user fatigue value U_fatigue > 0.65, then a "low fatigue relief waveform" is adopted. The specific steps of the low fatigue relief waveform are as follows: SC41: Set the starting voltage V_start to 15% of the rated voltage.

[0058] SC42: Set the voltage ramp-up time T_rise to 2.2 microseconds, which is between the enhanced waveform and the energy-saving waveform.

[0059] SC43: Set the overshoot voltage ratio O_vershoot to -2%, meaning that the voltage will first reach 102% before slightly dropping back to 100% before climbing to V_target.

[0060] SC44: Apply this waveform globally to the entire display screen.

[0061] The low-fatigue-relieving waveform, by eliminating overshoot and employing a slightly concave, gradual rise curve, generates a light signal with gentler visual stimulation and slightly weaker edge contrast. This waveform effectively reduces continuous strong stimulation of the user's visual nerves in a fatigued state, alleviates visual tension, and improves comfort during prolonged use, demonstrating the system's human-centered adaptive capabilities.

[0062] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A dynamic parallax rendering and optimization method, characterized in that, include: S10, Obtain and load the pre-built visual response prediction model; S20: When the dynamic parallax rendering engine starts rendering the current frame, it submits the predicted rendering completion timestamp of the current frame to the visual perception synchronization controller and calculates the dynamic parameters of the current scene. S30, the visual perception synchronization controller receives the prediction rendering completion timestamp and dynamic parameters, receives the actual line scan end timestamp of the previous frame, and calculates the optimal target display trigger time and waveform modulation parameter set for the current frame based on the visual response prediction model. S40 receives pixel data from the final frame buffer, waits for the clock to reach the optimal target display trigger time, and drives the pixel units of the display panel to emit light according to the voltage timing and waveform defined by the waveform modulation parameter set. S50 captures the actual waveform of the driving voltage in real time during the display driving process and compares it with the preset waveform in the waveform modulation parameter set. If the deviation between the real-time voltage and the preset curve is detected to exceed 1.8% of the calibrated voltage during the voltage ramp-up stage, dynamic compensation is performed by adjusting the input code value of the digital-to-analog converter. S60 measures the end-to-end photoelectric delay from the issuance of the drive command to the actual generation of photons, and feeds the measured value back to the visual perception synchronization controller for online calibration of the expected photoelectric delay parameters of the display hardware in the visual response prediction model.

2. The method as described in claim 1, characterized in that, The visual response prediction model in S10 defines a transfer function from the display light pulse to the visual cortex to generate stable motion perception. The parameters include the visual integration time window, the absolute perception delay reference, and the expected photoelectric delay constant of the display hardware.

3. The method as described in claim 1, characterized in that, The pre-establishment method for the visual response prediction model includes: presenting a standard test pattern at a standard viewing distance of 0.7 meters and an ambient illuminance of 200 lux, recording the threshold, collecting eye-tracking data and motor cortical potentials in electroencephalogram signals, and establishing the original dataset; Based on the original dataset, the functional expression of the visual response prediction model is obtained through nonlinear regression fitting. The data structure, which includes a lookup table of basic delay functions, dynamically adjusted function coefficients, and initial hardware parameters, is compiled. During system initialization, the visual perception synchronization controller loads the configuration file from non-volatile memory to complete the model establishment and readiness.

4. The method as described in claim 1, characterized in that, The calculation method for the predicted rendering completion timestamp includes: at the start of frame rendering, the dynamic parallax rendering engine first analyzes the command list and resource status of the current frame, calculates the rendering time of this frame, and the predicted rendering completion timestamp is obtained by adding the rendering time of this frame to the current system clock; the value of the scene dynamics parameter is the ratio of the maximum displacement of pixels in the scene between two consecutive frames to the length of the screen diagonal; when calculating the scene dynamics parameter, the average acceleration estimate of moving objects in the scene is also introduced to correct the scene dynamics parameter; when submitting the predicted rendering completion timestamp, the dynamic parallax rendering engine submits the rendering complexity estimate of the current frame, which is obtained by normalizing the weighted sum of the number of triangles in the scene, the amount of lighting calculation, and the number of parallax texture samplings.

5. The method as described in claim 1, characterized in that, The S30 also includes: The controller reads the actual line scan end timestamp of the previous frame and the target trigger time set in the previous frame, calculates the drift error, and uses the error value for feedback correction; Based on the dynamic parameters of the current frame, query the basic delay mapping table to obtain the basic delay compensation value and obtain the basic target time; Feedback and dynamic adjustments are made to calculate the final optimal target display trigger time: the base target time is corrected by a proportional-integral controller to eliminate systematic timing drift; The controller queries the waveform parameter mapping table based on the scene dynamic parameters and the user fatigue estimate to determine the waveform modulation parameter set; the waveform modulation parameter set is a structure that includes the initial voltage ratio, the target stable voltage ratio, the voltage ramp-up time, and the overshoot voltage ratio.

6. The method as described in claim 1, characterized in that, The S40 also includes: The waveform modulation parameter set is received, a matching voltage-time reference curve is selected, and after discretization, it is loaded into the waveform buffer of the high-speed digital-to-analog converter. Before the optimal target display trigger time arrives, the output voltage of the digital-to-analog converter is preset to the starting voltage level. When the system clock reaches the optimal target display trigger time, a high-resolution timer triggers the waveform generator to drive the digital-to-analog converter to output voltage, thereby controlling the charging current of the pixel capacitor and generating a drive voltage waveform. During the voltage stabilization phase, the junction temperature is monitored in real time by a temperature sensor inside the driver chip. If the junction temperature exceeds 85 degrees Celsius, the target stable voltage is automatically lowered, and the adjustment information is fed back to the controller in real time.

7. The method as described in claim 1, characterized in that, The S50 also includes: During the voltage ramp-up phase, the high-frequency monitoring circuit compares the real-time voltage with the preset voltage. When the deviation between the real-time voltage and the preset voltage exceeds 1.8% of the calibrated voltage, the integral value of the voltage error is calculated; Based on the integral value of the voltage error and the current differential error, the gate voltage control word of the subsequent driving transistor is adjusted by looking up the compensation table to perform real-time waveform shaping.

8. The method as described in claim 1, characterized in that, The S60 also includes: At the start of the vertical blanking period for each frame or every set number of frames, the test signal generation module in the display driver circuit sends a drive pulse to a dedicated test light-emitting diode integrated in the non-display area of ​​the display panel, and the driver circuit records the precise timestamp of the pulse transmission. The light pulses emitted by the test LED are received by a photoelectric sensor, converted and shaped to obtain digital pulses; Record the precise timestamp of the arrival of the rising edge of the digital pulse, calculate the measured value of the end-to-end photoelectric delay, send the measured value to the visual perception synchronization controller, and update the expected photoelectric delay parameters of the display hardware in the visual response prediction model through the calibration formula.

9. The method as described in claim 4, characterized in that, If the corrected scene dynamics parameter is greater than 0.7 and there are objects moving laterally in the scene, a high-motion-enhanced waveform is used. The specific steps include: Set the starting voltage to 10% of the rated voltage and the target voltage to 100% of the rated voltage; Set the voltage ramp-up time to 0.9 microseconds; The overshoot voltage ratio is set to 8.5%, the overshoot duration is 0.3 microseconds, and the voltage decays to the target stable voltage within 0.5 microseconds. The waveform parameter set is dynamically applied to the display area where the moving object is located.

10. The method as described in claim 5, characterized in that, If the estimated user fatigue level is greater than 0.65, a low-fatigue-relieving waveform is used; if the corrected scene dynamics parameter is less than 0.3 and the estimated rendering complexity is greater than 0.6, a high-quality energy-saving waveform is used. The specific steps for creating a low-fatigue-relieving waveform include: setting the initial voltage to 15% of the rated voltage; setting the voltage rise time to 2.2 microseconds; setting the overshoot voltage ratio to -2%, meaning that before the voltage rises to the target stable voltage, it first reaches 102% of the rated voltage and then falls back to 100%; and applying the waveform globally to the entire display screen. The specific steps for achieving a high-quality energy-saving waveform include: setting the initial voltage to 20% of the rated voltage; setting the voltage ramp-up time to 2.8 microseconds; setting the overshoot voltage ratio to 0%; and during the voltage stabilization phase, automatically reducing the target stable voltage to 95% of the rated voltage after the duration exceeds 2 milliseconds.

Images