A method to simplify the number of keyframes for discrete shaping effects in nonlinear editing

Abstract

This invention discloses a method for simplifying the number of keyframes for discrete shaping effects in nonlinear editing, belonging to the field of nonlinear editing technology. The method includes: acquiring an original discrete two-dimensional point sequence; setting an aliasing deviation threshold; identifying a first set of ideal polyline endpoints through forward search, wherein the search is achieved by continuously extending candidate line segments and determining whether the vertical distance from the midpoint to the line segment exceeds the threshold; obtaining a second set of endpoints through reverse search and matching and averaging it with the first set to correct endpoint offset; employing multi-threaded parallel processing of large-scale sequences; and for multi-parameter cases, taking the union of the time positions of the endpoints of each parameter to generate unified keyframe data. This invention solves the storage and performance bottleneck problem caused by recording massive keyframes for shaping effects frame by frame.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of non-linear editing, and particularly relates to a method and device for simplifying the number of key frames of discrete shaping effects in non-linear editing. BACKGROUND

[0002] In modern video non-linear editing (NLE) work, users often need to add various effects to video clips, such as color adjustment, blur, anti-shake, and frame adjustment. These effects usually contain one or more parameters that can change over time. In order to record the values of these parameters at different time points, the non-linear editing system introduces the effect key frame mechanism. The actual value of a parameter at any time can be calculated by interpolating the values of the two key frames before and after it.

[0003] In some application scenarios, such as intelligent frame reconstruction of long videos (automatically cropping a horizontal screen video into a vertical screen), the system may analyze each frame of the video and generate a key frame to record the position of the framing box. The position of the framing box is usually defined by the coordinate values of its left, right, top, and bottom margins, which are usually integer types, i.e., discrete integer-shaped data. For some video materials that are several hours or tens of hours long, with a frame rate of several dozen frames per second, this frame-by-frame recording method will generate millions or even tens of millions of key frame data. Such a large amount of data puts a heavy burden on the non-linear editing system, occupying a large amount of computer memory and disk storage space. When previewing, timeline dragging, or final rendering and exporting, the processor needs to calculate the massive key frame data, resulting in slow software response, operation lag, and greatly affecting editing efficiency and user experience.

[0004] How to simplify these framing box key frame data is related to user operation experience and storage economy. The framing box key frame is a two-dimensional data array, with the horizontal axis as the timeline and the vertical axis as the parameter data. Each parameter data saves 4 image margin positions, representing the left margin, right margin, top margin, and bottom margin image coordinate positions, respectively, such as Figure 3As shown. To simplify keyframes, it's necessary to calculate which keyframes can be indirectly derived from other keyframes. This requires calculating the endpoints of the original ideal polyline based on existing keyframe data, using these endpoints as new keyframes, and recording them. In video frame reconstruction, it's necessary to analyze the polyline data for each margin of the viewfinder and calculate the endpoints of the ideal polyline. Taking the left margin data as an example, we first construct a two-dimensional array with the left margin coordinate data and the timeline data. The horizontal axis represents the time series, and the vertical axis represents the left margin data of the image viewfinder. This is displayed on the image as a polyline that changes over time. However, due to aliasing caused by discrete data storage, the simple polyline becomes an aliased polyline composed of more aliased lines. The key to reducing stored data is to quickly and accurately approximate the aliased polyline back to a non-aliased polyline. Summary of the Invention

[0005] To address the aforementioned issues, this invention proposes a method for simplifying the number of discrete integer effect keyframes in nonlinear editing. This method can intelligently identify and eliminate redundant keyframes, improve nonlinear editing speed, reduce computer computational load and storage requirements, and overcome the shortcomings of existing technologies where the excessive amount of discrete integer keyframe data leads to a decline in the performance of nonlinear editing software.

[0006] To solve the above problems, the present invention adopts the following technical solution:

[0007] One approach is to simplify the number of keyframes for discrete shaping effects in non-linear editing, which includes the following steps:

[0008] S1. Obtain the original discrete integer two-dimensional point sequence, the sequence containing multiple data points sorted according to the first dimension coordinates, each data point including the first dimension coordinate and the corresponding integer parameter value y;

[0009] S2. Set a skew deviation threshold to distinguish between true inflection points and false inflection points;

[0010] S3. Perform the following steps to identify the endpoints of the ideal polyline by searching forward from the starting point to the last point of the two-dimensional point sequence, or by searching backward from the last point to the starting point:

[0011] a) Connect the current starting point i with the kth subsequent target point to form a candidate line segment, where the initial value of k is 2 and increases sequentially.

[0012] b) Calculate the linear equation of the candidate line segment.

[0013] c) Calculate the perpendicular distance from all intermediate points between the starting point and the target point on the candidate line segment to the candidate line segment;

[0014] d) If all calculated vertical distances are less than or equal to the aliasing deviation threshold, then let k = k + 1, and repeat steps a) to c).

[0015] e) If there is at least one intermediate point whose vertical distance is greater than the aliasing deviation threshold, then mark k-1 data points as an ideal polyline endpoint and set that point as the new current starting point i;

[0016] S4. Repeat step S3 until the entire two-dimensional point sequence has been traversed, and the first set of ideal polyline endpoints is obtained.

[0017] Furthermore, after obtaining the first set of ideal polyline endpoints, the process also includes a step to correct endpoint offset errors, including:

[0018] Perform a search with the same logic as step S3 but in the opposite direction to obtain the second set of ideal polyline endpoints;

[0019] Match the endpoints of the first set of ideal broken lines with the endpoints of the second set of ideal broken lines that correspond to the same ideal inflection point;

[0020] For each pair of endpoints that are successfully matched, the average value of their first-dimensional coordinates is taken as the first-dimensional coordinate position of the final ideal polyline endpoint after correction.

[0021] Furthermore, to process large-scale data point sequences, segmented parallel processing is employed, including the following steps:

[0022] The two-dimensional point sequence is divided into multiple subsequences;

[0023] Each subsequence is assigned to a different processing thread, and each thread independently executes steps S1 to S4 and the endpoint offset error correction method to obtain the local endpoint set of each subsequence.

[0024] Merge the local endpoint sets of all subsequences to form the intermediate endpoint set; and check and supplement or merge the boundary regions of adjacent subsequences to form the final ideal polyline endpoint set.

[0025] Furthermore, the merging process specifically includes:

[0026] Get the last endpoint A of the previous subsequence and the first endpoint B of the next subsequence;

[0027] Starting from point A and targeting point B, perform the search process from steps S3a) to S3e).

[0028] If a new ideal endpoint is found between endpoint A and endpoint B, it is added to the final endpoint set; otherwise, endpoints A and B are retained in the final endpoint set.

[0029] Furthermore, the value of the aliasing deviation threshold is 1.

[0030] Furthermore, the first dimension coordinate x is the time coordinate, and the shaping parameter value y is the value of the viewfinder margin parameter for the video frame reconstruction effect in non-linear editing.

[0031] Furthermore, the original discrete shaped two-dimensional point sequence is the viewfinder margin position data for the video frame reconstruction effect, and the shaped parameter value y corresponds to at least one of the left margin, right margin, top margin, and bottom margin of the viewfinder.

[0032] Furthermore, the shaping parameters of the original discrete-shaped two-dimensional point sequence include four parameters corresponding to the left margin, right margin, top margin, and bottom margin of the viewfinder, respectively.

[0033] The second aspect is a method for simplifying the number of keyframes in discrete reshaping effects during non-linear editing. Based on the above method, for video aspect ratio reconstruction effects with multiple margin parameters, the method further includes:

[0034] For each integer parameter, the above method is executed independently to obtain the set of ideal polyline endpoint time positions for each parameter. Each set contains multiple first-dimensional coordinate positions.

[0035] Take the union of the first-dimensional coordinates of the ideal polyline endpoints of all parameters to generate a unified set of keyframe master coordinates;

[0036] At each coordinate position in the keyframe master coordinate set, the specific values ​​of all shaping parameters at that coordinate position are recorded to form simplified keyframe data.

[0037] The beneficial effects of this invention are:

[0038] 1. Extremely high data compression rate: By accurately identifying and eliminating redundant data points (pseudo-inflection points) caused by data discretization and reshaping storage, and retaining only the real inflection points that represent changes in motion trends, the number of keyframes can be drastically reduced from tens of thousands or millions to tens or hundreds.

[0039] 2. Significantly improved computing performance: The sharp reduction in the number of keyframes directly reduces the computational load of non-linear editing software when performing interpolation calculations, rendering, and data loading, making timeline operations smoother and previewing and exporting faster.

[0040] 3. Effective saving of storage resources: The streamlined keyframe data reduces the size of the project file, saves users' disk storage space, and facilitates network transmission and backup of the project.

[0041] 4. Strong algorithm robustness: Through mechanisms such as bidirectional correction and boundary merging, it is ensured that the simplified data can highly restore the original motion trajectory under different data forms and scales, and the accuracy is guaranteed.

[0042] 5. Wide applicability: The method is not only applicable to video frame reconstruction, but can also be widely applied to other non-linear editing effects that generate discrete shaping keyframes.

[0043] The present invention will be further explained in detail below with reference to the accompanying drawings and specific embodiments. Attached Figure Description

[0044] Figure 1 This is a flowchart of a method for simplifying the number of keyframes provided by the present invention;

[0045] Figure 2 This is a schematic diagram illustrating the difference between a true inflection point and a false inflection point;

[0046] Figure 3 This is a diagram illustrating the viewfinder when converting a landscape video to a portrait video.

[0047] Figure 4 This is a schematic diagram of point misalignment caused by errors at the ideal endpoint in an embodiment of the present invention. Detailed Implementation

[0048] To explain the technical solution of the present invention in detail, the terms involved in the present invention are explained as follows:

[0049] Special effects: Special effects are functional modules that can perform algorithmic processing on video or audio data. A special effect has one or more parameters.

[0050] Effects keyframes: Each effect parameter has a different value at different times. The data that records the time position and value is called an effect keyframe. An effect parameter can have multiple keyframes. The value of an effect parameter at any given time is interpolated from the data of two consecutive keyframes.

[0051] Discrete integer data refers to data that is recorded in chronological order at regular intervals and whose data type is integer.

[0052] Video frame: In a video, each image is called a video frame.

[0053] Keyframe removal: Keyframe removal refers to removing keyframes that can be indirectly calculated. For example, for points on a line segment, you only need to record the positions of the two endpoints of the line segment to calculate the value of any point on the line segment. Therefore, all keyframes within a line segment (except for the endpoints) can be removed.

[0054] Aliasing: In computer graphics display, line segment aliasing refers to the stair-step or jagged appearance that occurs after a line segment is pixelated. This phenomenon is called "aliasing." Aliasing is mainly due to the fact that line segments on a raster display are composed of discrete pixels, and when line segments are drawn point-by-point by selecting the nearest pixel, the discontinuous pixel arrangement creates jagged edges. Similarly, in data recording, if the positions of each point on a line segment are recorded as integers, a stair-step pattern will appear in the recorded data.

[0055] Inflection point of a broken line: The point where the direction of the broken line changes is the inflection point.

[0056] False inflection points: When a straight line segment is displayed on the screen, it may appear as a stepped shape due to aliasing. The connection point between each step is a false inflection point. Similarly, when the points on a straight line segment are recorded using discrete shaping, many false inflection points will also appear.

[0057] Example: In the attached diagram Figure 2 In the example, for a square pixel block, the six points A, B, C, D, E, and F form five polyline segments. Due to aliasing caused by discretization, this results in line segments that are much larger than five polyline segments. For example... Figure 2 As shown, the actual inflection points of the broken line are A, B, C, D, E, and F, while the pseudo inflection points (distorted points) are each small line segment.

[0058] Video aspect ratio reconstruction: Video aspect ratio reconstruction is a special effect. (See attached image) Figure 3 Taking converting landscape to portrait footage as an example, to transform a 16:9 video into a 9:16 video, a portion of the image in the original video needs to be framed with a rectangle. This framed portion of the image will be used for subsequent editing and display; this frame is called the viewfinder. Figure 3 As shown, in this example, the viewfinder's aspect ratio is 9:16. The viewfinder's field of view changes with the video content, but the aspect ratio remains constant. Recording the position of the viewfinder is called keyframe data.

[0059] Video footage: Video footage refers to video files. Different video files have different frame rates. Frame rate refers to how many video frames are played per second. Currently, the mainstream video frame rates are 25 frames, 29.97 frames, 50 frames, 59.94 frames, etc.

[0060] Example 1: This example demonstrates a method for simplifying the number of keyframes for discrete shaping effects in non-linear editing, such as... Figures 1 to 4 As shown, it includes the following steps:

[0061] S1. Obtain the original discrete integer two-dimensional point sequence, the sequence containing multiple data points sorted according to the first dimension coordinates, each data point including the first dimension coordinates and the corresponding integer parameter value y.

[0062] S2. Set a skew deviation threshold to distinguish between true inflection points and false inflection points;

[0063] S3. Perform the following steps to identify the endpoints of the ideal polyline by searching forward from the starting point to the last point of the two-dimensional point sequence, or by searching backward from the last point to the starting point.

[0064] a) Connect the current starting point i with the kth subsequent target point to form a candidate line segment, where the initial value of k is 2 and increases sequentially.

[0065] b) Calculate the linear equation of the candidate line segment.

[0066] c) Calculate the perpendicular distance from all intermediate points between the starting point and the target point on the candidate line segment to the candidate line segment.

[0067] d) If all calculated vertical distances are less than or equal to the aliasing deviation threshold, then let k = k + 1, and repeat steps a) to c).

[0068] e) If there exists at least one intermediate point whose vertical distance is greater than the aliasing deviation threshold, then mark k-1 data points as an ideal polyline endpoint and set that point as the new current starting point i.

[0069] S4. Repeat step S3 until the entire two-dimensional point sequence has been traversed, and the first set of ideal polyline endpoints is obtained.

[0070] After obtaining the first set of ideal polyline endpoints, the process further includes a step to correct endpoint offset errors. Specifically, this involves performing a search process with the same logic as steps a) to e) of S3 above, but in the opposite direction, to obtain the second set of ideal polyline endpoints. The endpoints in the first set of ideal polyline endpoints and the second set of ideal polyline endpoints corresponding to the same ideal inflection point are matched. For each pair of endpoints that are successfully matched, the average value of their first-dimensional coordinates is taken as the first-dimensional coordinate position of the corrected final ideal polyline endpoint.

[0071] To improve processing efficiency, a multi-threaded parallel processing method can be used when processing large-scale data point sequences, including the following steps:

[0072] 1) Divide the two-dimensional point sequence into multiple subsequences;

[0073] 2) Assign each subsequence to a different processing thread, and have each thread independently execute steps S1 to S4 and the endpoint offset error correction step to obtain the local endpoint set of each subsequence.

[0074] 3) Merge the local endpoint sets of all subsequences to form the intermediate endpoint set;

[0075] 4) For the boundary regions of adjacent subsequences, check whether there are ideal inflection points that were missed due to the segmentation at the boundary points, and supplement or merge them to form the final set of ideal broken line endpoints.

[0076] The merging process specifically includes:

[0077] Obtain the last endpoint A of the previous subsequence and the first endpoint B of the next subsequence. With A as the starting point and B as the target point, execute the search process of steps S3a)-e). If a new ideal endpoint is found between endpoint A and endpoint B, add it to the final endpoint set; otherwise, keep endpoint A and endpoint B in the final endpoint set.

[0078] When a parameter has multiple line data points, the endpoints of each line are calculated, and then these endpoints are joined together. The keyframe only records the point data and time position of each thread at the time position of these joints.

[0079] The original discrete-shaped two-dimensional point sequence is the viewfinder margin position data for the video frame reconstruction effect in non-linear editing. The shaping data of the original discrete-shaped two-dimensional point sequence includes at least one of four parameters corresponding to the left margin, right margin, top margin and bottom margin of the viewfinder, respectively. In this embodiment, four parameters are preferred.

[0080] For each two-dimensional point sequence corresponding to each margin parameter, the ideal polyline endpoint time position set of each parameter is calculated independently. The union of the ideal polyline endpoint time position sets of all parameters is taken to generate a unified keyframe time point master set. At each time point in the keyframe time point master set, the specific values ​​of all margin parameters at that time point are recorded to form simplified keyframe data.

[0081] Example 2:

[0082] The following is a further detailed explanation of this invention. Based on the above embodiments, the first dimension coordinate x is a time coordinate, and the integer parameter value y is the value of the special effects parameter in non-linear editing. Figures 1 to 4 As shown, the method steps are as follows:

[0083] 1. Obtain the discrete integer two-dimensional point sequence. Input the discrete integer two-dimensional sequence data points (x, y), where x represents time and y represents the integer parameter value.

[0084] 2. Set the aliasing deviation threshold, d_Aliasing=1. This threshold is used to determine whether a point is a true inflection point or a pseudo inflection point caused by rounding error. Based on experience, the vertical distance between a pseudo inflection point (aliasing point) and the line segment does not exceed 1. Therefore, a vertical distance within 1 is a pseudo inflection point (aliasing point), and a distance exceeding 1 is a polyline endpoint.

[0085] 3. Array processing of discrete integer two-dimensional point sequences to identify polyline endpoints.

[0086] Starting from the first point: For the first and third points, calculate the parameters a, b, and c in the line equation ax + by + c = 0. Then calculate the perpendicular distance from the second point to this line. If the distance from the second point to the line segment is less than the aliasing deviation, it means the first, second, and third points are on a straight line. Next, calculate the parameters a, b, and c of the line equation between the first and fourth points. If the perpendicular distances from the second and third points to the line are both less than the aliasing deviation, it means the first, second, third, and fourth points are on a straight line. Next, calculate the parameters a, b, and c of the line equation between the first and fifth points, and calculate the perpendicular distances from all points between these two points to this line. Continue in this manner until a point between the starting point and the target point has a perpendicular distance greater than the aliasing deviation. At this point, the point before the target point can be considered an endpoint of the ideal line segment, and recorded. Use this new endpoint as the starting point for calculating the next line segment endpoint, following the above steps until the last element of the array. This will calculate all the true inflection points of the entire array. That is, starting from point A, find point B, then from point B find point C, then from point C find point D, and so on. Figure 2 As shown.

[0087] However, this algorithm encounters problems in practical applications. In continuous polylines, when the intersection angle between two adjacent line segments is very small, errors occur, and the intersection point shifts to 1 to n points in the search direction. This is because the line connecting the point after the endpoint of the ideal line segment and the beginning of the line may have an intersection point whose perpendicular distance to the line is less than the aliasing deviation, causing the position of the inflection point to shift backward. For example... Figure 4 As shown: A, B, and C are the three endpoints of the broken line, and X is the nth point after the ideal endpoint B. When endpoint A is connected to X, the perpendicular distance between the points within A and X and the line AX is less than the aliasing deviation. This causes the search for the inflection point to mistakenly identify point X as the ideal endpoint B. The point misalignment is XB.

[0088] 4. Perform offset correction on the identified polyline endpoints.

[0089] To address the aforementioned error issue, it's necessary to correct the offset of the identified polyline endpoints. The inflection point search mode is divided into forward search and reverse search modes: the forward search mode searches from the first point to the last, which may cause some inflection points to shift in the opposite direction. The reverse search mode searches from the last point to the first, but may cause some inflection points to shift in the forward direction. By performing both forward and reverse searches simultaneously, the forward search yields the first set of polyline endpoints, and the reverse search yields the second set. Averaging the corresponding points in the two sets of data significantly corrects for inflection point offsets.

[0090] 5. Based on the final set of polyline endpoints, generate a simplified keyframe sequence.

[0091] Example 3:

[0092] Based on Example 1, this example further optimizes the method in Example 1 by adding multi-threaded (segmented) search to the step of processing and identifying the endpoints of the polyline in the discrete integer two-dimensional point array.

[0093] The discrete integer 2D point array can be divided into several parts, each calculated independently. This method results in the generation of some additional keyframe inflection points, but the number of inflection points is less than or equal to 2*(N-1), where N is the number of parts. That is, each segment generates two additional endpoints. If the additional endpoint happens to be a kink, the split point will have only one extra endpoint; otherwise, it will have two extra endpoints. When merging the final set of endpoints of these subsequences, these extra endpoints need to be removed.

[0094] Example 4:

[0095] This embodiment is a further optimization of the method in Embodiment 1, to adapt to the case where the integer data in the discrete integer two-dimensional point array has multiple parameters.

[0096] If the data to be shaped has multiple parameters, i.e., multiple polyline data, then the endpoints of each polyline are calculated, and these endpoints are then joined together. The keyframe only needs to record the point data and time position of each curve at the time position of this union, and then deduplication is performed.

[0097] like Figure 3As shown, in the video aspect ratio reconstruction effect, there are four sets of aliasing keyframe data: left margin, right margin, top margin, and bottom margin. Each margin forms a polyline with the timeline. The endpoints of these four polylines are calculated and reconstructed. Each polyline has multiple endpoints in time sequence, and these endpoints are not necessarily consistent in time position. The union of these four polyline endpoints in time sequence yields a total of n time points, which are recorded as keyframe data points. For each set of data, based on the keyframes, the values ​​at the corresponding time positions are retrieved from the metadata and recorded. This significantly reduces the amount of keyframe data required for aspect ratio reconstruction.

[0098] Finally, it should be noted that the above is only used to illustrate the technical solution of the present invention and not to limit it. Although the present invention has been described in detail with reference to the preferred arrangement, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solution of the present invention (such as the order of various steps) without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A method for simplifying the number of keyframes of discrete shape effect in a nonlinear editing, characterized in that, The method comprises the following steps: S1. obtaining an original discrete reshaping two-dimensional point sequence, the sequence comprising a plurality of data points sorted by a first dimension coordinate, each data point comprising a first dimension coordinate and a corresponding reshaping parameter value y; S2. setting a deviation threshold for distinguishing between a real inflection point and a false inflection point; S3. performing the following steps to identify the end points of an ideal polyline by forward searching from the starting point to the last point of the two-dimensional point sequence or by reverse searching from the last point to the starting point: a) connecting the current starting point i with the subsequent target point, the kth point, to form a candidate line segment, wherein k is initially 2 and is sequentially incremented, b) calculating the equation of the straight line of the candidate line segment, c) calculating the perpendicular distance from all intermediate points between the starting point and the target point on the candidate line segment to the candidate line segment; d) if all calculated perpendicular distances are less than or equal to the deviation threshold, setting k = k + 1 and repeating steps a) to c); e) if there is at least one intermediate point whose perpendicular distance is greater than the deviation threshold, marking the k-1 data points as an end point of an ideal polyline and setting the point as a new current starting point i; S4. repeating step S3 until the entire two-dimensional point sequence is traversed to obtain a first set of ideal polyline end points.

2. The method of claim 1, wherein the number of keyframes of the discrete shape effect is simplified in the nonlinear editing. After obtaining the first set of ideal polyline end points, a step of correcting end point offset errors is further included, comprising: performing a search with the same logic as step S3 but in the opposite direction to obtain a second set of ideal polyline end points; matching the end points corresponding to the same ideal inflection point in the first set of ideal polyline end points and the second set of ideal polyline end points; for each pair of end points that match successfully, taking the average of their first dimension coordinates as the position of the first dimension coordinate of the final ideal polyline end point after correction.

3. The method of claim 1 or 2, wherein the number of keyframes of the discrete shape effect is simplified in the non-linear editing. To process large-scale data point sequences, segmented parallel processing is adopted, comprising the following steps: dividing the two-dimensional point sequence into a plurality of subsequences; allocating each subsequence to a different processing thread, and independently executing the method of claim 1 or 2 by each thread to obtain a local end point set for each subsequence; merging the local end point sets of all subsequences to form an intermediate end point set, and checking the boundary regions of adjacent subsequences for supplementary or merging processing to form a final set of ideal polyline end points.

4. The method of claim 3, wherein, The merging processing step specifically comprises: obtaining the last end point A of a previous subsequence and the first end point B of a subsequent subsequence; performing the search process of steps S3a) to S3e) with A as the starting point and B as the target point; if a new ideal end point is found between the A end point and the B end point, it is added to the final end point set; if not, the A end point and the B end point are retained in the final end point set.

5. The method for simplifying the number of keyframes of discrete shape effect in the nonlinear editing according to claim 1, wherein, The value of the deviation threshold is 1.

6. The method for simplifying the number of keyframes of discrete shape effect in the nonlinear editing according to claim 1, wherein, The first dimension coordinate is a time coordinate, and the reshaping parameter value y is the value of the frame margin parameter of a video frame reconstruction special effect in nonlinear editing.

7. The method of claim 6, wherein the number of keyframes of the discrete shape effect is simplified in the nonlinear editing, and the number of keyframes of the discrete shape effect is simplified in the nonlinear editing by using a keyframe of the discrete shape effect as a keyframe of the nonlinear editing. The original discrete reshaping two-dimensional point sequence is frame margin position data of a video frame reconstruction special effect, and the reshaping parameter value y corresponds to at least one of the left frame margin, the right frame margin, the top frame margin, and the bottom frame margin. ​ 8. The method for simplifying the number of keyframes of discrete shape effect according to claim 7, wherein, The shaping data of the original discrete shaping two-dimensional point sequence includes four parameters respectively corresponding to the left margin, the right margin, the top margin and the bottom margin of the framing box.

9. A method for simplifying the number of keyframes of discrete shape effects in a non-linear editing, based on the method according to any one of claims 1 to 8, characterized in that, For the video framing reconstruction special effect with multiple margin parameters, the method further comprises: For each shaping parameter, independently performing the method of any one of claims 1-8 to obtain a respective set of ideal spline end point time positions for each parameter, each set containing a plurality of first dimension coordinate positions; Taking the first dimension coordinates in the sets of ideal spline end point positions of all parameters in union to generate a unified set of key frame main coordinates; At each coordinate position in the set of key frame main coordinates, recording the specific values of all shaping parameters at the coordinate position to form the simplified key frame data.

Images