Camera Motion Retargeting for Human Quality Preservation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Generating motion capture video with realistic human camera motion is expensive, time-consuming, and often results in a mechanical look and feel due to the limitations in post-processing and re-recording requirements.
Innovation Solution
A camera motion data retargeting system and method that computes a second camera path with modified orientation data to place a target at the camera's center and removes camera roll, allowing for the extraction and application of subtle human movements from a first camera path to a third camera path, enabling adjustments to camera position and orientation without re-recording.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If motion capture video is generated using traditional methods with expensive equipment and facilities, then the quality of camera motion is improved, but the cost and time required increase significantly
Solution Approach 1:
The patent extracts human camera motion characteristics from recorded video footage and creates a digital model or representation of the operator's motion patterns. This copied motion data can then be applied to virtual cameras in post-production, eliminating the need for expensive motion capture facilities and multiple takes while preserving the authentic human quality of camera movement.
Solution Approach 2:
The system performs preliminary extraction and analysis of human motion characteristics during the initial recording phase, storing these characteristics for later application. This allows the motion data to be prepared in advance and reused multiple times without requiring additional recording sessions, reducing both time and cost for subsequent productions.
2Adaptability or versatility
If camera motion is fully controlled by automatic cinematography, then post-processing flexibility is improved, but the human quality and realism of camera motion deteriorates
Solution Approach 1:
The patent separates camera motion into distinct components: the mechanical/path-following portion and the human operator characteristics portion (such as subtle hand movements, breathing patterns, and stylistic choices). This segmentation allows automatic cinematography to handle the structured motion while the extracted human characteristics are overlaid to restore realism, achieving both post-processing flexibility and human quality.
Solution Approach 2:
The system combines automatic cinematography control with extracted human motion characteristics by merging the two data streams. The automated camera path provides the foundational motion, while the superimposed human characteristics data adds the subtle imperfections and stylistic elements that create a realistic, human-operated appearance, resolving the contradiction between automation and human quality.
3Adaptability or versatility
If camera motion data is extracted and modified in post-production, then creative freedom is improved, but the ability to preserve original human quality deteriorates
Solution Approach 1:
The patent creates a dynamic system where human motion characteristics are extracted as adjustable parameters that can be modified independently. Users can control the intensity, timing, and specific aspects of the human quality preservation, allowing creative freedom to adjust the level of realism or stylization while maintaining the ability to preserve the original characteristic feel when desired.
Solution Approach 2:
The system allows modification of specific parameters of the extracted human motion data, such as the amplitude of hand movements, frequency of adjustments, or specific stylistic elements. These parameter changes enable creative freedom to adapt the motion to different needs while preserving the essential human quality through controlled adjustment of individual characteristics rather than complete replacement.
Data Source
AI summary
A method of retargeting camera motion is disclosed. Data is received that describes a position of a target along a target path. Data is received that describes a position and an orientation of a first camera along a first camera path wherein the first camera is filming the target. A second camera path is computed, the second path corresponding to a second camera. The second camera path corresponds to the first camera path with one or more exceptions. The one or more exceptions include one or more of modification of orientation data corresponding to the second camera path to place the target at the center of a camera screen associated with the second camera and removing of a camera rotation roll for the second camera. A difference is computed between the first camera path and the second camera path. The computed difference is applied to a third camera path.