Bit depth-based clipping in adaptive loop filtering improves luma and chroma coding efficiency while preserving image quality in video encoding.
Parallel vehicle video encoders pre-encode frames at different bitrates, enabling fast switching without I-frame delays or buffer flushing.
Adaptive weighting of current and neighboring intra prediction blocks reduces bit rate while preserving image reconstruction quality.
Segmenting images into subpictures with selective filtering improves visual quality while reducing data volume and hardware resource use.
Conditional SPS and PPS sub-picture ID mapping reduces unnecessary syntax parsing while preserving boundary handling in video decoding.
Dynamic metadata carries color volume transform parameters and mastering display data so HDR video can be encoded for displays with limited color volume.
Adaptive intra prediction filters improve image coding efficiency while reducing bit rate and memory use for interpolation coefficients.
Adaptive deblocking sets filter length and boundary strength by block prediction and motion cues to smooth block edges efficiently.
Multiple IntraTMP matching blocks are selectively combined to improve video prediction accuracy and bitrate distortion with limited decoding overhead.
Adaptive LFNST scanning uses MIP parameters to select horizontal or vertical coefficient orders, improving VVC coding efficiency and picture quality.
Filtering the matched reference block corrects template-matching deviations, improving video prediction accuracy without losing encoding efficiency.
A flag and sub-picture IDs in an extracted bitstream let decoders handle partial sub-picture sets without coding errors or excess data.
Region-level coefficient flags let AVM decoders skip zero-only scan regions, cutting coding overhead and bitrate in transform blocks.
Threshold-based template selection adapts neighboring reference samples for CCCM chroma prediction, improving coding efficiency and bitstream size.
Extreme block aspect ratios trigger prediction mode disabling, cutting index handling and circuit load while preserving video coding quality.
Compact MPM indexing with MRL and planar flags improves intra prediction accuracy while cutting bitstream overhead in image and video coding.
Short-term reference pictures and separate L0/L1 motion vector signaling improve inter prediction efficiency with lower coding complexity.
Separating fully reconstructed and pending search regions improves Intra TMP block prediction by using more adjacent sample information.
Square sub-luma blocks and averaged neighboring motion vectors improve affine chroma prediction in high-resolution video coding.
Analytical 4-tap coefficients replace lookup tables in subpixel video prediction, reducing memory demands and latency while maintaining filtering accuracy.
High-resolution video increases data volume; joint chroma flags coordinate transform-skip decisions during decoding to reduce redundant signaling.
DIMD and TIMD derive non-MPM intra prediction modes to cut signaling bits while preserving prediction accuracy in video coding.
Deriving chroma scaling factors from neighboring blocks cuts bitstream overhead while preserving accurate CfL video prediction.
Offset-based regression and local illumination compensation adapt video-block prediction to improve coding efficiency and effectiveness.
Selective parameter set forwarding in multi-track video files cuts buffer use and decoding effort while preserving joint sub-stream access.
During sub-bitstream extraction, irrelevant scalable nesting SEI messages are removed while target-OLS messages remain, reducing bitstream and resource use.
Regression-based coding tools and offset removal improve video block to bitstream conversion by raising coding efficiency without added complexity.
Implicit tile entry offsets and omitted end-of-slice signaling reduce bit overhead and processing overhead in video decoding.
Sequence-level parsing adapts skip/direct motion index limits to candidate counts, reducing bit overhead in video coding and decoding.
Two-stage video transforms apply a secondary transform only to top-left coefficients, improving compression performance while limiting computational complexity.
Template matching refines GPM partition shapes and asymmetric blending regions for efficient image encoding/decoding and lower transmission and storage costs.
Pre-analysis sets feature compression and decoding controls so neural network inference stays reliable despite variable compressed data size.
Shift-based power-of-2 weighting replaces sine and cosine in geometric partition prediction, cutting video coding complexity and storage load.
SEI messages identify neural-network post-filters and reusable properties, reducing redundant signaling for efficient image bitstreams.
Reduced candidate scanning simplifies ATMVP motion prediction while maintaining video coding performance during encoding and decoding.
Deriving subblock transform flags from inter prediction modes reduces redundant signaling and improves encoding efficiency.
Flag relationships among TSRC, transform skip, and dependent quantization reduce residual-coding bits in image decoding.
A gating mechanism selects specialized neural models for each video block, combining spatial and temporal context to improve rate-distortion performance.
Timestamp-based output ordering aligns point cloud samples with display timing, reducing latency in point cloud coding.
Rice lookup tables derive transform-coefficient levels from bitstreams, helping preserve image quality while reducing video transmission and storage costs.
Encoding no-display frames and motion-vector-predicted macroblocks reuses image data, reducing overlay re-rendering and synchronization overhead.
Selecting a subset from neighboring motion vectors limits redundant searches, reducing computational complexity in video inter-prediction.
Fixed CABAC probability models can converge slowly as symbol rates change; adaptive context selection and updates improve coding efficiency.
Candidate comparisons and a history table select CCP models for each video block, improving coding efficiency and effectiveness.
Partition video blocks along edges, blend inter-prediction signals, and reduce bitrate and storage needs for HD video.
A decoder refines block vectors per subblock through intra template matching, improving prediction accuracy without an unnecessarily broad search.
Control-point-derived sample-unit motion vectors improve inter-prediction for rotated, zoomed, and deformed images while reducing coding data.
Conditional padded reference samples improve chroma intra prediction while balancing compression efficiency and picture quality.
Conditional generation of zero-motion, neighboring-block, and temporal candidates fills affine lists when initial entries fall short.
A control circuit generates mixed video predictions by combining primary and companion modes using adaptive lookup tables.
Horizontal geometry padding determines reference sample wraparound offsets to resolve face seams and improve motion compensation accuracy.
A point cloud encoding apparatus distinguishes patch types via a first index to selectively encode auxiliary information.
Builds an adaptive motion vector candidate list via histogram analysis of neighboring blocks, reducing decoder complexity while improving bit-savings.
Segmenting 3D volumetric video into independent atlas tracks resolves synchronization delays and reduces bandwidth consumption during decoding.
Signaling links video and depth components to exploit cross-component redundancy while maintaining backward compatibility with legacy decoders.
Reordering most probable modes via neighbor block analysis reduces bitstream overhead while maintaining decoding accuracy.
Segmenting search regions via a cost function reduces computational complexity while maintaining picture quality.
An image encoding apparatus modifies motion vectors to reduce flicker in flat areas.
A processing module adjusts quantization parameters based on residual energy to control encoded picture size.
Encoder derives motion vectors via selected modes to encode blocks without transmitting mode flags, reducing bitstream overhead.
Assigning shorter codewords to probable modes reduces bit usage while managing selection complexity through context-based probability estimation.
Encoding tile identifiers and slice addresses in headers enables sub-bitstream extraction without rewriting coordinates, reducing computational burden.
Segmenting images into blocks and applying an optimized pre-filter reduces boundary artifacts, preserving usable image quality despite high bit error rates.
FIR or IIR filtering of intra prediction samples improves coding efficiency while managing device complexity.
Adjusting pixel values reduces I-frame bitstream size while maintaining image quality by selecting intra-prediction modes with lowest residuals.
Dynamic playback mode selection resolves the contradiction between security against plugin attacks and compatibility with diverse streaming media formats.
Simplified position dependent prediction combination reduces computational complexity while maintaining prediction accuracy in inter prediction modes.
A video frame watermarking method modifies macroblocks and re-encodes output slices to embed data.
Segmenting the IBC virtual buffer into line buffers resolves memory complexity trade-offs while maintaining prediction accuracy for large video blocks.
A video encoder selects quantization steps to minimize remainders after division of group values.
A server device extracts and encodes a field of view image from video content for transmission to a client.
Deriving more than three most probable modes with context modeling reduces bitstream size while managing encoder complexity.
Adaptive multiple transform splits blocks into sub-blocks to apply distinct transform kernels.
Mapping intra prediction modes to target transform modes eliminates explicit signaling overhead while maintaining high compression efficiency.
Geometric partition mode reorders width candidates via template matching to determine adaptive blending area dimensions.
Selective transform coding preserves textual content sharpness by disabling transforms based on pixel value differences and rate-distortion constraints.
Segmenting coding units with multiple local illumination compensation models resolves inaccuracy from varying illumination changes across blocks.
A reduced secondary transform adapts coding kernels to intra prediction modes.
A parallel video encoder determines bit budgets using historical complexity distributions to allocate data efficiently.
Group sequence-level parameters in TLV units to reduce coding bits, eliminating redundant transmission across frames.
Neural network model determines lens position using encoded phase and image data, resolving autofocus accuracy and speed trade-offs in low light.
Partitioned video blocks use weighted prediction to align with object shapes, reducing spurious high-frequencies and computational resources.
Frequency mask tables selectively encode frequency-transformed blocks to improve video compression ratios.
A camera control unit dynamically adjusts bit rate allocation based on shake detection to manage encoding resources.
Encapsulates video frame data units with ordering information to eliminate reordering latency and processing resource requirements.
Assigning individual indices to DC and Planar modes resolves trade-offs between signaling overhead and coding performance in HEVC.
Transform index signaling selects coding parameters for blocks using matrix-based intra prediction.
A constrained block partitioning method infers partition types and directions based on parent block splits to reduce signaling overhead.