VQ Analyzer Update: V5.0 Launch
If you don’t already know, we’re ViCue Soft, the developers behind the bitstream analysis tool VQ Analyzer, and we’ve just released a fresh new update of our popular brainchild! Multiple extra features have been included with the new release; let’s run through them and take a look:
- Mac OS Dark Mode: this is a handy feature of newer Apple devices, and we couldn’t resist including it in our VQ Analyzer makeover. The tech giant invested big money in this dark design, and ViCue Soft are following their lead to allow any Mac fans this option.
Fig 1. The VQ Analyzer in Dark Mode on Mac OS
- AV1 4:2:2 and 4:4:4 chroma formats and high bitdepths are now fully supported. You can easily analyze high-quality contribution formats if you’re working in development. All the same features work just as they already do for 4:2:0, so there are no big operational changes.
Fig. 2 Analyzing AV1 bistreams
- Hex search is available if you know the byte sub-sequence you lost in megabytes of picture data.
Fig 3. Hex search function
- PSNR/SSIM metrics plot: in addition to legacy support of our per-block metric maps, you can invoke a metrics plot to oversee metrics dynamics, from frame to frame, for the whole sequence.
Fig 4. PSNR/SSIM metrics plot
- Statistics are now available in ‘main view’: you can pick out multiple statistics graphs and collate them together as an additional statistics map to suit your needs.
Fig 5. Statistics as shown by the VQ Analyzer
- Memory read access. During the decoding process, each block can have multiple reference prediction readings. For hardware and software decoders it may work a little differently, but in general this is associated with accessing a memory line in some way to fetch the relevant data. The VQ Analyzer 5.0 has a newly-introduced memory bandwidth map, designed to uncover any data hog blocks for further data prefetching optimization in the decoder.
Fig 6. VQ Analyzer memory bandwidth maps
- Last (but in no way least) on our list of shiny new features is VVC codec support. Bearing in mind that the VVC spec is not yet finalized and that VVC bitstream changes, this particular version is compatible with VTM8.0 of reference software codec. As many users may already be familiar with our HEVC and AV1 Analyzer functionality, we’ve kept our VVC Analyzer very similar in terms of layout, details, and statistics you might need to progress with codec development and optimization of new features beyond HEVC:
Partitioning: Explore a new quadtree with nested multi-type tree using a binary and ternary splits segmentation structure, which removes the separation of the CU, PU and TU concepts. And for I slices, the luma and chroma can have totally separate block tree structures. (For Inter they still have to share the same coding tree structure for the same CTU). The largest CTU size is 128x128.
Prediction: Beyond the coding features in HEVC, VVC includes several new and refined coding tools. To capture more characteristics in natural video, the non-square blocks are introduced and the number of directional intra modes extended up to 65. Using the VQ Analyzer, you can get insights about those state-of-art coding tools for intra:
Mode dependent intra smoothing (MDIS).
Cross-component linear model (CCLM) prediction.
Position dependent intra prediction combination (PDPC) method.
Multiple reference line (MRL) intra prediction.
Matrix weighted intra prediction (MIP) method.
And for inter:
- Symmetric MVD (SMVD) signalling.
- Affine motion compensated prediction.
- Subblock-based temporal motion vector prediction (SbTMVP).
- Adaptive motion vector resolution (AMVR).
- Bi-prediction with CU-level weight (BCW).
- Bi-directional optical flow (BDOF).
- Decoder side motion vector refinement (DMVR).
- Geometric partitioning mode (GPM).
- Combined inter and intra prediction (CIIP).
- Transform: To support ultra high definition large block-size, transforms up to 64×64 in size are enabled, as well as new transform matrixes which are quantized more accurately than in HEVC to keep orthogonality. Last but not least, the so called low-frequency non-separable transform (LFNST) is applied after conventional transform and before quantization. This transform is introduced to further compress the redundancy between low-frequency primary transform coefficients.
- Loop filters: Besides deblocking filter and SAO from HEVC, adaptive loop filter (ALF) and a new process called the luma mapping with chroma scaling (LMCS) (AKA the adaptive in-loop reshaper) were added. The order of the filtering process in the VVC is the LMCS, deblocking filter, SAO and ALF.
Fig 7. VQ Analyzer loop filters
Fig 8. SAO
We hope you enjoyed this tour of version 5.0 of the VQ Analyzer! We are constantly working to keep our products at the top of their game, and we know these state-of-the-art additional features will really make a difference for users.
Bye for now!
Thanks to Pavel Ditenbir.