International organisation for standardisation

6.3Source video test material

6.3.1.1.1.1.1.1.1JCTVC-F228 A study of SCC test sequences [X. Zhang (HKUST)] [late upload 07-08]

This report provides the coding results of submitted test sequences for SCC ad-hoc group. Results by both HM3.0 and HM3.0 SDIP are provided. Compared to HM3.0, it is reported that the average BD-rate gain from HM3.0 SDIP for AI-HE, AI-LC, RA-HE, RA-LC, LB-HE, LB-LC, LP-HE and LP-LC are 8.0%, 9.1%, 5.8%, 6.6%, 3.3%, 3.9%, 3.1%, and 3.7% respectively.

Coding artifacts are more obvious at large QPs (36/42), nothing visible at QP 22.

Gain was higher for the more artificial sequences.

The version of SDIP still included 16x1 (which is not the case anymore).

Is the content challenging enough? The overall rate ranges for the graphics-only content seem to be much lower than for natural video. This is different for the sequences with mixed video and graphics.

In general, the suitability of PSNR and BD measurements must be assessed.

Breakout activity was requested to review sequences and identify test points.

6.3.1.1.1.1.1.1.2JCTVC-F562 Updated Video Test Material Submission for "Screen Content" Coding Experiments: Scrolling Text, Overlays, Editing Window Switching and On-line Gaming [G. Cook, W. Gao, J. Zhou, H. Yu (Huawei Technologies)]

This contribution consists of a submission of twelve test sequences in response to JCTVC-D501, “Request for Video Test Material for ‘Screen Content’ Coding Experiments”; nine of the sequences are updates to previously submitted material and three are new sequences. The sequences consist of camera-view content with text overlays and scrolling text, and camera-view content with document editing, scrolling and window switching, high contrast text and gradient backgrounds and on-line gaming. The sequences were updated based on comments received at the previous meeting, and were asserted to follow the general guidelines given in JCTVC-D501. Simulations based on the test conditions agreed to by the Screen Content Coding Ad Hoc Group were made.

One computer graphics (CG) generated (games) sequence seems to be more challenging, includes effects such as edge alias and flickering apparently caused by cheap rendering (perhaps not completely state of the current art of rendering technology).

6.3.1.1.1.1.1.1.3JCTVC-F726 Video Test Material Submission for Screen Content Coding X. Zhang, O. Au, C. Pang, X. Wen [late reg. 07-12, upload 07-13]

Did not need to be presented – was essentially included in JCTVC-F228 presentation.

6.3.1.1.1.1.1.1.4JCTVC-F741 Test material update for screen content [Wenpeng Ding, Yunhui Shi, Baocai Yin] [late reg. 07-15, upload 07-16]

This contribution describes and introduces a set of test sequences for screen content which will be made available to JCT-VC and parent bodies by Beijing University of Technology. The test sequences are provided in both YUV4:2:0 and YUV4:4:4 formats. BJUT proposes these sequences for developing and testing HEVC in the JCT-VC.