RGBCube/serenity
1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-07-26 21:07:36 +00:00
Code Issues Activity Actions

LibVideo: Allow the VP9 decoder to decode ultra high resolution video

Browse source 
Previously, some integer overflows and truncations were causing parsing
errors for 4K videos, with those fixed it can fully decode 8K video.

This adds a test to ensure that 4K video will continue to be decoded.

Note: There seems to be unexpectedly high memory usage while decoding
them, causing 8K video to require more than a gigabyte of RAM. (!!!)
This commit is contained in:
Zaggy1024 2022-10-01 17:28:37 -05:00 • committed by Andrew Kaster
parent f894e8be62
commit 41cb705b47
6 changed files with 21 additions and 9 deletions
Download patch file Download diff file Expand all files Collapse all files

6
Userland/Libraries/LibVideo/VP9/Parser.cpp
View file

@ -979,8 +979,8 @@ DecoderErrorOr<void> Parser::decode_block(u32 row, u32 col, BlockSubsize subsize
// write out of bounds. This check seems consistent with libvpx.
// See here:
// https://github.com/webmproject/libvpx/blob/705bf9de8c96cfe5301451f1d7e5c90a41c64e5f/vp9/decoder/vp9_decodeframe.c#L917
auto maximum_block_y = min(num_8x8_blocks_high_lookup[subsize], m_mi_rows - row);
auto maximum_block_x = min(num_8x8_blocks_wide_lookup[subsize], m_mi_cols - col);
auto maximum_block_y = min<u32>(num_8x8_blocks_high_lookup[subsize], m_mi_rows - row);
auto maximum_block_x = min<u32>(num_8x8_blocks_wide_lookup[subsize], m_mi_cols - col);
for (size_t y = 0; y < maximum_block_y; y++) {
for (size_t x = 0; x < maximum_block_x; x++) {
@ -1337,7 +1337,7 @@ DecoderErrorOr<i32> Parser::read_mv_component(u8 component)
return (mv_sign ? -1 : 1) * static_cast<i32>(mag);
}
Gfx::Point<size_t> Parser::get_decoded_point_for_plane(u8 column, u8 row, u8 plane)
Gfx::Point<size_t> Parser::get_decoded_point_for_plane(u32 column, u32 row, u8 plane)
{
if (plane == 0)
return { column * 8, row * 8 };