michaelni changed the topic of #ffmpeg-devel to: Welcome to the FFmpeg development channel | Questions about using FFmpeg or developing with libav* libs should be asked in #ffmpeg | This channel is publicly logged | FFmpeg 7.0.1 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
compnn has joined #ffmpeg-devel
compn has quit [Read error: Connection reset by peer]
MrZeus has quit [Read error: Connection reset by peer]
<jamrial>
courmisch: alignment? overreliance on padding?
<courmisch>
jamrial: code assumes 8-bit alignment and no padding
<courmisch>
my money is on intermediate arithmetic overflow
<courmisch>
FWIW, checkasm fails to test strictly negative TC values
<courmisch>
pun unintended, but still funny
haihao has quit [Ping timeout: 268 seconds]
haihao has joined #ffmpeg-devel
unlord has quit [Changing host]
unlord has joined #ffmpeg-devel
Krowl has quit [Read error: Connection reset by peer]
<kurosu>
courmisch: tc offsets, or final tc values (used for clipping and look up tables)? I don't think the later would be defined. Not even sure it would be legal for this to happen
<kurosu>
anyway I guess the later is not much of a solace, and has to be handled anyway
<haasn>
Lynne: Without bitexact we could for example reassociate operations
<haasn>
Or enable SIMD paths that round differently from C
<Lynne>
nevcairiel: in the time domain it is random, but predictable
<Lynne>
quality settings can do that too
<courmisch>
kurosu: I dunno, but Lorren Merrit added a check for negative TC values whilst adding MMX support
<courmisch>
kurosu: it seems to me that the calling code forces negative values in "some cases", but I have zero knowledge of ffh264 insides
aphysically has quit [Ping timeout: 246 seconds]
Krowl has joined #ffmpeg-devel
jamrial has quit []
___nick___ has joined #ffmpeg-devel
Livio has quit [Ping timeout: 252 seconds]
Livio has joined #ffmpeg-devel
Krowl has quit [Read error: Connection reset by peer]
darkapex has quit [Remote host closed the connection]
darkapex has joined #ffmpeg-devel
Kwiboo has quit [Quit: .]
Kwiboo has joined #ffmpeg-devel
microchip__ has joined #ffmpeg-devel
microchip_ has quit [Ping timeout: 268 seconds]
microchip__ is now known as microchip_
___nick___ has quit [Ping timeout: 268 seconds]
compn has quit [Read error: Connection reset by peer]
compn has joined #ffmpeg-devel
<kurosu>
ok, looks like a signalling trick. I think the code sets bS for an "edge" to 0 to indicate it doesn't need filtering. The 0 entry in the tc table is always -1: chroma adds 1 to it, luma does nothing. In all cases, the value is indeed <= 0
<kurosu>
So the meaning of a <= 0 value is just a convenience to say "don't filter edge"
<kurosu>
maybe that could be caught earlier than in the leaf function, but ok, makes more sense, and you do have to implement it
Livio has quit [Quit: Reconnecting]
Livio has joined #ffmpeg-devel
___nick___ has joined #ffmpeg-devel
___nick___ has quit [Client Quit]
IndecisiveTurtle has joined #ffmpeg-devel
___nick___ has joined #ffmpeg-devel
compnn has joined #ffmpeg-devel
compn has quit [Read error: Connection reset by peer]
compn has joined #ffmpeg-devel
compnn has quit [Ping timeout: 256 seconds]
AbleBacon has joined #ffmpeg-devel
cone-892 has quit [Quit: transmission timeout]
IndecisiveTurtle has quit [Remote host closed the connection]