4 ms·

Yes, he'll fix the speedtest flag thing (which BTW, several people were accidentally using it wrong because the feature is unergonomic AF, but it's convenient t

Yes, he'll fix the speedtest flag thing (which BTW, several people were accidentally using it wrong because the feature is unergonomic AF, but it's convenient to blame the user for "holding it wrong" right?). No, single-threaded encoding/decoding is a valid use case and not a testing methodology flaw. > ... to be forthright about how significant codec optimization historically comes after adoption. Well, this trend has now been broken, so it's irrelevant to mention it. Codecs are now expected to be optimized well before adoption. Take the case of AV2: the AOM folks are currently improving the reference encoder, libavm. SVT-AV2 has just been announced. This is the reality we're now live in. The fact we're having this conversation is an instance of that trend change. > "I have no reason to believe they'd be better anyway". Why are you now selectively quoting fragments? That sentence goes "...and I have no reason to believe they'd be better *than dir-pred* anyway." The part you omitted makes the entire difference. Implementing splines will improve JXL's efficiency (again, this was never contested), but it IS unlikely that they'll fully compensate for the lack of dir-pred. You know why? I *did* take a shot at writing an automated splines implementation for libjxl, and I failed. Not because I didn't know what I had to do, but because I couldn't find a quick way to generate high-quality spline candidates that improve overall quality while making up for their size and encode compute overhead. I'm sure that there's a hypothetical clever way to do it, but the point is that the equivalent tool (dir-pred) is dead easy to implement in comparison, done countless times independently, and it just works. Efficiently leveraging JXL's splines into the encoding loop is legitimately a very hard problem. No, problems of this kind shouldn't be this hard, and it's healthy to call this stuff out instead of pretending that, some time in the future, the "potential" of "alien technology" coding tools will somehow be untapped. > You see, any unfilled area is specifically mentioned only to dismiss it as meaningless. Oh, you're still doing the weird "subtext" thing... you know what? I'm done. Have a nice day.

computerbuster 17d ago
Well-put.
BugsJustFindMe 17d ago
> Codecs are now expected to be optimized well before adoption. Maybe I hallucinated, but I seem to recall that your very fine AVIF work happened only well after major browser adoption. "Rules for thee, not for me", I guess. Talk about trying to pull up the ladder. > SVT-AV2 has just been announced Best of luck to it and AV2. That's a non-sequitur. > The part you omitted makes the entire difference. It really does not. Because... > Implementing splines will improve JXL's efficiency (again, this was never contested) your given position is that it will never be improved enough. That is the problem. That your position is that it will never be improved enough to be better than AVIF (I mean again, though it was previously until quite recently) is entirely the problem. You grip tight to this idea now that if JXL ever improves even one iota then you're vindicated in having graciously allowed it to improve, but you have not granted it the capacity to be improved any bit more than AVIF. > than dir-pred This is what I'm talking about. > I *did* take a shot at writing an automated splines implementation for libjxl, and I failed. I'm sure you're very good and very smart, and you've clearly done very good things. You're not the only very good and very smart person, of course. But you do have very strong opinions about what is and is not worthwhile for the world to pursue. > Oh, you're still doing the weird "subtext" thing. What can I say. I'm used to very good and very smart people knowing about subtext, so I guess I assumed you would.
- lmp88959 16d ago
  I don't know why you are being condescending to Julio. His work has pushed AV1 forward significantly. His opinions (specifically about dir-pred) are shared by others who are in the encoding space because they're objectively true. If a breakthrough for splines is invented, we'd be elated, but as it stands, dir-pred is in virtually every video/image coding standard since the late 90s for a VERY good reason. Dir-pred also has two decades of iterative improvement and research behind it. The argument is that JXL's lack of this well-established coding tool is a serious handicap that will make it very hard - if not impossible - to compete with AVIF within a reasonable amount of time. AVIF was performing better than JXL/WebP/JPEG for a long while, even before Julio and Gianni's tune IQ work happened. AVIF successfully met the bar, tune IQ further increased that efficiency gap. AV2 needs optimized encoders and decoders because it now needs to justify its spot among AV1 and JXL on the web. That's why avm, svt-av1 and dav2d are well underway.

computerbuster 17d ago

Well-put.

BugsJustFindMe 17d ago

> Codecs are now expected to be optimized well before adoption. Maybe I hallucinated, but I seem to recall that your very fine AVIF work happened only well after major browser adoption. "Rules for thee, not for me", I guess. Talk about trying to pull up the ladder. > SVT-AV2 has just been announced Best of luck to it and AV2. That's a non-sequitur. > The part you omitted makes the entire difference. It really does not. Because... > Implementing splines will improve JXL's efficiency (again, this was never contested) your given position is that it will never be improved enough. That is the problem. That your position is that it will never be improved enough to be better than AVIF (I mean again, though it was previously until quite recently) is entirely the problem. You grip tight to this idea now that if JXL ever improves even one iota then you're vindicated in having graciously allowed it to improve, but you have not granted it the capacity to be improved any bit more than AVIF. > than dir-pred This is what I'm talking about. > I *did* take a shot at writing an automated splines implementation for libjxl, and I failed. I'm sure you're very good and very smart, and you've clearly done very good things. You're not the only very good and very smart person, of course. But you do have very strong opinions about what is and is not worthwhile for the world to pursue. > Oh, you're still doing the weird "subtext" thing. What can I say. I'm used to very good and very smart people knowing about subtext, so I guess I assumed you would.

lmp88959 16d ago
I don't know why you are being condescending to Julio. His work has pushed AV1 forward significantly. His opinions (specifically about dir-pred) are shared by others who are in the encoding space because they're objectively true. If a breakthrough for splines is invented, we'd be elated, but as it stands, dir-pred is in virtually every video/image coding standard since the late 90s for a VERY good reason. Dir-pred also has two decades of iterative improvement and research behind it. The argument is that JXL's lack of this well-established coding tool is a serious handicap that will make it very hard - if not impossible - to compete with AVIF within a reasonable amount of time. AVIF was performing better than JXL/WebP/JPEG for a long while, even before Julio and Gianni's tune IQ work happened. AVIF successfully met the bar, tune IQ further increased that efficiency gap. AV2 needs optimized encoders and decoders because it now needs to justify its spot among AV1 and JXL on the web. That's why avm, svt-av1 and dav2d are well underway.