Re: [FFmpeg-user] libx265 a lot slower
On 2020-08-19 15:06 +0200, Moritz Barsnick wrote: > On Mon, Aug 17, 2020 at 22:53:12 +0200, Cecil Westerhof wrote: > > Maybe I am just dumb and I use the wrong questions, but when I > > searched I got a lot of hits that explained I needed tot switch > > because the space needed would be halved > > Well, if you're driving a compact, and someone writes "you need to > switch to a Jeep, because it can pull your trailer out of the mud", do > you buy a Jeep? ;-) > > I'm trying to say: Try to understand the benefits and the > disadvantages, and check against your requirements. > > Do you need less space? Are you willing to sacrifice encoding speed? Is > H.265 the right codec for your type of videos? As Mick Finn pointed > out, H.265 is more designed for higher resolutions and bit depths > (IIUC). Can your target players even decode H.265? (Does the Jeep even > fit into your parking spaces? Do your feet reach the pedals?) > > There's no matter of "must", "need to" - just of considerations. I > believe whatever you read, implied some other details, or forgot to > tell about them. > > BTW, if someone does magic hacking, or even more optimized GPUs emerge, > x265/H.265 *may* become "faster" than x264/H.264. It isn't right now. > > > (what is not true, I did see 'only' a 2/3) > > It depends on the material. I'm sure there are tons of comparisons out > there. > > > and a few explained that you could not use h265 > > everywhere, so you should evaluate if it was a good idea to switch. > > Absolutely! Even if I could encode my material to H.265, I probably > wouldn't, except when targetting a very specific player (modern smart > phones?). My PVR/STB, my TV, ..., cannot decode it. Possibly even my PC > doesn't have enough compute/GPU power to decode it in full HD. > > > In none of the hits I saw anything about a performance hit. Especially > > not a hit up to a factor three. > > They should. ;) Maybe also an interesting question to ask (I try to extend the car example): Do the cars available to you master the inclusion of new tech in a way that compares favourably to existing cars? Are you fine with the new feature - cost balance comparing to other cars? Are you willing to accept higher space and fuel usage of the big jeep for its extra features? Or without cars: Is there an encoder available to you that creates significantly smaller same quality output in less space while not taking ages to encode? Of course there are other similar questions to ask and it may depend on your use case. E.g. like Moritz mentioned which devices do you target or what is the enconding situation? (E.g. how much delay/energy usage can you accept?) Alexander ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
On Mon, Aug 17, 2020 at 22:53:12 +0200, Cecil Westerhof wrote: > Maybe I am just dumb and I use the wrong questions, but when I > searched I got a lot of hits that explained I needed tot switch > because the space needed would be halved Well, if you're driving a compact, and someone writes "you need to switch to a Jeep, because it can pull your trailer out of the mud", do you buy a Jeep? ;-) I'm trying to say: Try to understand the benefits and the disadvantages, and check against your requirements. Do you need less space? Are you willing to sacrifice encoding speed? Is H.265 the right codec for your type of videos? As Mick Finn pointed out, H.265 is more designed for higher resolutions and bit depths (IIUC). Can your target players even decode H.265? (Does the Jeep even fit into your parking spaces? Do your feet reach the pedals?) There's no matter of "must", "need to" - just of considerations. I believe whatever you read, implied some other details, or forgot to tell about them. BTW, if someone does magic hacking, or even more optimized GPUs emerge, x265/H.265 *may* become "faster" than x264/H.264. It isn't right now. > (what is not true, I did see 'only' a 2/3) It depends on the material. I'm sure there are tons of comparisons out there. > and a few explained that you could not use h265 > everywhere, so you should evaluate if it was a good idea to switch. Absolutely! Even if I could encode my material to H.265, I probably wouldn't, except when targetting a very specific player (modern smart phones?). My PVR/STB, my TV, ..., cannot decode it. Possibly even my PC doesn't have enough compute/GPU power to decode it in full HD. > In none of the hits I saw anything about a performance hit. Especially > not a hit up to a factor three. They should. ;) Cheers, Moritz ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
On 2020-08-17T18:07:54+0200, Phil Rhodes via ffmpeg-user wrote: > The reason people are complaining is nothing to do with the logic of what > you're saying. > It's because you're so incredibly impolite about it. And you're being rude for not following this mailinglist's netiquette. You're subscribed here long enough to know that top-posting is not allowed. > To put it succinctly, you are a very good example of what is wrong with open > source software in general. Was this really necessary? -- Reino Wijnsma ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
Am 17.08.20 um 22:53 schrieb Cecil Westerhof: > Reindl Harald writes: > >> we have 2020 - there is "performance h264 versus h265" which you can >> type in any search engine - disclaimer: educate yourself for free may >> make you smarter >> >> i don't get questions which can be answered by goggole at all because >> it#s so much faster and very few questions are not asked and answered >> doezns of times > > Maybe I am just dumb and I use the wrong questions, but when I > searched I got a lot of hits that explained I needed tot switch > because the space needed would be halved *need* to switch is nonsense you don't re-encode existng stuff with a losy codec anyways > (what is not true, I did see > 'only' a 2/3) and a few explained that you could not use h265 > everywhere which is also common sense your device from before H265 existed which don#t get firmware updates won't play H265 encoded stuff as devices from before H264 don't play H264 stuff > My question was not about that there was a performance hit, but about > that it was so high. in pure software be happy when you can *decode* it in realtime ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
Reindl Harald writes: > we have 2020 - there is "performance h264 versus h265" which you can > type in any search engine - disclaimer: educate yourself for free may > make you smarter > > i don't get questions which can be answered by goggole at all because > it#s so much faster and very few questions are not asked and answered > doezns of times Maybe I am just dumb and I use the wrong questions, but when I searched I got a lot of hits that explained I needed tot switch because the space needed would be halved (what is not true, I did see 'only' a 2/3) and a few explained that you could not use h265 everywhere, so you should evaluate if it was a good idea to switch. In none of the hits I saw anything about a performance hit. Especially not a hit up to a factor three. My question was not about that there was a performance hit, but about that it was so high. -- Cecil Westerhof Senior Software Engineer LinkedIn: http://www.linkedin.com/in/cecilwesterhof ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
Am 17.08.20 um 18:34 schrieb Carl Zwanzig: > > On 8/17/2020 1:53 AM, Phil Rhodes via ffmpeg-user wrote: >> People, be polite. Harald, you are incredibly impolite. This is a user >> list. It is reasonable for people to ask basic questions here. Your >> assumption that everyone is an expert is unreasonable. > Exactly. And many people will readily say that "common sense" isn't very > common. sadly yes, a shame in days where you can find every iformation on the internet within seconds i wonder how we survived the 1970's and 1980's > On 8/17/2020 8:12 AM, Reindl Harald wrote: >> i don't need knowledge about cars to guess that one needs more fuel >> when he drives double as fast while put a big stone in the luggage space >> >> i don't even need to own a car for that > Maybe not, but you _do_ need to have even the vaguest idea of > mass/energy equilibriums (even without knowing that term). Many people > have very little idea what goes on in a video codec, let alone why one > might be faster than another on some hardware or why some streams are > more compressible than others. > > Instead of saying "this is obvious! why are you asking??" you could just > patiently explain the answer. Or not answer at all and leave it to > someone else. we have 2020 - there is "performance h264 versus h265" which you can type in any search engine - disclaimer: educate yourself for free may make you smarter i don't get questions which can be answered by goggole at all because it#s so much faster and very few questions are not asked and answered doezns of times ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
On 8/17/2020 1:53 AM, Phil Rhodes via ffmpeg-user wrote: People, be polite. Harald, you are incredibly impolite. This is a user list. It is reasonable for people to ask basic questions here. Your assumption that everyone is an expert is unreasonable. Exactly. And many people will readily say that "common sense" isn't very common. On 8/17/2020 8:12 AM, Reindl Harald wrote: i don't need knowledge about cars to guess that one needs more fuel when he drives double as fast while put a big stone in the luggage space i don't even need to own a car for that Maybe not, but you _do_ need to have even the vaguest idea of mass/energy equilibriums (even without knowing that term). Many people have very little idea what goes on in a video codec, let alone why one might be faster than another on some hardware or why some streams are more compressible than others. Instead of saying "this is obvious! why are you asking??" you could just patiently explain the answer. Or not answer at all and leave it to someone else. z! ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
Am 17.08.20 um 18:07 schrieb Phil Rhodes via ffmpeg-user: > The reason people are complaining is nothing to do with the logic of what > you're saying. > It's because you're so incredibly impolite about it. so what aren't you the guy which wanted to modify his operating systems file dialog from within a webmail? :-) > On Monday, 17 August 2020, 16:12:37 BST, Reindl Harald > wrote: > > > > Am 17.08.20 um 16:47 schrieb Eduardo Alarcón: >> El lun., 17 ago. 2020 a las 7:19, Reindl Harald () >> escribió: >> >>> Am 17.08.20 um 10:53 schrieb Phil Rhodes via ffmpeg-user: Your assumption that everyone is an expert is unreasonable. >>> >>> you don't need to be an expert for common sense, thast's why it's called >>> that way neiter am i an expert for video codecs >>> >>> It's common sense that common sense does not exist > > don't let appear something like it has written by who you respond to! > >> People are not born with intrinsic knowledge that is applicable to >> everything. > > the next one which confuses common sense with knowledge > > i don't need knowledge about cars to guess that one needs more fuel when > he drives double as fast while put a big stone in the luggage space > > i don't even need to own a car for that > ___ > ffmpeg-user mailing list > ffmpeg-user@ffmpeg.org > https://ffmpeg.org/mailman/listinfo/ffmpeg-user > > To unsubscribe, visit link above, or email > ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe". > ___ > ffmpeg-user mailing list > ffmpeg-user@ffmpeg.org > https://ffmpeg.org/mailman/listinfo/ffmpeg-user > > To unsubscribe, visit link above, or email > ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe". > -- Reindl Harald the lounge interactive design GmbH A-1060 Vienna, Hofmühlgasse 17 CTO / CISO / Software-Development m: +43 676 40 221 40 p: +43 1 595 3999 33 http://www.thelounge.net/ GPG-Public-Key: https://arrakis-tls.thelounge.net/gpg/h.reindl_thelounge.net.pub.txt ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
The reason people are complaining is nothing to do with the logic of what you're saying. It's because you're so incredibly impolite about it. To put it succinctly, you are a very good example of what is wrong with open source software in general. No, it is not okay to do this because you think it somehow produces technical rigour. P On Monday, 17 August 2020, 16:12:37 BST, Reindl Harald wrote: Am 17.08.20 um 16:47 schrieb Eduardo Alarcón: > El lun., 17 ago. 2020 a las 7:19, Reindl Harald () > escribió: > >> Am 17.08.20 um 10:53 schrieb Phil Rhodes via ffmpeg-user: >>> Your assumption that everyone is an expert is unreasonable. >> >> you don't need to be an expert for common sense, thast's why it's called >> that way neiter am i an expert for video codecs >> >> It's common sense that common sense does not exist don't let appear something like it has written by who you respond to! > People are not born with intrinsic knowledge that is applicable to > everything. the next one which confuses common sense with knowledge i don't need knowledge about cars to guess that one needs more fuel when he drives double as fast while put a big stone in the luggage space i don't even need to own a car for that ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe". ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
Am 17.08.20 um 16:47 schrieb Eduardo Alarcón: > El lun., 17 ago. 2020 a las 7:19, Reindl Harald () > escribió: > >> Am 17.08.20 um 10:53 schrieb Phil Rhodes via ffmpeg-user: >>> Your assumption that everyone is an expert is unreasonable. >> >> you don't need to be an expert for common sense, thast's why it's called >> that way neiter am i an expert for video codecs >> >> It's common sense that common sense does not exist don't let appear something like it has written by who you respond to! > People are not born with intrinsic knowledge that is applicable to > everything. the next one which confuses common sense with knowledge i don't need knowledge about cars to guess that one needs more fuel when he drives double as fast while put a big stone in the luggage space i don't even need to own a car for that ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
El lun., 17 ago. 2020 a las 7:19, Reindl Harald () escribió: > > > Am 17.08.20 um 10:53 schrieb Phil Rhodes via ffmpeg-user: > > Your assumption that everyone is an expert is unreasonable. > > you don't need to be an expert for common sense, thast's why it's called > that way neiter am i an expert for video codecs > > It's common sense that common sense does not exist. People are not born with intrinsic knowledge that is applicable to everything. Common sense comes from the experience and knowledge of an individual, so it can be teached because you can explain and teach another why you reached a certain conclusion based on what facts and experiences. Common sense for you it's not common sense for another, and there are a lot of other people that will think that you don't have any common sense. it's simply logical that smaller files with the same or better quality > need more computing power and examples where it's possible are very rae > in the history > > > Am 17.08.20 um 02:07 schrieb Mark Filipak: > >> On 08/16/2020 05:30 PM, Reindl Harald wrote: > >>> Am 16.08.20 um 19:14 schrieb Carl Zwanzig: > Reindl is known for acerbic and unhelpful answers, AFAICT most readers > ignore them. > >>> > >>> there is nothing acerbic or unhelpful point to common sense that a new > >>> codec with better quality or smaller files (and if both is ecpected) > >>> comes with a logical cost > >> > >> That is illogical. There is no reason why one should expect that libx265 > >> will take longer to encode. If you have information to the contrary, you > >> could cite it instead of being insulting. > > > > i can't teach you common sense > > ___ > ffmpeg-user mailing list > ffmpeg-user@ffmpeg.org > https://ffmpeg.org/mailman/listinfo/ffmpeg-user > > To unsubscribe, visit link above, or email > ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe". ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
Reindl Harald writes: > Am 17.08.20 um 10:53 schrieb Phil Rhodes via ffmpeg-user: >> Your assumption that everyone is an expert is unreasonable. > > you don't need to be an expert for common sense, thast's why it's called > that way neiter am i an expert for video codecs I was not offended myself, but I have seen myself (albeit not with video compression) that changing things can save space and be faster. ;-) By the way: I did not expect it to be faster. I would not have been surprised if it was slower. But 2.5 to 3 times seemed a lot to me. But it seems it is not. -- Cecil Westerhof Senior Software Engineer LinkedIn: http://www.linkedin.com/in/cecilwesterhof ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
Am 17.08.20 um 10:53 schrieb Phil Rhodes via ffmpeg-user: > Your assumption that everyone is an expert is unreasonable. you don't need to be an expert for common sense, thast's why it's called that way neiter am i an expert for video codecs it's simply logical that smaller files with the same or better quality need more computing power and examples where it's possible are very rae in the history > Am 17.08.20 um 02:07 schrieb Mark Filipak: >> On 08/16/2020 05:30 PM, Reindl Harald wrote: >>> Am 16.08.20 um 19:14 schrieb Carl Zwanzig: Reindl is known for acerbic and unhelpful answers, AFAICT most readers ignore them. >>> >>> there is nothing acerbic or unhelpful point to common sense that a new >>> codec with better quality or smaller files (and if both is ecpected) >>> comes with a logical cost >> >> That is illogical. There is no reason why one should expect that libx265 >> will take longer to encode. If you have information to the contrary, you >> could cite it instead of being insulting. > > i can't teach you common sense ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
H.265 produces larger savings for larger resolutions than h274 )4k and higher). H265 requires much higher computation levels and works best when mapped to GPU. Sent from my iPhone > On Aug 17, 2020, at 12:02 AM, Cecil Westerhof wrote: > > Reindl Harald writes: > >>> Am 16.08.20 um 16:48 schrieb Cecil Westerhof: >>> I heard a lot that you should libx265 instead of libx264. I did not, >>> because at some places that went wrong. (I think uploading. But it is >>> several years ago, so I am not sure.) >>> I am again playing with ffmpeg and creating new scripts. >>> When using libx265 the file size is about a third smaller, but it >>> takes about 2.5 to 3 times longer to generate the file. Is this >>> normal, or a quirk at my side? >>> For the moment I stay with libx264 >> >> what do you expect? > > I did not expect anything, just noticed something. > > >> H264 is also a lot slower and more expensive comapred ot codes from the >> 1990s and you can't expect better quality and smaller files falling free >> from heaven > > For the moment I will keep with 264. Especially because these files > are only played once. Just wanted to make sure I was not overlooking > something. > > > By the way: when searching on the internet, I saw often said that 265 > would be half as big as 264, but I see 'only' a third less space > taken. Are the people saying 50% overly optimistic, or do I just have > 'strange' videos? > > -- > Cecil Westerhof > Senior Software Engineer > LinkedIn: http://www.linkedin.com/in/cecilwesterhof > ___ > ffmpeg-user mailing list > ffmpeg-user@ffmpeg.org > https://ffmpeg.org/mailman/listinfo/ffmpeg-user > > To unsubscribe, visit link above, or email > ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe". ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
For christ's sake. People, be polite. Harald, you are incredibly impolite. This is a user list. It is reasonable for people to ask basic questions here. Your assumption that everyone is an expert is unreasonable. P On Monday, 17 August 2020, 07:24:33 BST, Reindl Harald wrote: Am 17.08.20 um 02:07 schrieb Mark Filipak: > On 08/16/2020 05:30 PM, Reindl Harald wrote: >> Am 16.08.20 um 19:14 schrieb Carl Zwanzig: >>> Reindl is known for acerbic and unhelpful answers, AFAICT most readers >>> ignore them. >> >> there is nothing acerbic or unhelpful point to common sense that a new >> codec with better quality or smaller files (and if both is ecpected) >> comes with a logical cost > > That is illogical. There is no reason why one should expect that libx265 > will take longer to encode. If you have information to the contrary, you > could cite it instead of being insulting. i can't teach you common sense ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe". ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
Am 17.08.20 um 08:30 schrieb Mark Filipak: > On 08/17/2020 02:24 AM, Reindl Harald wrote: >> Am 17.08.20 um 02:07 schrieb Mark Filipak: >>> On 08/16/2020 05:30 PM, Reindl Harald wrote: Am 16.08.20 um 19:14 schrieb Carl Zwanzig: > Reindl is known for acerbic and unhelpful answers, AFAICT most readers > ignore them. there is nothing acerbic or unhelpful point to common sense that a new codec with better quality or smaller files (and if both is ecpected) comes with a logical cost >>> >>> That is illogical. There is no reason why one should expect that libx265 >>> will take longer to encode. If you have information to the contrary, you >>> could cite it instead of being insulting. >> >> i can't teach you common sense > > Troll well, try to encode H264 or even *decode* it on hardware from20 years ago where other video codecs existed nobody right in his brain can expect a codec based on H264 with better quality *and* lower filesize coming without any cost ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
On 08/17/2020 02:24 AM, Reindl Harald wrote: Am 17.08.20 um 02:07 schrieb Mark Filipak: On 08/16/2020 05:30 PM, Reindl Harald wrote: Am 16.08.20 um 19:14 schrieb Carl Zwanzig: Reindl is known for acerbic and unhelpful answers, AFAICT most readers ignore them. there is nothing acerbic or unhelpful point to common sense that a new codec with better quality or smaller files (and if both is ecpected) comes with a logical cost That is illogical. There is no reason why one should expect that libx265 will take longer to encode. If you have information to the contrary, you could cite it instead of being insulting. i can't teach you common sense Troll. ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
Am 17.08.20 um 02:07 schrieb Mark Filipak: > On 08/16/2020 05:30 PM, Reindl Harald wrote: >> Am 16.08.20 um 19:14 schrieb Carl Zwanzig: >>> Reindl is known for acerbic and unhelpful answers, AFAICT most readers >>> ignore them. >> >> there is nothing acerbic or unhelpful point to common sense that a new >> codec with better quality or smaller files (and if both is ecpected) >> comes with a logical cost > > That is illogical. There is no reason why one should expect that libx265 > will take longer to encode. If you have information to the contrary, you > could cite it instead of being insulting. i can't teach you common sense ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
On 08/16/2020 05:30 PM, Reindl Harald wrote: Am 16.08.20 um 19:14 schrieb Carl Zwanzig: Reindl is known for acerbic and unhelpful answers, AFAICT most readers ignore them. there is nothing acerbic or unhelpful point to common sense that a new codec with better quality or smaller files (and if both is ecpected) comes with a logical cost That is illogical. There is no reason why one should expect that libx265 will take longer to encode. If you have information to the contrary, you could cite it instead of being insulting. there is also nothing acerbic or unhelpful point to common sense that every "look here how good xyz is" is based on best-case and never reflects the reality That is also illogical. ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
Am 16.08.20 um 19:14 schrieb Carl Zwanzig: > Reindl is known for acerbic and unhelpful answers, AFAICT most readers > ignore them. there is nothing acerbic or unhelpful point to common sense that a new codec with better quality or smaller files (and if both is ecpected) comes with a logical cost there is also nothing acerbic or unhelpful point to common sense that every "look here how good xyz is" is based on best-case and never reflects the reality > On 8/16/2020 10:02 AM, Cecil Westerhof wrote: > >> For the moment I will keep with 264. Especially because these files >> are only played once. Just wanted to make sure I was not overlooking >> something. > > Probably not; for a one-use file, I'd take whichever one is easier to > deal with (which might mean quickest to encode). And depending on the > source material, and well, everything in the chain, you might use > another codec anyway, there's nothing mystical/magical about x264 (and a > few decidedly unfriendly things- ref "moov atom location"). > >> By the way: when searching on the internet, I saw often said that 265 >> would be half as big as 264, but I see 'only' a third less space >> taken. Are the people saying 50% overly optimistic, or do I just have >> 'strange' videos? > > "Never generalize." > > I'd take any size estimate as a guess since your content and encoding > parameters are probably different. If my own tests of x265 showed 30% > smaller but 2x the encode time, I wouldn't bother. ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
> Am 16.08.2020 um 19:14 schrieb Carl Zwanzig : > > Reindl is known for acerbic and unhelpful answers There are several reasons why you should be extremely careful with such comments, please consider this a little warning. Carl Eugen ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
> Am 16.08.2020 um 19:02 schrieb Cecil Westerhof : > > By the way: when searching on the internet, I saw often said that 265 > would be half as big as 264, but I see 'only' a third less space > taken. Are the people saying 50% overly optimistic, or do I just have > 'strange' videos? You completely misunderstand what „half as big“ means: It is easy to get an even smaller file with FFmpeg‘s mpeg1 video encoder. Carl Eugen ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
Am 16.08.20 um 19:02 schrieb Cecil Westerhof: > Reindl Harald writes: > >> Am 16.08.20 um 16:48 schrieb Cecil Westerhof: >>> I heard a lot that you should libx265 instead of libx264. I did not, >>> because at some places that went wrong. (I think uploading. But it is >>> several years ago, so I am not sure.) >>> I am again playing with ffmpeg and creating new scripts. >>> When using libx265 the file size is about a third smaller, but it >>> takes about 2.5 to 3 times longer to generate the file. Is this >>> normal, or a quirk at my side? >>> For the moment I stay with libx264 >> >> what do you expect? > > I did not expect anything, just noticed something. > >> H264 is also a lot slower and more expensive comapred ot codes from the >> 1990s and you can't expect better quality and smaller files falling free >> from heaven > > For the moment I will keep with 264. Especially because these files > are only played once. Just wanted to make sure I was not overlooking > something. > > By the way: when searching on the internet, I saw often said that 265 > would be half as big as 264, but I see 'only' a third less space > taken. Are the people saying 50% overly optimistic, or do I just have > 'strange' videos? people sell the typical much higher costs for encoding combined with untypical best-case results - nothing new the past 20 years no matter what topic ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
Reindl is known for acerbic and unhelpful answers, AFAICT most readers ignore them. On 8/16/2020 10:02 AM, Cecil Westerhof wrote: For the moment I will keep with 264. Especially because these files are only played once. Just wanted to make sure I was not overlooking something. Probably not; for a one-use file, I'd take whichever one is easier to deal with (which might mean quickest to encode). And depending on the source material, and well, everything in the chain, you might use another codec anyway, there's nothing mystical/magical about x264 (and a few decidedly unfriendly things- ref "moov atom location"). By the way: when searching on the internet, I saw often said that 265 would be half as big as 264, but I see 'only' a third less space taken. Are the people saying 50% overly optimistic, or do I just have 'strange' videos? "Never generalize." I'd take any size estimate as a guess since your content and encoding parameters are probably different. If my own tests of x265 showed 30% smaller but 2x the encode time, I wouldn't bother. Later, z! ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
Reindl Harald writes: > Am 16.08.20 um 16:48 schrieb Cecil Westerhof: >> I heard a lot that you should libx265 instead of libx264. I did not, >> because at some places that went wrong. (I think uploading. But it is >> several years ago, so I am not sure.) >> I am again playing with ffmpeg and creating new scripts. >> When using libx265 the file size is about a third smaller, but it >> takes about 2.5 to 3 times longer to generate the file. Is this >> normal, or a quirk at my side? >> For the moment I stay with libx264 > > what do you expect? I did not expect anything, just noticed something. > H264 is also a lot slower and more expensive comapred ot codes from the > 1990s and you can't expect better quality and smaller files falling free > from heaven For the moment I will keep with 264. Especially because these files are only played once. Just wanted to make sure I was not overlooking something. By the way: when searching on the internet, I saw often said that 265 would be half as big as 264, but I see 'only' a third less space taken. Are the people saying 50% overly optimistic, or do I just have 'strange' videos? -- Cecil Westerhof Senior Software Engineer LinkedIn: http://www.linkedin.com/in/cecilwesterhof ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] libx265 a lot slower
Am 16.08.20 um 16:48 schrieb Cecil Westerhof: > I heard a lot that you should libx265 instead of libx264. I did not, > because at some places that went wrong. (I think uploading. But it is > several years ago, so I am not sure.) > I am again playing with ffmpeg and creating new scripts. > When using libx265 the file size is about a third smaller, but it > takes about 2.5 to 3 times longer to generate the file. Is this > normal, or a quirk at my side? > For the moment I stay with libx264 what do you expect? H264 is also a lot slower and more expensive comapred ot codes from the 1990s and you can't expect better quality and smaller files falling free from heaven ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
[FFmpeg-user] libx265 a lot slower
I heard a lot that you should libx265 instead of libx264. I did not, because at some places that went wrong. (I think uploading. But it is several years ago, so I am not sure.) I am again playing with ffmpeg and creating new scripts. When using libx265 the file size is about a third smaller, but it takes about 2.5 to 3 times longer to generate the file. Is this normal, or a quirk at my side? For the moment I stay with libx264. -- Cecil Westerhof Senior Software Engineer LinkedIn: http://www.linkedin.com/in/cecilwesterhof ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".