that's right folks, now your life will be complete, because I have decided to review divx's h265 encoder.
for this review i tested with a variety of sources, including adult HD content sourced from blu-ray's, main stream blu-rays, cartoons like south park and some dvd sourced SD content. for the sake of a fair comparison, all my test clips were created by firing up xvid4psp, loading the source, clipping atest segment and exporting it with x264 lossless, which resulted in testing clips of about 5 gigs for about 4 minutes of content. audio was not included.
i then loaded the clips into one of the following applications for encoding:
divx converter, using the hevc 1080p profile
handbrake, using x264 with custom settings, such as sub me = 11, tesa. 8 reference frames, no dct decimate, all the b frame options at the highest settings including auto, deblocking 0,0; i also used 3 b frames, psy-rd, aq and mb-tree were all at there default settings and trellis was at 2; partition type was set to "some", here is the encoder options line:
ref=8:b-adapt=2:direct=auto:me=tesa: subme=11:merange=24:analyse=i4x4,i8x8:trellis=2:no-dct-decimate=1
for this test i wanted to stick with legally free software, so nothing from encoders sourced via questionable methods.
so who wins this shootout, divx's hevc or x264 with the settings maxed out? i bet the people that are reading this are expecting me to use this as an opportunity to bash x264 and it's developers and proclaim divx's offering the clear winner.
i guess then you guys haven't been paying attention. i don't bash x264 and/or it's developers just to be a dick or to troll, i offer my honest opinion. and my honest opinion in divx hevc vs x264 is that there is no clear cut answer.
in test after test, depending on the source, the results in many ways were close enough that i could make an argument to use either encoder.
if you're starting with a really high quality source, like say the blu-ray of Lincoln (the 2012 release) some of the adult blu-rays released by Digital Playground or Private, and used ample bit rate, say 12mb/s for 1080p or went down to 8mb/s for 720p, then i would say the two test encoders trade punches while the divx hevc encoder may retain some additional detail the difference isn't that great that a casual viewer would notice or would care.
if you start bit starving your encode and say try and go down to something completely silly like 4mb/s for 1080p, then the differences become quite clear and the divx offering wins hands down but since i don't have a bit staring fetish the point is a moot one for me.
likewise with SD spec cartoons, like south park, even at less than 2mb/s one really needs to split hairs to say one is better than the other.
where the divx hevc encoder is the clear winner is when you have a source that's of less than perfect quality and you need to retain as much detail as possible, in those cases you would be insane to go with x264. for this test i took an adult dvd named Sodomania 3 and a couple of greek comedy dvd's. these sources have a few things in common: the cameras used to film them were clearly budget cameras even in their era, the lighting guy needed to go back to school, the encoding software used was one step above pathetic and the dvd's need a lot of work to clean them up.
i loaded these dvd's into xvid4psp, cut a segment of each, applied qtgmc to deinterlace, 10bit Denoise MD for denoising and exported the results with x264 lossless. as before these became my test clips and i loaded them into divx and handbrake and created 3mb/s 720x480 encodes. in this scenario divx hevc blew x264 away, when retaining as much detail as possible is of the utmost importance and any loss of detail is easily noticed due to the poor quality of the source, there is no choice but to use the latest codec technology.
there are some limitations to this divx hevc encoder and they may be a deal breaker for some people. first things first, since rovi/divx/main concept sell high end encoding solutions such as totalcode studio and license sdk's for use by various vendors, they understandably don't include all the bells and whistles in the free version that's included in the paid versions. that means other than setting the bit rate and the encoding resolution you have zero control over the encoder settings, you can't set a gop size, you can't set the number of b frames, you can't even set a quality setting.
there is a also a problem under some circumstances with rate control, in a handful of tests where i loaded a 1080p source and targeted 720p as output, i set the bit rate at 4mb/s but divx hevc only used 3mb/s. interestingly enough, i didn't notice at first, but i had done a 1080p->720p test using handbrake and the above mentioned x264 settings at 4mb/s and i was comparing the results and i was saying to myself how surprising it was that the divx hevc encode was only slightly better than the x264 encode. then i noticed a big size discrepancy in the outputted files and upon investigation noticed that the hevc encode had undershot the 4mb/s target rate by 1mb/s, so with 25% less bit rate divx hevc still beat x264 with the most aggressive settings.
in terms of speed, divx hevc is slow, even with an i7 3770k. it's well threaded as it loads up all cores, but as slow as x264 is with subme=11 + tesa + 8 reference frames, divx hevc "feels" slower still. it's not possible to tell an actual encoding speed as divx converter doesn't show fps or encode time but my gut feeling is that it's slower than x264 + placebo probably by quite a bit. it also offers an overall superior quality encode so...
all in all how you view divx's hevc encoder is a matter of prospective. an argument could be made that Strongene's hevc encoder offers higher quality. what can't be argued is that divx's offering is easily the most important hevc encoder on the market from a market penetration standpoint, i.e. from the ability to bring hevc to the masses.
where Strongene has decided to make it's money by marketing it's encoder to God knows who, as evidenced by it offering both a 32 and 64 bit variant, only offering it as a direct show filter not a stand alone encoder, they license it to vendors at $2 a license with a minimum of 20,000 licenses required, they clearly are targeting ISV's.
x265 meanwhile is run by retarded chimps that know jack about managing a software project, less about marketing the results and fronted by a man that will probably run the company into the ground. they're trying to make money off of "free" by copying x264llc's licensing scheme, a scheme that is good enough if you want to make enough money to buy a loaf of bread but not if you wish to build a big company.
the divx people meanwhile are experts in marketing. say what you will about there encoders but they know how to market a product. thanks to there efforts divx became synonymous mpeg-4 asp encoding, makers of dvd and blu-ray players felt obligated to support divx encoded content whether it was asp or avc, and competitors encoders such as xvid and x264 had guides written for them on how to create divx complaint streams using open source variants. thanks to this market penetration divx earned hundreds of millions of dollars and was able to buy main concept for something over 150 million and then was bought out by rovi for just under 100 million. this is some serious cash, generated by smart men and women who understood how to market a codec to the public.
this is why divx's hevc is destined to become the go to hevc encoder of choice. divx already has worked out licensing agreements with hardware vendors for soon to be released divx certified hardware players that support divx's hevc, so any worries about the content you encode using divx hevc being playable only on a computer are alleviated. that's something neither Strongene nor x265 can offer.
more importantly, the underlying technology is licensed by numerous vendors, in the past sony, elecard, cyberlink, pegasys and cyberlink has already announced that in 2014 it will be releasing an updated power director with hevc encoding support. the tmpg people always are among the first to include a new encoding technology, they were the first to include a spurs engine plug in, the first to feature a cuda encoder, the first to offer support got intel's QS technology so there's no reason to think they won't be offering support for the divx/main concept hevc encoder with configurable parameters. MC also has made Open CL and CUDA powered h264 encoders and considering hevc is said to be easier to gpu accelerate there is no reason to think that the divx/rovi/main concept sdk won't be featuring support for a gpu powered hevc encoder sooner rather than later.
all in all divx's hevc encoder is a solid start, offering great quality albeit at a high speed cost. as i said an argument could be made either way as to whether one should use x264 with all the settings maxed out or divx hevc and i certainly wouldn't argue if one were to advise a newbie to stick with x264 + one of the slower settings for the time being.
that being said, for me, the time for any h264 encoder is done, the hevc era is here and i plan on doing all my encodes using divx hevc.
besides, it gives me a good excuse to upgrade my computer sometime as soon as i get a few bucks.
for this review i tested with a variety of sources, including adult HD content sourced from blu-ray's, main stream blu-rays, cartoons like south park and some dvd sourced SD content. for the sake of a fair comparison, all my test clips were created by firing up xvid4psp, loading the source, clipping atest segment and exporting it with x264 lossless, which resulted in testing clips of about 5 gigs for about 4 minutes of content. audio was not included.
i then loaded the clips into one of the following applications for encoding:
divx converter, using the hevc 1080p profile
handbrake, using x264 with custom settings, such as sub me = 11, tesa. 8 reference frames, no dct decimate, all the b frame options at the highest settings including auto, deblocking 0,0; i also used 3 b frames, psy-rd, aq and mb-tree were all at there default settings and trellis was at 2; partition type was set to "some", here is the encoder options line:
ref=8:b-adapt=2:direct=auto:me=tesa: subme=11:merange=24:analyse=i4x4,i8x8:trellis=2:no-dct-decimate=1
for this test i wanted to stick with legally free software, so nothing from encoders sourced via questionable methods.
so who wins this shootout, divx's hevc or x264 with the settings maxed out? i bet the people that are reading this are expecting me to use this as an opportunity to bash x264 and it's developers and proclaim divx's offering the clear winner.
i guess then you guys haven't been paying attention. i don't bash x264 and/or it's developers just to be a dick or to troll, i offer my honest opinion. and my honest opinion in divx hevc vs x264 is that there is no clear cut answer.
in test after test, depending on the source, the results in many ways were close enough that i could make an argument to use either encoder.
if you're starting with a really high quality source, like say the blu-ray of Lincoln (the 2012 release) some of the adult blu-rays released by Digital Playground or Private, and used ample bit rate, say 12mb/s for 1080p or went down to 8mb/s for 720p, then i would say the two test encoders trade punches while the divx hevc encoder may retain some additional detail the difference isn't that great that a casual viewer would notice or would care.
if you start bit starving your encode and say try and go down to something completely silly like 4mb/s for 1080p, then the differences become quite clear and the divx offering wins hands down but since i don't have a bit staring fetish the point is a moot one for me.
likewise with SD spec cartoons, like south park, even at less than 2mb/s one really needs to split hairs to say one is better than the other.
where the divx hevc encoder is the clear winner is when you have a source that's of less than perfect quality and you need to retain as much detail as possible, in those cases you would be insane to go with x264. for this test i took an adult dvd named Sodomania 3 and a couple of greek comedy dvd's. these sources have a few things in common: the cameras used to film them were clearly budget cameras even in their era, the lighting guy needed to go back to school, the encoding software used was one step above pathetic and the dvd's need a lot of work to clean them up.
i loaded these dvd's into xvid4psp, cut a segment of each, applied qtgmc to deinterlace, 10bit Denoise MD for denoising and exported the results with x264 lossless. as before these became my test clips and i loaded them into divx and handbrake and created 3mb/s 720x480 encodes. in this scenario divx hevc blew x264 away, when retaining as much detail as possible is of the utmost importance and any loss of detail is easily noticed due to the poor quality of the source, there is no choice but to use the latest codec technology.
there are some limitations to this divx hevc encoder and they may be a deal breaker for some people. first things first, since rovi/divx/main concept sell high end encoding solutions such as totalcode studio and license sdk's for use by various vendors, they understandably don't include all the bells and whistles in the free version that's included in the paid versions. that means other than setting the bit rate and the encoding resolution you have zero control over the encoder settings, you can't set a gop size, you can't set the number of b frames, you can't even set a quality setting.
there is a also a problem under some circumstances with rate control, in a handful of tests where i loaded a 1080p source and targeted 720p as output, i set the bit rate at 4mb/s but divx hevc only used 3mb/s. interestingly enough, i didn't notice at first, but i had done a 1080p->720p test using handbrake and the above mentioned x264 settings at 4mb/s and i was comparing the results and i was saying to myself how surprising it was that the divx hevc encode was only slightly better than the x264 encode. then i noticed a big size discrepancy in the outputted files and upon investigation noticed that the hevc encode had undershot the 4mb/s target rate by 1mb/s, so with 25% less bit rate divx hevc still beat x264 with the most aggressive settings.
in terms of speed, divx hevc is slow, even with an i7 3770k. it's well threaded as it loads up all cores, but as slow as x264 is with subme=11 + tesa + 8 reference frames, divx hevc "feels" slower still. it's not possible to tell an actual encoding speed as divx converter doesn't show fps or encode time but my gut feeling is that it's slower than x264 + placebo probably by quite a bit. it also offers an overall superior quality encode so...
all in all how you view divx's hevc encoder is a matter of prospective. an argument could be made that Strongene's hevc encoder offers higher quality. what can't be argued is that divx's offering is easily the most important hevc encoder on the market from a market penetration standpoint, i.e. from the ability to bring hevc to the masses.
where Strongene has decided to make it's money by marketing it's encoder to God knows who, as evidenced by it offering both a 32 and 64 bit variant, only offering it as a direct show filter not a stand alone encoder, they license it to vendors at $2 a license with a minimum of 20,000 licenses required, they clearly are targeting ISV's.
x265 meanwhile is run by retarded chimps that know jack about managing a software project, less about marketing the results and fronted by a man that will probably run the company into the ground. they're trying to make money off of "free" by copying x264llc's licensing scheme, a scheme that is good enough if you want to make enough money to buy a loaf of bread but not if you wish to build a big company.
the divx people meanwhile are experts in marketing. say what you will about there encoders but they know how to market a product. thanks to there efforts divx became synonymous mpeg-4 asp encoding, makers of dvd and blu-ray players felt obligated to support divx encoded content whether it was asp or avc, and competitors encoders such as xvid and x264 had guides written for them on how to create divx complaint streams using open source variants. thanks to this market penetration divx earned hundreds of millions of dollars and was able to buy main concept for something over 150 million and then was bought out by rovi for just under 100 million. this is some serious cash, generated by smart men and women who understood how to market a codec to the public.
this is why divx's hevc is destined to become the go to hevc encoder of choice. divx already has worked out licensing agreements with hardware vendors for soon to be released divx certified hardware players that support divx's hevc, so any worries about the content you encode using divx hevc being playable only on a computer are alleviated. that's something neither Strongene nor x265 can offer.
more importantly, the underlying technology is licensed by numerous vendors, in the past sony, elecard, cyberlink, pegasys and cyberlink has already announced that in 2014 it will be releasing an updated power director with hevc encoding support. the tmpg people always are among the first to include a new encoding technology, they were the first to include a spurs engine plug in, the first to feature a cuda encoder, the first to offer support got intel's QS technology so there's no reason to think they won't be offering support for the divx/main concept hevc encoder with configurable parameters. MC also has made Open CL and CUDA powered h264 encoders and considering hevc is said to be easier to gpu accelerate there is no reason to think that the divx/rovi/main concept sdk won't be featuring support for a gpu powered hevc encoder sooner rather than later.
all in all divx's hevc encoder is a solid start, offering great quality albeit at a high speed cost. as i said an argument could be made either way as to whether one should use x264 with all the settings maxed out or divx hevc and i certainly wouldn't argue if one were to advise a newbie to stick with x264 + one of the slower settings for the time being.
that being said, for me, the time for any h264 encoder is done, the hevc era is here and i plan on doing all my encodes using divx hevc.
besides, it gives me a good excuse to upgrade my computer sometime as soon as i get a few bucks.