the best video encoding app currently available?

[deadrats]

it may just be cyberlink's MediaEspresso 6:

http://www.cyberlink.com/products/mediaespresso/overview_en_US.html

i downloaded the 30 day trial (h264 has a further limitation of 30 trial runs, but this little app offers things i haven't seen any where else yet:

you have the option of having the cpu handle everything or you can choose gpu assisted decoding AND encoding of not only h264 (which can be found in at least half a dozen other apps) but mpeg-2 (the only other place i have seen gpu assisted mpeg-2 is in cyberlink's power director 7 and in a vfw mpeg-2 codec who's name escapes me at the moment) as well as wmv and i think divx as well, though the trial doesn't allow you to use the divx export option and thus i can't confirm my suspicions.

it also has 2 video enhancement technologies which i tested and they seem to work pretty well, also gpu assisted.

it doesn't appear to do all the encoding on the gpu as my quad core was still maxed out but it was definitely faster on my 9600 gso with gpu assist turned on.

one major downside is that it doesn't seem to have any option to de-interlace but you may be able to get around this with avisynth (didn't try to feed a script to it and see if it will accept it).

regardless, since this app supports both cuda and ati's stream technology, and an ati 5570 with 400 stream processors and 1 gig of video ram costs about $100, this little app, at least according to my initial tests, seems to allow one to build a pretty formidable computer, from a video encoding standpoint, without having to spend an arm and a leg, likewise it will allow the casual encoder that doesn't need professional grade encoding the ability to upgrade to a faster video card for a very reasonable price and have a pc that can encode the most common format fairly quickly.

i'd like to hear from anyone else that has tested this app and whether or not they liked it.

[jagabo]

How does the quality compare to x264 and Divx (their h.264 encoder, not the ASP encoder)?

[deadrats]

How does the quality compare to x264 and Divx (their h.264 encoder, not the ASP encoder)?

that's kind of a loaded question, if you're of the mind set to use the least amount of bit rate possible, then nothing is going to beat x264, that's what it's designed for, using bit rates that are absurdly low (just look at all the 1080p movies floating around the net encoded at 4 mb/s).

once you start getting into what i would consider "normal" bit rates any difference in quality effectively ceases to exist (i have 1080p flv files encoded at 12 mb/s and xvid avi's encoded at 10 mb/s that are stunning).

what this app shines at is gpu assisted mpeg-2, mpeg-4 (asp and avc) and wmv encoding and decoding as well as some pretty nice video enhancing technologies at a reasonable price, so i guess the answer is at low bit rates, say under 5 mb/s for 720p and higher rez files it can't compete with x264 though from my tests it seems to stack up pretty nicely with main concept's avc encoder (which is what the divx h264 is based on).

the above is based on an apples to apples comparison, i.e. a straight gpu assisted encode using espresso 6 versus using x264, the answer becomes quite a bit more complicated if you compare a low bit rate encode (say 720p at 3 mb/s) using x264 vs using espresso and activating the video enhancing technologies they have built into the app and you are starting with a high quality source, an argument could be made that espresso comes out on top.

in fact, in test after test with mediocre adult dvd sources from the late '90's, the 720x480 mpeg-2 at 3 mb/s encodes i was able to produce could not be matched, much less beaten, by either tmpg express using it's denoise and color correction filters and outputted to avi using the x264 vfw plug in nor could i beat it with virtual dub mod, using any of the vdub denoise or color correction filters i have (i have the pack of about 200 filters) coupled with x264.

so for me, at least so far, this little app has been pretty impressive, that's one of the reasons i wanted others to test it to to see if they had similar experiences.

[poisondeathray]

once you start getting into what i would consider "normal" bit rates any difference in quality effectively ceases to exist (i have 1080p flv files encoded at 12 mb/s and xvid avi's encoded at 10 mb/s that are stunning).

I'm going to disagree and say that you hold that advantage even at normal bitrates. The quality difference is visible to the naked eye, even normal playing. You don't even have to "pixel peep" or use objective measures like psnr or ssim. The quality at a given bitrate will still be higher with x264 i.e. it will only require maybe 60-80% of the filesize for similar quality if you had used x264 instead of xvid . If you use objective measures, it's a lot worse for xvid/divx . You can tune x264 for ssim or psnr (they score higher, but don't look as good subjectively in most people's eyes) . If you use the tunings ,it destroys xvid - you typically only need 50-60% of the filesize for equivalent quality at normal bitrate ranges.

Regarding Cyberlinnk Expresso, it was tested at Doom9 awhile back. I'm not sure if they had improved it, but it was based on the same nvcuenc.dll (same as the GPU assisted one mediacoder uses, and the cuda SDK ) - basically crap quality, even worse than badaboom. Maybe the one you're using is new and improved ?

what this app shines at is gpu assisted mpeg-2, mpeg-4 (asp and avc) and wmv encoding and decoding as well as some pretty nice video enhancing technologies at a reasonable price, so i guess the answer is at low bit rates, say under 5 mb/s for 720p and higher rez files it can't compete with x264 though from my tests it seems to stack up pretty nicely with main concept's avc encoder (which is what the divx h264 is based on).

I've read a bit about the "video enhancing" technology . The reviews I saw when it came out basically said it was garbage, but these were magazine reviews, and not really good ones. Can you shed some light on that ?

the above is based on an apples to apples comparison, i.e. a straight gpu assisted encode using espresso 6 versus using x264, the answer becomes quite a bit more complicated if you compare a low bit rate encode (say 720p at 3 mb/s) using x264 vs using espresso and activating the video enhancing technologies they have built into the app and you are starting with a high quality source, an argument could be made that espresso comes out on top.

in fact, in test after test with mediocre adult dvd sources from the late '90's, the 720x480 mpeg-2 at 3 mb/s encodes i was able to produce could not be matched, much less beaten, by either tmpg express using it's denoise and color correction filters and outputted to avi using the x264 vfw plug in nor could i beat it with virtual dub mod, using any of the vdub denoise or color correction filters i have (i have the pack of about 200 filters) coupled with x264.

That's not really an apples to apples comparison . You are testing the effects of filters, and confounding the analysis (again ) . You want to test 1 variable at a time. e.g. just the encoders in a given test scenario , like measuring quality (psnr or ssim) at certain bitrate, or encoding speed at a given set of encoding parameters, and then do a Quality/speed analysis.

Post some examples

[deadrats]

I'm going to disagree and say that you hold that advantage even at normal bitrates. The quality difference is visible to the naked eye, even normal playing. You don't even have to "pixel peep" or use objective measures like psnr or ssim. The quality at a given bitrate will still be higher with x264 i.e. it will only require maybe 60-80% of the filesize for similar quality if you had used x264 instead of xvid . If you use objective measures, it's a lot worse for xvid/divx . You can tune x264 for ssim or psnr (they score higher, but don't look as good subjectively in most people's eyes) . If you use the tunings ,it destroys xvid - you typically only need 50-60% of the filesize for equivalent quality at normal bitrate ranges.

as i pointed out in the other thread, starving a video of bit rate is no longer necessary when you can buy a huge hdd at a very reasonable rate, thus the fact that you can use just half the bit rate with x264 is irrelevant. in my own tests, as long as you start with clean source, once you hit the 10 mb/s range x264's quality advantage effectively disappears, and once you get into the 15 mb/s range even mpeg-2 becomes a viable option. of course this also has a lot to do with your source, if your source is a 4 mb/s 1080p encode of a blu-ray rip you downloaded off the pirate bay or an HDTV capture of a prize fight that some guy recorded using a $80 HDTV tuner, then yes, you pretty much have to use x264 if you want to maintain quality.

if however you are starting with high quality source that you recorded on your own $1000+ HD camcorder or a blu-ray rip from a properly, professionally encoded commercial blu-ray (especially some of the "private label" adult blu-rays that originate in europe, they are so crystal clear you would swear you were in the room where the filming is taking place), then the reality is that i don't think anyone will be able to see any difference between x264 and even an mpeg-2 transcode (<--so long as you use a good mpeg-2 encoder, like cce).

Regarding Cyberlinnk Expresso, it was tested at Doom9 awhile back. I'm not sure if they had improved it, but it was based on the same nvcuenc.dll (same as the GPU assisted one mediacoder uses, and the cuda SDK ) - basically crap quality, even worse than badaboom. Maybe the one you're using is new and improved ?

i'm not sure were the chaps over at doom got their information but looking through the espresso install directory shows no sign of nvcuenc.dll but as i'm sure you know nvidia's reference h264 encoder doesn't support mpeg-2 or wmv gpu assisted encoding nor does it work on ati video cards, this app uses cuda to do hardware accelerated encode and decode of h264, mpeg-4, mpeg-2 and wmv on nvidia cards and open cl to do the same on ati cards.

furthermore part of me suspects that some of the poor encoding quality seen on other cuda enabled apps may be due to improperly tuned drivers, the hardware not being able to take full advantage of the software (i suspect that the latest DX11 cards from ati and nvidia would provide much cleaner encodes) as well as the app passing incorrect parameters to the encoder.

in a test of ImToo, which features cuda accelerated h264 encoding, i found that using the default settings invariably resulted in shitty quality, no matter how much bit rate i threw at it but when i sat and manually tweaked various settings (for instance choosing the "same quality" option) and some other tweaks, the quality of the encode sky rocketed from shitty to pretty good. depending on the quality of the source and the target audience of the final output "pretty good" might be good enough.

I've read a bit about the "video enhancing" technology . The reviews I saw when it came out basically said it was garbage, but these were magazine reviews, and not really good ones. Can you shed some light on that ?

what you read was true, up until this version of espresso, cyberlink's "enhancing" technology, at least in power director, was pure shit, it's color correction filters resulted in colors that seemed artificially enhanced, kind of like the effect seen in many wmv encodes were the color saturation seems way to high, it would blend and or blur pixels in an effort to eliminate noise, it really was about the worst at "enhancing" video.

with this release, at least in my early tests, it seems like they finally figured out how to actually enhance video, they claim that they applied the technology they use in their media player that cleans up and upscales dvd to this encoding app and if you're curious to see how well it works (or, if at all), cyberlink has what it's calling "TrueTheater Enhancer" that is a browser plug in that increases the quality of youtube videos:

http://www.cyberlink.com/products/truetheater-enhancer/overview_en_US.html

Post some examples

i don't know, i have a strange feeling that posting a clip of some chick face down and ass up may be against the rules of this forum, mods i'm going to need a ruling on this

[deadrats]

just an update for anyone that was following or may have stumbled onto this thread. i ran a few more tests, using a test file from an adult site with a massage theme and a parlor (that should be enough of a hint for anyone that wants to see the videos for himself), this file is VC-1 in WMV container, video is 8 mb/s, audio is wma 9.2 192 kb/s 48 khz 18:44 min long, 1440x1080 and it seems to have some weird header problem, as most players play the file pillar boxed with the incorrect aspect ratio and media info reports an odd resolution, kmplayer however has no problem playing the file back properly.

i used the latest version of tmpg express (downloaded the 14 day trail) and the latest trial version of espresso (6), i chose the following settings:

espresso:

mp4, h264 at 10 mb/s, 128 kb/s 44 khz aac, 1440x1080 16:9, 12:48 min (can't choose 8 mb/s as a target bit rate)

mpeg-2, 10 mb/s, 128 kb/s 44 khz ac3, 1440x1080 16:9, 13:03 min (same as above)

both hardware decode and encode enabled

tmpg express:

mpeg-2, 10 mb/s, 128 kb/s 44 khz ac3, 1440x1080 16:9, 23:32

h264, used the main concept avc profile, 1 hour 24 min

with both software the resulting mp4 had the incorrect aspect ratio (was pillar boxed) and i couldn't find a way to correct it, the mpeg-2 produced with both pieces of software was fine.

in terms of quality, i could not see and discernible difference between any of the outputted files nor could i see any real difference from the source.