Video Tutorial: Video Encoding - Smallest File Size & Highest Quality w/ Free Tools

Hey long long requested is a tutorial on how I encode my videos. They always come out visually identical to the original and super small in size (small enough to email as an attachment in many cases)

It was way too complicated back in the day when I used to do everything via command line and download/use each program individually but now with great GUI front ends that use the same tools I do I was able to make a user friendly tutorial.

This information is good stuff, I am sure you can learn something and increase your quality of videos many times over with my techniques unless you are already very savvy with video editing/encoding.

This covers info on both using a video editor like Sony Vegas, and how to go directly from Fraps files to a final encoded single file ready for youtube with no editing.

Enjoy.

Download this file directly - http://rapidshare.com/files/445877124/Video_Encoding_Tutorial.rar (95mb)




<embed src="http://www.youtube.com/v/ciIHfUOn8Xw?fs=1&hl=en_US" type="application/x-shockwave-flash" allowscriptaccess="always" allowfullscreen="true" height="385" width='640'>

YouTube - Video Encoding Tutorial: Smallest File Size & Highest Quality Using All Free Tools

 

Great tutorial, Vicious. I've always wondered what steps encoders took to scrunch down their file sizes. Now I know. I do notice that you're doing this on a desktop, and it took about fifty minutes for the encode. What kind of hardware are you packing, so I can have an idea. I mean, I'm not going to be encoding anytime soon, but I'd like an idea.



Also saw you have Shakugan No Shana 2 nestled somewhere. Looking forward to Funimation's release.

 

Q6600 @ 3.6ghz

the x264 encoding will max all 4 cores @ 100% during the entire duration of the encode as it is infinite threaded and will use all available cpu resources.

The Lagarith encode uses like 60/70% of my CPU on all 4 cores.

If you do not see the CPU @ 100% during x264 encoding that means something else is holding back the encode (like HDD write speed) I use pretty high quality settings, so it would be less cpu intense to use more lax settings.

I was on the Magicka forums and some guy was asking for encoding help, using the default H264 encoder settings from Sony Vegas for a 10minute long 1080p clip he got almost a 4GB final file.

Also I found the default H264 encoding for Vegas changed the video colors some (made them darker) so I had to make this tutorial to show I can make a 14 minute video also @ 1080p well under 1GB else people would call me out as a liar.

Since this 28min video is only just over 100MB when I get home I will probably upload it to a filehosting site so it can be downloaded for direct viewing and highest quality, youtube kind of kills off the quality/clarity of the small text even in 1080p and it has a horrible time steaming 1080p for me.

One more minor recommendation from me. Dont be bold and assume a 64bit video editor is the best. If your like me and use codecs that are not "the norm" many only have x86 variants you can not load your files in the editor. I captured this tutorial with Camtasia using MSU's Lossless Codec (better than Camtaisas default, its a free download) but not realizing that it is ONLY 32bit and on my desktop I only have 64bit video editors.

So I was not able to edit this tutorial footage, I had to encode it directly into MeGUI just like I showed how to do with the FRAPS files at the end.


I have to say honestly to this day I have not seen files floating around compressed as good as mine, even "scene" videos and stuff usually are at least 2x larger than they need to be for the same quality.

I must sound pretty full of myself lol, not trying too but being honest. Guys at like Doom9 know 50x more than me, but I just never see any videos from them. Also just like any enthusiast forum they take things to a whole different level. Like a normal encode setting for them might be like 10 for quality on Placebo preset. Takes 2 full days to encode one movie.

Same way you ask for headphones here and get a recommendation for $100 and go to headfi and get one for $1000 lol. So here is my "middle man" settings that knows enough about the stuff to be talking about it and yet not obsessed with it.

The MeGUI guys are the ones that deserve the biggest praise IMO, I just was the guy that brought the information to public use and simplified it into a tutorial.

Other free encoders similar to MeGUI you can look at.

HandBrake, StaxRip, AviDemux, RipBOT

 

Great post im also encoding using MeGui in my core2duo laptop and im looking forward for the new SB laptops. In a 24 minute 720p anime at high settings for MeGui how many minutes will it take when I am in a for example Core i7 SB laptop?

 

Just depends, but it should be about 2x faster if your i7 is close to the same clock speed and nothing else is slowing you down.

If you use a lot of filters or have a slow HDD that can slow it down and keep the cpu from reaching 100% load.

Something to note there is a 64bit version of x264 and even a 64bit version of MeGUI & Avisynth (need them all) and if you use 64bit it can be like 15% faster.

I just use 32bit because its more compatible, as many Avisynth filters are only 32bit.

 

bookmarked. thanks for the info and +1

 

I prefer the term "visually transparent"

 

Thanks to Vicious, I'm encoding like nuts. It's not nearly as complicated as I thought.

 

so wait, can i get the nutshell process here? I'm not sure what is special about having a 1080p video come in under 1 GB for 14 min... what are we comparing it against?

Some things to note:

- The fraps recorder is not lossless. It is advertised as "near lossless". That is exactly the same as "not lossless" and "lossy". All encoders have to strike a balance between encoding time, image quality, and file size. Fraps finds its balance on fast encoding, large file size, and high image quality. It is absolutely lossy, though.

- as long as your final upload to youtube is under 20 GB, you should not re-encode anything if you can avoid it. You can use virtualdub (free, open source) to stitch together fraps files perfectly without reconversion.

- x264 is the best thing since sliced bread if you want to do lossy video compression. It is the answer as far as I'm concerned. It has the flexibility and control to let you decide the balance of the encoder. If you can wait a long time to encode, you can get very high quality files and very small file size. Generally, the results with x264 are significantly better than other encoders and the resultant files have extremely high compatibility. Handbrake, in particular, is a good x264 encoding tool with GUI and CLI variants.

- just as a metric, x264 encoded blu rays with very high quality and very long encode times at full resolution can come in under 4 GB depending on your audio settings. On average, maybe 8 GB. The original blu ray format video is usually about 25-30GB.

 

Thanks a lot for the video and for sharing all your tips. + rep

What would you use for ripping Blu Ray (legally owned) into smaller files with full mutli channel audio, without losing to much quality?

 

Rip it and just use pass thru for the audio and re-code the video to x264

I would probably use DVDFAB for ripping.

FRAPS does have lossless compression if you enable it, I do not like to have large files uploaded to youtube takes forever and I feel that they will not do as good of a job converting it if its a large file vs a file already in a format they endorse.

 

Since my upload speed is about 2mbps (that's, what, 250 KB/s?) I'm going to be there a long time if I were to upload a 20GB video file. Or even a 1GB file. And during that time I'll be choking my upload speeds for other things, like my servers.

 

ppro and ae come with the mainconcept codec for h.264. is x264 much better than this? im concerned more for speed rather tha quality.

 

You can still use x264 for speed just use more lax settings.

I have actually started to use slightly more lax settings just this week for youtube because youtube started to not transcode properly when I had too many reference frames enabled.

 

ok. so if i wanted to try x264 i d/l from here?

x264.nl

and get the 8 or 10 bit-depth 64-bit encoder exe?

 

x264 itself is a commandline encoder, your much better off using one of the various gui frontends for it.

> Staxrip
> Handbrake
> Megui
> Avidemux
> Others

I prefer Megui the most, its updated very regularly and will download everything for you when you install it and keep it up to date (with the exception of NeroAAC that you have to manually d/l)

 

if speed was your primary concern, what workflow would be faster in your experience? doing everything (rendring, compression, muxxing) in sony vegas pro or doing what you described in the video? what finishes projects faster?

 

Depends on if editing is needed. If no editing is needed you can do it all through megui.

Like right now I am ripping some dvd's and I just rip them and encode them and never touch vegas.

I use vegas when I need to edit. It will of course take longer to double encode it. Once with the lagorith codec and then again with x264 but the quality and compression is well worth it.

If you go directly from vegas it will be faster but I was not happy with any of the built in compression options.

 

i never do rips just edits. im not worried about the quality of adobe's h.264 since it does a fairly accurate preview of the export, where i can correct ant gamma, etc deficiencies.
im more concerned about speed. i have no problems with ppro and the built in codec, it is fast. im just owndering whether the same can be used in after effects where the inherent faster speed of the x.264 renderer may be of use- provided i get it to work. also its lack of audio is not important in ae since i usually export with no audio with ae.
im also a little apprehensive about installing megui, since ive had codec uninstallation issues in the past(ffdshow or xvid, i cant remember). have you tried uninstalling it and did it leave remnants?

 

MeGUI does not install any codecs, its not a codec pack it just is a front end for several encoders.

Example:

AviSynth - A frame server for many editors like VirtualDub
MP4Box - Best MP4 container multiplexer/demultiplexer command line only
MKVToolnix - Best MKV container multiplexer/demultiplexer
DGIndex - File indexing tool
X264 - Best H264 encoder
NeroAAC - Best AAC encoder

All of these are the best of the best imo and all of them for the most part are command line only so to use them by themselves is a lot of hassle and you gotta be super pro to know the different commands for each of them.

MeGUI simply is a GUI to let you use all these tools easily and it downloads current versions of them all that synergize together well.

I just got done with Spy Kids 3D and funny thing is it came out to exactly 700MB lol but I was not shooting for target size just quality. The quality I have is visually identical to the original, and I think about those popular pirated 700mb files on the internet that everybody thinks are great and they dont even hold a candle to my encodes.

 

Man you make it sound so good..... btw not doubting you at all. Maybe when you get some time and feel like it, do a tutorial into using those tools into ripping a BR, i know you did one on the first post and uses a lot of of this, but its more oriented into a Sony vegas/fraps and encoding a self made video, i know very little about all those stuff you posted, but really interested into trying, but those code commands might scare me away into something like DVDFAB or MakeMKV.

 

DVDFab is what I use to rip the DVD files (the free HD Decrypter) after that I just use MeGUI to index the .vob files together and encode it (with BR you index the .m2ts files)

I cant make a BR Rip tutorial unless somebody donates me a BR drive. Id be happy to go buy a BR disk and make the tutorial if that is the case.

Im sure I can find a way to make a very easy to follow and understand BR/DVD rip tutorial as you dont have to do any fancy manual editing of the .avs file. That was just my way of working around not having an editor to put fraps files together because this tutorial was supposed to be a way to use all free tools and help get people to get FRAPS encoding on youtube.

 

ok i get it now, i thought it downloaded/updated that whole list of codecs on your megui interface by default!
ive been searching a bit on megui/x.264 and ae functionality. so fare have only come up with one post saying its a bad idea. ill do a lil more searching. if you have any info on this id be happy to know.

cheers

 

Just search doom9 forums if you want info on the program(s) but I think its best just to give it a shot yourself.

Its a pretty light download.

I think handbrake wins overall in popularity, but I found MeGUI to offer me a lot more functionality. It didnt want to hold my hand through everything and let me have control over my project so I could get the results I needed and prevent certain errors that may occur in a fully automated process.

AVIdemux use to be my favorite since it had its own lightweight built in editor so you could cut/trim and easily crop, however at some point it started to not work for most of my files. MeGUI around that point added its own preview for crop/resize and that pretty much added the only feature I needed that it was missing.

 

Vicious, I only just saw this tutorial. Thanks +1. Will be trying this out!

 

If you want to put video on youtube disable b-pyramid and limit b-frames and reference frames to 3.

I had some videos not transcoding properly and found b-pyramid was the reason, and youtube only uses a max of 3 reference frames in most cases so I just do the same.

This will make the videos not compress as good, so do not do it for your own vidoes, just stuff you intend to put on youtube.


I may come up with another tutorial on this soon that shows how to extract subtitles, convert them to text. Also how to rip a DVD with more than one episode/movie on the disk and have it as its own file instead of being part of a single larger file.

I had to rip some Hellsing DVD's that had subs and 3 episodes per disk so I found that my method to do this was rather unique and more efficient than other methods I have seen.

 

I'm a complete noob at this but i just recorded some SSF4 replays at 60 fps. 60 fps is a must since the game runs at that speed. When i went to meGUI and did everything you did, the preview window makes the video run at like half that speed. Everything is slow motion. Now the encoding is complete and the video speed is faster than it was recorded. How can i fix this?

 

Based on our PM's

> Try 30FPS the game should still play well and you will cut your file size & encoding work in half.

> Do not de-interlace your video it should be progressive.

> Increase your motion estimation if you feel things are a bit jumpy

> Try to get HD resolutions if you want good quality on youtube, they do not allocate much bandwidth to non-HD videos.

> Preview window will be slow its a full nearly lossless file of huge size so it wont playback full speed not even on my desktop.

 

This is more desktop related, but I guess it can be relevant

Vicious, how well does H264 encode when comparing clocks vs threads?

Say, let's say I want to run three 64-bit encoding sessions at once (and assume I have enough RAM and HDD speed is good enough that they will not be bottlenecks).

What would run them faster, a 4-core or 6-core processor, assuming they have the same clocks and basic architecture?

 

x264 is almost perfectly linear in my testing.

So a 4ghz Core2Duo would get almost exactly the same as a 2ghz Core2Quad

2Ghz Quad is exactly half of 4Ghz Quad.

Hyperthreading is a bit of a guess I am not sure how much it helps.

But just having one instance running will use the cpu 100% so no extra speed running more than one at a time, infact it may slow stuff down due to needing more HDD resources and stuff.

Again I use really high settings that can use 100% cpu though, if you use lesser it may not use 100% due to another bottleneck.

 

Okaydokes. Looks like it's the Phenom X6 and not the X4, then.

I've got an old AM3 motherboard and case lying around so might as well not let them go to waste. Going to form an encoding box.

 

Somewhat off topic but this seems like a good place to ask.

I've got a dumb question, our digital camera records videos in .AVI format, the resolution is only 640x480 @ 30FPS and for a 3 minutes clip it takes up ~380Mb

Is there some way to re-encode this to get it small enough to send videos to family through email etc.

It would be nice to be able get the size down for storage purposes I've got ~500 videos so far that take up ~31Gb.

 

For quick and dirty I use the free Windows Live Movie Maker. You have to download it from Microsoft, but just google (or bing ) windows live movie maker. Just make sure you pay attention on install so you don't install a bunch of crap you don't want.

 

I would say just use MeGUI and encode using h264. It's quick once you get the hang of the interface, and it does a great job. It's no joke that you can get an awesomely low size and nearly the same quality.

 

Sorry to necro this thread (awesome as it is) but I've been seeing talk of 10-bit depth encoding, which gives better quality at the same bitrates, although it's a bit slower during the encoding process.

Anyone try it out? I've seen some screen comparisons, and the 10-bit encoding does control color banding much more nicely.

 

I am not sure what I am using. I have never seen any banding though in any of my videos H264 does a good job with its filters as well to prevent it.

 

Probably 8-bit unless you're specifically calling up 10-bit.

It's supposed to be superior to 8-bit, but of course encoding is slower and decoding software isn't as prevalent as 8-bit decoders.

For example, look at this comparison.

 

Remember if your source is not atleast 10bit that is pointless to "upgrade" it, also if your viewer is not using a display capable of 10bit they wont see the difference.

But 10bit contains more information so it should make the file larger not smaller.

Even that comparison shot shows the 10bit being larger, and it was shot in 8 bit and then shot in 10 bit. Not "upgraded"

If you take an 8bit image and change the color space to 10bit in photoshop it gets larger and nothing happens to the file no new data is created and no magic happens.

Even when I am working with 24bit images I always produce a final result of 8bit in photoshop for said reasons above.

 

Don't know about that. Chroma bug notwithstanding, side-by-side comparisons imply that 10bit is an improvement, even if the source isn't natively 10-bit. I assume that's because the encoder has more color depth to work with to prevent banding without the need of additional filters.

 

Thanks for the encoding tutorial.
Same here, the best CPU for encoding is the Intel chips with the SSE4 instruction set, anyone chip with those instructions will beat when it comes ot encoding.

 

I'm already watching your tutorials video.

When I created avs Script Creator and input the video. There's pop-up that asking what method I'm using.

One click, File indexer, Directshow ( I choose this one )

And following your each steps.

But when I start to encode it, my files is 700MB+ mkv but the projected size is about 400MB+. I don't finished the encode because it takes 1 hours +.

Why you can have 39GB+ and become < 1 GB ?

Whats the different ?

Thanks.

 

The material determines the compression ratio. In the video I used in the tutorial there was a lot of duplicate frames and similar colors allowing better compression.

Still no matter what the contents the compression and quality will be great, and yes it takes a long time you need a strong computer or a lot of patience to produce professional material like this.

 

Thanks mate

 

Someone stated previously in this topic that he could turn ~30GB blue-ray movies into 8 GB ones.

Can I know what encoding settings exactly he uses for the video encoding?

 

Depends on your patience and your eyes.

Personally, if I was going to encode a movie to 1080p, I would probably do Very Slow preset and a CRF of 18, plus some filters. Like I said, though, it depends on your tolerances for time and quality. I don't mind spending a day an half encoding, and I like those nice crispy details.

However, you could easily get away with Slow and a CRF of 20 for 1080p, depending on your source.

 

Need to get the GPU working with something like this...

And thanks for the awesome tutorial. I've always tried to achieve this for my videos but AVIdemux fails on a lot of the modern DSLR codecs and I never really found other options that worked as well.