Handbrake vs new TMPGEnc?

Comments

MikeLV wrote on 4/30/2014, 6:07 PM
Well, I got the trial going on TMPGEnc Video Mastering Works 5 (VMW5) I'm not sure I like the layout of the programs, seems really confusing to me, lots of options.. Although I cannot even find that settings section for x264 as seen in the screen shot I posted. Nice to see that VMW5 makes use of CUDA though!
musicvid10 wrote on 4/30/2014, 6:09 PM
Actually, Handbrake uses Yadif, along with many other algorithms in combination. It's their own concoction. Also now has a "bob" setting for 60i->60p (similar to Yadif 2X). Only rarely does a frame get handled incorrectly. IMHO, only the soup in Nick Hope's cauldron smells sweeter. The option to show the decomb mask in Handbrake is really an education.

"

MikeLV wrote on 4/30/2014, 6:14 PM
Unless some guru writes a tutorial for VMW5, there's so many settings and tweaks in this software, I wouldn't even know where to start, or how to duplicate the settings in musicvid10's handbrake tutorial. I'm sure this software is probably worth the price, but knowing how to use it to its potential is another story, at least for me. Guess I will have a go at Nick's tutorial as at least I can follow all his instructions. With VMW5, I'm flying blind!
musicvid10 wrote on 4/30/2014, 6:19 PM
Hey Mike, take heart. The original GUI for TMPGENC (including the revered 12b Beta, which I still have somewhere) was all written in Japanese! Someone finally started porting it to English, kind of . . .
dxdy wrote on 4/30/2014, 6:24 PM
Yes, the GUI layout is very un-standard for Windows. Things like Save Project, etc., are under the Options button in the upper right corner.

H.264 is the encoder under the Output Format; MPEG-4 AVC file output; you'll see it on the second page of the format section.


flyingski wrote on 4/30/2014, 9:53 PM
"... was all written in Japanese! " And it was a freeware "beta" good for 29 days to skirt license issues with the mpeg-2 codec if memory serves. I used it to make XVCD and SVCD disks when DVD's and burners were too expensive for the common man. I was in heaven when that first TMPGEnc English version arrived. Ahh, the good old days!
musicvid10 wrote on 4/30/2014, 10:14 PM
And it got them in a lot of trouble too, iirc.
It made some really good mpeg-2, for its day.
Back then, most mpeg-2 codecs were on hardware.
We must be about the same age . . .
MikeLV wrote on 5/1/2014, 6:27 PM
Has anyone else downloaded the trial? I'm trying to figure how how to duplicate the settings musicvid came up with for the handbrake tutorial, in this trial of VMW5. What I like about VMW5 is that it supports CUDA, encoding is pretty fast, it even tells you what % is CPU and GPU. Plus the fact that it accepts frameserved file, this might be the best way to go for speed if I can get the quality and bitrate part happening.
Laurence wrote on 5/2/2014, 9:18 PM
>The one thing that I would like however is to be able to use a high quality MXF format for my intermediate without losing one of the stereo channels. Before I download and try it, can somebody tell me if TMPGEnc can use an MXF intermediate and not lose one of the stereo channels? That one thing would win me over.

I just tried it with an XDcam ,mxf format video. While mxf files aren't in the list of supported file types, if you change the options so it sees all file types and select an mxf file it will open it. It has the same problem as Handbrake though in that it will read just one of the stereo channels.

It looks like an interesting program and I do like all those output options, but it really has no advantage over Handbrake for me if all I'm doing is writing h264.
MikeLV wrote on 5/2/2014, 9:40 PM
I've hunted down someone who appears to be intimately familiar with all things x264 so I've solicited his help in setting up VMW5 to try and obtain same or similar results as handbrake or nick's method, we shall see! I may also try what you've done Laurence, with the mxf as I understand that format encodes much faster than DNxHD?
Laurence wrote on 5/2/2014, 9:51 PM
On the other hand, it does work very well with Cineform which is something I always wished Hanbrake would do.
MikeLV wrote on 5/2/2014, 9:52 PM
"with cineform" meaning it accepts Cineform as an input file?
Laurence wrote on 5/2/2014, 10:11 PM
yes.
MikeLV wrote on 5/4/2014, 5:47 PM
I looked into Video Mastering Works 5 a bit more and found there are many more settings than in the first screenshot from their site, have a look below. I also printed to PDF the help file topic from the program associated with these settings, link below the screen shots.

Video tab:


Settings button on video tab:


GOP tab:


Advanced tab (part 1):


Advanced tab (part 2):


Advanced tab (part 3):


Quantization Matrix Settings button:


Link to PDF download: http://www.datafilehost.com/d/9d41d952

Based on all these available settings for encoding an MP4, my question to the gurus: How would you set all these variables to achieve high quality results such as in the DNxHD -> Handbrake video tutorial, or Nick Hope's workflow?
NickHope wrote on 5/5/2014, 3:15 AM
I did quite a lot of work on this and my conclusion was to leave everything at x264 defaults except for using Constant Quality 18.0 for my YouTube uploads, and Automated 2pass for a video to self-host in something like JWPlayer or Flowplayer.

In your case I would try Constant Quality and do some testing at values between around 18-23 to see what file sizes and quality you're happy with.

Now then, whether TMPGEnc is just going with the x264 defaults is debatable. When I encode a video for YouTube upload in MeGUI, my "Advanced Settings" in the x264 configuration dialog are as follows (this is basically showing you what the x264 command line options would be):

program --crf 18.0 --output "output" "input" 


When I open the resultant mp4 file in MediaInfo's tree or text view, the encoding settings are:

cabac=1 / ref=3 / deblock=1:0:0 / analyse=0x3:0x113 / me=hex / subme=7 / psy=1 / psy_rd=1.00:0.00 / mixed_ref=1 / me_range=16 / chroma_me=1 / trellis=1 / 8x8dct=1 / cqm=0 / deadzone=21,11 / fast_pskip=1 / chroma_qp_offset=-2 / threads=6 / lookahead_threads=1 / sliced_threads=0 / nr=0 / decimate=1 / interlaced=0 / bluray_compat=0 / constrained_intra=0 / bframes=3 / b_pyramid=2 / b_adapt=1 / b_bias=0 / direct=1 / weightb=1 / open_gop=0 / weightp=2 / keyint=250 / keyint_min=25 / scenecut=40 / intra_refresh=0 / rc_lookahead=40 / rc=crf / mbtree=1 / crf=18.0 / qcomp=0.60 / qpmin=0 / qpmax=69 / qpstep=4 / ip_ratio=1.40 / aq=1:1.00


When I encode an "Automated 2pass" video for self-hosting, the advanced options are:

program --pass 2 --bitrate 2000 --stats ".stats" --output "output" "input" 


So if I were you I would start with the default options except for changing constant bitrate to constant quality, choose a crf between 18 and 23, render a file then look at the resultant encoding settings in MediaInfo and see how they differ from mine, if at all, and adjust as necessary.

If you're selling downloads, I reckon the standard x264 are probably fairly universally playable these days. That might not have been the case a few years ago.

You should also take a look at this old thread, starting at Arthur.s's post of 2/16/2012 1:03:45 AM so that you are aware of possible luminance shifts in TMPGEnc. Without testing, I don't know if it will occur in your AVC workflow, as opposed to MPEG-2, but scrutinise the input vs the final file on your video scopes.
musicvid10 wrote on 5/5/2014, 8:37 AM
x264 defaults with CRF are fine. You won't be able to predict file size though.
MikeLV wrote on 5/5/2014, 10:17 AM
The PDF file lists what all the defaults are, I don't know if those are TMPGEnc defaults, or defaults of the x264 encoder itself. With so many settings, it's hard to do thorough experimentation but I'm giving it a try. One thing I noticed is that the expected level change in the encoded file looks like it needs contrast, as if it did a cRGB to sRGB leveling, not the other way around, kind of weird....
MikeLV wrote on 5/11/2014, 2:33 PM
I just noticed that VMW5 changes the levels in the encoded MP4, contrast is lost for some reason. Another thing I noticed is that the encoded file does not play back properly in WMP, but does in VLC.