Editing without Recoding

[criggs]

For your FLV that "failed" in TMPGEnc Smart Renderer, can you post the mediainfo report (view=>text)
https://www.videohelp.com/software/MediaInfo

Sure. Here it is:

General
Complete name : C:\Users\chrig_000\Videos\ProblemFile.flv
Format : Flash Video
File size : 86.5 MiB
Duration : 5 min 49 s
Overall bit rate mode : Variable
Overall bit rate : 2 074 kb/s
Encoded date : UTC 2016-12-02 21:32:26
Writing application : Lavf56.30.100

Video
Format : AVC
Format/Info : Advanced Video Codec
Format profile : Main@L3.1
Format settings, CABAC : Yes
Format settings, ReFrames : 4 frames
Format settings, GOP : M=1, N=60
Codec ID : 7
Duration : 5 min 49 s
Bit rate mode : Variable
Bit rate : 1 891 kb/s
Maximum bit rate : 10 000 kb/s
Width : 1 280 pixels
Height : 720 pixels
Display aspect ratio : 16:9
Frame rate mode : Constant
Frame rate : 29.970 FPS
Color space : YUV
Chroma subsampling : 4:2:0
Bit depth : 8 bits
Scan type : Progressive
Bits/(Pixel*Frame) : 0.068
Stream size : 78.9 MiB (91%)

Audio
Format : AAC
Format/Info : Advanced Audio Codec
Format profile : LC
Codec ID : 10
Duration : 5 min 49 s
Bit rate : 106 kb/s
Channel(s) : 2 channels
Channel positions : Front: L R
Sampling rate : 48.0 kHz
Frame rate : 46.875 FPS (1024 spf)
Compression mode : Lossy
Stream size : 4.42 MiB (5%)

You said videoredo handled it, but what operations did you do in videoredo ?

Did some editing/deletions/changes to it, and then saved the result. It came out fine, same parameters as the original, and the size was slightly smaller than the original, which made perfect sense considering that my deletions were minor.

[poisondeathray]

Well, I do have Adobe Premiere Elements, and I'm very happy with it for what it does (no smart rendering as far as I can tell, plus a full complement of transitions (dissolves, wipes, text, etc. etc.)). So perhaps my mixing and matching should dump the concept of smart rendering completely.

If tmpgenc works for you , by all means use it

You have to weight the pros/cons of the operations . I'm suggesting that automatic "conversion" details are usually less than ideal, even with NLE's. If you had to match frame rates for example, it might blend frames by default. Sometimes that might be appropriate but other times it's plain horrible. Sometimes you don't care and go with what's easier... up to you

videoredo will work with 99.9% of AVC videos. But there is that 0.1% where it will either outright fail, or have problems like glitches at the seams . It's very reliable for MPEG2 sources, but not perfect for AVC . I just wanted you to be aware just because tmpgenc "failed" on a certain source, be prepared that videoredo can and willl as well.

So I guess the question is whether it's just a random accident that I bumped into this file which Smart Renderer 5 can't handle but have not yet bumped into a similar problem with VRD. Am I A) likely to bump into just as many rogue files, over the long-term, with VRD as with SR5? Or will B) VRD always have a slight edge over SR5 in that regard?

If the former, A), then that is a strong argument for SR5, inasmuch as it is both cheaper than VRD and has this mix-and-match capability which VRD apparently lacks.

If the latter, B), then I might be better off dumping the mix-and-match capability of SR5, going with VRD, and simply accepting the fact that going forward my mixing and matching will be done Premiere Elements-style without smart-rendering.

I can't answer head to head (I only have videoredo and solveigmm), but an earlier poster had all 3 software

It depends on what types of sources you are using. I'm 100% certain there will be instances where one or all will fail. If all you use are self CFR captures and AVC or MPEG2 blu-ray (there are VC-1 BD's too which aren't support by vrd or tmpgenc ) , then you will likely have 100% success. It really depends on the source - The more "crap" or more variety of sources, the higher the chance of failure. People use all kinds of weird and wonderful settings, so if you're not using actual acquisition footage, or a standard like BD - the chances are higher of failure

Another fairly common scenario where all of them will fail is VFR (or at least VFR that has signficant changes) . Many streaming sites use VFR to reduce bandwidth, and MP4 and FLV are common VFR containers. Cellphone/portable device videos are mostly VFR. It's nearly impossible to edit VFR as VFR and keep in sync. Almost always you have to convert to CFR by inserting duplicate frames first (requires re-encoding).

Your specific FLV wasn't VFR according to mediainfo, but earlier your VLC post showed a display dimension difference

Resolution: 1280x738
Display Resolution: 1280x720

That strikes me as very odd . It might have some bizarre AR signalling information which is causing SR5 grief. You can find more details using a bitstream parser, or ffprobe, or h264_parse, but if it's in the video bitstream, you might be able to reset it strip it and it might play nice with SR5 . If you can streamcopy a piece out and upload it I can have a look and maybe provide some suggestions . ffmpeg would be the best option to cut a flv , but it's commandline. VRD will change container and possibly some other stuff such as metadata or container flags

e.g cut a 10 sec segment

ffmpeg -i input.flv -c:v copy -c:a copy -t 00:00:10 cut.flv

[criggs]

If you can streamcopy a piece out and upload it I can have a look and maybe provide some suggestions . ffmpeg would be the best option to cut a flv , but it's commandline. VRD will change container and possibly some other stuff such as metadata or container flags

e.g cut a 10 sec segment

Sure, no problem. Here you go, and thanks! If you solve this, there'll be something nice under your Christmas tree (theoretically speaking, of course).

[poisondeathray]

I think that's actually a VLC bug. Checking some random 1280x720 videos, they all say that in VLC.

And bitstream checks out fine too, I don't see any abnormalites with the actual video stream

But I think there SR5 might be having some issue with the timestamps and time scale. Also related - it's supposed to be CFR, but some programs see the framerate "off" as 30.3 . The timecodes are slightly different depending on what program is reading them. So when you just remux it with xmediarecode (which is ffmpeg based) , I' m guessing the output was VFR ? (you can check with mediainfo) .

I would try mp4box to force the framerate and will also fix the container timebase. Test this when you get a chance:

[criggs]

it's supposed to be CFR, but some programs see the framerate "off" as 30.3 . The timecodes are slightly different depending on what program is reading them. So when you just remux it with xmediarecode (which is ffmpeg based) , I' m guessing the output was VFR ? (you can check with mediainfo) .

I'm afraid I was not able to get you an answer, due to my ignorance. When I remux in XMedia Recode I don't see any listing for anything like VFR in the mediainfo for the new file copy created by XMedia Recode. I'm guessing that by VFR you mean variable frame rate. If I'm right, then the answer is no; the frame rate is listed as being constant. I am seeing the same 30.3 fps that you're seeing in the mediainfo, actually 30.303 to be exact.

I would try mp4box to force the framerate and will also fix the container timebase. Test this when you get a chance:

No dice, I'm afraid. Your new clip, mp4box.mp4, is still triggering the No Smart Render-able clip message in SR5.

[TimA-C]

Back to VRD and encoding multiple files with one set of settings. If you use the batch manager then you can add a number of files for processing all with the one profile. You'd still have to find the stats of the "first" file you want to join the others to, and you'd still have to enter those stats into a profile at the point of adding them (or predefine a profile beforehand) but at least you'd only have to enter them once per batch.

As for relying on VRD's help . . . it's usually not even close to being up to date. Much better to try for oneself and then ask in the forum.

[criggs]

Back to VRD and encoding multiple files with one set of settings. If you use the batch manager then you can add a number of files for processing all with the one profile. You'd still have to find the stats of the "first" file you want to join the others to, and you'd still have to enter those stats into a profile at the point of adding them (or predefine a profile beforehand) but at least you'd only have to enter them once per batch.

Out of the question, I'm afraid. Here are the parameters for my master clip, according to VRD:

File: Name : C:\Users\chrig_000\Videos\Watch the Live Stream - 3.mp4
Size : 0.090 GB
Duration : 00:02:49.28
Mux type : MP4
Video: Encoding : H.264
VideoStreamID : x201
Frame rate : 29.97 fps
Encoding size : 1920 x 1080
Aspect ratio : 16:9
Header bit rate : 20.000 Mbps
VBV buffer : 380 KBytes
Profile : High/4.0
Progressive : Progressive
Chroma : 4:2:0
Entropy mode : CABAC
Bit rate : 4.102 Mbps
Captioning : EIA 608
Audio Stream: 1 (Primary) Codec : AAC
Format : RAW
Channels : 2.0
Language : und
PID : x202
PES Stream Id : xC0
Sampling rate : 44100

But when I opened up the Output Profile menu in VRD, at least half of the above stats were not even listed, so it was impossible to fill them in correctly. Not to mention the fact that getting through just a third of the above stats to enter them in VRD had taken ten minutes at least. And I'm supposed to do that every time I have a master clip which i want to use as reference? Fuggedaboutit.

As for relying on VRD's help . . . it's usually not even close to being up to date. Much better to try for oneself and then ask in the forum.

As previously mentioned, I've posted a query in the VRD forum at http://www.videoredo.net/msgBoard/showthread.php?36042-Re-Encoding-Files-to-Match-Prim...-File&p=124342 , as well as an appropriate query regarding SR5 at http://bbs.pegasys-inc.com/bbs/list/lang/en/board/TMSR5/messid/70555 .

[TimA-C]

OK. Hopefully you'll get more helpful suggestions at one or both of the other forums.

Good Luck.

[usually_quiet]

@criggs You have forgotten about audio. The audio must match too. VideoReDo provides limited audio encoding options when you are setting up a profile. I have only seen Automatic (no re-encoding will be done), LPCM, MPEG-1 Layer 2, and AAC as available choices (and LPCM isn't always available). This means it is likely that you will need to encode audio outside VRD at times.

Here is an example: OTA TV (and cableTV) use AC3 audio. If I wanted to use one of my HTPC's recordings (either AVC video and AC3 audio or MPEG-2 video and AC3 audio) as my "master" in VideoReDo, I would be unable to re-encode non-conforming audio to AC3 within VideoReDo.

Given the limited encoding options for both audio and video, I can't in good conscience recommend VideoReDo for combining clips from random sources.

[TimA-C]

Good point.

[johnmeyer]

@criggs You have forgotten about audio. The audio must match too. VideoReDo provides limited audio encoding options when you are setting up a profile.

Really good point. I had completely forgotten about the audio. When I've done things like the OP, I often have to do the audio through a completely separate chain and then remux as the last step in the process.

[TimA-C]

Ditto. I tend to use either AviDemux or MeGui these days and then remux with either VRD or MKVToolnix depending upon whether it's H264 or Mpeg2 and what I'm planning on doing with it. Not that this helps Criggs much.

[criggs]

@criggs You have forgotten about audio. The audio must match too. VideoReDo provides limited audio encoding options when you are setting up a profile. I have only seen Automatic (no re-encoding will be done), LPCM, MPEG-1 Layer 2, and AAC as available choices...Here is an example: OTA TV (and cableTV) use AC3 audio. If I wanted to use one of my HTPC's recordings (either AVC video and AC3 audio or MPEG-2 video and AC3 audio) as my "master" in VideoReDo, I would be unable to re-encode non-conforming audio to AC3 within VideoReDo. Given the limited encoding options for both audio and video, I can't in good conscience recommend VideoReDo for combining clips from random sources.

Totally critical point, thank you; I was completely unaware of that.

Just for the halibut, I grabbed a recording I had made on my Channel Master DVR+ over the air, which was, as you said, mpeg-2 video/ac3 5.1 audio in a ts file, and pulled that into Smart Renderer 5. Then I pulled in a standard Youtube HD 1080 Mp4 with AAC stereo audio into Smart Renderer 5. Then I told SR5 to combine the two files into one, using the OTA recording as the Master.

It did it without turning a hair, cleanly and perfectly converting the Youtube MP4 into mpeg 2 video/5.1 AC3 audio while smart-rendering the first ota mpeg-2/ac3 clip. Really impressive! Now if only I could get SR5 to handle all the files that VRD can handle. It's a conundrum.

[criggs]

I think there SR5 might be having some issue with the timestamps and time scale. Also related - it's supposed to be CFR, but some programs see the framerate "off" as 30.3 . The timecodes are slightly different depending on what program is reading them...Test this when you get a chance:

The file you gave me has yielded what may be interesting information in Smart Renderer 5. While I am still getting the message "No Smart Render-able clip," I believe I've found out why, though I don't know what to make of it. You may know what to make of it, or someone else here may spot this and know what to make of it.

Here goes.

When I pull a clip into Smart Renderer 5, there's a tiny link in the upper right-hand corner labeled Clip Properties which I never noticed before (for some reason the openable menus in the user interface are not clustered in the upper left, as is normal with most Windows aps; instead, they're scattered all over the window, in the upper left, the upper right, etc. etc.). I clicked those Clip Properties and got a ton of information.

I then scrolled to the bottom of this ton of information.

And there, in tiny fine print, was the following: "This file cannot be smart rendered due to the following reason."

Not very informative, right? Why wasn't the sentence followed by the reason? Why did the sentence end in a period rather than a colon?

Regardless, below that sentence, in a little box, FINALLY, the reason was listed!

"low_delay_hrd_flag = 1 (Not Supported)"

What on Earth does that mean? And is there any way I can fix it without re-encoding the whole clip??????? If I can fix it, then I'd be a very happy camper!

[poisondeathray]

You can patch certain VUI parameter flags, but not that one. In theory it should be possible, but you need someone to write a program. It would be easier for tmpgenc to fix it on their end

[criggs]

You can patch certain VUI parameter flags, but not that one. In theory it should be possible, but you need someone to write a program.

Thank you for getting back to me, and for answering that question. So I guess I can forget about correcting the problem myself.

It would be easier for tmpgenc to fix it on their end

Unfortunately tmpgenc's policy is not to respond to tech support requests, or to analyze them, unless the customer has bought the product. They say repeatedly all over their web site that there will be no support for free trial periods.

On the plus side, I may be close to solving the problem with videoredo. Dan has told me he will look into a system of writing profiles that can be based on specific clips, and will get back to me shortly. If I can teach myself how to write profiles for non-matching clips in a Joiner list, that will solve my problem and I won't need Smart Renderer 5.

Of course, that still leaves me needing to find an ap which can do the same thing, audio-wise, with clips that have AC3, which apparently VRD cannot yet handle as an encode function (though I gather it's fully capable of smart-rendering AC3 audio, since I've done that myself a few times with VRD with no problems).

[usually_quiet]

Of course, that still leaves me needing to find an ap which can do the same thing, audio-wise, with clips that have AC3, which apparently VRD cannot yet handle as an encode function (though I gather it's fully capable of smart-rendering AC3 audio, since I've done that myself a few times with VRD with no problems).

All VideoReDo can do with AC3 is cut and copy. VideoReDo has no ability to encode AC3/Dolby Digital audio for smart rendering either.

[criggs]

All VideoReDo can do with AC3 is cut and copy. VideoReDo has no ability to encode AC3/Dolby Digital audio for smart rendering either.

Understood, thanks. In most cases, I won't be needing to do dissolves, so that should work O.K. for me most of the time.

Speaking of which, I was just playing with the transition options in Smart Renderer 5; it really is powerfully impressive. I just threw in an mpeg-2/AC3 file along with an H.264/AAC file and told it to do some transitions between the two clips. It did a perfect slide, perfect dissolve, and, in every case, it did not only a clean video transition but also a clean audio transition. It may be not able to handle every file I throw at it but, when it likes the files, it sure is powerful, almost like a full-fledged editor with the Smart Render thrown in as a bonus. I've been using Premiere Elements for a few years now but boy, I could get used to this. This baby even does titles.

Of course, audio would be a bit of a challenge. I suppose what one could do is edit your video with reference audio, render that, then do a completely different project for the purpose of editing your audio which would include reference video (Smart Renderer 5 won't take audio-only files; I tried), and then pull both files into tsMuxer and tell tsMuxer to take the video from one clip and the audio from another. It would be a bit of a brain tease to keep everything in sync, but an experienced editor like me MIGHT be able to handle it, -- maybe. Hmmm.

And all this just to save a generation of quality and take advantage of smart-render. Like I said, hmmm.

[criggs]

I just wanted you to be aware just because tmpgenc "failed" on a certain source, be prepared that videoredo can and willl as well.

Well, I finally decided that I would attempt a comprehensive series of tests with a variety of files to see whether Smart Renderer 5 continued to choke on more files than VideoReDo.

As I believe I have mentioned previously, mpeg-2 files with AC3 audio appear to need to be re-contained through XMedia Recode as mp4 copies in order for SR5 to handle them properly. In the case of the mpeg-2/AC3 files I've pulled into SR 5, SR5 misreads the frame rate as 59.94 rather than 29.97.

In addition, there was this 29.97fps h264/AAC 1280 x 720 file which SR5 could not handle at all because of the really deep in-the-weeds fact that the low delay HRD flag was set to 1, whatever that means (and this was the glitch that triggered my latest round of inquiries and questions, of course).

There are a round ten sources of video clips which I work with regularly. So I sliced off small sections of sample clips from all these sources and pulled them into both VRD and SR5 to see what would happen.

There were two more problems.

An mpeg file with H.264 video, 1280x720 resolution, and AAC stereo audio with 48k sampling, would not render properly in VRD; the length was wrong; the render cut off with only about four-fifths of the file rendered (about four minutes worth of the five-minute file). After re-containing as an mp4 copy through XMedia Recode, the problem disappeared. SR5 had no problem with the original file.

And SR5 was able to pull in another file for which it read the wrong frame rate. It was an mpeg file with H.264 video, 1920 x 1080 resolution and AAC stereo audio with 48k sampling. I re-contained it in an mp4 copy with XMedia Recode and the problem disappeared. VRD had no problem with the file.

So the bottom line is that SR5 had problems with three files, two of which were fixable, one of which was not, where VRD had no problem. VRD, on the other hand, had a problem with one file which was fixable, a file with which SR5 had no problem.

Again, not, perhaps, a comprehensive sample. But my results certainly tend to continue to suggest that VRD is capable of handling more flavors of files than is SR5.

On the other hand, the fact is that my search for a Smart Rendering editor began because I suddenly found myself in a situation where I would need to edit mpeg-2 files on a regular basis without going down a generation. Previously, my only need was a GOP h.264-capable editor, so Total Recorder fit the bill fine; frame-accurate edits were not a critical consideration for me, so I was O.K. with the limitation of only editing on i-frames.

With this new need, I was in the market for a smart rendering editor of more than one codec. Clearly both VRD and SR5 fit that bill. So I need to weigh the fact that VRD seems to be less finicky about files against SR5's rather awesome editing capabilities for a smart render editor.

The fact is that all the files from the source which SR5 cannot handle at all -- let's call that source LS -- are h264 material which, heretofore, I've been content to edit in my GOP editor. If I give up on frame-accurate edits from that one source, then there's no question SR5 is a better deal for me. And, of course, if worse comes to worse, I can always take LS files down a generation by editing them in Premiere Elements if I have to. In addition, if I do any mixing and matching with AC3 sources, where the AC3 clips are my smart rendering standard, then SR5 has to be the choice since VRD cannot encode AC3. So those are the main arguments in favor of SR5.

On the other hand, now that I've decided I want to be able to do frame-accurate smart render editing, I know that getting SR5 means that I am permanently prevented from doing frame accurate edits with files from LS where the files retain first-generation. Moreover the number of times that I will need to do transition editing (dissolves, wipes), titles, and so on, as well as the number of times that I will need to do mix and match to marry mismatched files, is probably less than the number of times I might need to do frame-accurate editing with clips provided by LS. As for situations where I may want to daisy-chain a group of files, with those having AC3 being the smart render standard, I may just have to accept that I will need to do my audio editing in Premiere Elements, which can create 5.1 clips, and my video editing in VRD if I am to retain the video first-generation. If I accept all of those facts, then VRD is the way to go.

One irrelevant but compelling fact: Despite this very annoying, possibly deal-killer problem handling clips where the low delay HRD flag is set to 1, the fact remains that, in my estimation, SR5 is a far more impressive application than VRD. It is very well-designed, works very well, was very ambitious, and has a very professional and powerful feel. VRD, on the other hand, is no more than it pretends to be, a reliable utility that can do Job A, but not really spectacular in any way. Smart Renderer 5, on the other hand, is a unique and very ambitious concept; the developer or developers thought big, and they produced big. As someone who hasn't got much experience with this particular technology but who edited professionally for thirty years, I took to this tool like a duck to water; it has an impeccably elite feel, and can not only do Job A but also Job B, C, and D. It's fun to operate and fun to create products with it. If I had my druthers, I'd get this program.

But the bottom line is that the strengths of SR5 don't really dovetail with my needs the way VRD does. So, if I follow my head, VRD's the way to go; if I follow my heart, SR5 is the unquestionable choice.

As I said before, hmmm.

You know, earlier in this thread someone suggested I try both aps and do an in-depth analysis and comparison. At the time, I pooh-poohed the suggestion as involving the kind of time I didn't want to expend. I wanted to pick a product and get to work ASAP.

Well, something appears to have happened on the way to the office. I've wound up spending a lot of time, and doing a lot of analysis, of these two aps after all. While I continue to be skeptical that this report, based as it is on less than two weeks of testing, is particularly authoritative when compared to the accumulated experience of all the participants in this discussion thread and/or in this forum, nevertheless it has helped me crystallize the pros and cons of the two aps, so I'd say it has proven somewhat helpful after all.

Of course, I still can't decide what to do.....

[TimA-C]

As I said earlier, you have to find what works for YOU with YOUR files. You still have to make a decision but at least it will be a heck of a lot more informed than it would have been a couple of weeks ago. One question, the files that didn't work in VRD, you did try running them through the QuickStream Fix feature in VRD first, didn't you? (If you add the files via the Batch Manager then you can get VRD to automatically match the profile to the clip so you can add multiple files of different formats in one go.)

TBH I'm quite glad you started this thread as I suspect that the team at VRD may well (at some point) give us the option of creating a profile from a clip which, although I probably wouldn't use it often, would still be a useful option to have. I guess TMPGenc may come up with a solution but based on my past dealings with them, I won't be holding my breath.

[criggs]

One question, the files that didn't work in VRD, you did try running them through the QuickStream Fix feature in VRD first, didn't you?

Nope. I didn't even know there was such an animal. This is why experimenting and testing on my own, while useful, needs to be supplemented by feedback and advice from folks like you who've been working with this specific technology for a while.

In any event, I followed up on your suggestion and VRD crashed. I got the message "Mpeg stream error: H264, error in replenish buffer, details in the log file."

So I looked at the end of the log file, and here's what it said:

* 25 PTS: 0.00 (00:00:00.00), [ 101981.00 (00:01:41.24 ) 0], , pict_num: 56 ( 13), P Ref, Diff:101981.00 P 0 54777 62445513 QP: 0, PPS:0, slices:0
** End of Video buffer, 26 frames, video bytes: 0.434 MB, average bit rate: 3.099 Mbps, buffer allocation: 4.850 MB

2016-12-10 13:55:38 >> Audio Buffer: H.264 ReplenishBuffer: MaxGOPsInBuffer: 0, VideoBufferMax: 1638, count:3831, StreamNumber: 0, Codec: AAC, PID: x101, next audioPTS:181879.67
- 0, A-PTS: 100151.67 (00:01:40.03), [ 20132.67 (00:00:20.03)], PTS Diff: -80019.00 366 0
- 1, A-PTS: 100173.00 (00:01:40.04), [ 20154.00 (00:00:20.03)], PTS Diff: -80019.00 334 1920
- 2, A-PTS: 100194.33 (00:01:40.04), [ 20175.33 (00:00:20.04)], PTS Diff: -80019.00 357 1920
- 3, A-PTS: 100215.67 (00:01:40.05), [ 20196.67 (00:00:20.04)], PTS Diff: -80019.00 314 1920
- 4, A-PTS: 100237.00 (00:01:40.05), [ 20218.00 (00:00:20.05)], PTS Diff: -80019.00 331 1920
.... 3822 similar frames not shown ...
-3827, A-PTS: 181794.33 (00:03:01.19), [ 101775.33 (00:01:41.19)], PTS Diff: -80019.00 340 1920
-3828, A-PTS: 181815.67 (00:03:01.20), [ 101796.67 (00:01:41.19)], PTS Diff: -80019.00 332 1920
-3829, A-PTS: 181837.00 (00:03:01.20), [ 101818.00 (00:01:41.20)], PTS Diff: -80019.00 355 1920
-3830, A-PTS: 181858.33 (00:03:01.21), [ 101839.33 (00:01:41.20)], PTS Diff: -80019.00 347 1920
**End of Audio Buffer(s)

2016-12-10 13:55:38 Graph, monitoring thread received terminate signal.
2016-12-10 13:55:38 Graph monitoring thread finished.
2016-12-10 13:57:47 Image support lib: 8.0.1 (r40318), ippIP AVX (g9)
2016-12-10 13:57:48 Decoder support lib: 8.0.1 (r40318), ippIP AVX (g9)
2016-12-10 13:57:48 VideoReDo TVSuite started. Licensed to: TRLKEYA0-H2NWB2-, Version: 5.1.3.741 - Mar 14 2016, AVP On (0051003f-5} Windows 8
2016-12-10 13:57:48 Checking reg: 101

TBH I'm quite glad you started this thread as I suspect that the team at VRD may well (at some point) give us the option of creating a profile from a clip which, although I probably wouldn't use it often, would still be a useful option to have.

Interesting you should say that. Dan got back to me early this morning with a custom-built profile he made based on the Video Program Info I copied and pasted from a file I wanted to use as the template for a Joiner List render. And it worked beautifully finally. I threw in the file in question that I wanted Smart Rendered, plus another file that didn't match it at all and would need to be complete re-encoded. And it worked beautifully, just like I've always wanted. It zoomed through the first file, obviously smart-rendering only, and then took its time with the second file to match the first one perfectly. When I opened up the new Joined file, it was exactly what I wanted, with the first half smart-rendered exactly as the original and the second one meticulously re-encoded to match it perfectly.

So I sent a message to Dan asking how he managed to create a profile that matched my clip so perfectly that the Joiner knew it could get away with smart-rendering the first file. I await his reply. If he explains the formula he used, and the guidelines he followed, to create this perfectly matched profile to my reference master clip, and if I can then create a brand new profile of my own based on an entirely different clip that can also function as an encode profile for mismatching clips, then I'm on my way and I can go with VRD. On the other hand, if he tells me that it was a laborious trial-and-error process to create the profile, and that creating such a profile will always be that sort of laborious trial-and-error process, then I think I may go with SR5.

[criggs]

I sent a message to Dan asking how he managed to create a profile that matched my clip so perfectly that the Joiner knew it could get away with smart-rendering the first file. I await his reply. If he explains the formula he used, and the guidelines he followed, to create this perfectly matched profile to my reference master clip, and if I can then create a brand new profile of my own based on an entirely different clip that can also function as an encode profile for mismatching clips, then I'm on my way and I can go with VRD. On the other hand, if he tells me that it was a laborious trial-and-error process to create the profile, and that creating such a profile will always be that sort of laborious trial-and-error process, then I think I may go with SR5.

Someone named Jim got back to me, who apparently worked with Dan on that profile they made for me to match this particular clip. Jim explained to me what was important, and how he arrived at his profile parameters.

So I grabbed two entirely different clips, and attempted to do the same thing that Jim had succeeded in doing.

SUCCESS!!!!!!!!!!!!!!!!!!!!

No, it is not a laborious trial-and-error hit-or-miss process. Yes, he explained the formula he used, and the guidelines he followed, to create this perfectly matched profile to my reference master clip. Yes, I was able to create a brand new profile of my own based on an entirely different clip that could also function as an encode profile for mismatching clips.

This time, I chose as my master clip an mpeg-2 1080 29.97/AC3 5.1 clip and I chose as my mismatched clip an h264 720 x 418 25/AAC stereo clip.

The bottom line was that I even managed with VRD to preserve the AC3 5.1 audio kinda sorta by converting it to the lossless wav format (obviously ballooning the file size somewhat but nothing's perfect).

First, here were Dan's initial instructions for the first pair of clips: "Ok so what you want to do is go to Tools->Edit profiles list. Select the H.264 MP4 profile and then click Copy & Edit. In the profile dialog set the profile mode to "force recode". Set the resolution to 1920x1080. Set the average bitrate to 6,000 and the max bitrate to 8,000. Set the deinterlace mode to "smart". And set the frame rate to 29.97. Also set the aspect ratio to 16:9 and the aspect correction mode to "letter box/pillarbox". In the audio section set the audio codec to AAC, set the bitrate to 192, set the sample rate to 44.1KHz, set the channel mode to "stereo", and uncheck the option to allow multiple audio streams. That should recode any video fed into it to match the one you posted here."

I then asked him to elucidate on what basis he came up with the parameters outlined above and whether, following those same concepts, I would then be able to make such a profile for any clip under the sun. Here's the gist of his reply: "If you are going mp4 to mp4 then some settings must be set to 'force recode" or they will not change...The bitrate is your personal choice (Quality vs size)...I hate interlacing and always deinterlace files...192 is a common dvd stereo setting."

And now, in case anyone's curious, here were all the procedures I did to create a successful join today with these two new files. This is the report I sent back to Jim announcing success. It looks like I'll be buying VRD! My report to Jim and Dan "translated" what Dan and Jim did with my first clip to doing the same thing with the pair of clips I selected today. Hopefully anyone else out there looking to use VRD to combine mismatched clips with a minimum of generational loss and a maximum of smart render will find the following of long-term use.

Here goes.

"So, armed with your operating principles as outlined above, plus the parameters that were determined to be important in your previous message, I took two completely different clips, an mpeg-2/ac3 5.1 1920 x 1080 29.97 fps clip and a 720 x 418 h264/AAC 25 fps clip. I created a profile that matched the first clip. An audio caveat there: As I'm sure you know better than I do, VRD does not encode to AC3 (although it smart renders it fine, of course). So I tried making the profile both ways, first with Automatic to see what it would do and then with Mpeg layer 2. I then re-encoded the second clip twice using those two versions of the profile.

"I then put the first clip into the Joiner, followed by the second clip as audio Automatic, and told VRD to create the joined clip. VRD automatically picked the mpeg-2 transport stream profile and rendered the joined clip.

"It was very good. The video rendered super-fast, indicating that I had successfully made the two clips close enough that the smart render was able to kick in. The audio was re-coded however, and it was changed from 48k to 44k; VRD chose to go with the sampling rate of the second clip rather than the sampling rate of the first clip. I don't know why VRD did that.

"I then replaced the second clip in the Joiner list with the version that was rendered with the audio set to mpeg layer 2, the bit rate set to match the first clip and the sampling rate set to match the first clip at 48k. I then rendered that joined clip.

"Again it was super-fast and the result indicated that the video had again been smart-rendered. However there was an audio recode needed since the second clip had stereo mpeg layer 2 while the first clip had AC3 5.1. This time the clip ended up again with mpeg layer 2 stereo but this time with 48k sampling.

"This is all very good news, for the most part. It means that you have now taught me a system I can use which will quickly and easily match mis-matched clips in a Joiner list, and which will trigger Smart Render for those clips which conform closely to the utilized profile.

"So then it occurred to me that if I want to preserve the first clip's five audio channels, I could re-code both clips with an audio codec with 5.1 which VRD can handle, like AAC; of course, that would mean taking that first clip out of the mpeg 2 world, so the video would have to be re-coded. Then create a joined clip of those encodes. Then pull both the joined clips described above into tsMuxer and grab the video from the mpeg-2 join and the audio from the AAC 5.1 join.

"Which is exactly what I did, and it worked perfectly! Yes, it was a bit disappointing that I can't preserve the AC3 5.1 first generation from the first clip, but I was 95% of the way toward my goal now, and VRD had now proved, by and large, that it could do what I wanted it to do, so I was already a happy camper!

"And then it hit me: Just go with a lossless format, like wav, and you don't have to lose any quality in the 5.1 AC3 audio at all. So I looked around and noted that VRD permits wave codec with four mpeg 2 formats, Program mpg, elementary m2v, Tivo tivo and Matroska mkv.

"So I did a smart render of the first clip in mpeg 2 matroska, changing the audio to 5.1 wav. Then I created a copy of that profile, as you taught me, for the second 720 x 418 clip, and modified it to match the first clip, as you taught me, and then re-rendered the second 720 x 418 file with this new profile.

"Then I added both clips to the Joiner List and pulled the trigger.

"PERFECTION! The video for the first clip was still first-generation smart-rendered, the audio, while it had ballooned admittedly in file size, had been preserved losslessly in 5.1 wav, and the second file was converted beautifully to match with a smooth join.

"So I've now successfully joined two different combinations in VRD, yesterday joining H264 1080/AAC as master with mpeg-2 720/AC3 stereo and today joining mpeg-2 1080/AC3 5.1 as master with h264 720 x 418/AAC stereo. As far as I'm concerned, I'm satisfied that I can now make VRD do what I need it to do, and that I've now got this puppy licked.

"Hurray!!

"Thanks muchly for the guidance. I think I'm good to go!"

[johnmeyer]

Glad to hear it is working out. If you go back to my second post, I recommended VideoRedo and did so not only because of its feature set, but because of the wonderful support they provide. As you now understand, this makes a huge difference in anyone's ability to get the most out of a piece of software.

[criggs]

I recommended VideoRedo...because of the wonderful support they provide.

And they certainly came through here.

As you now understand, this makes a huge difference

Actually I've always understood that product support makes a difference. I remember, back when I was two years old, when my parents trained me on, er, doing my business, they said the following unforgettable words to me: "Son, whatever you do, always post the 800 number for your toilet paper in the bathroom; and always make sure it's staffed 24/7/365; you never know."

[criggs]

As indicated above, I now have a full procedure for matching unmatched clips for the purpose of joining a collection of such into one clean file using VideoReDo.

It occurred to me that I'm probably not the only one out there who's had to figure this out. While some very experienced hands probably already know this stuff, newbies like me do not.

So, in hopes of saving others, who may be in my shoes, a lot of time spent reading, researching and searching, I pulled together a video tutorial of my own on the VideoReDo procedure that accomplishes this task.

I have posted it at https://www.youtube.com/watch?v=hVV7ZG40Q_c . The first seven minutes are merely a spoken overview of the premise of the tutorial, and have no meaningful video instruction, so you may skip that portion if you wish. This video is not particularly polished or fancy. My goal was to get the message out, rather than to wow the viewer with a whiz-bang presentation that would have taken more time and effort than, frankly, I thought the subject warranted. However I hope that anyone who needs to know how to do this, as I did a month ago, will come away from the tutorial with a rather clear idea of how to go about accomplishing this task.

As far as I know, this video is the first such attempt to provide instructions on this procedure online. But it need not be the last. Perhaps others with more experience and knowledge than me, after viewing this video, may be moved to do a better job. I would welcome that. As far as I'm concerned, the more widely these procedures are disseminated, taught and learned, the better.

So feel free to watch the tutorial, and to post here any questions, comments or suggestions you may have. Don't hold back! My skin is reasonably thick, I assure you.

[zamar27]

I wanted to add some observations to this useful thread. Many folks now use smartphones to record MP4 video clips, and usually they are auto recorded with variable framerate. VideoRedo can't join such VFR clips together without re-encoding to the Master clip framerate. Also, it can't import in right orientation smartphone taken clips recorded in vertical phone position, so they need to be rotated first as part of re-encoding process before been joined. In contrast, TMPGEnc Smart Renderer 5 imports such clips in correct vertical orientation, and can auto identify what segments of these clips should be smart rendered, and what segments re-encoded to the Master clip parameters, when joining clips together after cutting unneeded pieces from them. SR5 also auto identifies the biggest clip as Master to save more time at joining by smart rendering.

However, SR5 seems to offer only one pass MP4 re-encoding, while VRD optionally offers two-pass re-encoding, thought I didn't notice any serious video degradation at editing smartphone taken and then cut and joined in SR5 video & audio clips into one video file in one pass.

When it comes to smartphone taken clips, showing date & time stamp on the video is often important, but the stamp is not burned into the video when using Android default Camera app, despite preserved in each frame's metadata. To burn date & time stamp into the video while recording it, I found Timestamp Camera being the best app. If a video was taken by default Camera app, its original frame date & time stamps can be extracted from the MP4 metadata and saved as Subtitles, imported back and burned into the clips in correct orientation and position in my tests best in Sony Vegas Pro using Vegasaur plugin. Then such rendered and saved video clips with optional video effects added can be imported to SR5 or VRD for smart cutting and joining, since rendering to join the cut clips in Sony Vegas may require more time and result in some artifacts & few extra frame repetitions at joint points, especially when done by an amateur hobbyist.

[criggs]

Many folks now use smartphones to record MP4 video clips...TMPGEnc Smart Renderer 5 imports such clips in correct vertical orientation, and can auto identify what segments of these clips should be smart rendered, and what segments re-encoded to the Master clip parameters, when joining clips together after cutting unneeded pieces from them. SR5 also auto identifies the biggest clip as Master to save more time at joining by smart rendering.

As a matter of fact, since I posted that tutorial, I no longer use VideoReDo for most of these jobs. I not only found myself repeatedly bumping into h.265 clips but also into h.264 clips with the 4:2:2 encoding. Only VideoReDo Pro ($300) can handle those, so I switched a few months ago to Smart Renderer 5. Smart Renderer 5 is not perfect. For example, the other day I bumped into what I thought was a perfectly ordinary h.264 clip which Smart Renderer 5 could not handle. It was High profile at Level 4 with CABAC/4 Ref frames. VideoReDo handled it fine. So I pretty much need both nowadays.

[TimA-C]

After all that effort . . . ! Oh well. Funnily enough I went from the forerunner of SR5 (Mpeg2 Editor?) to VRD 3.?? I think when I got my Hauppauge USB Freeview tuner and it couldn't cope with my Freeview recordings. Then I had to upgrade again when I started playing with HD Freeview and needed to be able to deal with H264 video. There has been a post or two on the VRD forums, about the next version of VRD which, IIRC, should be able to deal with H265 video, and supposedly isn't that far away.

Anyway, glad you got a solution in the end.

p.s. I was joking about the video tutorials!

[criggs]

There has been a post or two on the VRD forums, about the next version of VRD which, IIRC, should be able to deal with H265 video, and supposedly isn't that far away.

Probably too late for me. I've only got 7 days left of free usage on Smart Renderer 5. When my time runs out, I'll take one last look at the VideoReDo site to see if they've released the new version with H.265. If it's still not released, I guess I'll be paying Smart Renderer 5 my $70. I may even get it anyway. Even though it's more unpredictable in terms of which files it won't accept, it's got so many really convenient features, even effects transition options of which VideoReDo has none, that I've found myself getting spoiled by it, and increasingly dissatisfied with VideoReDo's limitations.

Of course, one of the biggest headaches I keep bumping into these days are VP9 videos, which NEITHER app can handle. There are rumors Solveig may add VP9 support soon to their Splitter ap (see http://www.solveigmm.com/forum/index.php?topic=5327.0 ), in which case I'll probably be getting that instead. That said, when I tried the free Splitter, my computer got really slow; it seems that Splitter uses a lot of CPU unfortunately. So I would probably only use it for VP9 files, nothing else.

I was joking about the video tutorials!

Hey, it was a good idea!