Currently I am using Cyberlink's PowerDirector (with Hauppage) to capture video from my TV to my laptop. When I render the MPEG files, it uses SVRT. After that process, I convert the files using Handbrake to prepare them for my PLEX server.
I was thinking about plunking down a few dollars to purchase VideoRedo. The only advantage I can see with VideoRedo is that it can edit MPEG-2 frame-by-frame which is very handy but I rarely need that feature. According to their website, they also use SVRT.
Is there a comparison chart somewhere that can explain to me the differences among SVRT's or are they all basically the same? Or is there a way that I can try both and evaluate the outcome myself?
+ Reply to Thread
Results 1 to 10 of 10
What do you mean by "capture from my TV", because very few TVs can output video from any connection? Are you capturing the output from a set top box (cable box/dvr, or satellite receiver/dvr) using a USB capture device, or recording the output from a USB digital TV tuner? Which Hauppauge product are you using?
As editors, VideoReDo Plus and VideoReDo TV Suite H.254 use smart rendering that only re-encodes one or two GOPs on each side of cuts that would otherwise leave GOPs incomplete. If there are missing/corrupt frames in the audio and video streams, as is often the case when recording the output of a digital TV tuner card, they will also re-encode the affected sections of the streams to preserved audio and video synchronization.
Perhaps I was imprecise in stating my question. My question is this. Is there a standard algorithm for Smart Video Rendering Technology or does each software maker use their own proprietary version of SVRT?
To indulge you, I will describe my "TV hookup" (which has nothing to do with my question). My cable OUT (composite) is connected to an old VCR's IN (composite). One VCR OUT (component) is connected to my TV. One VCR OUT (composite) is connected to my Hauppauge USB Live 2 (which I know is bottom-of-line but adequate for my needs) which is in turn connected to my laptop (DUO processor if you must know) via USB.
Now back to my original question -- are the algorithm's for SVRT common or unique?
I doubt there is much confusion or debate about Smart Rendering. You keep all the frames it is possible to keep and only render the frames that have been affected by any processing. For simple edits with no FX, Color Correction, Resizing, Titling, Compositing, Colorspace change or other similar processing, this means I-frame-based-media projects don't have to render ANYTHING unless the edit includes a transition (where the transition is the only thing that has to be rendered). Same thing if one adds a title layer above portions of a simple track: the portion including the title would need to be rendered, but nothing else. That's all very straightforward.
GOP-based-media projects throw in extra layers of complexity, but only in that now it has to deal with dependent frames as well. So the area to be affected has similar processing constraints, plus non-I-frame editing, and that area has now expanded to the 1 GOP (in the case of Closed GOP sequences) or 2 GOP (in the case of Open GOP sequences) level (double these if your are talking about a transition between 2 non-I-frame sequences).
So the answer ought to be that the algorithms for deciding if something is affected IS UNIVERSAL. However, we know that companies and their programers have limits on their time, budget, and knowledge, so you expect there to be some variation in that (and unfortunately, we can only tell by testing - they rarely divulge their code to us).
Thank you. So the simple answer is that SVRT algorithms differ among software developers and there is no way of knowing which is best. Right?
As an engineer, it's doubtful that I would give you a straight up answer without some equivocations, restrictions & dependencies added on.
In a perfect world, they shouldn't differ. They probably DO in the real world. I wouldn't say there is NO way of knowing, you just have to do some thorough tests to find out. I would start, however, by taking their claims at their word (after having read fully on their advertised scope & limitations). It helps if you have a good understand yourself of the possible categories and being able to read between the lines (because usually it is the marketers who are putting out app descriptions, not the engineer/programmer, and they may not get it themselves).
The thing is, even if they do differ somewhat, they are still basically the same. You are making this out to be a bigger deal than it really is. Stuff either smart-renders or it doesn't, and their smart-rendering capability also depends on which codecs/formats they are able to smart-render. Looking at the history of it: 1st smart-rendering was DV, then MPEG2, now MPEG4 & AVC (in most of their flavors). There may be other codecs (e.g. ProRes in FCP, DNxHD in AVID), but those are the main ones. And anytime an editor can edit on non-I-frame or non-GOP-boundary frames, it ALWAYS has to render, whether it be smart-rendering (limited area of rendering) or non (whole file rendered).
Of course, you can always CHEAT!
Here's a tip: You can pre-split (on GOP boundaries) into chunks, work on only the chunks that need rendering, render them in a non-smart-rendering app (using IDENTICAL encoding parameters), then join the files back together afterward. BINGO, it just smart-rendered (since those files all started & ended on GOP boundaries and are of identical format and all you are doing now is concatenating them, there should be no need to render anything). Use a known-good GOP-based cutting & joining app for the first & last parts, of course.
I thank you for your time and patience with my question. As of now, the only program I am aware of that can designate GOP boundaries is XMedia Recode. When I have time, I will play with that by program by chunking the original and working with the chunks to eliminate the advertisements.
In the meantime, I think I will go with VideoRedo, just to be sure that I have the correct frame without having to worry if the frames are I P or B. Basically I want to render the files with as little recoding as possible to avoid deterioration for display on a larger screen. (Don't forget, I still plan to use Handbrake as the final step.)
Since you have been so helpful thus far, one last question. Single or Double pass?
Re: 1 or 2pass encoding, 2pass almost always is better quality (because 1pass is a wild guess and 2pass is a calculated, educated guess), but does take twice as long. However, do you need to be targeting a specific filesize/bitrate limit? If not, why not skip the cbr/vbr methods and use cq/crf? It is 1pass and gives you the quality you desire at whatever the filesize/br turns out to be.
To educate you... The video source and the identity of a Hauppauge capture device is somewhat important to know when recommending an editor to use for editing capture files. The Hauppauge HD-PVR series of USB capture devices output an H.264 transport stream, and VideoReDo TV Suite H.264 does a better job editing them than most other H.264 editors. Saying that you don't have much need for frame-accurate MPEG-2 editing doesn't entirely rule out the possibility that you might be using an HD-PVR series device instead of the USB-Live 2. ...and there was the possibility that you were capturing the output of a Hauppauge USB digital TV tuner, which can output an MPEG-2 transport stream in need of error correction. In that case either VRD Plus or VRD TV Suite H.264 would be preferable to PowerDirector
Last edited by usually_quiet; 6th Apr 2014 at 01:56.
Ya see. That's what you folks are there for. To tell us the things that we don't know and I do appreciate it.