Best format for VHS capture, MP4 or MPEG2?

[BarryTheCrab]

I did an internet search but could not find the exact answer.
Given available equipment and no way out, and in need of SOME editing, would you prefer to capture tape as MPEG2 or MP4? Capture profiles can be bumped up to higher bitrates than "normal". I know, it's a choice in most minds here twixt gas chamber or electric chair, and I agree, but I ask anyway.
Along the same lines, compression artifacts...if the math is the math, why does high bitrate MPEG2 (for example) of around 12-15MB/s rather than customary 8MB/s-ish bitrates, why are there fewer artifacts? Just curious, and fearful the explanation will make my head explode and I'll miss the Pat's game at urgent care.

[jwillis84]

I wrestled with the same problem.

Let the source material guide your decision.

MPEG2 was designed for film and NTSC for playback on big screens.

MPEG4 was designed for OTA and Game capture, and moonlighted by hypercompressing NTSC for streaming on small screens.

Editing can mean simple "snip n clip" to edit out commercials with no re-encoding, or it could refer to "audio and video restoration with format conversion from ntsc/pal or pal/ntsc" for which re-encoding is unavoidable.

"Snip n clip" editors like VideoRedo are available for a descent price, more fuller function DVD Authoring suites like TMA are also available. Even "free" lossless (or non re-encoding) editing tools are available for MPEG2 so its not as big of an issues as it use to be.

Its sad but new stuff available for capture is better tailored towards creating MPEG4 output, be that as a result of capturing RAW and then compressing it down after editing, or compressing on the fly direct to disk.

The result of that is the signal conditioners, time base correctors and filters on the older MPEG2 equipment is gone, or being disabled by the MPEG4 equipment vendors even if its designed into the capture hardware.

The expectation is the source signal will always be "perfect" and stable and not in need of correction... which means that the MPEG4 capture hardware will be very unstable or intolerant of "Real-world" VCR signals..(even at high bit rates).. and makes the capture process that much harder.

The MPEG2 world had its problems too.. with DVD recorders only rare recorders allowed offloading the MPEG2 video without first burning it to disc, and that forced using an intermediary DVD-RAM/DVD-RW and cut and splice to get it to a PC for burning to a Blu-Ray. - DVD burning was and "is" slow.. much slower than people think acceptable today, and thats time added (on top) of the 1:1 time spent playing the VHS tape for capture in the first place.

Until very recently it wasn't easy to get MPEG2 straight off the HDD, though salvaging and video recovery tools were used. And its no secret that MPEG2 hardware encoding gear is quickly turning into vaporware, or tossed out because people don't see it as useful gear anymore or don't even bother to put it up for sale.

Windows Media Center Edition 2005 kind of took over the role of MPEG2 DVD recorders with a PVR leaning, and ATI Wonder Elite (or the 550) chip "kind of" simplified that all down into a "new" metadata wrapper for MPEG2 in the dvr-ms file format.. which was as close to DVD-Video title sets on a PC or Network as you could get. Thats important for tracking your videos and presenting them nicely in a playback device or streamer.

In a way that should have been the "Peak" as that Elite PCI card had hardware MPEG2 and both hardware audio and video capture all on one card. It could be used in an HTPC mode with MCE or in full blown PC Capture and Edit suite mode.. but things conspired against it.. and it never caught on. Reviewers of the time torpedoed it since (like BetaMax) it wasn't as good at compressing down timeshifting video to MPEG4 levels for network streaming. But it also had a reported AGC problem causing "flickering" which drove some people away (some people thought it was only Macrovision interference and not really a thing if you weren't trying to copy Copyrighted videos)... the same year ATI was sold and a couple years later the markets tanked, an after that NTSC was replaced with ATSC and it all just went to heck.

AMD did prop up the ATI idea of the 550 chip on a USB dongle in cable card tuner, and few "butter dish" designs, but they were too expensive compared to slimmer less capable USB capture dongles, and people still had access to MPEG2 DVR recorders.. so they basically failed in the market and disappeared.

[lordsmurf]

MP4 (MPEG-4) is just a container, not a format.

MPEG-1/2 is both format and container (though the container can only hold that format). Something like AVI can also containize MPEG, as we get with the Matrox codecs.

[sanlyn]

The following are vast oversimplifications:

Lossy inrterframe codecs (MPEG/h.264) are the lowest quality choice for VHS and other analog sources. These are final delivery codecs not designed for quality output from noisy sources analog sources such as VHS, VHS-C, SVHS, etc., and not designed for editing without further lossy-stage re-encoding and serious quality loss. They encode VHS noise as imbedded digital artifacts that are difficult or impossible to remove without severe damage and further data loss. Data loss from lossy interframe encodes cannot be recovered, and additional stages of lossy re-encoding result in cumulative, unrecoverable loss and image distortion. These codecs are the poorest choice for repair and restoration work. Even at high bitrates they are a poor choice for archival use and will always look as bad as, or worse than, the original source.

Lossy intraframe encodes (consumer DV, MJPEG) are shoot-and-view codecs not designed for re-encoding without further data loss or quality degradation. They can undergo simple cut-and-join editiing without re-encoding, but any other image modifications (resizing, color correction, transitions/special effects, masking, title overlays, etc.) require lossy re-encoding and causes further loss, degradation, and added artifacts. VHS source noise and other defects are interpreted as additional digital artifacts and compression noise that don't exist in the original source. Their YUV color systems are not identical to the YPbPr color storage systems used in VHS/SVHS and will involve visible color inaccuracies and other undesirable effects. Besides color effects, NTSC sources will undergo 50% chroma loss using consumer DV codecs. These codecs are designed for PC-only or camera-only playback and require further lossy encoding to other lossy codecs for non-PC playback systems and media. As usualy, "lossy" means "you lose" -- data lost through lossy encoding cannot be recovered.

So-called "virtually lossless" intraframe codecs such as ProRes are lossy neveretheless and require skill and experience in their configuration to avoid further damage through editing, re-encoding, repair and restoration work. They are designed primarily for virtually noiseless digital sources.

Lossless codecs (Huffyuv, Lagarith, Magic, UT Video, etc.) are (usually) intraframe codecs designed primarily for archiving, repair, and restoration work involving compression and recompression with zero data loss. They are the nearest method available for achieving a virtually 1:1 analog-to-digital copy of analog sources. They are considered free of added digital compression effects.

interframe encoding consists of compressed video frames arranged as Groups of Pictures (GOP)'s containing one or more Key or Index frames and several to hundreds of additional partial images whose data and contents are based on changes between Key frames and subsequent index frames, predictive partial frames, and connecting GOP's. Edits involve decoding and re-encoding of entire GOP's. The only complete full-frame images in interframe GOP's are key frames. Because frames and entire GOP's are inter-dependent on one another, edits or image changes with intraframe encodes involve re-encoding of the entire video and all GOP's, except for specially designed "smart rendering" apps that can re-encode only a few frames or GOP's n the immediate area of cuts or other edits. Some smart-rendering apps are "smarter" than others.

Intraframe encoding consists entirely of complete images, i.e, made up entirely of Key or I-frames. Each frame is encoded independently of the content of other frames.

[BarryTheCrab]

Thank you for your time and answers. There is no site like this.

[lordsmurf]

Quote:

Originally Posted by BarryTheCrab

Thank you for your time and answers. There is no site like this.

Thanks.

[dpalomaki]

Lossy vs. lossless, and bit rates, all go to precision, If you have more bits available you can specify the image more precisely. With compression you have fewer bits to describe the image, and at some point the compression math becomes lossy. The compression routines are designed to discard the portions of the image that most users cannot see first. But once lost it is gone forever. Lossless compression is a bit like using PKZip (or RAR or Arc or PKArc) for data files, the data is compressed in a way that it can be recovered without loss when unZipped.

As time goes by compression routines have become more "efficient" in that they can achieve the same nominal image quality using fewer bits. This has been driven in large part by having more processing power available for the encode/decode process as PC evolve. Thus for the same overall end viewing quality, given the same uncompressed input image. H.264 is more efficient than H.264 and it is more efficient than MPEG. Further this is driven by the move to more streaming delivery and display system.

The gear to support old legacy analog formats like VHS is disappearing because there is no significant market for it; i.e., if you can't sell it at a fair price, why produce it. Modern digital sources are not subject to the mechanical variables and noise issues of the analog formats, so there is no demand for the gear needed to resolve those analog issues, especially at the consumer/prosumer levels.

The lossless formats will give you better end results, especially if you need to do any correction/restoration, with analog tape source material. A simple dump from a good S-VHS VCR direct to a DVD recorder (with TBC if needed) will probably give a better end use product than trying to play the same tape in a generic home VCR. That is why many consumers are content with results of the $20 transfer services. Only you can decide how good is good enough for your purposes. These forums provide information on how to do it better.

[sanlyn]

VHS recorded directly to lossy formats, whether MPEG, h.264, or whatever else that users can devise to re-invent the wheel along lower-performance terms, really looks like crap. It can never look better than the source and usually looks worse. Having tried to fix up a few of these lossy transfers over the past couple of years, I simply have to ignore this junk the next time I see it posted. The cleanup effort from lossy just isn't worth it and is always disappointing.

I agree, a lot of people think lossy garbage looks just fine. What a shame. It looks like an oil painting somebody cleaned with a rag soaked in turpentine and scrubbed with a wire brush, and the user is right proud of having been clever enough to think up all that damage. Visual illiteracy is the order of the day. Thank heaven they're not making the movies or the tv shows, although lately I've seen some cable broadcasts that look as if they were downloaded off PoohTube. I'm afraid this low-tier mediocrity and visual down-conditioning is gradually drifting into the mainstream. Thank goodness I have thousands of well made DVD's to replace the dreck that's on the way, and fond memories of growing up when professional film was king of the silver screen. Today's video neanderthals have no idea what they've missed.

[dpalomaki]

When noise becomes art...

"Blair Witch Project" was a horror film wasn't it? (I've only seen trailers for it.)