Should deinterlace VHS captured material ?

[macmithos]

hi all ..
I'm sending this because of something I've read yesterday ....
A guy saying that after capture a VHS material, before send it to Compressor to do it's job (as a QT video), we should first deinterlace the video to avoid "blocks" on final compressed video. Does it proceeds ?
Moreover ... "There are really no diference between the DVD presets "Best Quality and Fast encoding" for a VHS captured material" ....
If those arguments are right, I'm wasting time doing the Best Quality and making no good DVD out of a QT movie interlaced ... lol

[jagabo]

If you're making DVDs you should NOT deinterlace. Just encode as interlaced MPEG 2. The DVD player and/or TV will play the video properly. Deinterlacing may throw out half the temporal resolution and as much as half the spacial resolution -- depending on the source and the deinterlacing algorithm used.

If you are making video primarily for internet streaming you'll want to deinterlace.

The noise and sync problems in VHS caps makes it harder for encoders to find motion vectors (the major difference between different quality settings in the encoder). Depending on how good your VCR and processing is you may not see much difference. Try it and find out.

[lordsmurf]

If you are making video primarily for internet streaming you'll want to deinterlace..

... using a good method, not some checkbox in crap software.

Know that blend, odd, even, and drop-frame (same as odd and even) are all CRAPPY METHODS.

[lordsmurf]

hi all ..
I'm sending this because of something I've read yesterday ....
A guy saying that after capture a VHS material, before send it to Compressor to do it's job (as a QT video), we should first deinterlace the video to avoid "blocks" on final compressed video. Does it proceeds ?
Moreover ... "There are really no diference between the DVD presets "Best Quality and Fast encoding" for a VHS captured material" ....
If those arguments are right, I'm wasting time doing the Best Quality and making no good DVD out of a QT movie interlaced ... lol

If this is something you've read online, please link to it here.

[kreg]

Know that blend, odd, even, and drop-frame (same as odd and even) are all CRAPPY METHODS.

I do VHS to Xvid. I deinterlace with blend in Virtualdub. What do you recommend instead?

[Vidd]

There is no one suitable deinterlace method for all material.

The higher quality methods using AviSynth include MCBob, TempGaussMC, and Yadifmod+NNEDI.

[terryj]

A guy saying that after capture a VHS material, before send it to Compressor to do it's job (as a QT video), we should first deinterlace the video to avoid "blocks" on final compressed video. Does it proceeds ?
Moreover ... "There are really no diference between the DVD presets "Best Quality and Fast encoding" for a VHS captured material" ....
If those arguments are right, I'm wasting time doing the Best Quality and making no good DVD out of a QT movie interlaced ... lol

jagabo's post in the forum is the way to go.

Having done thousands of VHS to DVD transfers from TBC VHS VCR to my Canopus ADVC-110 to the Mac,
you do NOT deinterlace the video while it is being captured, but when you encode to
DVD ( MPEG-2 assets), then the interlacing will occur in the Compressor Preset.

As for the "Best Quality" argument, your looking at it the wrong way.
"Best Quality" preset in Compressor is for the END RESULT, not the Beginning result.
In this case, "Best Quality" will refer to closely matching WHAT WAS ORIGINALLY
INPUTTED INTO THE MAC TO BEGIN WITH, and encoding that footage at a high
enough Bitrate to match the original footage input, as so that compression
artifacts do not WORSEN the footage being transcoded to MPEG-2.

How you affect "Best Quality" on the capture front end is by having the following:

1. Best Source of the movie possible ( Factory VHS over On Air Dub to tape,
quality recording in SP mode vs. One off recording in ELP mode, etc.)
2. A TBC ( Time Based Correcting) VCR with CLEAN HEADS to keep the tape stablized
3. Proper Transfer units to get the footage captured via FireWire, such as a Canopus ADVC-110

For most VHS footage, if you did everything you could at the outset to insure against GIGO
(Garbage in Garbage Out), then your end result should focus on what you
need on the DVD format, as opposed to trying to make quality better,
because on VHS transfers, you really can't make them better ( overall Quality wise as far as playback)
against what you have to work with at the input stage.

Focus at the end stage, as far as your preset, what you need and set your encoding preset to that:

1. How many films do you need on the DVD?
2. Are you making a Fancy menu with Main Menu, Chapter Menus, extras, etc, or
just a simple Play All menu?
3. How long is the film? 2hrs, 4hours, etc?
4. What kbps rate is the audio? 224? 192 ( acceptable) or 160?

[lordsmurf]

You don't deinterlace when encoding the DVD, either.
The ONLY reason to deinterlace is to forever commit to streaming video.

Anything viewed on a television must be left interlaced for best quality.

DV further degrades VHS quality, as opposed to uncompressed methods, and sometimes even direct-to-MPEG methods. DV was intended for shooting, not conversion. You can get away with it on the high quality pro devices, but the consumer grade Canopus DV boxes don't qualify.

[macmithos]

Thank you all for your advise. You did ensure what I've learned here at this site .... and once more we have to be more carefull about things we read on internet writen by "curious people" ...

[orcus]

Anything viewed on a television must be left interlaced for best quality.

Does this apply even with progressive displays like HDTVs?

I am about to convert VHS tapes (analog camcorder footage) and was not sure if I should deinterlace or not. I mainly watch on a projector or pc monitor, both getting their signal from the PC. So unless I want to watch stuff with lines all through it I've got to at least turn on ffdshow's deinterlacing, so it seems like using a superior non-realtime deinterlace during encoding would make more sense, or am I wrong? I haven't watched anything sent from the PC to my plasma tv in a long time so I'm not sure if it's the same situation there. At any rate I don't think I'll ever again be using an interlaced display.

[jagabo]

Anything viewed on a television must be left interlaced for best quality.

Does this apply even with progressive displays like HDTVs?

I am about to convert VHS tapes (analog camcorder footage) and was not sure if I should deinterlace or not. I mainly watch on a projector or pc monitor, both getting their signal from the PC. So unless I want to watch stuff with lines all through it I've got to at least turn on ffdshow's deinterlacing, so it seems like using a superior non-realtime deinterlace during encoding would make more sense, or am I wrong?

The problem is this: if you deinterlace now you are forever locking in whatever losses and artifacts the deinterlacer is producing. Whereas if you leave the video interlaced it may not look as good now using ffdshow's deinterlacer, but in the future, as on-the-fly deinterlacers improve, playback of your interlaced video will improve, eventually getting better than you can do now in software now.

The very best software deinterlacers now are pretty good (AviSynth's TempGaussMC_beta(), for example) but are also very slow -- typically a few frames per second. On-the-fly deinterlacer probably won't get significantly better than that for a long time.

Some exceptions to the "don't deinterlace" rule are reversing the 3:2 pulldown of NTSC telecined film sources and realigning the fields of out-of-phase PAL captures. Both of these can usually be done very cleanly and give you better encoding with lossy codecs like MPEG 2. Especially IVTC of NTSC video where you can encode 23.976 fps progressive rather than 29.97 fps interlaced.

The other time you want to deinterlace is when uploading to on-line video sites like YouTube. They don't deal well with interlaced video.

[orcus]

Thanks for the response. This is pure interlaced, not telecined or anything like that. I will give the Avisynth deint you mentioned a try. The thing is that being old camcorder VHS footage, mostly shot in low-ish light, this footage is pretty crappy looking anyway, so I doubt there'd be much (any?) noticeable improvement down the road with superior deinterlacing, as that's the least of its problems.

[jagabo]

this footage is pretty crappy looking anyway, so I doubt there'd be much (any?) noticeable improvement down the road with superior deinterlacing, as that's the least of its problems.

Deinterlacing from 30i to 30p will noticeably reduce the smoothness of motion. AviSynth's TempGuassMC_beta() is a bob'er. It will give 60p from 30i. If you need 30p you can use SelectEven() to throw out half the frames. This post has a comparison of several AviSynth deinterlacing filters on some difficult material:

http://forum.videohelp.com/topic363647.html#1935596

Also, if you don't have a VCR with a line time base corrector the deinterlacing algorithms may have trouble determining what is motion and what is simply time base errors.

[Whuntmore]

A guy saying that after capture a VHS material, before send it to Compressor to do it's job (as a QT video), we should first deinterlace the video to avoid "blocks" on final compressed video. Does it proceeds ?
Moreover ... "There are really no diference between the DVD presets "Best Quality and Fast encoding" for a VHS captured material" ....
If those arguments are right, I'm wasting time doing the Best Quality and making no good DVD out of a QT movie interlaced ... lol

jagabo's post in the forum is the way to go.

Having done thousands of VHS to DVD transfers from TBC VHS VCR to my Canopus ADVC-110 to the Mac,
you do NOT deinterlace the video while it is being captured, but when you encode to
DVD ( MPEG-2 assets), then the interlacing will occur in the Compressor Preset.

As for the "Best Quality" argument, your looking at it the wrong way.
"Best Quality" preset in Compressor is for the END RESULT, not the Beginning result.
In this case, "Best Quality" will refer to closely matching WHAT WAS ORIGINALLY
INPUTTED INTO THE MAC TO BEGIN WITH, and encoding that footage at a high
enough Bitrate to match the original footage input, as so that compression
artifacts do not WORSEN the footage being transcoded to MPEG-2.

How you affect "Best Quality" on the capture front end is by having the following:

1. Best Source of the movie possible ( Factory VHS over On Air Dub to tape,
quality recording in SP mode vs. One off recording in ELP mode, etc.)
2. A TBC ( Time Based Correcting) VCR with CLEAN HEADS to keep the tape stablized
3. Proper Transfer units to get the footage captured via FireWire, such as a Canopus ADVC-110

For most VHS footage, if you did everything you could at the outset to insure against GIGO
(Garbage in Garbage Out), then your end result should focus on what you
need on the DVD format, as opposed to trying to make quality better,
because on VHS transfers, you really can't make them better ( overall Quality wise as far as playback)
against what you have to work with at the input stage.

Focus at the end stage, as far as your preset, what you need and set your encoding preset to that:

1. How many films do you need on the DVD?
2. Are you making a Fancy menu with Main Menu, Chapter Menus, extras, etc, or
just a simple Play All menu?
3. How long is the film? 2hrs, 4hours, etc?
4. What kbps rate is the audio? 224? 192 ( acceptable) or 160?

Now here's the person I should be talking to...

I have an iMac (leopard) 2.4 gig, 4 gig ram and 1.5 TB ex. drive. I also have the Canopus ADVC-300 via firewire. I'm using the S-video IN. Two RCA plugs to my VCR/DVD combo for Audio (Toshiba DVD Hi-Fi VHS D-VR4SC) I have been using the ADVC software to clean up the video Settings are usually on weak or low and the rest is on default. I'm currently using iMovie '06 HD, but I've ordered Final Cut Express 4.x.

I'm editing it iMovie, then sharing with iDVD. It seems to use iDVD '08, and I've got that setting on Professional Quality. I'm keeping movie length under one hour, and keeping chapters and splash screen to a min. then straight to either a .img file (so I can burn on my other pc), or straight to burn.

When I capture into iMovie, it seems to look pretty good, but when I burn, and then try to watch on the DVD player, it's blocky and poor.

Anything you can recommend (settings for the Canopus software, using FCX) other then expensive hardware, or expensive software? Should I stay under or about the one hour amount on a single layer disk? Should I (or not also) deinterlace? What editing software do you use? FCX, or Final cut Pro?

Do you have a step by step on what you actually do?

Any help is much appreciated.

[sphinx99]

A lot of conventional knowledge about whether to deinterlace or not is a bit outdated. Sure, maintain interlacing if encoding to DVD and don't if uploading to youtube... unfortunately it's 2010 and those no longer are the only two options.

For example a lot of what I capture gets encoded to h264 in a mkv container and watched on my BD player which streams and upscales MKV files from a network share. Encoding h264 natively interlaced can be a pain, and it's amazing the number of hardware h264 decoders that don't do a good job deinterlacing, so I'll deinterlace and encode in that situation.

I also capture quite a bit of content that will be viewed on a computer. Sure my receiver has a HQV Reon-VX chipset and my Blu-ray player has an ABT2010 and both do a terrific job deinterlacing, but that doesn't do me much good when I'm encoding material to watch on my laptop using VLC or Windows Media player. So that's another situation where I'll deinterlace.

Obviously you have to pick your battles. But remember, if you're encoding VHS at 720x480, you have a lot of extra resolution with which to dull the edges of that "deinterlacing causes permanent information loss" argument.

[M-O]

No wonder i was thinking "why haven't i heard of this?" What's with Avisynth and WMP talk in the Mac forum?
Seems to me that whether it's windoze or dual-boot, talk about that junk should not be here. I googled around and couldn't find anything on a working version of Avisynth for OS X.

[lordsmurf]

A lot of conventional knowledge about whether to deinterlace or not is a bit outdated.

No. The information is as accurate now as it was 10 years ago.

if uploading to youtube... For example a lot of what I capture gets encoded to h264 in a mkv container and watched on my BD player which streams and upscales MKV files from a network share.

Youtube is streaming. MKV from network share (computer files!) is streaming. You're just not using the right terms. The container doesn't change the methodology. It's either viewed conventionally or streaming. That's it. There are ONLY still two choices.

Encoding h264 natively interlaced can be a pain, and it's amazing the number of hardware h264 decoders that don't do a good job deinterlacing, so I'll deinterlace and encode in that situation.

You have to make the best choice for the situation. You may very well be making the correct (or only) choice. The problem comes where people don't know what choice to make, where, how or why. Discussing formats and containers just confuses the issue.

I also capture quite a bit of content that will be viewed on a computer. Sure my receiver has a HQV Reon-VX chipset and my Blu-ray player has an ABT2010 and both do a terrific job deinterlacing, but that doesn't do me much good when I'm encoding material to watch on my laptop using VLC or Windows Media player. So that's another situation where I'll deinterlace.

Again, streaming video. Local stream, mind you, but still a computer-based stream -- meaning "not conventionally" viewed. (Conventional = TV with player, or TV from signal)

, if you're encoding VHS at 720x480, you have a lot of extra resolution with which to dull the edges of that "deinterlacing causes permanent information loss" argument.

That doesn't make any sense to me. If you want to deinterlace VHS, I suggest it be done at a 4:3 interval, either 768x576, 720x540 or 640x480. The odd-sized 720x480 and 352x480 DVD-Video resolutions suck for making these mid-workflow alterations. Standard non-4:3 disc-format resolutions create a lot of aliasing. The jaggies are not caused simply by the deinterlace, but by the re-stretch between storage and display AR.

I've discussed video capture and interlacing / deinterlacing in past posts both here and elsewhere in the past 6-7 years. It's a topic I follow quite closely, actually, to stay on top of the newest hardware and software methods. The professional community struggles with this too, it's not just the hobby/home crowd. (And some pros are just as clueless as Best Buy shoppers, to be fully honest!)