What would be a waste of bits on VHS

[Marco33]

When caping an original VHS, what would be a waste of bits?

Since DVD cap setting at 704X480 is 6400 bits. Anything more than that on vhs would be a waste.

No?

[Marco33]

I'll take that as a yes.

[txpharoah]

Your resolution is a waste too.

Televisions have a max of about 352x480, which is true s-video resolution, and a middle compliant DVD resolution.

Wasted bits depends on your source quality. I sometimes encode at 8MB/S, sometimes at 3MB/S. All footage has different needs. In general 6MB/S is fine. Just saying "VHS" doesn't mean too much.

The old industry standard on DVD was 352x480 or 720x480 at 5.71MB/S.

With DVD9 discs, they changed it to the max allowable of 8MB/S, up to ~9MB/S if encoded with AC3 audio. Resolutions still vary from 704x480 to 720x480 to 352x480.

[energy80s]

For a good, clean VHS encode use 352x576 (480) at a bitrate of 3 or 4Mbps.

[Marco33]

I get a better picture with 6000 than 4000.


txpharoah?

When you say TV, do you mean tub or projection or both?

So let's see.

VHS 356X480 6400 tops, even WS videos.


hmm

[mpb]

I get a better picture with 6000 than 4000.
txpharoah?

When you say TV, do you mean tub or projection or both?

So let's see.
VHS 356X480 6400 tops, even WS videos.

hmm

Well, OK, I see no mention of bitrate ranges here so I hope I can be helpful to you. VHS and SVHS tapes are not holding interpretable analog data that really exceeds 352x480 in a practical sense. I could show you captures from VHS at that resolution, encoded to DVD, left interlaced, that look EXACTLY like the source - in every respect, including that "smoothness" that seems to be absent in captures sometimes (and is generally a function of vertical resolution and your interlacing options).

I tend to capture everything at 720x480 through DV now, but for VHS, I still "encode down" to 352x480. If you capture analog, cap at 352, 480, or 640x480 and you'll get a perfectly useful source file. If you can, I recommend 720x480 for broadcast television (including satellite and such) and laserdiscs. It's a bit near the top of their resolution approximation, and will yield a good picture.

Here are my bitrate ranges for saving file space while not losing any visible quality whatsoever. Macroblocks and stutters are NOT acceptable to me and I'm guessing you won't want any either. I gauge the average by the amount of motion/action in the video clip. Music videos are deceptive; they tend to require lots of bitrate because I can tell you that drums being played are very fast motion and look BAD if you skimp on the bitrate. Also I pad the minimums for maximum DVD player compatibility. My Apex 1500 doesn't care if the bitrate drops way below that... but many DVD players do, and you want to be careful of this.

-
TMPGEnc - template NTSC-DVD

720x480
min: 2000 padded
max: 8000
average: 3000 for no-motion, 4000 for normal, 5000 for high-motion

352x480
min: 2000 padded
max: 4900
average: 2600 for no-motion, 3200 for normal, 3600 for high-motion
-
-
TMPGEnc - template NTSC-SVCD*

480x480
min: 1150 padded
max: 2520
average: 1900 for no-motion or normal, 2200 for high-motion
-
TMPGEnc - template NTSC-VCD**
352x240
1150cbr only
Quality: "Highest Quality"

* Notice for SVCD that the CVD resolution (352x480) which is also the low-end DVD res... is better to use; you burn to a DVD and take advantage of the higher bitrates with slightly less image resolution, the end result is just a better, cleaner, and smoother looking video. I only use SVCD when I want to save money by burning something that isn't all that important to a CDR instead of to a DVD-R. 23 cents vs. $1.70 is a big difference.

** You can't control bitrate so crank up the rest of your quality settings. A helpful hint: use the clip frame and MASK to letterbox over existing letterboxes in your source material. This is common on widescreen stuff like you mentioned, as well as music videos. The letterbox on your source will actually be a grainy black, sucking up precious bitrate, bitrate we can't turn up or else we go non-standard. If you mask them over, "hard-matting" them, minimal bits will then be allocated to the black areas and they'll look much more black and less a part of the picture anyway. Two identical videos, one masked and one not, captured uncompressed, both CBR-1500 VCD resolution - the one masked will look FAR better because it could use all 1150kps on the part of the image that mattered.

Good luck in your efforts!!

[Marco33]

Real tests.

Thanks for your effort.

[Nelson37]

When the increased bitrate (or resolution) does not make it look any better, then you are wasting bits. The variability of source quality and viewer perception does not allow for any other simple answer.

I cap everything at 720x480, 12 MB/s. Multiple tests confirmed that, for me, 12 MB/s looked better than 10, and 720x480 looked better than a lesser resolution. I then re-encode that source file.

It is easy to get caught up in a numbers game. You can not mathematically prove that one file looks better than another, any more than you can prove that one cherry pie tastes better than another. If it looks better to you, then it IS better.

If you are going to re-encode, then there is no practical limit, get that source cap as good as you can. If you wish to cap straight to final file, then size versus quality becomes a meaningful consideration.

[Marco33]

To each his own. I agree.

But, I would hate to waste bits on capping VHS when the resolution is lower than what I'm capping at. It will not improve. So why bother.

[mpb]

To each his own. I agree.

But, I would hate to waste bits on capping VHS when the resolution is lower than what I'm capping at. It will not improve. So why bother.

I should have clarified.

With TMPGenc I am usign 2-pass VBR at all times to encode. For some reason I am finding that this delivers quality that is very hard to match by any other method.

All captures should be either:
1. Uncompressed,
2. Huffyuv lossless compressed, or
3. DV-AVI via 1394.

Capturing to mpeg on the fly is always a bad idea. Not only do you lose quality even at high bitrates, but it's often off-spec MPEG 1 or 2 requiring a pass through TMPG anyway if you plan to burn it to a DVD and have it work properly in as many players as possible.

If you plan to compress to 352x480, such as with VHS, it's perfectly alright not to capture a whole lot higher than that. Plenty of caps I've run at 352x480 right on the nose have turned out great after compression and burning.

Good luck!!

[txpharoah]

On an ATI card, you can capture to MPEG2 with I-frames only, and it won't look any better or any worse than any AVI with any codec. Best part: it's already MPEG2. Ready to dump to disc. All that fancy AVI that everybody loves to push is not needed if the source is just VHS and you have a decent hardware MPEG encoder on your card, of which ATI has...

[mpb]

On an ATI card, you can capture to MPEG2 with I-frames only, and it won't look any better or any worse than any AVI with any codec. Best part: it's already MPEG2. Ready to dump to disc. All that fancy AVI that everybody loves to push is not needed if the source is just VHS and you have a decent hardware MPEG encoder on your card, of which ATI has...

Actually... it's not MPEG2. Not strictly speaking. ATI MPEG2 is off spec. You have to run it through TMPGenc again to get it technically right for DVD burning.

There's nothing better, really, than using DV with a cam passthru (with tbc)... and then compressing from there, in terms of net ease of use vs quality.

[energy80s]

Very true. I used to cap with my ATI AIW card at 15mbps MPEG2 I Frame only and the quality still looked shite. RAW (and even Huffy) AVI's are just too bloody big to deal with in any meaningful way (unless its a five minute piece!) so DV AVI is the only way to go. The final quality is so much better with a DV AVI cap than with an MPEG cap that I would never go back to MPEG capture again.

[txpharoah]

Yes true, just reverse the fields to TOP A FIELD in TMPGenc with closed GOPs, and its fine. It will realtime or less transcode.

[Marco33]

I frames.

My cap software has a setting for GOP & Sub Gop. I know that has to do with I frames and the like. But, on my machine it wants #'s not I's or B's. So, I set my GOP to 6 and the Sub GOP to 3. This seemed to fix the pixelation at scene change. Not too sure what the best setting for me would be. I don't care how it helps me edit later on.

AVI or MPG

I always felt that It would be better to cap at MPG then, encode down then burn. With AVI you have to encode down then convert to MPG. You have to be losing something durring conversion to MPG.

[geekzapoppin]

The problem with capping in MPEG and then reincoding to another size or bitrate is that any artifacting present in the original capture will be amplified in any succesive encodes. Unless you have a reallllly good hardware encoder, you're bound to get minor artifacting. You're always better off getting as clean a source as possible for encoding.

[hwoodwar]

Marco33:

Your capturing program may allow you to watch the data rate that the video is captured at, VirtualDub does, on the right side of the capture window. If your capture program allows for this I suggest that you use that number, as any more would definitely be a waste of bits.

I have tried capturing VHS at 720 x 480, 704 x 480, and 352 x 480 and then encoding to 352 x 480 with mixed results. Sometimes the higher resolution captures look better when encoded to 352 x 480, sometimes I can't tell a difference. I suggest you try to make a test of a, capture and encode at 352 x 480 to see how you like it. Capturing at 352 x 480 I rarely see the bitrate go above 4300 kbps.

You mention not wanting to "waste bits". If you don't want to waste bits than use 2-Pass VBR instead of CBR. CBR does waste bits, as it applies the same bitrate to scenes that do not need that high a bitrate to maintain quality.

[SatStorm]

The trick ain't the bitrate, the trick is the filters you have to use...

For NTSC and because of the nature of this system, grabbing the higher you can and then add filters to stablise the picture and clean the random noise, not mentions adjust the colours.
For PAL any analogue grabb beyond 384 X 576 is questable. For me and because of today's mainstream limitations, it is a waste. Others may have other opinion...


Why filtering? Filtering ain't only for cleaning the picture visualy. It is also for cleaning the picture for things we can't see but the encoders really see.
The theory, say that 352 X 576 needs 2000Kb/s to support perfect picture. The praxis show us that without filtering, any source at this framesize needs double this bitrate for a "perfect" transfer".

Without filtering, it is like this: Encode with 2 Pass VBR (or more) with 1000 minimum (CVD's suggested lower value), 4000 average and 7000 maximum. That gonna give you VHS/SVHS virtual identical picture.

With the correct filtering, you encode with 2 Pass VBR (or more) with 1000 minimum (CVD's suggested lower value), 2800 average and 4600 maximum bitrate and get results like the non filtered picture.

Now the question rise: How to filter.
This is a huge subject and needs really hard reading and testing. It is always much easier to encode to a higher bitrate and be OK that going the hard way. But if you go the hard way, you have the satisfaction that you succeed something really difficult for the average enthusiast of this hobby, don't mention that you can keep more per DVD / CD disc.

A good start for those interesting for filtering, is this: You need at least 2 filters for VHS: Random Noise Reduction and Static Noise Reduction (with this order...). Use them wise and you save 1000kb/s average bitrate for any encoding, without loosing any detail or sharpness!
Advance methods includes the reconstruction of field to progressive frames and later re-encoding to interlace mpeg 2 (if your TV set is interlace of course).

[houtx67]

If you don't have a DV cam with TBC to pass through, but do have a Panasonic DMR-E30 (or other with TBC), then you can use that as a passthrough. I have found that it noticeably cleans up the cable signal that I pass to my VCR (versus a straight connection). I originally tried a splitter with a booster, but found that the pass through works better (and even when the Panasonic is off or recording another channel). Of course, if you have the DMR-E30, then you might take the easy route and just capture with it in the first place . You could then do any re-encoding/authoring you felt necessary later with those VOBs.

[doug1937@hotmail.com]

I dunno...just last nite I captured some VHS via Firewire at 720x480 and encoded a few minutes of it to 352x480 and also to 720x480, both with a 3700 bitrate. I burned those two small clips to a DVDRW and viewed in on my TV...the 352x480 technically looked fine, but there was something about the 720x480 that made me like it better...the colors were a tad richer, the lines a tad crisper, etc. The content was animated (cartoon), but I don't know if that would make a difference.

[SatStorm]

The 720 X 576 gonna look better if your DVD standalone don't do great digital to analogue convertion.

You see, 352 X 576 is anamorphic horizontly, that means an extra step for the standalone players. This extra steps ain't always be done the best possible way. Expecially on cheap standalones.

[vcd4ever]

The trick ain't the bitrate, the trick is the filters you have to use...

For NTSC and because of the nature of this system, grabbing the higher you can and then add filters to stablise the picture and clean the random noise, not mentions adjust the colours.
For PAL any analogue grabb beyond 384 X 576 is questable. For me and because of today's mainstream limitations, it is a waste. Others may have other opinion...


Why filtering? Filtering ain't only for cleaning the picture visualy. It is also for cleaning the picture for things we can't see but the encoders really see.
The theory, say that 352 X 576 needs 2000Kb/s to support perfect picture. The praxis show us that without filtering, any source at this framesize needs double this bitrate for a "perfect" transfer".

Without filtering, it is like this: Encode with 2 Pass VBR (or more) with 1000 minimum (CVD's suggested lower value), 4000 average and 7000 maximum. That gonna give you VHS/SVHS virtual identical picture.

With the correct filtering, you encode with 2 Pass VBR (or more) with 1000 minimum (CVD's suggested lower value), 2800 average and 4600 maximum bitrate and get results like the non filtered picture.

Now the question rise: How to filter.
This is a huge subject and needs really hard reading and testing. It is always much easier to encode to a higher bitrate and be OK that going the hard way. But if you go the hard way, you have the satisfaction that you succeed something really difficult for the average enthusiast of this hobby, don't mention that you can keep more per DVD / CD disc.

A good start for those interesting for filtering, is this: You need at least 2 filters for VHS: Random Noise Reduction and Static Noise Reduction (with this order...). Use them wise and you save 1000kb/s average bitrate for any encoding, without loosing any detail or sharpness!
Advance methods includes the reconstruction of field to progressive frames and later re-encoding to interlace mpeg 2 (if your TV set is interlace of course).

You get no arguments from me, using filters is the only way to get quality at low bitrates. I have used filters now in about 16 month time and i will never encode anything without using them.

vcd4ever.

[Marco33]

Here's something.

Say I capture at 7000 max bit rate. Then re-encode with CCE. There's no reason I should set the max rate to higher than 7000 since the file it's looking at doesn't go higher than that?

[macleod]

there was a post on dvdrhelp several months ago where a person did a test on the highest resolution for final output. He/she tested various resolutions and bitrates above and beyond what "TV" does and he did confirm that there were slight improvements the higher (res or bitrate) you went above the standard TV resolution size.

I cant remember the poster or when or whatever, I just remember that was the outcome. I think that "waste" should be determined by:

1. source (IE tape, and type of video). It is something that you want to preserve for all time, then I would ramp it up as high as you can go for the capture and then by determining the source type (motion/non motion) ramp down the encoding to answer #2 and #3.
2. expected results-Is this an old cartoon and not a big deal or is it the wedding video, kids home movies, etc.
3. output - (is it going to go on a DVD-R (4.3gig) or a CD ,etc. If you have a 30 minute tape, cap it as high as you can go and then use a bitrate calculator to determine the max that will fit on your output media.

My personal preference is to cap everything as high as my machine will let me (res and bit rate) and then use CCE (frameserved to with scripts) to convert at VBR 3500 - 5000.

[SatStorm]

The nature of NTSC needs to capture the highest possible and then resize while encoding to your target framesize.
With PAL, you can't capture for real beyond ~390 X 576, because of cable/hardware limitations of today's technology. With really good cables and hardware, you can grabb higher from TV signals or Laser Discs. RGB is the best solution, but needs special hardware.
An alternative for PAL is capture any analogue source though DV bridge. But this is not "capture". It is a whole different thing and also needs a DV camera. That way, you succeed better results than trying capture the orthodox way (with a PCI card for example)

[Marco33]

Just made a dry run. ROJ WS to DVD. Captured @ 704X480 7000 bit rate. Captured 2 seperate files. Cut & merged through Tmpegnc. Split with Tmpegnce, Encoded with CCE, Autho'ed with Spruce up.

I'm very happy. Looks just like the VHS. Captions are cut off when using the fill option on WS TV but overall, it's killer.

[Marco33]

Ran the file through Bitrate calc. The max bit rate was 7000 while capturing at 7500. The movies was Return Jedi. So I guess that answers that.