Am I doing it right? Video8 and VHS Capture

[Marto2008]

Hi.

Thanks to you all I managed to capture using my ATi TV Wonder 650 USB without sync problems and get the most of it.

My system is:
Core i5 750
4gb
Ati 4870 1gb
2tb Hdd
and the ATi 650 USB

I am capturing with Windows XP, Virtualdub and Lagarith lossless codec YUY2 at 720x480.

With these configuration I am getting a 50+gb file for 2hrs. video, which it's OK if I have 1 or 2 but tapes, I do have around 60 tapes and thou hdd are cheap, I would need like 3 more disks...

So what I did is... first of all "deinterlace" the source using "TempGauss Deinterlacer" (which doubles the framerate = cool) , apply unsharp mask and after that I that I use x264 codec; shrinking my file to around 5gb...

I am really not an expert and can't notice any "big" difference on the video file... so before continue my project and capture all 60 tapes I would like to knwo if i am doing it right... or almost right...
I do want my videos no be easily accessible in the future and to watch them on my lcd screen, I don't expect to be HD or better that the source... just equal.

Will be uploading snapshot of my configuration so you can check if it's OK.

Thanks and any comments will be appreciated.

[edDV]

Hi.

Thanks to you all I managed to capture using my ATi TV Wonder 650 USB without sync problems and get the most of it.

My system is:
Core i5 750
4gb
Ati 4870 1gb
2tb Hdd
and the ATi 650 USB

I am capturing with Windows XP, Virtualdub and Lagarith lossless codec YUY2 at 720x480.

With these configuration I am getting a 50+gb file for 2hrs. video, which it's OK if I have 1 or 2 but tapes, I do have around 60 tapes and thou hdd are cheap, I would need like 3 more disks...

So what I did is... first of all "deinterlace" the source using "TempGauss Deinterlacer" (which doubles the framerate = cool) , apply unsharp mask and after that I that I use x264 codec; shrinking my file to around 5gb...

I am really not an expert and can't notice any "big" difference on the video file... so before continue my project and capture all 60 tapes I would like to knwo if i am doing it right... or almost right...
I do want my videos no be easily accessible in the future and to watch them on my lcd screen, I don't expect to be HD or better that the source... just equal.

Will be uploading snapshot of my configuration so you can check if it's OK.

Thanks and any comments will be appreciated.

First a TBC will help alot.

The ATI-650 has a nice 3D combfilter* and built in hardware MPeg2 codec. Try capturing to high bit rate MPeg2 using ATI program or Graphedit, then deinterlace in Vdub and encode. Then compare results. I bet they are similar and will save you all that disk space.

*This works even better for composite TV captures.

[jagabo]

The ATI-650 has problems with its automatic gain function -- you'll find the brightness and saturation varying quite a lot. You need to use a "video clarifier" to fix that.

https://forum.videohelp.com/threads/326560-Which-is-better-usb-stick-vhs-cap-or-hd-pvr-...=1#post2023227

A full frame TBC should work too.

Your processing is ok if you need a progressive result. If any of your tapes are telecined film you should inverse telecine them back to 23.976 fps progressive frames. That will be faster and give better results than TGMC. Also consider encoding as interlaced MPEG 2 for DVD.

[edDV]

If you don't use a TBC, pixels on every other line move relative to each other. That reduces encoder efficiency.

[usually_quiet]

Hi.

Thanks to you all I managed to capture using my ATi TV Wonder 650 USB without sync problems and get the most of it.

My system is:
Core i5 750
4gb
Ati 4870 1gb
2tb Hdd
and the ATi 650 USB

I am capturing with Windows XP, Virtualdub and Lagarith lossless codec YUY2 at 720x480.

With these configuration I am getting a 50+gb file for 2hrs. video, which it's OK if I have 1 or 2 but tapes, I do have around 60 tapes and thou hdd are cheap, I would need like 3 more disks...

So what I did is... first of all "deinterlace" the source using "TempGauss Deinterlacer" (which doubles the framerate = cool) , apply unsharp mask and after that I that I use x264 codec; shrinking my file to around 5gb...

I am really not an expert and can't notice any "big" difference on the video file... so before continue my project and capture all 60 tapes I would like to knwo if i am doing it right... or almost right...
I do want my videos no be easily accessible in the future and to watch them on my lcd screen, I don't expect to be HD or better that the source... just equal.

Will be uploading snapshot of my configuration so you can check if it's OK.

Thanks and any comments will be appreciated.

First a TBC will help alot.

The ATI-650 has a nice 3D combfilter* and built in hardware MPeg2 codec. Try capturing to high bit rate MPeg2 using ATI program or Graphedit, then deinterlace in Vdub and encode. Then compare results. I bet they are similar and will save you all that disk space.

*This works even better for composite TV captures.

If you want to try edDV's suggestion of using a filter graph for capture to utilize your card's hardware MPEG-2 encoder, this is a screen capture of a filter graph I made some time ago using GraphStudio for the TV Wonder 650 PCI-e. ATI's filters can be found under Graph->Insert Filter in the WDM Streaming Capture section in the drop-down filters list. I think the Dump filter used is Moonlight DumpPOS which I think was installed along with with DVRMSToolbox. Just drag and drop the pins as shown and right-click on the filters to configure their settings

I hope this proves useful. Unfortunately, my TV Wonder 650 PCI-e died a few weeks ago, and I have uninstalled the card and its drivers, so I can't load the real graph I saved to check anything for you.

[jagabo]

Don't capture as MPEG 2 if you plan on heavy filtering afterwards. MPEG 2 encoding will harm the video quality, reducing the amount of detail and adding blocky artifacts.

[Marto2008]

Thank you all very much for the info...

Regarding MPEG2 compression... time is not really a problem so I will test MPEG2 vs h264 and I will compare...

By high bitrate on MPEG2 what do you mean? Like 10mbps?

Thanks for the Graphedit graph...

Also a TBC is not an option as I don't want to spend too much money... also these are "1st gen" videos so it might not be too much time base problems, right?

Thanks and I will report results.

[jagabo]

All VHS has time base jitter. The helical scan heads cannot maintain a constant spin rate so the length of scanlines varies causing horizontal jitter. All that irregular horizontal movement is a killer of MPEG compression.

https://forum.videohelp.com/threads/306272-Computer-video-capture-vs-vcr-to-dvd-combo?p...=1#post1882662
https://forum.videohelp.com/threads/319420-Who-uses-a-DVD-recorder-as-a-line-TBC-and-wh...=1#post1983288

[Brad]

No one is going to comment on the Unsharp mask step? I would skip that, and if you find that you want extra "sharpness" just turn up the control on your TV or video player. Instead of baking in something you might later regret.

[edDV]

The main quality add will be the TBC. Timebase jitter line to line causes still areas of the frame to appear to be in motion. This combined with VHS noise which also is detected as false motion means Mpeg2 interframe compression will be minimally applied forcing most compression to intraframe. This in turn causes blockyness unless high bit rates are used.

In other words, noisy-jittery video causes MPeg (MPeg2 or h.264) to fail unless very high bit rates are used.

Next issue for MPeg performance is deinterlace or inverse telecine (for film rate source) if high compression is desired. Deinterlace to 59.94p will preserve full motion. Deinterlace to 29.97p removes half the motion samples and can introduce motion artifacts. Alternative is to leave the video interlace and let the player deal with deinterlace or inverse telecine. Player processors continue to advance whereas software deinterlace-IVTC during encode is "baked in" to your encode with no improvement possible.

[jagabo]

There's talk of a software TBC that seems to have a little success:
https://forum.videohelp.com/threads/340426-DeJitter-help

[Marto2008]

The main quality add will be the TBC. Timebase jitter line to line causes still areas of the frame to appear to be in motion. This combined with VHS noise which also is detected as false motion means Mpeg2 interframe compression will be minimally applied forcing most compression to intraframe. This in turn causes blockyness unless high bit rates are used.

In other words, noisy-jittery video causes MPeg (MPeg2 or h.264) to fail unless very high bit rates are used.

Next issue for MPeg performance is deinterlace or inverse telecine (for film rate source) if high compression is desired. Deinterlace to 59.94p will preserve full motion. Deinterlace to 29.97p removes half the motion samples and can introduce motion artifacts. Alternative is to leave the video interlace and let the player deal with deinterlace or inverse telecine. Player processors continue to advance whereas software deinterlace-IVTC during encode is "baked in" to your encode with no improvement possible.

But I do need compression, as I can't manage 50+gb files...

I have read H264 is not good for interlaced videos, that's the only reason I am deinterlacing which results in a 59.94fps video.

Which compression is good for interlaced video? Any as good as H264? So I can keep it interlaced but with a good compression, thus taking advantage of future deinterlacing methods/hardware

Also,

I know the TBC is THE way to go... and like some expert users say in this forum, I will never lose money with it as I buy one and after use I can sell it for the same price or with a marginal loss... but the problem is that where I live there are no TBC to buy and I don't want to spend money...

[Marto2008]

No one is going to comment on the Unsharp mask step? I would skip that, and if you find that you want extra "sharpness" just turn up the control on your TV or video player. Instead of baking in something you might later regret.

I think I will follow your advices... as nowadays, software video players take advantage of vertex shaders and can apply very nice sharp filters...

[usually_quiet]

I don't remember whether ATI's software encoder offered a higher bit rate than the hardware encoder or not. I think the software encoder filter came with Calalyst Media Center. Here's another capture graph employing the software encoder if you want to take look. Encoding closed captions is optional.

[Marto2008]

I have been trying to capture in MPEG2 using Graphedit but it seems there is a problem as I always get an error...

Anyway...

I am capturing / editing using lagarith, afterwards I use "AMD Video Converter" (ATi Video Card utility that use GPU for encoding similar to nVidia CUDA) to compress it to MPEG2 at 8.28mbps. The RATE is high enough?

I am getting good quality, maybe not as good as Tempgauss Deinterlaced at 59.90fps and then compressed to x264... but it's close enough... in much lesser times.

Will I get any benefits If I encode it 720P or 1080P MPEG2? Remember my source is 720x480.

I am not deinterlacing using Tempgauss and I'm not using x264 because I have an interlaced source.

Best Regards.

[jagabo]

I am capturing / editing using lagarith, afterwards I use "AMD Video Converter" (ATi Video Card utility that use GPU for encoding similar to nVidia CUDA) to compress it to MPEG2 at 8.28mbps. The RATE is high enough?

Yes. I haven't really looked at ATI's MPEG 2 encoder, but it's h.264 encoding was very poor. You might do better with HcEnc or some other encoder.

I am getting good quality, maybe not as good as Tempgauss Deinterlaced at 59.90fps and then compressed to x264... but it's close enough... in much lesser times.

You can encode interlaced in x264. And if you use the veryfast preset it may be faster (and still better quality) than ATI's hardware encoder.

Will I get any benefits If I encode it 720P or 1080P MPEG2? Remember my source is 720x480.

Generally, no. In some cases, especially anime and cartoons, you can upscale better in software than your TV can. But you have to get into some advanced AviSynth filtering.

[Marto2008]

jagabo, thanks for your answer...

I don't even know how to start with HcEnc... input files are d2v and avs... Do I need an Avisynth script?

UPDATE: created a standard script and I am already encoding... but there are sooo many options...
Any special one that I need to take care?

The Ati MPEG2 encoder seems to be "low" quality but it's very fast... taking approx 30min for compressing a 60GB+ file at 8.68mbps

I thought x264 was not recommended for INTERLACED sources... was I wrong?

Best Regards.

[jagabo]

Have you tried HcGUI? x264 isn't as well optimized for interlaced encoding but it works. Make sure your intended playback device/software supports interlaced h.264 playback.

[Marto2008]

Have you tried HcGUI? x264 isn't as well optimized for interlaced encoding but it works. Make sure your intended playback device/software supports interlaced h.264 playback.

Yes, I am using HCGui right now... testing it with the best quality at 10mbps to see how it compares to ATI...

Quality is INDEED much better than ATI encoder... but I HC Gui takes much longer and:

1-. What happends with the sound?
2-. My final video ends with a lot of "information" over it.

I will keep looking to find the correct options.

Best Regards.

[jagabo]

1-. What happends with the sound?

HcEnc only handles the video. You have to use something else for the audio.

2-. My final video ends with a lot of "information" over it.

Does your AVS script include info()?

[Marto2008]

1-. What happends with the sound?

HcEnc only handles the video. You have to use something else for the audio.

2-. My final video ends with a lot of "information" over it.

Does your AVS script include info()?

Yes... my bad...

This is my script / Already removed "Info ()":
AVISource("LISTO nº 27 año 97 - 98.avi") #open the source video
#ConvertToYV12(Matrix="PC.601")
ConvertToYV12(Interlaced=True)

Regarding audio... I have to mux it with another encoder?

Also, is there any GOOD MPEG2 encoder that do video and audio together?

Bye and thanks for your time

UPDATE: I will update some pictures, so you can appreciate if HC025 encoder is much better to the ATI encoder... Another opinion would help me decide.

[Marto2008]

Here are the captures...

What do you think? (THE ONES WITH "NO AUDIO" ICON ARE HC025, the other ATI encoder...

[manono]

Regarding audio... I have to mux it with another encoder?

No, not an encoder. You bring in the audio when authoring the DVD. Good MPEG-2 encoders do just that - encode for MPEG-2 video. Why expect a video encoder to handle audio? Sure, you can use an all-in-one program, a good one such as Avs2DVD which also uses HCEnc, for example. But the audio encoding will be handled by a different and audio-specific program.

[Gavino]

is there any GOOD MPEG2 encoder that do video and audio together?

QuEnc works off an Avisynth script like HCEnc, but also encodes the audio.
(However, HCEnc is generally said to be a better MPEG encoder.)

Also, the SoundOut plugin can be used directly in your Avisynth script and will encode to a number of formats, including AC3. So you can use the script with HCEnc and get the audio from SoundOut at the same time.

[Marto2008]

Thanks for your replies... I will try those programs.

But, do you clearly see a difference in the pictures? Is there any way to compare?

Best Regard

[jagabo]

But, do you clearly see a difference in the pictures? Is there any way to compare?

It's very hard to judge frames captured with a screen cap like that. The frame has been resized by the player to match the display aspect ratio. You can't tell which artifacts are in the original video and which were caused by the resizing. I recommend you use a program like VirtualDubmod to export as PNG (Video -> Snapshot Source Frame... save as PNG). That will produce a pixel-for-pixel image from your source.

It's also important you export exactly the same frame. And, if you can, the same frame type (I, P, B).

One easy way to compare images is to use Windows Photo Viewer (right click, select Preview). You can then switch back and forth with the left and right arrow keys. It's harder with Videos. I usually use an AviSynth script that interleaves frames from the two videos. You can then open the script in an editor like VirtualDub and use the arrow keys to switch back and forth between frames of the two videos. And I use a screen magnifier to zoom in about 4x.

v1=WhateverSource("video1.ext")
v2=WhateverSource("video1.ext")
Interleave(v1,v2)

You'll have to learn a little AviSynth to use it.

[Marto2008]

OK, here they are... what do you think?

Is it worth it to go with HCENC or I keep using ATi encoder which is the fastest for me?

[jagabo]

I see several differences:

1) The HcEnc encodes are sharper. But since I don't know what the videos looked like before encoding I don't know if the ATI encoder is blurring the image or HcEnc is sharpening. I think it's the former. But VHS caps are often over sharpened by the VHS player (because VHS has inherently low resolution) so a little softening may be called for.

2) There's a color problem with on of the encodings. Or maybe just the way you produced the screen caps. Notice how the red car and the red door are different shades of red. This is probably because one of them is using rec.709 color, the other rec.601. Without seeing the video before compression, and both compressed videos, I can't tell you which is correct. But standard definition video should be coming out of the ATI 650 with rec.601 color. And the MPEG 2 file should also be rec.601. DgIndex will tell you if the MPEG video is flagged as rec.601 or rec.709.

3) With so much macroblocking it's hard to say which video has more. Especially since one is producing less sharp video. The frame type (I, B, P) may also differ (B frames are encoded at lower quality settings and will show more macroblocking).

See if you can sort out the color problem, then go with whichever you prefer. Be sure to watch them on TV at normal playback speeds as well as looking at still frames.

[Marto2008]

Thanks for your answer.

Here I upload one image of the red car which is lossless lagarith. I also uploaded other two images in lossless lagarith codec.

The ATi encoder seems to be the correct red...
Also that the HCenc is sharpening.

I'am trying AVStoDVD which seems to be in the middle of both encoders. Good quality and easy of use... Taking longer times thou.

[jagabo]

The ATi encoder seems to be the correct red...

Note that you can specify the colorimetry in HcEnc (on the Settings 2 tab of HcGUI). Use BT.470-2M or BG.

Also that the HCenc is sharpening.

That's odd because I've never really noticed it in my own encodings (I don't do much MPEG 2 encoding anymore though, and when I was doing lots of it I was using CCE). Check which Matrix you used.