I-o DATA GV-USB2 + Panasonic DMR-ES15 first test capture

[boolian2]

Also using JVC HR-S3800U S-VHS VCR with S-video. Is there any way to get rid of the artifacts along the bottom other than cropping it out? I assumed the TBC would take care of that.

[davexnet]

I haven't looked at this latest video, but the interference at the bottom is usually the head switching noise,
Not much you can do except crop/add a clean border back to keep the same frame size. Easy to do in Avisynth

[Alwyn]

I'm confused. It doesn't appear to be interlaced (I can't get any back and forth motion with TFF or BFF) but there are obvious interlacing artefacts. Are they baked-in?

[Sharc]

It's standard 3-2 telecined material and one has to IVTC it to get the original progressive 23.976fps frames back. See the other recent threads of the same poster.

assumeTFF()
TFM().TDecimate()

[Alwyn]

Thanks Sharc, I'll keep away in PAL land.

[Sharc]

In PAL land you may occasionally find the 2:2:2:2:2:2:2:2:2:2:2:3 "Euro-Pulldown" to convert 24fps film to 25fps instead of "speedup"

https://forum.doom9.org/showthread.php?t=151372
https://en.wikipedia.org/wiki/Telecine

[davexnet]

The video has some strange ghosting, here as the guy moves to the right, you can see 3 or 4 shadows of where he was in previous frames.
Could this be on the tape?

[Sharc]

Don't know, maybe.
I would suggest lowering the brightness somewaht to make the luma better fit the 16 .... 235 range.
The black level seems to be elevated.

Edit:
The level correction can also be done in post processing instead of fiddling with the "ProcAmp", as the signal is not clipped.
These color and levels adjustments are to a certain extent a matter of personal preference though.

[boolian2]

Here is the inverse telecined version.

[Alwyn]

It's coming over as 3:2, but it should be 4:3, IMO.

[boolian2]

Here is another capture with this setup. I inverse telecined it with avisynth/vdub then handbraked it to reduce file size. Notice the stutter when he moves the eyepatch. What is causing that? I captured the whole tape, and I don't see it anywhere else. Also is there any way to let avisynth/vdub use more CPU? It's only using like 2% CPU, my CPU is 16 cores. It's taking like an hour when it should be taking a couple of minutes.

[sambat]

@boolian2
FWIW re head noise on vhs capture.
https://forum.videohelp.com/threads/343940-Resolved-Use-Subtitle-to-Create-Border-on-D...ut-Re-encoding

[boolian2]

It's coming over as 3:2, but it should be 4:3, IMO.

Should I capture in 640x480 or 720x480? Or should I capture in 720 and then somehow convert to 640?

[lollo]

The video has some strange ghosting, here as the guy moves to the right, you can see 3 or 4 shadows of where he was in previous frames.
Could this be on the tape?

Maybe. It could also be a denoise of the chroma across a temporal radius (cannot check now), even if this appears more on scene change.
OP must be sure that all Noise Reduction is turned off in Panasonic ES-15.

The level correction can also be done in post processing instead of fiddling with the "ProcAmp", as the signal is not clipped.

I agree with Sharc. The level correction with the card procamp is "rudimentary", and should be done only if the input levels are outside the range that can be captured by the card to avoid crushing the blacks and clipping the whites. IIRC GV-USB capture in range 16-250, then when the input levels are inside this limits there is no need to act.

Later in postprocessing the level correction is more accurate, but is necessary only if a filtering in RGB colorspace is needed, to compress the YUV range to 16-235 for proper RGB conversion. If you stay in YUV colorspace even this is not needed. And be careful that many of the AviSynth filters expands internally the level at 0-255 to be more efficient, so ideally the check of the levels should be done after each step.

Always check the levels at the end of the postprocessing: according to your display options you may need to stay inside 16-235 range, or you may want to change the look of your video to your personal taste.

[DB83]

It's coming over as 3:2, but it should be 4:3, IMO.

Should I capture in 640x480 or 720x480? Or should I capture in 720 and then somehow convert to 640?

There is a whole topic on here about capturing @ 720x480. 640x480 will not be a native capture but a correctly resized output format. Which you can do when you create a delivery format from your archive one (a full UT capture will be HUGE)


Not truly relevant here but I trust you have many tapes that dvds are not available - the one you sample is so well known and has many dvd releases which will now give you the full aspect ratio and not a scanned 4:3 one and at much better visual quality. You have invested quite heavilly in this kit and the same money could have acquired the dvds.

[boolian2]

It's coming over as 3:2, but it should be 4:3, IMO.

Should I capture in 640x480 or 720x480? Or should I capture in 720 and then somehow convert to 640?

There is a whole topic on here about capturing @ 720x480. 640x480 will not be a native capture but a correctly resized output format. Which you can do when you create a delivery format from your archive one (a full UT capture will be HUGE)


Not truly relevant here but I trust you have many tapes that dvds are not available - the one you sample is so well known and has many dvd releases which will now give you the full aspect ratio and not a scanned 4:3 one and at much better visual quality. You have invested quite heavilly in this kit and the same money could have acquired the dvds.

Yeah I can get the 4K of a lot of tapes but that's not as fun. I have a big collection of VHS from my childhood. And they're in surprisingly good condition since they've been stored in a hot attic for years.

By the way yes the initial captures are huge. They are from 40-80GB. Then when you de-telecine them with avisynth they are doubled or quadrupled in size for some reason. But I have big hard drives so it's not that big a deal.

[Sharc]

Should I capture in 640x480 or 720x480? Or should I capture in 720 and then somehow convert to 640?

Always capture the tapes as 720x480.
Once you have the progressive frames (e.g. after IVTC or after deinterlacing), you have the basic options:
a) Resize to square pixels like 640x480, or 1440x1080, and encode.
b) Leave as 720x480 (anamorphic), and encode with the Pixel Aspect Ratio (also called Sample Aspect Ratio) of 8:9, and/or the DAR (display aspect ratio) of 4:3. How to do this in detail depends on the encoding software and its GUI.
c) Leave as 720x480, encode it and simply force the player (software or TV) to play it as 4:3.

[DB83]

The ONLY thing that determines file-size in your scenario is bitrate. Filesize = runlength x bitrate

You can confirm what the IVTC is doing by comparing the original to the new via mediainfo. Is the frame rate doubled ? Do you now have true progressive video as against interlaced ? And is the bitrate substantially higher for these ?


And since you have the bitrate then a resize from 720*480 to 1280*960 (or even 1440*1080) at a substantially lower bitrate than you have right now is a viable proposition.


I can concur with your reasoning about even capturing tapes you can get on much better quality now. But equally, I saw many a film butchered due to the constraints of display and I would always seek the true display over the underlying sentiment.

[lollo]

You can confirm what the IVTC is doing by comparing the original to the new via mediainfo. Is the frame rate doubled ? Do you now have true progressive video as against interlaced ? And is the bitrate substantially higher for these ?

We are in NTSC world here: IVTC of the 3:2 pulldown will not double the frame rate, but rather change it from 29.970fps to 23.976fps. (duration in time is the same before and after, number of frames is changed, audio is then in synch)

edit

And since you have the bitrate then a resize from 720*480 to 1280*960 (or even 1440*1080) at a substantially lower bitrate than you have right now is a viable proposition.

Not sure I understand this. Higher resolution requires higher bitrate. But probably you meant that going from 'lossless 720x480' to 'x264 1440x1080' the bitrate can be substantially decreased, which is true.

[Alwyn]

a) Resize to square pixels like 640x480, or 1440x1080, and encode.

@Sharc, is this a satisfactory way of doing this in VDub2?

-Activate Resize filter as follows:

[Attachment 65594 - Click to enlarge]
-Export using the x264 8bit codec from the main video codec list (Configure>Quality factor 18 or a bit less), Audio in Full Processing mode, using AAC.
-Rename video to MP4 (not AVI, which is the default).

I regularly change the aspect ratio from 5:4 to 4:3 in my NLE (720x576 PAL VHS captures), but I have never been able to work out how to use VDub2's Resize filter to change a 5:4 video to a 4:3 export without just disabling everything and exporting at 768x576.

[Sharc]

Looks right to me, yes, if you want to resize the NTSC 720x480 capture to square pixels. In this case make sure that when configuring the encoder, set the SAR width=1 and the SAR height=1.
And you can't just rename the .avi to .mp4. When you export it via File->Save video ..... select the file type in the dropdown list as MP4 (MPEG-4 Part 14) (*.mp4). Then it gets correctly muxed into the .mp4 container.

You could also encode it anamorph (e.g. for DVD). Then you don't need to resize the video but set in the encoder configuration the SAR width=8 and SAR height =9 for 4:3 NTSC. Or crop 8 pixels from left and 8 pixels from right so that the frame becomes 704x480, and set the SAR width =10 and the SAR height=11 in the encoder configuration.

(For PAL captures you would have to adapt the values accordingly).

Edit:
Attached here the 3 variants:
- square pixel 640x480
- anamorphic 720x480 (SAR 8:9)
- anamorphic cropped 704x480 (SAR 10:11)

[Eric-jan]

capture @ 800x600

[Cornucopia]

capture @ 800x600

No.

[boolian2]

The ONLY thing that determines file-size in your scenario is bitrate. Filesize = runlength x bitrate

You can confirm what the IVTC is doing by comparing the original to the new via mediainfo. Is the frame rate doubled ? Do you now have true progressive video as against interlaced ? And is the bitrate substantially higher for these ?


And since you have the bitrate then a resize from 720*480 to 1280*960 (or even 1440*1080) at a substantially lower bitrate than you have right now is a viable proposition.


I can concur with your reasoning about even capturing tapes you can get on much better quality now. But equally, I saw many a film butchered due to the constraints of display and I would always seek the true display over the underlying sentiment.

Here is mediainfo on a capture that I de-telecined with avisynth and vdub. I "saved as avi" in vdub. The bit rate is 199Mbps. The file size is 169GB - the original AmarecTV capture is 41GB.

assumeTFF()
TFM().TDecimate()

General
Complete name                            : Z:\A03EC7193EC6E7\Goonies.avi
Format                                   : AVI
Format/Info                              : Audio Video Interleave
Format profile                           : OpenDML
File size                                : 161 GiB
Duration                                 : 1 h 55 min
Overall bit rate                         : 200 Mb/s
Writing library                          : VirtualDub build 35491/release

Video
ID                                       : 0
Format                                   : RGB
Codec ID                                 : 0x00000000
Codec ID/Info                            : Basic Windows bitmap format. 1, 4 and 8 bpp versions are palettised. 16, 24 and 32bpp contain raw RGB samples
Duration                                 : 1 h 55 min
Bit rate                                 : 199 Mb/s
Width                                    : 720 pixels
Height                                   : 480 pixels
Display aspect ratio                     : 3:2
Frame rate                               : 23.976 (24000/1001) FPS
Color space                              : RGB
Bit depth                                : 8 bits
Bits/(Pixel*Frame)                       : 24.000
Stream size                              : 160 GiB (99%)

Audio
ID                                       : 1
Format                                   : PCM
Format settings                          : Little / Signed
Codec ID                                 : 1
Duration                                 : 1 h 55 min
Bit rate mode                            : Constant
Bit rate                                 : 1 536 kb/s
Channel(s)                               : 2 channels
Sampling rate                            : 48.0 kHz
Bit depth                                : 16 bits
Stream size                              : 1.24 GiB (1%)
Alignment                                : Aligned on interleaves
Interleave, duration                     : 42  ms (1.00 video frame)
Interleave, preload duration             : 500  ms

[davexnet]

It's because you didn't select a compression codec, vdub defaults to uncompressed RGB.
If you want to save a lossless intermediate or encode to h264 right in vdub, select video (at the top)/compression
and select the codec you want

[boolian2]

a) Resize to square pixels like 640x480, or 1440x1080, and encode.

@Sharc, is this a satisfactory way of doing this in VDub2?

-Activate Resize filter as follows:

[Attachment 65594 - Click to enlarge]
-Export using the x264 8bit codec from the main video codec list (Configure>Quality factor 18 or a bit less), Audio in Full Processing mode, using AAC.
-Rename video to MP4 (not AVI, which is the default).

I regularly change the aspect ratio from 5:4 to 4:3 in my NLE (720x576 PAL VHS captures), but I have never been able to work out how to use VDub2's Resize filter to change a 5:4 video to a 4:3 export without just disabling everything and exporting at 768x576.

Is it better to resize in vdub using a filter or resize using avisynth+vdub? All I see under File>Export is "Raw video..."

[davexnet]

Not File/export,
File/save video ( use the newer vdub2)

[boolian2]

Not File/export,
File/save video ( use the newer vdub2)

Ohh ok thanks a lot

[boolian2]

Not File/export,
File/save video ( use the newer vdub2)

Can I make vdub2 use more than 3% CPU? Right now that's all it's using, it's only rendering ~70FPS

Also I'm currently resizing and cropping the head noise in avisynth, here is my script.

assumeTFF()
TFM().Tdecimate()
BilinearResize(640, 480, 0, 0, 720, 474)

It resizes the 720x480 to 640x480 and crops the bottom 6 rows of pixels off.

[dellsam34]

capture @ 800x600

That's not capturing, it's called resizing on the fly, Capture cards' ADC captures at 720x480 or 720x576, You can only change the output, but for what reason?