Is it worth it recapturing videos?

[Sharc]

@Alwyn:
So you know: if one captures into the <16 and >235 range and watches it on the PC one should use the "PC.601 Matrix" to see the full range and the gradients on the PC monitor. Nothing is lost - except out of gamut RGB (when viewing on the PC monitor) which can nevertheless happen in either case but are usually less for the "safe" (16....235) approach.

[Attachment 88367 - Click to enlarge]
But independent from the current case I agree that the brights are slightly too high for the I-O Data GV-USB2. Using reference patterns I have to step the proc-amp contrast about 5 steps back from default (105->100) to capture reference white at Y=235 and reference black at Y=16 if I want to capture "spot on". At least for my setup.

Or if you eventually like it more blue or whatever you can use any editor/NLE (including DaVinci) and tweak it to your taste, or use a mask to select specific areas you want to change etc.

[Attachment 88368 - Click to enlarge]

Main point: Do not clip the signals, and decode/watch it correctly.

[Riker0007]

Thank you. I've got the script for ffmpeg to change to prores as suggested, but I'm having issues with setting up the QTGMC de-interlace. I thought I put the correct files in the right locations, but I keep getting an error. I wish there was a simple way to just copy the needed files to where they need to be. Ugh!

test.avs
AVISource("signing.avi")
ConvertToYV16(interlaced=true)
QTGMC(Preset="Slower")
ConvertToYV12()

run:ffmpeg -i "test.avs" -c:v prores_ks -profile:v 2 -pix_fmt yuv422p10le -c:a pcm_s16le "test.mov"

I get this:
[avisynth @ 00000129bcc29340] Script error: There is no function named 'mt_makediff'.
(C:/Program Files (x86)/AviSynth+/plugins64+/QTGMC-3.32.avsi, line 776)
(C:/Program Files (x86)/AviSynth+/plugins64+/QTGMC-3.32.avsi, line 386)
(test.avs, line 3)
[in#0 @ 00000129bcc0dac0] Error opening input: Unknown error occurred
Error opening input file test.avs.
Error opening input files: Unknown error occurred

[Alwyn]

Just open that AVS in VDub2 and export as ProRes.

While you're there, apply the CCD filter.

[Sharc]

Thank you. I've got the script for ffmpeg to change to prores as suggested, but I'm having issues with setting up the QTGMC de-interlace. I thought I put the correct files in the right locations, but I keep getting an error. I wish there was a simple way to just copy the needed files to where they need to be. Ugh!

Why are you using QTGMC v3.32? Try the updated version and the dependencies (plugins) recommended here under 'Core Plugins and Scripts':
http://avisynth.nl/index.php/QTGMC

Is the masktools2.dll plugin installed and updated?
http://avisynth.nl/index.php/MaskTools2#Filters


Also, for a change try a faster setting of QTGMC, like QTGMC(preset="fast") to begin with.

If you don't succeed you could eventually also get away with a simpler deinterlacer ....... (BWDIF, yadifmod2)

[Sharc]

While you're there, apply the CCD filter.

AFAIK this filter is one of those requiring a YUV->RGB conversion, kicking out (clipping) any Y>235 luma data when using the default Rec.601 matrix for conversion.
See 'BUT1' here
https://forum.videohelp.com/threads/418254-Info-Hauppauge-USB-Live-2-capturable-range#post2780708

[Alwyn]

I've had a blank: how does one find line 776 of QTGMC.AVSI?

[Sharc]

I've had a blank: how does one find line 776 of QTGMC.AVSI?

Using notepad for example. Edit->GoTo..... It indicates the line and column numbers of the cursor position in the status bar.
The latest QTGMC has 1342 lines.

[Alwyn]

Using notepad for example.

What about the wrapping? There's 4 lines that wrap on a 2560 pixel screen.

Edit: Zoomed out so the text is almost unreadable and got no wraps.

Edit 2: aha! The line count doesn't count line wraps.

[rgr]

run:ffmpeg -i "test.avs" -c:v prores_ks -profile:v 2 -pix_fmt yuv422p10le -c:a pcm_s16le "test.mov"

I suggest converting to BT.709 in the AviSynth script right away, because ProRes files are often treated as Rec709.

[lollo]

A question to OP: which driver versions are you using for the IOData GV-USB? (I am doing some histogram experiments across 112-115 driver releases)

[Riker0007]

Thank you. I've got the script for ffmpeg to change to prores as suggested, but I'm having issues with setting up the QTGMC de-interlace. I thought I put the correct files in the right locations, but I keep getting an error. I wish there was a simple way to just copy the needed files to where they need to be. Ugh!

Why are you using QTGMC v3.32? Try the updated version and the dependencies (plugins) recommended here under 'Core Plugins and Scripts':
http://avisynth.nl/index.php/QTGMC

Is the masktools2.dll plugin installed and updated?
http://avisynth.nl/index.php/MaskTools2#Filters


Also, for a change try a faster setting of QTGMC, like QTGMC(preset="fast") to begin with.

If you don't succeed you could eventually also get away with a simpler deinterlacer ....... (BWDIF, yadifmod2)

Thanks.
Using what you said here. worked. Maybe the links I was using were old code??
That being said, the script I've used is to convert my videos to prores :
@echo off
setlocal enabledelayedexpansion

REM Output folder name
set outdir=Converted

REM Make sure the output folder exists
if not exist "%outdir%" mkdir "%outdir%"

for %%a in ("*.avi") do (
echo Converting: %%a
ffmpeg -i "%%a" -c:v prores_ks -profile:v 2 -bits_per_mb 8000 -pix_fmt yuv422p10le -vf "scale=interl=1,setfield=bff" -c:a pcm_s16le "%outdir%\%%~na.mov"
)

echo Done!
pause


It converted my videos it seems to prores, but using this:

AviSource("your_capture.avi") # load your VHS capture
AssumeBFF() # VHS is usually Bottom Field First
QTGMC(preset="slow") # apply QTGMC deinterlacing


The video did not look right. Not sure where to go from here?
Thanks

https://files.videohelp.com/u/139455/horse%20jumping.mov

Is there a way to take the converted files to prores from above and just use the QTMC? Or what should I do? Scrap it all and start recapture and use???
Thanks again!

[Alwyn]

AssumeBFF() # VHS is usually Bottom Field First

No it's not. From the GV-USB2 and most other USB digitisers, it's TFF.

It will be obvious when you step thru your processed video: with BFF, you'll get a backward forward motion eg the horse. With TFF, it will always be forward. You need a frame stepping program to observe the frame by frame motion eg VDub. VLC will just show stuttery motion.

[Sharc]

AssumeBFF() # VHS is usually Bottom Field First

No it's not. From the GV-USB2 and most other USB digitisers, it's TFF.

It will be obvious when you step thru your processed video: with BFF, you'll get a backward forward motion eg the horse. With TFF, it will always be forward. You need a frame stepping program to observe the frame by frame motion eg VDub. VLC will just show stuttery motion.

His former uploads were all BFF. That's why I suggested assumeBFF() in my script.
(I'll take a look later, have to leave now ......)

[jagabo]

VHS (all interlaced analog video, really) is neither top field first nor bottom field first. It's simply a stream of alternating top and bottom fields. It's the capture device (or firmware/driver/software) that weaves pairs of fields together to produces TFF or BFF frames. Ie, it can start with a bottom field then weave in the next field (a top field) to produce a bottom field first frame. Or it can start with a top field then weave in the next field (a bottom field) producing a top field first frame.

[Riker0007]

How do I fix it? The original captures had to be deleted to make room for the converted ones.

[Alwyn]

Just use AssumeTFF()

[Riker0007]

AssumeBFF() # VHS is usually Bottom Field First

No it's not. From the GV-USB2 and most other USB digitisers, it's TFF.

It will be obvious when you step thru your processed video: with BFF, you'll get a backward forward motion eg the horse. With TFF, it will always be forward. You need a frame stepping program to observe the frame by frame motion eg VDub. VLC will just show stuttery motion.

His former uploads were all BFF. That's why I suggested assumeBFF() in my script.
(I'll take a look later, have to leave now ......)

Thank you.

[Riker0007]

I was able to recover the original captures and converted them all to prores? BFF OR TFF, I have no idea. If I use the same settings and hardware/software, it should remain either TFF or BFF right?
Here is another capture. I tried in virtdub trying TFF filter and then BFF filter and I really could not tell the difference going through frame by frame.
Thanks

https://files.videohelp.com/u/139455/horse%20tape2.mov

[Alwyn]

That one's TFF, a normal capture (after it's been ProRes'd). That's good.

Except that your code is using QTGMC, which should be deinterlacing your AVI; it currently isn't. It's still interlaced.

Your actual QTGMC code part (I changed BFF to TFF and added the red bit) looks OK:

AviSource("your_capture.avi") # load your VHS capture
ConvertToYV12(interlaced=true) # not sure if this is needed
AssumeTFF() # VHS is usually Top Field First
QTGMC(preset="slow") # apply QTGMC deinterlacing

This will result in a Progressive file with the frame rate doubled to 59.94.

One of the others will have to help with your other coding and specifically why you're still getting a interlaced ProRes file.

Re analysing a file, try this (the experts will criticise this analysis technique but it works well for what we're doing here):

Set your VDub deinterlace filter up like this:


[Attachment 88384 - Click to enlarge]

That setup is called double-rate deinterlacing because each field has been turned into a frame, hence your 29.97 capture changes to 59.94 (same with QTGMC on normal settings).

OK out of that filter setup and then you'll see the motion. If it's on BFF, the horse will be going backwards and forwards. It's quite obvious.

You'll also see the jaggies in the left pane, which indicate the file is interlaced, have been cleared in the right pane.

If your file is Progressive, when you step through the video, the right pane will step only on every second frame like this: 1122334455667788 because VDub is double-rate "deinterlacing" a file that is already deinterlaced. And, of course, there will be no jaggies evident in the left pane in areas of motion.

[Alwyn]

Some other issues:

-The frame rate of the Horse tape2.mov is a bit wonky, according to MediaInfo. It should be spot on 29.97 (or 59.94 if the deinterlacing had worked). It's showing:

Frame rate mode : Variable
Frame rate : 29.494 FPS
Minimum frame rate : 14.985 FPS
Maximum frame rate : 29.970 FPS

The raw capture won't be like that, nor would QTGMC do that, so it must have been introduced with that code chain that converts it to ProRes/MOV.

-The ProRes conversion has encoded the display aspect ratio incorrectly at 3:2. It should be 4:3.

-There's a glitch near frame 137, where the deinterlaced version does a field reversal for a couple of frames. It looks like a glitch on the tape and is hardly noticeable. A stabiliser of some type might help. Did you get that ES-15?

[Sharc]

How do I fix it? The original captures had to be deleted to make room for the converted ones.

LWLibavVideoSource("horse jumping.mov")
selectevery(2,1,0)
convertbits(8).converttoYV16() #added for subsequent x264 encoding only

(But better do it right from beginning than fixing after)

[Sharc]

Or using your 'horse tape2.mov' of post#78:

v=LWLibavVideoSource("horse tape2.mov")
a=LWLibavAudioSource("horse tape2.mov")
Audiodub(v,a)
AssumeTFF()
QTGMC() #or use a deinterlacer of your choice

Now save this as script.avs and encode the script with the encoder of your choice, for example with x264 or with ProRes for further editing in DaVinci.
Commandline for ProRes encoding:

ffmpeg.exe -i "script.avs" -c:a copy -c:v prores_ks -profile:v 3 -vendor apl0 -bits_per_mb 8000 -pix_fmt yuv422p10le -vf setsar=10/11 "proRes422__.mov"

(As mentioned by Alwyn there are 2 glitches in the source, around frames 10 and 274)

[Riker0007]

Some other issues:

-The frame rate of the Horse tape2.mov is a bit wonky, according to MediaInfo. It should be spot on 29.97 (or 59.94 if the deinterlacing had worked). It's showing:

Frame rate mode : Variable
Frame rate : 29.494 FPS
Minimum frame rate : 14.985 FPS
Maximum frame rate : 29.970 FPS

The raw capture won't be like that, nor would QTGMC do that, so it must have been introduced with that code chain that converts it to ProRes/MOV.

-The ProRes conversion has encoded the display aspect ratio incorrectly at 3:2. It should be 4:3.

-There's a glitch near frame 137, where the deinterlaced version does a field reversal for a couple of frames. It looks like a glitch on the tape and is hardly noticeable. A stabiliser of some type might help. Did you get that ES-15?

Is there anyway to fix this, or do I need to just recapture. The original captures are gone. I only have the prores converted ones. I have no idea why the aspect was changed, nor this variable frame rate. UGH!!!.
Thanks for your help!

[Riker0007]

I just noticed, after figuring out mediainfo app that the files I thought I recovered were the original converted prores ones have in fact been converted to BFF. If I have to recapture, then I will if it is better than trying to fix, esp if the outcome will be better. I really thought there is no way to recover once the field has been written??? Also, not sure where in the code converting from .avi ut video codec to prores it would change the aspect and frame rate?
Yes, I did snag a dmr-es15 off ebay, it is coming.

General
Complete name Horse Tape 2.mov
Format : MPEG-4
Format profile : QuickTime
Codec ID : qt 0000.02 (qt )
File size : 7.06 GiB
Duration : 12 min 26 s
Overall bit rate mode : Variable
Overall bit rate : 81.2 Mb/s
Frame rate : 29.801 FPS
Writing application : Lavf61.7.100

Video
ID : 1
Format : ProRes
Format version : Version 0
Format profile : 422
Codec ID : apcn
Duration : 12 min 26 s
Bit rate mode : Variable
Bit rate : 79.7 Mb/s
Width : 720 pixels
Height : 480 pixels
Display aspect ratio : 3:2
Frame rate mode : Variable
Frame rate : 29.801 FPS
Minimum frame rate : 2.997 FPS
Maximum frame rate : 29.970 FPS
Color space : YUV
Chroma subsampling : 4:2:2
Scan type : Interlaced
Original scan type : Progressive
Scan type, store method : Interleaved fields
Scan order : Bottom Field First
Bits/(Pixel*Frame) : 7.738
Stream size : 6.92 GiB (98%)
Writing library : Lavc
Matrix coefficients : BT.470 System B/G

Audio
ID : 2
Format : PCM
Format settings : Little / Signed
Codec ID : sowt
Duration : 12 min 26 s
Bit rate mode : Constant
Bit rate : 1 536 kb/s
Channel(s) : 2 channels
Channel layout : L R
Sampling rate : 48.0 kHz
Bit depth : 16 bits
Stream size : 137 MiB (2%)
Default : Yes
Alternate group : 1

[Sharc]

Some other issues:

-The frame rate of the Horse tape2.mov is a bit wonky, according to MediaInfo. It should be spot on 29.97 (or 59.94 if the deinterlacing had worked). It's showing:

Frame rate mode : Variable
Frame rate : 29.494 FPS
Minimum frame rate : 14.985 FPS
Maximum frame rate : 29.970 FPS

The raw capture won't be like that, nor would QTGMC do that, so it must have been introduced with that code chain that converts it to ProRes/MOV.

-The ProRes conversion has encoded the display aspect ratio incorrectly at 3:2. It should be 4:3.

-There's a glitch near frame 137, where the deinterlaced version does a field reversal for a couple of frames. It looks like a glitch on the tape and is hardly noticeable. A stabiliser of some type might help. Did you get that ES-15?

Is there anyway to fix this, .... I have no idea why the aspect was changed, nor this variable frame rate.

Framerate and aspect ratio are correct in the attached files of post #82. Fixing the 2 glitches around frame 10 and 274 - if you notice these at all - would require some extra work. Just being too lazy ...... maybe the ES-15 in passthrough will solve it.

[Riker0007]

This is what my captured files are pretty much showing. Here is the info from the one below. Not sure why the aspect is showing 3:2 and from the looks of it, they are still interlaced. If there are slight glitches here and there, I can handle that. Just not sure on this TFF or BFF. If my captures are BFF and need to be changed to TFF, and aspect is wrong can they be changed or just re-capture? Will they all be TFF?

Format : MPEG-4
Format profile : QuickTime
Codec ID : qt 0000.02 (qt )
File size : 129 MiB
Duration : 13 s 130 ms
Overall bit rate mode : Variable
Overall bit rate : 82.4 Mb/s
Frame rate : 29.970 FPS
Writing application : Lavf60.3.100

Video
ID : 1
Format : ProRes
Format version : Version 0
Format profile : 422
Codec ID : apcn
Duration : 13 s 130 ms
Source duration : 13 s 113 ms
Bit rate mode : Variable
Bit rate : 77.6 Mb/s
Width : 720 pixels
Height : 480 pixels
Display aspect ratio : 3:2
Frame rate mode : Constant
Frame rate : 29.970 (30000/1001) FPS
Standard : NTSC
Color space : YUV
Chroma subsampling : 4:2:2
Scan type : Interlaced
Original scan type : Progressive
Scan type, store method : Interleaved fields
Scan order : Bottom Field First
Bits/(Pixel*Frame) : 7.496
Stream size : 122 MiB (94%)
Source stream size : 127 MiB (98%)
Writing library : Lavc
Matrix coefficients : BT.470 System B/G

https://files.videohelp.com/u/139455/Video.mov

[Sharc]

Your video is captured as 720/480 = 1.5 = 3:2. This is correct and exactly what MediaInfo reports. However, MediaInfo has no information that the picture is stored "anamorphic" and should be viewed as 4:3. This is done later when resizing it for encoding, or by signalling the pixel aspect ratio as 10/11 which is standard NTSC video practice.
The instruction for the player how to play it is communicated in the ffmpeg commandline (see post#82) using -vf setsar=10/11. (Alternatively you could also put -vf setdar=4/3 there which is probably easier to understand). When you check the encoded files of post#82 (attachments) with MediaInfo it will report the Display Aspect Ratio as 4:3 and the player should play it accordingly as 4:3.
If this does not work for your player you should always find a player setting to force the playback as 4:3.
Plenty of posts in the forum about this. Keywords: anamorphic encoding, non-square pixels, square pixels, resizing, pixel aspect ratio, sample aspect ratio, display aspect ratio.

Why it is still interlaced after QTGMC I don't know. Something must be wrong or missing on your side.

[Riker0007]

I think/hope I Got it. So will all my other captures use the same avs script and ffmpeg cmd, just change file name?
Thanks again for the help!

[Sharc]

Yes (as far as your videos follow the NTSC format).

[Riker0007]

So I hope this is right. I did one at 29.97 and the other at 59.94.

https://files.videohelp.com/u/139455/29.97.mov

https://files.videohelp.com/u/139455/59.94.mov