Horizontal Red and Green Lines (GV-USB2)

[Sharc]

....... I did end up trying AmarecTV and whaddaya know - it didn't drop a single frame in a 3 hour video! .....

You have been given this advice many times. Glad it worked for you now.

Question about that while I'm here - should I be doing resizing and rendering at the end of my QTGMC script so it's all done in one step? Or should I run QTGMC and save as a lossless HuffyUV AVI again, and then perform resizing + rendering to H.264 in a separate script?

You can do it in one step (one script). First QTGMC before any resizing is what matters in the sequence, otherwise the field structure would be destroyed.
(And better use Avisynth and ffmpeg rather than some confusing GUIs. But it is up to you of course what you feel more comfortable with.)

[WhiteboyUltima]

....... I did end up trying AmarecTV and whaddaya know - it didn't drop a single frame in a 3 hour video! .....

You have been given this advice many times. Glad it worked for you now.

Question about that while I'm here - should I be doing resizing and rendering at the end of my QTGMC script so it's all done in one step? Or should I run QTGMC and save as a lossless HuffyUV AVI again, and then perform resizing + rendering to H.264 in a separate script?

You can do it in one step (one script). First QTGMC before any resizing is what matters in the sequence, otherwise the field structure would be destroyed.
(And better use Avisynth and ffmpeg rather than some confusing GUIs. But this is up to you what you feel more comfortable with.)

Yes well, I wanted to keep eliminating other possibilities before having to learn yet another piece of software


I already have AVISynth+ and my script is pretty much ready, I've been using Notepad and AvsPmod to write it. I was just going to run the script in VirtualDub2 (64-bit), is that alright?

And is this renderer ok? I wanted a standard H.264 mp4 for maximum compatibility across devices/software, but looks like none of my installed ffmpeg codecs are H.264 so I'll have to install it otherwise


[Attachment 90231 - Click to enlarge]

[Alwyn]

Well there you go! Waiting for the Amarec haters to arrive...

Whether you do the resizing (and cropping ) in the script or in VDub is personal preference; while you can get quick feedback using the VDub AVISynth script editor, I prefer to keep the script simple and do cropping and resizing in VDub, while you are doing your trimming. You can then export out of VDub with H264; no need to go via hUFFYUV. Others may comment on the quality of the H264 encoder in VDub2; it's fine for me. If you're going to do more editing in a proper NLE, you can export out of VDub into an AVI that will be accepted by your NLE (which is what I do).

I do value @Sharc's input and knowledge, but IMO, using FFMpeg is like practicing bleeding. There is a reason Bill replaced DOS with Windows. The flexibility and visual feedback (eg colour and levels adjustments), that even the lowly VDub gives makes it far more productive than mucking around with arcane command lines. It's horses for courses though; it's just that that particular horse ain't for me!

In my view, with these long tapes, a very important aspect of achiving these videos is making them watchable. That is, there might be 5 minutes of car ride to the beach, then 15 minutes of a party. You don't want those joined together, you want them in separate files. That is much easier to do in an NLE. In one of my projects in Magix Video Deluxe, I have 30 timelines/movies from 3 tapes, with bits moved from one tape to another to combine like-subjects, and batch export them to H264 all in one hit. You could never do that in FFMPeg, not to mention fades/transitions, zooming and panning and titling.

[Sharc]

I already have AVISynth+ and my script is pretty much ready, I've been using Notepad and AvsPmod to write it. I was just going to run the script in VirtualDub2 (64-bit), is that alright?

That's all right.

And is this renderer ok?

Yes it is ok. Press configure to tweak the encoder parameters as needed, for example:

[Attachment 90232 - Click to enlarge]



@Alwin: AFAIK Vdub x264 encoder is based on an ffmpeg library (libavcodec). Quality is fine.

[WhiteboyUltima]

Well there you go! Waiting for the Amarec haters to arrive...

Whether you do the resizing (and cropping ) in the script or in VDub is personal preference; while you can get quick feedback using the VDub AVISynth script editor, I prefer to keep the script simple and do cropping and resizing in VDub, while you are doing your trimming. You can then export out of VDub with H264; no need to go via hUFFYUV. Others may comment on the quality of the H264 encoder in VDub2; it's fine for me. If you're going to do more editing in a proper NLE, you can export out of VDub into an AVI that will be accepted by your NLE (which is what I do).

There are Amarec haters? Hmm. Anyway, I didn't consider cropping. Why do I need to do that? My videos don't seem to have black vertical lines on the sides like some people get. Is it due to the aspect ratio?

I have captured at 720x576 (5:4 aspect for non-square pixels). My target final resolution is 1440x1080 (4:3). I was just going to resize straight to that, but some sources (including you apparently) say I should first be cropping the video (to 704x576...which isn't even 4:3 but is advised) before I resize to 1440x1080... what is best?
Why must I crop first when I can just resize in a single command? Is it because you get a higher quality resize if your source and target are the same AR?

Also, is it best to do QTGMC deinterlace > crop > resize all in the same AVISynth script (as Sharc said is possible), or is it better if they done in 2 or 3 separate scripts? I kind of want to keep a video that is just deinterlaced so I can check the result first, but if that lowers the quality (because I'm saving in HuffyUV twice (first time during capture)), then that is bad.

Also I'm not sure where in the script to do a crop if I'm going to do one, should it go before QTGMC or after? I understand the resize comes last.

I don't even know what NLE is I am just a simple man trying to get the best digitisation I can from a few old VHS tapes haha

[WhiteboyUltima]

@Alwyn also mentioned trimming, I do need to take off some seconds at the end of the video which I've figured out how to do in VDub. But can I do that without having to make another save/compress using HuffyUV? I was just going to go straight to h.264 mp4 when saving my next script. Is it possible to drag the script into VDub AND edit the video (takes some frames off the end) and save the video in h.264, all in one go?

[Alwyn]

My videos don't seem to have black vertical lines on the sides like some people get.

If your edges are clean, I'd just leave it. But don;t you have that annoying head-switching noise at the bottom? I crop that off (or you can blank it with a border).

including you apparently

Ouch!

Why must I crop first when I can just resize in a single command?

You can, without loss. The 704 bit (ie crop 16 off the sides) is allegedly to correct the way the capture standard works. And yes, you then go straight from 704x576 to 1440x1080.

Also, is it best to do QTGMC deinterlace > crop > resize all in the same AVISynth script (as Sharc said is possible), or is it better if they done in 2 or 3 separate scripts?

Believe @Sharc. I'm just an enthusiastic amateur.

Also I'm not sure where in the script to do a crop if I'm going to do one, should it go before QTGMC or after? I understand the resize comes last.

Don't. It's too hard. Just crop using the Crop filter in VDub. And once you've interlaced with QTGMC via your script, you can crop and resize to your heart's content, although it is more logical to crop first, then resize. The correct order is only important when dealing with interlaced footage because if crop or resize incorrectly, you may mess up the field order. That obviously doesn't apply to Progressive video ie post-QTGMC.

Also I'm not sure where in the script to do a crop if I'm going to do one, should it go before QTGMC or after? I understand the resize comes last.

QTGMC>Crop>Resize.

Is it possible to drag the script into VDub AND edit the video (takes some frames off the end) and save the video in h.264, all in one go?

Yes. I have a speil on VDub here. I've also done a quick and dirty Youtube on VDub editing here.

[Sharc]

I do value @Sharc's input and knowledge, but IMO, using FFMpeg is like practicing bleeding. There is a reason Bill replaced DOS with Windows. The flexibility and visual feedback (eg colour and levels adjustments), that even the lowly VDub gives makes it far more productive than mucking around with arcane command lines. It's horses for courses though; it's just that that particular horse ain't for me!

Gentle reminder: Not so long ago you argued similarly about using QTGMC against Vdub deinterlacers/plugins. Now since you have got QTGMC working you recommend it all the time

In my view, with these long tapes, a very important aspect of achiving these videos is making them watchable. That is, there might be 5 minutes of car ride to the beach, then 15 minutes of a party. You don't want those joined together, you want them in separate files. That is much easier to do in an NLE. In one of my projects in Magix Video Deluxe, I have 30 timelines/movies from 3 tapes, with bits moved from one tape to another to combine like-subjects, and batch export them to H264 all in one hit. You could never do that in FFMPeg, not to mention fades/transitions, zooming and panning and titling.

There are excellent free NLEs like Shotcut or Kdenlive btw, certainly good for all the effects you mention.

[Alwyn]

Gentle reminder: Not so long ago you argued similarly about using QTGMC against Vdub deinterlacers/plugins. Now since you have got QTGMC working you recommend it all the time

Ha ha! I thought about mentioning that! You are right, but I will say that that is the only normal use I have for AVISynth. AVISynth has no features that I can do as well as or better in VDub and Video Deluxe, and those two are far easier and faster to use. Processing cine is a different story, where the tools that work are only in AVISynth.

[Alwyn]

There are excellent free NLEs like Shotcut or Kdenlive btw, certainly good for all the effects you mention.

Unlimited, separate timelines?

[Sharc]

There are excellent free NLEs like Shotcut or Kdenlive btw, certainly good for all the effects you mention.

Unlimited, separate timelines?

Sure. Perhaps not unlimited (I didn't count), but when would you need thousands of timelines?

[Alwyn]

I use the term "unlimited" with tongue in cheek. The most I've used is around 40. And I'm not talking about tracks, I'm talking about timelines.

[WhiteboyUltima]

My videos don't seem to have black vertical lines on the sides like some people get.

If your edges are clean, I'd just leave it. But don;t you have that annoying head-switching noise at the bottom? I crop that off (or you can blank it with a border).

My left and right edges are clean yes, and as for that weird head noise at the bottom... I actually like it, seems more authentic. Plus just below that you can see actual footage so I'd be getting rid of that if I cropped off the head noise.

So that leads to my next point... can I just not crop at all? If I just do a resize from 720x576 (5:4) to 1440x1080 (4:3) with no crop whatsoever, will that give me the correct looking proportions e.g. a circle looks like a circle?

including you apparently

Ouch!

Wasn't meant as a jab

Also I'm not sure where in the script to do a crop if I'm going to do one, should it go before QTGMC or after? I understand the resize comes last.

QTGMC>Crop>Resize.

Thanks.

Is it possible to drag the script into VDub AND edit the video (takes some frames off the end) and save the video in h.264, all in one go?

Yes. I have a speil on VDub here. I've also done a quick and dirty Youtube on VDub editing here.

Will give those a watch, cheers

EDIT: Your guide said "Resizing using this method will only work properly if your crops were made in the 4:3 ratio, as explained above."
So that suggests that if I just Resize to 1440x1080 then it won't work unless I crop first? What if I don't want to crop?

[Sharc]

So that leads to my next point... can I just not crop at all? If I just do a resize from 720x576 (5:4) to 1440x1080 (4:3) with no crop whatsoever, will that give me the correct looking proportions e.g. a circle looks like a circle?

Not exactly, but "almost". For captured VHS tapes you will end up with a ~2.3% aspect ratio error. It's little, but trained eyes may notice the slight squashing. Very most viewers won't notice the slight error for "natural" scenes though.
The reason for the distortion is because your PAL capture does not have square pixels but a pixel aspect ratio of ~12:11 (see the screenshot in post#64 for the x264 encoder settings).

Or in other words: If you don't crop to 704 wide and want to keep the circles intact you would have to resize to 1474x1080 which is not exactly 4:3. Choose your poison.

[WhiteboyUltima]

So that leads to my next point... can I just not crop at all? If I just do a resize from 720x576 (5:4) to 1440x1080 (4:3) with no crop whatsoever, will that give me the correct looking proportions e.g. a circle looks like a circle?

Not exactly, but "almost". For captured VHS tapes you will end up with a 2.3% aspect ratio error. It's little, but trained eyes may notice the slight squashing.

Hmm so it sounds like I do need to take off 8 pixels either side to get 704x576, would that be:

video.Crop(8, 0, -8, -0)

I hope I don't lose any actual footage by doing this Hard to even tell if I've even got black lines at the sides of my capture, as every video player has a black background...

I also still don't understand that doing that crop gives an aspect that is not actually 4:3. Why is this required in order to accurately resize to 1440x1080 (which is 4:3)?

[Sharc]

So that leads to my next point... can I just not crop at all? If I just do a resize from 720x576 (5:4) to 1440x1080 (4:3) with no crop whatsoever, will that give me the correct looking proportions e.g. a circle looks like a circle?

Not exactly, but "almost". For captured VHS tapes you will end up with a 2.3% aspect ratio error. It's little, but trained eyes may notice the slight squashing.

Hmm so it sounds like I do need to take off 8 pixels either side to get 704x576, would that be:

video.Crop(8, 0, -8, -0)

I hope I don't lose any actual footage by doing this Hard to even tell if I've even got black lines at the sides of my capture, as every video player has a black background...

Don't you see any black sidebars in Vdub or AvsPMod (on the gray background)?

I also still don't understand that doing that crop gives an aspect that is not actually 4:3. Why is this required in order to accurately resize to 1440x1080 (which is 4:3)?

It's confusing at the beginning, I agree. 1440x1080 maps to 4:3 display aspect ratio for square pixels only. Your captured PAL VHS video is however not square pixels but has a pixel aspect ratio of ~12:11, (or in another approximation used by Vdub ~59:54 which is virtually the same). You would have to understand anamorphic (non square pixels) formats and the analog capturing process as per international standards. Plenty of posts about it, I am just lazy to search links. For explanation see here
https://en.wikipedia.org/wiki/Pixel_aspect_ratio.
For now just believe that the 4:3 PAL picture is contained within ~704x576 (on basis of international standards)
Anyway, no sleepless nights because of a ~2% aspect ratio distortion if you don't crop.

[WhiteboyUltima]

Don't you see any black sidebars in Vdub or AvsPMod (on the gray background)?

Ya know what, I just watched all 3 of my captured tapes today and yes, they all do indeed have black sidebars! Looks like I have no reason not to crop then.

So I'd just like to run this all past you guys, tell me if I'm wrong about anything please. I'll write a script later and also post it here but here's my plan: Given that these are all interlaced PAL captures at 720x576 HuffyUV, I am keeping these as my masters. I will then run a single script in VDub on each tape, which does QTGMC > Crop 8 pixels off either side > Resize, in that order. I will encode/save the result using VDub's x264 H.264 8bit filter (will play with those settings later). Using mp4 and AAC for audio of course. It will be a 1440x1080 resize, which is 4:3 which is what (I think) I want for my final finished video file.

Is all of that right? Would that produce a video that displays in the exact correct proportions (i.e. same as the original tape would have looked on a CRT TV) on all (or most) TVs, computers, youtube etc.?


Assuming this is all well and good to proceed, my next question is, should I resize with Bilinear, Lanzcos or Spline64? Again I am new to all this and the advice AI gives you is sometimes conflicting.


Side note, since I'm worried about how much quality might be lost in doing H.264 compression, I might just create a script that only does QTGMC processing and no other steps first, saving using HuffyUV, just so I can check I'm getting the best output from my QTGMC settings before any crops/resizing/H.264 compression. But my final video will not be based on that intermediary video file, it will come straight from my raw capture file and do all the processing in one script. Hope that makes sense!
Edit - since I just discovered you can live preview all the QTGMC settings etc. in AvspMod, maybe saving this extra video file isn't necessary. I guess it's a good way to also preview different resizers... but I'd be keen to hear your recommendations for resizing.

[Alwyn]

Is all of that right? Would that produce a video that displays in the exact correct proportions (i.e. same as the original tape would have looked on a CRT TV) on all (or most) TVs, computers, youtube etc.?

Yes and Yes.

[Sharc]

Don't you see any black sidebars in Vdub or AvsPMod (on the gray background)?

Ya know what, I just watched all 3 of my captured tapes today and yes, they all do indeed have black sidebars!

No surprise. That's how it should be.

I will then run a single script in VDub on each tape, which does QTGMC > Crop 8 pixels off either side > Resize, in that order.

Sometimes the black borders are not symmetric. You just have to crop 16 (or 18 for 702 width for PAL would be more accurate even, just nitpicking) in total left+right. So you can crop 8+8 or 10+6 or 6+10 or 4+12 or 10+8 or 0+16 or 8+10 etc. etc. whatever fits. Crop in even numbers (=mod2).
Also, you might want to add some denoising filter, preferably preceeding the upscaling.

I will encode/save the result using VDub's x264 H.264 8bit filter (will play with those settings later). Using mp4 and AAC for audio of course. It will be a 1440x1080 resize, which is 4:3 which is what (I think) I want for my final finished video file.

Is all of that right? Would that produce a video that displays in the exact correct proportions (i.e. same as the original tape would have looked on a CRT TV) on all (or most) TVs, computers, youtube etc.?

Yes to all.
In the x264 encoder settings you can adjust the CRF parameter which controls the compression vs quality.

Assuming this is all well and good to proceed, my next question is, should I resize with Bilinear, Lanzcos or Spline64?

For upscaling Bicubic, Lanczos and Spline64 are all good. Trust your eyes.

[WhiteboyUltima]

Thanks guys. Sharc, what Denoise filter do you recommend (if any)? Part of me wanted to just preserve that original VHS look but I'm open to seeing what this can do.

One question, you told me it's more accurate to crop 18 (to 702 pixels) - but then agreed with my comments about a 704 crop bring the correct proportions anyway. So which is correct?

[jagabo]

702x576 is more accurate than 704x576 for PAL video. But the latter is most often used because it is a mod 4 frame size (some codecs require mod 4) and because it's similar to the 704x480 frame size (which is also an approximation) for NTSC video. In any case, you'll never notice the difference. Even with a 720 frame it isn't generally noticeable.

[Sharc]

Thanks guys. Sharc, what Denoise filter do you recommend (if any)? Part of me wanted to just preserve that original VHS look but I'm open to seeing what this can do.

QTGMC includes denoising already which may be adequate. Study its settings. Additional denoising - if needed at all - depends on the source. SMDegrain, MCDegrainSharp and TemporalDegrainn2 to name a few .... Try and experiment and gain experience. Same for possible other defects which you may notice and want to address. Just don't overdo. You may damage more than what you gain.
http://avisynth.nl/index.php/External_filters#Restoration_Filters

One question, you told me it's more accurate to crop 18 (to 702 pixels) - but then agreed with my comments about a 704 crop bring the correct proportions anyway. So which is correct?

18 is slighly more correct for PAL. But no worry, the difference does not matter in practice (<0.3%). If you crop 16 and still see some residual black borders or fuzzy edges you can crop to 702 without bad feelings. 704 (16) or 702 (18) is nitpicking. I am sure your video has more important issues to be addressed.

[lollo]

702x576 is more accurate than 704x576 for PAL video.

https://forum.videohelp.com/threads/404626-How-do-i-upscale-PAL#post2646849

I tend to agree with Skiller, I see that half scan line in my captures.

[Sharc]

702x576 is more accurate than 704x576 for PAL video.

https://forum.videohelp.com/threads/404626-How-do-i-upscale-PAL#post2646849

I tend to agree with Skiller, I see that half scan line in my captures.

Oh yes, I remember this discussion
For info I attach the EBU document which says 702x576 shall represent the "active" picture
I mentioned the 702 mainly because the OP may find this value somwhere else, e.g. here
https://en.wikipedia.org/wiki/Pixel_aspect_ratio

[Sharc]

@Whiteboy:
I forgot to mention that IF you upscale to 1440x1080 think about converting - or at least signalling - the colorspace correctly. A player or YouTube may by default assume 709 colorspace for the upsized picture while your captured VHS source is 601. So there is a certain risk that the colors are slightly off. Most visible for reds and greens. Some people notice it immediately, some only in direct comparison. Just in case you stumble across it.

[WhiteboyUltima]

702x576 is more accurate than 704x576 for PAL video.

https://forum.videohelp.com/threads/404626-How-do-i-upscale-PAL#post2646849

I tend to agree with Skiller, I see that half scan line in my captures.

What is a half scan line and does it affect PAL? Reading that thread in your link by Skiller was quite over my head - stuff about padding and centering and formulas and whatnot, that I don't think I need to do? So is the conclusion, I should just crop to 704x576 for the best proportions before I upscale to 1440x1080?

I do notice a small line at either side when cropping to 704, that maybe cropping to 702 would fix, so is there any downside to doing that (aside from some codecs apparently not working with it since it's not a multiple of 16 or so i read)? I.e. will proportions still be accurate or should I just leave it at 704 or will it not really matter. I am looking for the correct answer here, but there seems to be so many ways to do this, it's confusing.

[WhiteboyUltima]

@Whiteboy:
I forgot to mention that IF you upscale to 1440x1080 think about converting - or at least signalling - the colorspace correctly. A player or YouTube may by default assume 709 colorspace for the upsized picture while your captured VHS source is 601. So there is a certain risk that the colors are slightly off. Most visible for reds and greens. Some people notice it immediately, some only in direct comparison. Just in case you stumble across it.

Thanks, I will just use the denoising built into QTGMC I think, and keep it light but also not smooth over everything too much. I kinda like the original look. Are these settings ok?

QTGMC(Preset="Fast", SourceMatch=3, Lossless=2, Sharpness=0.1, MatchEnhance=0.75)

Not sure if I should keep MatchEnhance at the default 0.5 or 0.75. And if I put Sharpness=0.0, will it not apply any sharpening whatsoever, or is there inherent sharpening going on anyway with the whole QTGMC process?
I guess my aim is to keep things as close to the capture as possible; I don't want that plastic waxy look, and I don't mind a tiny bit of sharpening as long as it doesn't introduce any strange artifacts. Any suggestions for my settings?

QTGMC includes denoising already which may be adequate. Study its settings. Additional denoising - if needed at all - depends on the source. SMDegrain, MCDegrainSharp and TemporalDegrainn2 to name a few .... Try and experiment and gain experience. Same for possible other defects which you may notice and want to address. Just don't overdo. You may damage more than what you gain.
http://avisynth.nl/index.php/External_filters#Restoration_Filters

I am definitely upscaling to 1440x1080; I have no idea about colourspace, never thought about that before. What does 'signaling' it do and how do I do that? (I assume I need it to retain a 601 colourspace after conversion?)

[Sharc]

Thanks, I will just use the denoising built into QTGMC I think, and keep it light but also not smooth over everything too much. I kinda like the original look. Are these settings ok?

QTGMC(Preset="Fast", SourceMatch=3, Lossless=2, Sharpness=0.1, MatchEnhance=0.75)

They are ok. Just keep in mind that QTGMC settings (it has about 200 parameters) depend on your source and on your personal preference. No one here has seen your capture nor has your eyes, nor do we have a crystal ball.

I guess my aim is to keep things as close to the capture as possible

So don't deinterlace and don't upscale. Both of these processes introduce artifacts. Video is full of compromises! Apart from obvious flaws much depends on your personal preference though.

I am definitely upscaling to 1440x1080

In this case you have to deinterlace.

I have no idea about colourspace, never thought about that before. What does 'signaling' it do and how do I do that? (I assume I need it to retain a 601 colourspace after conversion?)

Signalling: Means the encode has to include instruction for the player how to decode it (means applying the same color matrix as the source). In the hope that the decoder respects these instructions.
Conversion: Means you convert the colorspace to the "HD" standard (720+ pixels vertical) so the decoder - when using its "HD defaults" (709 colorspace) - will return the correct RGB colors.
Leave as is, and revert when you have doubts about your encode, or post your full script so someone may comment.

I think you are good to go now.

[Alwyn]

"Pixel format" in VDub2 x264 encoder:


[Attachment 90260 - Click to enlarge]

Choose 709 if your video is 720 high or higher.

[WhiteboyUltima]

"Pixel format" in VDub2 x264 encoder:


[Attachment 90260 - Click to enlarge]

Choose 709 if your video is 720 high or higher.

Ah ok that's easy, any downsides? And my source is not 720 high (or were you meaning the output video, which will be resized to 1440x1080?)

EDIT: Should I have actually done this step for my original capture? Or does changing the colourspace only apply when I am processing that capture i.e. resizing to HD.