DV-AVI best format for viewing on a TV

[lollo]

A resize to 768x576 is useless, unless you map a screen with SD resolution. The player of the TV or the PC will resize to fullHD or higher resolution anyhow, so yur intermediate step is not necessary (but I understand you like SAR=4:3 and PAR=1)

If you want to map the fullHD resolution bypassing the TV or PC resizing (or for YT) the upscale to 1440(1920)x1080 is appropriate.

Not optimized, because the choice of the resizer depends on your source. When is very sharp, a filter that sharpens (Bicubic, Lanczos, SplineXX) introduces halos and enhaces artifacts. I ignore what you use in VirtualDub2, because I do not use it.

Just nitpicking

[Alwyn]

I ignore what you use in VirtualDub2, because I do not use it.

I see. In that case, I'll will treat your criticisms of VDub (and GUIs in general) with a grain of salt.

[lollo]

I do not use does not mean i never used nor I do not test. No criticism, I prefere to not use GUIs when is not necessary and, more important, to have full control of the processing.
There are many topics on how Handbrake, StaxRip, VirtualDub, etc (even NLEs) have hidden wrong manipulation of data. But that’s me.

[Alwyn]

Lollo, that's fine, you're free to use whatever you think is best. What I object to is you continually saying GUIs are useless and nonsense. I'll bet there are plenty of newbys on here who are misled by your comments and struggling with the ridiculously arcane world of command line editing. For a lot of work, it just isn't necessary because there are good GUIs that take 5% of the time to achieve near-identical results. Perhaps, doing TV ads and shows, you find one or two AVS scripts work great. But when you are doing dozens of tapes, perhaps hundreds of hours of video, it is just crazy to poo poo GUIs.

A 90s home movie I am working on has 29 different objects over 12 minutes, each clip with it's own noise reduction settings (have you ever edited a 60min Digital 8 tape with half the scenes in low light?), half with colour correction settings and probably another half with individual, tailored cropping. It's just not viable to edit this with AVISynth, but you continually sound like it is the only option.

Yes, I did use AVISynth to sort out the decimation, but I exported that to an AVI and did the rest in my NLE. Oh, and I did use ChromashiftSP on two four-second scenes because I can't do that in my NLE.

The reason I jumped in here suggesting VDub to Skyblues was because I know trimming DV is easier in VDub than frame-numbering in AVISynth, which is certainly not the impression you are giving.

So by all means spruik AVISynth, but please don't sledge GUIs/NLEs. They can achieve almost the same (and, I might add, are more fun because they are so much more efficient).

Ironically, you sound a lot like another spruiker on here: My Way Or The Highway...

[lollo]

Stay on topic, without all this useless blah blah. Because OP uses AviSynth, I simply suggested to stay there for the trimming without introducing the useless (in this case) VirtualDub, as you did. Just as simple as that.

[SkyBlues2021]

I've been trying to decide for months whether to keep my analogue captures interlaced or deinterlace. I've made a few test clips and watched them many times on my smart TV, I think I prefer the appearance of the deinterlaced videos. I decided to do the Edit>Delete with Virtualdub2 and then Save >Direct stream copy, then I deinterlaced in Avisynth. I also did a version with trim in Avisynth and deinterlaced. The interlaced version I only used Virtualdub2. Both the deinterlaced version looked the same to me. Hopefully I haven't made any mistakes. Thanks.

[jagabo]

The trims are slightly different between the two deinterlaced videos. The deinterlacing with QTGMC is better than any other deinterlacer I've seen If you leave the video interlaced you'll be at the mercy of whatever deinterlacer is used in your playback chain. The difference isn't big in this particular video but will be more noticeable in others -- especially those with lots of near-horizontal edges (things like horizontal blinds in the background, shirts with horizontal stripes, etc.).

[poisondeathray]

The only reason AVISynth is even needed here is to run QTGMC. If it wasn't for that, all trimming, cropping and resizing could be done more easily in VDub.

Having to open the file in VDub, then open the AVS, type in the trim points, only to then re-open in VDub again really is making life more difficult than it needs to be.

Open the QTGMC AVS in VDub, set up the edit/trim with the In and Out selection on the visual timeline, crop visually and resize, then export as H264 (optionally, do all your segments at once using the Queue batch feature). Done.

Did you know there was a "GUI" of sorts for avisynth ? Avspmod.

You can mark in/mark out sections on the timeline with the home/end keys with the visual preview - the regions are marked in blue on the timeline - you can keep or delete marked regions (inclusive or exclusive editing) . vdub is better if you're going to direct stream copy (you can't with avisynth, audio/video gets decompressed)

There is also an interactive crop editor in avspmod if you prefer to do it that way. You drag the edges in/out with the mouse (visually, using the GUI), and it translates them into crop command

You can setup CLI encoders and presets in avspmod , you don't even need vdub, you can encode directly from avspmod if you wanted to. One problem with Vdub bundled encoders are static snapshots, and do not get updated often (for vdub you can use external encoder to get around that)

If there was a vdub-only filter (e.g deshaker), you can use LoadVirtualDubPlugin in the avs script to access them

It's just another option for those who like editing in vdub, because the mark in/mark out is about the same in terms of GUI functionality



For more complex editing, multiple layers and clip references, transitions, a dedicated NLE is going to be much easier, no question

[SkyBlues2021]

Thanks jagabo. I think I will go with deinterlaced. I would prefer to do the trim()++ with Avisynth but I was worried there would be a limit on consecutive trim()++ trim()++ I could have? as some of my videos have approx 18 sections to remove. If there is a limit I would probably have to use the Virtualdub2 Edit>Delete>Direct stream copy>Save.
The trims were just an approximation, when I trim the actual videos I will use the Bob() ShowFrameNumber() you taught me. Thanks.

[jagabo]

Keep your original interlaced video as archival backups. Watch and share the deinterlaced video.

I don't know what the limit for concatenating clip is in AviSynth. But I've done dozens at a time in the past.

[SkyBlues2021]

Thanks poisondeathray. I use Avspmod as a script editor I didn't know you could do encoding on it but I haven't really explored this. My abilities are somewhat limited. My understanding was, probably wrongly that in VirtualDub2 I could do a simple Edit>Delete >Direct stream copy>Save then open the video in Avisynth and deinterlace?

[poisondeathray]

Thanks poisondeathray. I use Avspmod as a script editor I didn't know you could do encoding on it but I haven't really explored this. My abilities are somewhat limited. My understanding was, probably wrongly that in VirtualDub2 I could do a simple Edit>Delete >Direct stream copy>Save then open the video in Avisynth and deinterlace?

I just wanted to point out that you can also use a GUI for trimming with avisynth without vdub

The encoding can be CLI or VFW. The CLI method is the most powerful by far, but you have to learn commandline arguments. You can setup presets, so you don't really have to "learn" them, you can get someone to help you set it up

You can edit in vdub2 first, or choose what works best for you . Many ways to do something

I would always keep the original stored away

If you were going to keep the edits separately to the originals, and encoded versions - then I might do it the way you outlined, otherwise, I wouldn't bother saving an intermediate only to open in avisynth to deinterlace - it seems like many extra steps and intermediates that you are just going to discard

If you were going to use vdub2 to edit (edit - as in just trimming out sections) , and QTGMC, you could open the script with QTGMC and edit that encode directly in vdub2 too, without the intermediate step

[SkyBlues2021]

Thanks jagabo. I will attempt a few trim()++ on my videos in the next couple of days and let you know how many I've been able to use, hopefully there isn't a limit. Thank you for the advice regarding backing up the interlaced AVI files. I have saved all the AVI files I've captured on 4 Fantom drives and I will keep them safe. I also have kept a back up at one of my son's home. He thinks it's a temporary arrangement!

[jagabo]

My understanding was, probably wrongly that in VirtualDub2 I could do a simple Edit>Delete >Direct stream copy>Save then open the video in Avisynth and deinterlace?

You can for intra-frame codecs.

[SkyBlues2021]

Thanks poisondeathray. I think I misunderstood. Sorry. That is probably worth me exploring. I'm very rigid in my thinking, if something works I don't like to stray too far! I will have a look at that tomorrow, especially the GUI for trimming with Avisynth. I'm guessing there is information on the Avisynth page. I found something in Utilities but I need to research it more. Thanks.

[SkyBlues2021]

Thanks jagabo. Are analogue YUY2 videos intra-frame codecs? Thanks.

[jagabo]

Thanks jagabo. Are analogue YUY2 videos intra-frame codecs? Thanks.

If they are uncompressed, or compressed with most lossless codecs, yes.

[SkyBlues2021]

Thanks jagabo. I used Huffyuv V2.1.1 lossless compression when I captured the video. Thanks.

[poisondeathray]

Thanks poisondeathray. I think I misunderstood. Sorry. That is probably worth me exploring. I'm very rigid in my thinking, if something works I don't like to stray too far! I will have a look at that tomorrow, especially the GUI for trimming with Avisynth. I'm guessing there is information on the Avisynth page. I found something in Utilities but I need to research it more. Thanks.

That's ok too - "If it ain't broke, don't fix it"

There are also sliders you can setup for basically any filter in avspmod . Lets say you wanted to change brightness, saturation etc... . This is all GUI based for those GUI lovers, and (essentially) realtime with feedback in the preview. Lets say you want to setup a slider for "sharpness", no problem

You have the option to monitor scopes, waveforms, histogram, vectorscopes in the script which show up in the GUI.

More advanced functions like HSV vs SAT , LUM vs SAT, HSV vs HSV etc.. are simlar in a way to curves in Davinci Resolve . (e.g. say you wanted to adjust a range of hues , or range of saturations. Resolve is much more powerful for color work, but it's nice to see avs+ getting extended functions ) . The video below shows example of sliders, some of the scopes available, and HSV x HSV filter



The best function of avspmod IMO is tabs. You can compare multiple versions of scripts, and they are all aligned on the timeline spatially and temporally. It would be like soloing a layer in a NLE, and each layer has a different filter chain. Or nodes in a nodal editor. What settings or combinations of filters should you use ? Set them up in different tabs, and use the number keys to hotswap. Navigate to different frame, all versions are on the same frame, zoom in and examine.

You have a status bar with color picker - it reveals YUV values, or converted RGB values (or vice versa, depending on what the source pixel format was), HEX values

In some ways the GUI is actually better than vdub's in terms of functionality . Spend some time playing with it, if you don't like it, that's ok too - They are all just tools, no need for anyone to get all bent out of shape

[Sharc]

I've been trying to decide for months whether to keep my analogue captures interlaced or deinterlace. I've made a few test clips and watched them many times on my smart TV, I think I prefer the appearance of the deinterlaced videos. .....

Keep your original interlaced captures in a safe place. In another couple of months (or years?) you may come to new conclusions how to process your captures for best results, be it due to your learning curve, changing preferences, new or improved filters, different filter settings (parameters) ....., and you may wish to redo some of your videos. (Happened to many of us, I believe).

Edit:
The deinterlaced (QTGMC) variant is about half the file size of the interlaced encode, despite the double number of frames. This size reduction is mainly due to the relatively strong denoising effect of QTGMC with its default settings. Too much denoising tends to make skins look 'plastic'. One has to find the (personal) balance.

[SkyBlues2021]

Thanks Sharc. I've been watching the videos after I've encoded them, everything so far looks good but thank you for alerting me to the fact the skin could look plastic. I will look out for that when I'm watching them. I'm keeping my interlaced copies, hopefully once I have a copy of all the videos completed for my children. I want to practice more and generally improve the videos and understand and experiment more.

[SkyBlues2021]

Thanks poisondeathray. I've watched the video. However, I didn't get much of a chance to research this today but I will take a look at the weekend. I'm guessing you find the sliders somewhere in edit. I've seen the examples of them in Avisynth's User Sliders page.

[poisondeathray]

Thanks poisondeathray. I've watched the video. However, I didn't get much of a chance to research this today but I will take a look at the weekend. I'm guessing you find the sliders somewhere in edit. I've seen the examples of them in Avisynth's User Sliders page.

There are so many 3rd party user plugins, that you cannot possible have them all as sliders. Also some functions have like 60 parameters, they would be too complex for sliders (way too much clutter)

You can enable your own sliders
http://avisynth.nl/users/qwerpoi/UserSliders.html

Or there is a filterdb.dat (filter database) , that has pre-programmed sliders for many of the commonly used functions. Avspmod already comes with this version -
https://github.com/gispos/AvsPmod/releases

But there user modded versions, that include their own functions . And of course you can save your own database version
eg. dogway's version, which has his filter definitions enabled in the filterdb.dat
https://github.com/Dogway/Avisynth-Scripts/tree/master/Tools/AvsPmod
https://github.com/Dogway/Avisynth-Scripts/blob/master/Tools/AvsPmod/filterdb.dat

Personally, I don't use the sliders. It's just an option for people that like them in GUI's

[Alwyn]

The deinterlaced (QTGMC) variant is about half the file size of the interlaced encode, despite the double number of frames. This size reduction is mainly due to the relatively strong denoising effect of QTGMC with its default settings. Too much denoising tends to make skins look 'plastic'. One has to find the (personal) balance.

All my QTGMCs are around double the size of the original AVI (FPS Divisor=1). I see virtually no denoising with "Fast". What settings are you using to end up with 1/2 file size and significant denoising? I would have thought to reduce the file sizes by that much a lot of the frame would have to be the same colour.

[Sharc]

The deinterlaced (QTGMC) variant is about half the file size of the interlaced encode, despite the double number of frames. This size reduction is mainly due to the relatively strong denoising effect of QTGMC with its default settings. Too much denoising tends to make skins look 'plastic'. One has to find the (personal) balance.

All my QTGMCs are around double the size of the original AVI (FPS Divisor=1). I see virtually no denoising with "Fast". What settings are you using to end up with 1/2 file size and significant denoising? I would have thought to reduce the file sizes by that much a lot of the frame would have to be the same colour.

I referred to the OP's x264 recompressed .mp4 files of post#156. The deinterlaced variants are about half the size of the interlaced variant using the same encoder settings (crf 12.0). Apparently the noise reduction of QTGMC resulted in about 1/2 files size due to better compressibility of the denoised source. The noise suppression is clearly visible.
(The default speed setting of QTGMC() is 'medium'. Also, the OP set sharpness to 0.0 to avoid oversharpening which contributes to the compressibility).

[SkyBlues2021]

I've been reading a few things today, this always makes me jittery! I didn't specify Assume BFF() in my scripts for my DV Digital8 videos as my understand is that this is the default for Avisynth and it's not required. However when I've seen it in other scripts today I became a bit concerned. The videos look great but I just wanted to check I haven't made a mistake. Thanks.

AviSource("DV clip 2.mp4")
ColorYUV(gain_y=-15, off_y=-10)
QTGMC(Sharpness=0.0)
dehalo_alpha(rx=2.0, ry=2.0)
Prefech(4)

[poisondeathray]

I didn't specify Assume BFF() in my scripts for my DV Digital8 videos as my understand is that this is the default for Avisynth and it's not required.

Native DV will usually be BFF. If you get the wrong field order, you will know, because the motion will go forward/back/forward/back jerky after double rate deinterlacing.

[SkyBlues2021]

Thanks poisondeathray. That's a relief! I've have seen so many other people putting AssumeBFF() in there DV scripts today that I thought I must have interpreted this incorrectly and I should have put this in my scripts. It sent me into a panic, Thank you for the putting my mind at rest.

[poisondeathray]

It doesn't "hurt" to include the AssumeBFF() (or TFF for TFF sources) - it doesn't slow down the script or add overhead or anything like that. In some ways, it "reminds" you to consider the field order, sort of a double check. You can have rare occurrences of TFF DV, and mislabled as BFF or vice versa. It's always good practice to check the actual field order, it takes a few seconds - better than "blindly" running a script, only to waste hours encoding the wrong field order and have a botched file

[SkyBlues2021]

Thanks poisondeathray. I'm just trying to find a way to test it. I guessing that you have to eyeball it. I've just found something Avisource("filename.avi")
Assumebff().SeparateFields() save> then watch it in virtualdub. Unless there is better way?