Quality of captured VHS tapes, Video8 records?
 

Hello Team!
I'm absolutely beginner from Hungary, and spent the last 1-2 month to build my capture station The details of the system:
VHS recorder, Panasonic NV-HS 950EE (TBC off, SVHS ON)
DVD player, Panasonic DMR-ES15
ATI All-In-Wonder Radeon 9600 PRO 64MB
Intel Pentium4, 3066 Mhz

I have video8 tapes and VHS tapes, what are the copies of the "deleted" video8 tapes. Most of the video8 tapes was recorded by a Kyocera KX-77 cam.
For the capturing of VHS's I use the suggested methode from this site: Virtualdub 1.9.11 and based on the "guide", of course to lossless avi.

I have attached some samples to show to you the lossless captured video quality.
Video1-2-3-4: Original tape recorded with Kyocera KX-77 video8 camera, after it was transfered to VHS. I captured with AIW system described above
Video4_DV:Original tape recorded with Kyocera KX-77 video8 camera. I "saved" with Sony DCR-TRV 345E via DV (from video8 tape, with WinDV)

My concern is, that the captured lossless quality is not, what I'm looking for. I would like to be ensured, that the capturing what I did, is OK (I can imagine, that with tapes recorder with this Kyocera camcorder and after transfered to VHS with "who knows what kind of" VHS recorder, that is the max quality, but I do not know really). If it is theoritically OK, than I would like to make some post-process, filtering on it, because from my point of view it is not in good state. What I do not know, how many possibility do I have to filter these. Which kind of filters should I use, ect. I'm beginner.

My second concern is the DV "saving". If we compare the Video4 and Video4_DV file, that I can state, that Video4_DV much more better, than Video4. And that is not the conclusion, what I can read from expertised peoples.
If I can reach the same quality, with AIW captured file with filtering as the sampled DV save in attached, I would reach my goal.

Concerning de-interlacing, I'm not sure, that it is what I have to do, but that part is concerning to the "necessary post process, filtering question.

What is your opinion ?

The Kyocera KX-77 looks like a clone/badge engineered or maybe modified version of a Sony camcorder.

The quality difference here is not between DV and "not DV" capture, but between the original and a copied tape. Dubbing an analog tape was always a operation that would lose quality - in this case both due to the dubbing process itself, and that the VHS format has slightly lower resolution than Video8 - how much quality impact would of course depend on the gear used. Digitizing the videotape via S-video output from the camcorder and the the AIW instead of using the DV output should give similar quality but without the DV compression artifacts. You want to use the original tapes when possible though dubbing the video to VHS for easier viewing and re-using the camcorder tapes was a common thing so ye.

You can also play around with the settings on the VCR to see to what extent TBC, EDIT and 3D DNR settings work better or not. This vcr has much more advanced noise reduction than the average VCR so not sure what works best here.

Avisynth/Vapoursynth can be powerful tools for noise filtering, deinterlacing and processing on vhs captures, Hybrid front-end/graphical program that makes it a bit easier to use them if you don't want to delve into scripting, so it might be easiest to start on that.

Thanks for your comments.

I have made the exercise with the VCR setup. Maybe I have to make much deeper investigation on it, but during the test, what I made, the conclusion is, that there is not any significant change if I variate with the Edit, TBC, 3D and mix of them.

Could you give me a support, that based on the attached videos, what is the basic filters, what I have to start to insert firstly? Is it temporal smoother, HSV adjust, and sharpen? I really did not do it before, so I'm looking for some starting point.

Thanks :)

I'm confused here. I agree 4_DV is far better than 4.

To produce Video4, did you capture the Video4 VHS tape with the AIW?

And to produce Video4_DV, did you play the Video4 VHS tape using the TRV 345 in passthrough mode to a Firewire capture?

Or do you still have the Video4 Video 8 tape, which you played in the TRV 345 and captured via Firewire?

You have right, that can be confusable.
Video4 is captured from VHS via AIW. Video4_DV captured from video8 tape by TRV 345 via Firewire. So yes, thats material I have today on video8 and VHS as well, but it is mostly not the case with the rest one.
So maybe there I'm dumb, and try to compare the apple with the pear.

Or it can represent, what is the "quality loss" during the transfer from video8 to VHS. In this term what is your opinion? Could we made any conclusion on this way?

But that would be magically if I would be able to reach that Video4_DV quality with the VHS tapes. And I understand, that due to the VHS transfer step there is lost quality, but I would like to do something, to improve at least something :).

Referring to Video4, during the DV to VHS conversion (or during capture by the AIW), somehow the saturation got boosted. Worse, the blacks got crushed.

Here's a quick restored version, with the saturation reduced to match the DV, the edges of the frame masked out, and the video deinterlaced using QTGMC. Here are the relevant lines in my Avisynth script:

I didn't bother to adjust the contrast to match the DV because it would have turned the crushed blacks to dark grey, so it's up to you whether the contrast accuracy is worth the tradeoff.

Thanks traal!
Honestly speaking I just start (first time) to use AviSynth. The needed staff is installed, but when I try to open the script, got a bloody message: couldn't locate a decompressor for fourcc HFYU. Uninstall everything, reinstall, same issue. I have stucked here, and google do not want to be my friend still.
Due to it I try to do some alternatives, as using Hybrid. It is in progress to figure out how is working.

This means it couldn't find a codec to decompress your Huffyuv file.

Presumably you do have a Huffyuv codec installed, since you compressed the file in the first place. Usually what causes this error is that you only have the 32-bit version installed, and you're trying to open a 64-bit script, or vice-versa.

I haven't tried it, but the simplest solution should be hofmand's new installer posted here: http://www.digitalfaq.com/forum/vide...html#post81619

Could explain to me, what does it mean? How it looks like crushed and not crushed black?

Could I get some of suggestions to the Video1-2-3 as well? I do not need really too much, just to see where do i have to play: contrast, saturation, denoising, ect., and just get a simple script as traal did with video4 :). (do not spend the time with the borders).

Video1: Virtual Dub YADIF Deinterlace Double Frame Rate, Colormill, Color Balance and Neat Video noise reduction (NV is a commercial plugin).

I cropped the edges of the frame, then created a histogram which you can see on the right:

Attachment 14995

The minimum luma of this frame is 17 (it should be 18+) which means there could be some clipping of the darker greys. Then looking at the luma histogram, you can see an abrupt falloff right at the 17 luma point (the vertical line) where the falloff should be farther to the right.

The VirtualDub Settings Guide says, "slight spillover at left is usually OK", but I think there was too much for this particular frame. The DV version of the video doesn't have the same frame, but a nearby frame has dark gray falling off farther to the right, well before it hits the luma 17 vertical line:

Attachment 14996

As a result, the VHS capture has less dynamic range than the DV video. Hopefully this happened at the time of capture and so all you have to do is adjust your brightness/contrast settings in VirtualDub and capture again.

How can I open the histogram dring video editing? During capture mode it is trivial, but by editing I do not find it. Is it a filter? if yes, which one?

Traal: thanks to shared with me these informations, it is really helpful!

The result is mesmerizing, and as I see know, the key is to make white from the yellow :).
May you share with me, how to play with colormill and color balance? I spent time check both, but it is not trivial, what is the direction with them.
I just checked NV, and I decided to look for another alternatives :). Could you advice something noise reduction possibility what is free, but not too far from NV in term of quality?

Here's the saved Virtual Dub script file (I'm using Virtual Dub 2 here):

https://drive.google.com/file/d/19wl...ew?usp=sharing

Open Video1 in Virtual Dub then go File>Load Processing Settings and browse to that "vdscript" file.

I've left both the colormill and the 6-axis color correction filters in; just tick or untick to see their effects. Open up the filters to see what I've set them on.

Re noise reduction, I've always struggled with noise in Virtual Dub. AVISynth can do wonders with noise filtering and will certainly match, and exceed, NV's capability but I've never had the time to sort through all the filters and work out how they work. For me, Neat Video is money well-spent.

Thanks for it, that is a good way for me to learn.
Unfortunately I have an issue with virtualdub2 plugins. 6-axis seems ok, but maybe colormill is crashing since with your setup I see a "blue" picture (I'm using the original VD2 pack from sourceforge.net, and colormill is not a basic plugin, I made a copy from VD 1.9.X.colormill.vdf., and that is maybe crashing. What is your advice for VD2 plugins?

That's a bummer. Make sure you don't have both selected; when I have both selected, there is a blue haze over the whole srceen.

Here's the colormill I'm using in VDub 2:

https://drive.google.com/file/d/1V_f...ew?usp=sharing

My VDub 2 Help screen is attached.

I'm just starting to convergate to the same conclusion :).
What is your workflow with neat video? Do you deinterlace it before, so using 2 filtering "steps"? I would like to deinterlace with Avisynth. NV is the last step in the workflow? Thanks to advice based on your experience.

Yes, I deinterlace first, either with QTGMC in AVISynth (if I can be bothered!) or the VDub internal deinterlacer (Yadif and double TFF [yours might be BFF, determine field order using this technique]), colour adjustments then Neat Video.

Re deinterlacing in AVIsynth, there's no extra steps involved because VDub will open the file "already deinterlaced" (I think you know how AVISynth works).

Re NV, add the NV filter, then hit Build Noise Profile.

Hit Progressive (assuming you've deint prior to the NV filter).

If possible, find a frame that has a section of "constant" eg sky. Drag out the selection box and hit Build Profile.

Hit Adjust and Preview, then over on the right, you can adjust the noise reduction settings. The main ones are in the Temporal Noise section: the Noise Level and the Noise Reduction amount.

When happy, click "Apply", then Close.

One thing that is not logical (to me) is saving the Neat Video settings in VDub for later. When you're going to exit VDub, open the Neat Video filter and do a Profile>Save as... and Filter>Save Filter Preset... I haven't worked out how all that works but that will mean, when you save your VDub Processing settings, you'll keep your NV settings and won't get warning messages from VDub.

I don't think I'm scratching the surface of what NV can do but I am very happy with it as is.

Yes, for VirtualDub I use a filter called Color Tools by trevlac.

I see no Y levels below 12 in video4.avi video, but no crushed blacks:

Attachment 15037

Attachment 15036