Comparing JVC VCR settings? (TBC, picture mode, calibration)
7 Attachment(s)
This will be fun. I've attached capture samples - the same 10 second clip captured several times - showing each possible filter setting on my JVC SR-VS30.
Please share your opinion on which setting you believe to provide optimal quality and hopefully minimize the amount of restoring needed in Avisynth and VirtualDub. All ran through my TBC-3000 (with all proc-amp settings at default) via S-video cord to my capture card. The various settings in the VCR menu include: Digital TBC/NR, Picture Control, Video Calibration, Digital R3 and Video Stabilizer. Let me know if you want me to upload any different setting combinations. I did not use Video Stabilizer at all, instead opting for Digital TBC/NR on all examples except for the first (control) sample. Sample 01: No filters (not even Digital TBC/NR), Picture Control: edit, Sample 02: Digital TBC/NR: on, Picture Control: edit, Video Calibration: off, Digital R3: off Sample 03: Digital TBC/NR: on, Picture Control: edit, Video Calibration: on, Digital R3: off Sample 04:Digital TBC/NR: on, Picture Control: edit, Video Calibration: off, Digital R3: on Sample 05:Digital TBC/NR: on, Picture Control: norm, Video Calibration: off, Digital R3: off Sample 06:Digital TBC/NR: on, Picture Control: auto, Video Calibration: on, Digital R3: off Sample 07:Digital TBC/NR: on, Picture Control: auto, Video Calibration: on, Digital R3: on |
Part of this will be for you to understand what the settings actually do.
Picture Mode = NR - edit = OFF - sharp = minimal NR + edge sharpening (causes halos) - soft = heavy NR - auto/norm = standard NR, mostly to remove chroma, some softening may occur (depends on source) Digital R3 = edge correction, without actual sharpening (causes halos) VHS is not precise. The signal recording wobbles. Calibration tries to navigate this imperfect path to give the best playback. It should never be used for recording, and is hit-or-miss (usually miss) for other JVC decks. This specific model is one of the few where it helps. Also read: http://www.digitalfaq.com/forum/vide....html#post9965 So, really, the only thing you should focus on is picture modes. edit = NR off auto/norm = NR on Nobody ever disagrees on these other settings. - R3 = OFF - calibration = OFF (usually) ... but not for you, not this model - sharp = never useful - soft = generally useful only an animation You need to remember that NR is needed to allow for compression in later steps. So you either do it now, or later, but it's going to happen. This is really good, and realtime. Later is slow, software, and something like chroma is harder to deal with in software. It's all about choices. I'll look at the samples a bit later. |
Thanks, that is helpful.
I also found it interesting how certain settings created a larger file size, and some a smaller file size, than the one without any settings selected. (You will see how I had to shorten the larger files to fit under the 99mb limit) Quote:
|
- For lossless, noise makes larger files.
- For compression, noise requires more bitrate -- or create more artifacts. Larger files only if CRF. And as you're seeing it's generally at least 10% difference, if not more. |
2 Attachment(s)
Quote:
Thanks for the samples. Quote:
IMO 02 and 03 are your best bet for detail retention and overall definition after some mild to standard post-processing -- and in post you can get far more sophisticated and far less destructive than you will with a VCR's primitive filters. 04: Digital R3 kills this one. You'll never get rid of the halos (which result from improper edge sharpening) or the annoying buzz from sharpened noise. With 05 things get foggy, contrast looks relatively curtailed and the image lacks snap. Fine detail is visibly disappearing. The face of the little girl on the floor looks out of focus. No post-capture sharpener can retrieve lost details or textures. 06 and 07 represent typical destruction from these types of players. The picture is denuded and hazy, motion smearing gets worse. The little girl on the floor looks as if she's wearing a silk stocking over her face. This kind of damage can't be repaired and isn't for archiving, mainly because the results are avoidable in the first place. The attached 03_sample_DVD.mpg presented some color problems. JVC does odd things to color sometimes, especially reds, and so does cheap s-video wire (the latter due to uneven impedance curves and transmission noise). A Panasonic has cleaner reds but which unfortunately get pumped-up out of proportion. VHS itself is anything but pure . Then there is color reflected off the walls and the cyan daylight from the background window. The overall color cast was yellow-green, with even more green in skin shadows. The corrections seem tolerable except for the dark green of the plant in the left corner -- but, then, it's usually best to correct for skin tones. If your plants look pretty but the people are the same green, something's wrong. Try a trip down memory lane to recall how the walls and furniture looked back then. The sample's color balance changes from cool green at the start to warmer when the bright window leaves the picture frame. And gamma's a littler tricky because of lens flare. Some color elements could be tweaked here and there, such as the boy's hair which still looks a bit green at times. But VHS color is impossibly murky anyway. I corrected in VirtualDub with gradation curves, Colormill, and temporal smoother applied to the Avisynth output while the script was running. The attached .vcf file has the VirtualDub settings. Relatively moderate denoising was used. Code:
AviSource("E:\forum\faq\ragu0012\A\03 - Video Calibration on.avi") |
Thanks Sanlyn. Quick question.. did you change all your .avs scripts to .avsi? I remember in the past you said you would put a load command at the top of the script because the autoload avsi were troublesome, but I don't see that here.
As for the video, I believe there is a florescent overhead light that is probably causing issues with the color in the room, especially when combined with the daylight it is competing with. Believe it or not, those walls are definitely supposed to be white. Many of my home movies are shot in this same living room, so I have to deal with the issue in many of these tapes. Is my cheap s-video cable really working against me?? Quote:
|
Quote:
Quote:
Quote:
But I don't think changing cables would make a great difference under most circumstances for most people. Variations are different but subtle. It's difficult to make conclusions without precisely calibrating monitors and TV's using optical probes and calibration software, and working with and watching the same few problematic videos dozens of times (or more) using different wire and high-end players. Testing includes asking others for their detailed responses to different viewing setups. And as I say, the devil is in the details -- big and small, every component affects the sum of the capture to one extent or another. |
Lordsmurf I'd still be interested to hear your take as well.
|
3 Attachment(s)
Replying to this old thread where it's more relevant, so that I can keep the pics where they belong for future reference. Gonna link here from the current sorta-OT discussion in another thread.
These are individual fields from 3 of the posted videos, with the two chroma channels shown underneath. "Raw" version. Attachment 10820 EDIT with TBC/NR. Lines are straightened. There is no luma NR. There is mild chroma NR. Looks like temporal averaging and maybe some spatial. Attachment 10821 NORM with TBC/NR. Now spatial & temporal NR are both in effect, for both luma & chroma. The chroma averaging now includes more frames, which causes some trailing of colors during scenes with large movement. Her face appears to lose some color because it's been blended in with the color of the objects that were in that pixel position in the previous frames. Same with the table. And her hand gains a color ghost of its previous position. Meanwhile, the luma NR is mainly to blame for the disappearing couch detail. Attachment 10822 The differences are most obvious if you open each attachment in a new tab and flip between them. The Avisynth script I used to compare and grab the images: Code:
V1 = AVISource("01 - no filters.avi").Crop(2,0,-0,-0).AddBorders(0,0,2,0).AssumeTFF().SeparateFields() |
Site design, images and content © 2002-2024 The Digital FAQ, www.digitalFAQ.com
Forum Software by vBulletin · Copyright © 2024 Jelsoft Enterprises Ltd.