I can send you a 1 minute snippet for you to test/process - I need the absolute best result for this project - money no issue if results are excellent
+ Reply to Thread
Results 1 to 16 of 16
Send? Why not link to to your 1-minute sample in this thread? Or is it a large lossless AVI? In which case you might make available something shorter and smaller.
And you might explain your capture process. The chances are good you've already ruined any chance for an "absolute best result for this project".
I don't really know but......
1. Where is the sound that the clip preports to include ? Could it be the 24 bit PCM whereas typically this would be 16 bit or even 12 bit ?
2. I have NEVER* witnessed such interlacing artifacts in an interlaced (VHS) source. There is no report as such but could this have been captured/converted as progressive since in my early capping days days* I did then witness it and alltered my capture workflow accordingly. If this is so then the capture is compromised already.
I split it into fields, and the attached picture shows one field from the video. I see no interlacing or other artifacts.
As for improving it, I might be tempted to do a very small amount of gamma correction, to make it a little brighter, but that's about all I'd do. It looks like a pretty standard VHS tape capture and I don't think the OP could do a whole lot more to improve it.
Last edited by johnmeyer; 9th Jul 2020 at 10:51. Reason: added quote
I think it's the same clip (no evidence of the OP uploading anything else)
I just could not get my head around all the interlacing artefacts >> combing.
And, again, where is the sound ?
But this 'dangling' of money since all on here would suggest means for the OP to do the improvements himself as 'BEST' is a relative remark and anyone taking the time and trouble to improve the whole thing on a promise could end up with no reward anyway.
Cropping the crud and applying some denoising (MCDegrainSharp) doesn't do any harm, I think. Denoising strength is a matter of taste though, as well as leaving it interlaced rather than bobbing.
I am now on my main editing computer so I took another quick look. I won't change anything I said in my previous post, but now that I've looked at it more critically, there is a LOT of sharpening, and also a lot of chroma bleed/bloom. If there is a chance to re-do the transfer, the OP needs to make sure that the Edit Switch on the playback device is turned ON. This turns off the deck or camera's sharpening circuits.
The attached photo, from a single field of the video, shows the two problems.
Last edited by johnmeyer; 9th Jul 2020 at 18:12.
I agree with johnmeyer. The video is normal interlaced PAL (top field first) and needs to be re-captured with the sharpener turned off and the TBC enabled (or a better line TBC).
To the gentleman above, there is no 'dangling' of money - you will invoice me and I will pay you, up front if required
There is no audio, as I am looking after that part of it. I just want vision fixed.
EDIT mode was in fact ON for this capture. I tried it with EDIT OFF and the results were identical.
This may mean the footage is not first generation, and the problems may be baked into the tape?
Here is another capture with FS100 > S-video > DataVideo TBC4000 > Osprey > AVI
This second capture in post #10 is inferior in every way to the clip provided in post #3: it has less detail in the grass; the white levels are almost clipped; and if you watch the left side of the screen, just beyond the edge of the picture, you will see obvious time base errors rippling down that left edge.
If you can figure out how to avoid those sharpening halos in your first capture setup, and if you can improve the midtone levels a bit (if your capture hardware has a proc amp), I think you can actually get some really good captures.
So, setup #1 is definitely better in every way to setup #2.
The first example has lots of ugly sharpening artefacts besides the halos and ringing, as I see it. These might be seen as 'details' though. The second example is softer and looses perhaps some details in the grass (if this matters to the viewer of the game). I would just add some grain to the 2nd example, like AddGrainC(3.0), to make it look more 'crispy'.
Thanks for responses so far.
That first capture is as good as it gets - I've tried everything.
So if anyone else feels like having a crack at minimising all the problems, be my guest.
I will then decide based on samples received if I go ahead or not, and you can then quote me on the whole job.
I don't have the time nor experience to attempt any of this myself.
Well that's the first time on here that someone has called me a 'gentelman'
I have not looked at the second sample - I am, alone it seems, concerned with another aspect of the first one.
But there are people more experienced than me in these matters (I have only been transferring VHS since the late 90's) but as I said before I only witnessed such interlacing artifacts when an interlaced sourced was actually captured as progressive (even if it was flagged as interlaced). And I am even more surprised to read that the second sample, with additional equipment in the chain, is worse.
But I see another potential hurdle in your attempt to have this capture 'restored'. I'll keep the maths simple although the net result will be a little larger. Based on your first sample, one sees 10 secs of video for 200 mb. 1200 mb for 1 minute. 72000 mb for 1 hour. 144000 mb for 2 hours. So, putting it into context can you honestly expect anyone to download >> 140 gb ? Or maybe you have an alternative proposal for getting this to the restorer. It may also be useful to anyone who might be considering this task if you were to state your preferred delivery format. But if you are going to process the audio yourself then my guess you want it like for like else the result would be re-encoded with a loss of quality.
A much better option in my book is to have the VHS recaptured by the restorer - PAL based will restrict your options though - or even have that done by a professional transfer house assuming that this is a non-commercial tape.
And, finally, uncompressed is not the be all and end all. A lossless capture will be a fraction of the size and be of equivalent visual quality.
I do get somewhat tired of armchair restoration experts, a term I generally retain for those that just blast errors without considering the source. (BTW, not referring to anybody specific here, just a general comment.) These are tapes, after all. The source wasn't some pristine digital file that poor video editing has butchered. I get really irritated after considerable work, in making something enjoyably watchable, only to have some @sshat complain about how it's not perfect (and yet, far better than it existed on the original tape). I know it's not perfect. Duh.
Pointing out potential capture/hardware issues is fine, but (as an example) things like haloing is often very baked in with nth gen VHS tapes.
Yes, it can be caused by a VCRs (or even TBCs, other hardware, etc), but I find this far less common on nth gens. And Panasonic decks do have a tendency to oversharpen at default settings, leading to halos (or more/intenser haloing). JVCs and Panasonics often give different output for both luma and chroma, sometimes sharpness, often with neither looking better (just different).
EDIT mode often has zero effect, and indeed can make image quality worse. Again, subjective. My deference is to AUTO(NORM) modes on both JVC and Panasonic, as the quality is least-worst to better.
This is all stuff that must be considered.
Restoration is an art, and subjective choices often have to be made. Perfection is not possible. For restoration, as intended for watching, choices are made. For forensic recovery, not really "for watching", entirely different choices must be made.
So retaining sharpness, luma, chroma, reducing noise, etc -- ALL AT ONCE -- is simply not going to happen. You must do as best as possible, with ranked priorities of importance. Sometimes folks want "clean", others "sharp", yet others "no noise". You have decide which takes priority.
Those who would insist you can have everything ("have cake, eat it too") really have no extensive experience in this area.
Avisynth advances have helped in the past decade. Things once considered impossible are now possible, though to varying degree. Again, choices must be made, ranked.
In years past, I'd sometimes have to capture and/or restore clips multiple times for people, and they were never happy, instead chasing wild gooses, the untameable unicorn of unachievable video perfection. It's just not going to happen -- ESPECIALLY with an nth gen tape (copy of copy, ad infinitum, ad nauseam).
Sometimes restoration itself gets ad nauseam. Learn its limitations. Perfection ain't it.
TL;DR: garbage it, less-garbage out, is the goal of restoration.
Make it better, not make it perfect. (Though close to perfection can be a tenable goal, sometimes.)
Best result for what playback device?
TV? Pc ? iPhone, utube?