How to determine what to improve on TV recordings

[Prelude]

I have been, like many before me no doubt, been playing around with my capture setup to just get some content to start playing around with. Perhaps not the best workflow but I personally need to get a feel for things so I tend to do a bit of everything and return in the end once lessons have been learned (if I ever )

I have recorded a 5 minute clip from a TV broadcast from the second half of the ninetees and by then fast scene cuts and lots of special effects had already become the norm especially in the 5 minute preview for that evenings special I just recorded.

My problem is that from cutscene to cut scene there is a wide variety of problems. It could be just sloppy editing back then or the problem could be in my capture, the VHS tape, the recording and so on. Although I could cut the entire recording up into their respective clips and process them one by one I do not feel like making my own "despecialized" version here

5 minutes of uncompressed video is way to big to post or download for anyone so I'll just make a few screen grabs and take it from there. I know that avisynth and virtualdub are the defacto standards here but for now I am using FCP to work on my video's, although using avisynth to process a video is not a problem since I can run that in vmware.

I have only uploaded images from the same show out of the clip. I could also view a histogram or vectorscope but I am not sure what is preferred. It seems to me that although there is a lot of blue and red in the picture that the value's should not constantly exceed 100? Also, it's seems that the blacks are not dark enough. If I take however a scene which looks fairly ok to me in terms of darkness and I apply that to the entire recording, there are other parts that look off to me.

So, how do I judge a recording of a tv show and how do I decide to clean something like that up? (and how). Is it basicly a question of picking a middle of the road solution for such material (this is over 4 hours of video) and reserve the scene by scene restauration for my wedding video for example, or is there a better approach?

Thanks!

-P

[sanlyn]

Can't help you with FCP, mate. Your screen grabs are a nice demo of your FCP display, which appears to need adjustment. FCP isn't budgetware, so surely it must allow images directly from video frames. You shouldn't have to upload 5 solid minutes of video, when a few seconds of a segment or two would do to help you get a feel for things. I thought FCP could make smart-rendered edits? You'll get better results if someone can look at video instead of an uncalibrated display. Sometimes you have to work in a video's initial YUV colorspace to fix some problems. I understand FCP can do that, as well.

For a look at how to read FCP's waveform and RGB parade graphs, try the user guide or popular mainstay books like Color Correction Handbook. That particular book makes special mention of how to work with FCP's color controls and 'scopes, such as the waveforms in your images that show problems with saturation and brightness levels.

[Prelude]

Thank you for the reply.

I realize that FCP is not budgetware. At this moment I am in the weird situation that I had to dig deep for a TBC and a secondhand videocard, money is tight, but IT stuff is not really a problem. I help a friend of mine who owns a business from time to time and he slips me some nice IT related stuff in return, be it a mac pro, headset, software license...

Anyway, I am not opposed to using the standard supported tools in this site but I was intent at translating the lessons I could learn from those platforms to others, like FCP so please ignore the fact that I try to use it where necessary just know that I try to use it in the background.

You say "appears to need adjustment" Are you refering to my display? It is calibrated actually, I use a datacolor spyder elite 4 http://spyder.datacolor.com/portfoli.../spyder4elite/ but it has been a few weeks since I last calibrated so I will do that again next week.

Or are you referring to the image I'm sorry but english is not my native language so sometimes I get confused

I am sure FCP can do a lot of nice things, I am still trying to figure it all out. I will try to extract a few seconds of video, that would be a nice first target to learn

-P

[Prelude]

I managed to extract a few clips using ffmpeg on the commandline.

The first clip is from the same part I got the screengrabs from. The second shows another problem I encounter: artificial noise. I would like to clean up the video but keep the noise as was intended in the footage. The third shows what I beleive to be to light darks. (look at the starfield) Also there is an echo next to the warp nacelle and the "O" of "go". Not sure what causes this...

[NJRoadfan]

The echo in clip 3 is from the original broadcast. That type of ghosting happens from multi-path interference or coax cable issues, particularly with amplified cable TV systems.

With noise reduction, I tend to avoid doing much it on the luminance portion of the signal as that is more likely to result in the "plastic" look. Usually the chrominance/color signal of a VHS tape is a complete mess.

Your graphs show that some clipping of the signal is happening, you have video that is "whiter than white". This is usually resolved by adjusting the capture device's brightness and contrast settings before capture, or with an external proc amp. Is it possible to show the graphs split out in luminance and chrominance?

[Prelude]

Ah, yes I understand the echo thing now that you explain it. I am also a licensed ham radio operator so I know a thing or two about coax etc. Thanks for clearing that up! I was so into the video side of things that that did not occur to me.

I have included 4 images, 3 new screenshots and a 4th with the options I can choose. I Choose the YCbCr graph but I can have luma and chroma graphs as well, just not combined.

Regarding the adjustment of the brightness and contrast; I have a AVT-8710 and it has controls on there to adjust but I beleive I read on here somewhere (LS said that I think?) that in general one does not touch those. I can also adjust the capture software (virtualdub).

If I want to (try) to calibrate the my capture card input, are the colorbars the AVT produces any good? I can capture the image produced and compare it to a calibrated chart I found online (doom9 I believe) or is this process not reliable.

Thank you

-P