Brightness, contrast levels for studio tape captured?

[Feedbucket]

A philosophical question, I guess, for people who've spent time capturing studio-produced tapes. I recently picked up a JVC SR-MV45 and I'm noticing that it renders luma a bit brighter than the HR-S3912.

This got me wondering how safe it would be to just capture things without adjusting the proc amp settings (AIW 2006) and leaving everything at 128 - I guess with the MV45 in particular. With the S3912 I bumped the brightness setting up to ~145 to prevent crushed blacks. With the MV45, in my quick unscientific testing so far (sample size of 2 tapes), the luma for everything including the head switching noise seems to sit between 0 and 255.

Curious to hear others' thoughts if they have any kind of experience with the above. With a prosumer unit, is it generally safe to assume that studio-produced tapes will have levels that are in range?

[lordsmurf]

Is the MV45 lighter than source, or is the 3912 darker than source? Because those lower end S-VHS, and especially VHS, are known to not be 100% accurate. (The 3800/4600/4800 were highly accurate, the later 29xx/39xx/59xx not so much.)

AGC may vary on the MV series vs. HR series. I've never compared in detail, but possible.

No, that's not a safe assumption. Not at all.

Most professionals (and hobbyists too!) will find the overall best exposure, and capture wholesale. Nobody ever wants to pay for scene-by-scene type adjustment, unless it's NASA or Bigfoot footage. Nobody does it. (Unless it's a hobbyist with only a few tapes, and unlimited time.) Only when scenes have sustained stark changes are partial re-captures required. As long as the capture values are not a total mess, post-capture color correction is what happens. Pretty much everything needs mild restoration anyway.

[Feedbucket]

Quote:

Originally Posted by lordsmurf

Is the MV45 lighter than source, or is the 3912 darker than source? Because those lower end S-VHS, and especially VHS, are known to not be 100% accurate. (The 3800/4600/4800 were highly accurate, the later 29xx/39xx/59xx not so much.)

It's not clear to me at this point which is most representative of the source since I haven't done any kind of comparisons with a reference. The MV45 appears lighter than the 3912. I guess my question is whether the MV45 can be (or ought to be) considered more accurately representative of the source than the 3912?

Quote:

Originally Posted by lordsmurf

Most professionals (and hobbyists too!) will find the overall best exposure, and capture wholesale. Nobody ever wants to pay for scene-by-scene type adjustment, unless it's NASA or Bigfoot footage. Nobody does it. (Unless it's a hobbyist with only a few tapes, and unlimited time.) Only when scenes have sustained stark changes are partial re-captures required. As long as the capture values are not a total mess, post-capture color correction is what happens. Pretty much everything needs mild restoration anyway.

Sure, and this is really what I was getting at. I'd like to not have to keep changing things on a per-tape basis - it's not the way I want to spend my time. I was thinking that if the default MV45 output can be considered standard, than it would be a good boilerplate place to start any future captures. If it works, it works, keep on doing it until it doesn't and then figure things out from there.

[aramkolt]

Unpopular opinion, but if you capture with a full 10 bit chain, you can purposely capture at lower levels and have more freedom to adjust after the capture without losing as much color/brightness information.

I'd propose that a 100IRE ramp that is pre-scaled down to 50IRE (with an analog proc amp) and then captured in 10 bit (and then later scaled back up to 100IRE in post production) will still be much smoother than a 100IRE ramp captured at the correct 100IRE in an 8 bit capture.

10 bit has 1.07 billion possible values per pixel whereas 8 bit has 16.7 million possible values for each pixel. So capturing even at 50% of the 1B colors and then expanding back up to the correct brightness still gives 500 million possible unique values in post which is still far greater than the 16.7 million possible values in a best case scenario with 8 bit.

10 bit is usually overkill for SD video, but that is less true if you plan to do post production color/brightness adjustments, particularly in very dark or very uniform scenes like sky or otherwise low dynamic range scenes.

[mts1]

10-bit capture gives 4x better precision and lower digital noise for your luma/chroma, but not the wider range.
So, if your gear is not calibrated and it gives you wrong brightens/luma, then 10-bit is not going to help you in any way.