Optimizing AIW proc amp levels for every capture
Is there any way to log the histogram in VDub? Or is there a way to enable that same histogram view on already captured footage?
Approximating proc amp settings based on a few minutes worth of footage per tape doesn't seem like a well-thought out plan, when a tape (in my case, a single capture) could be up to 6 hours in length. Basically, my idea is to set the levels very low at a known fixed value and log histogram data with timecode as a "1st pass" through the tape, akin to indexing a DV tape. This would reveal the maximum values for the entirety of the tape, which could then be used to scale the levels in preparation for a 2nd (video capture) pass. What do you guys think? Is this feasible? Realistic? Is it even worthwhile? Also, if the editing, filtering, and final encoded product are all planned to be done in the YUY2 colorspace, is there still any reason to confine the capture to the "safe" (16-235) range? |
Quote:
vhs test 01 - no tbc.avi and vhs test 01 - v10u tbc.avi have unsafe brights. If the frame hopping isn't actually on the tape, you should start your capture earlier in the tape. If the tape itself really does have these disturbed frames at the start, there's not much you can do about it. You shouldn't try frame-perfect "editing" during capture. It won't happen. Quote:
If the tape dubs are all you have available, that's what you'll have to work with. Quote:
Histograms don't "fix" problems. They just show graphically what's happening. Quote:
One can set levels for worst-case scenarios and tweak the results later. Most users do this. For my sister's insane home videos I have to make two captures, one with relatively normal levels, another with compressed levels for offending scenes, then tweak and assemble them later. Now you know why Hollywood spends exorbitant amounts of time setting up scenes before exposing a single frame. Home camera users don't do that. They just aim, pull the trigger and expect miracles, usually in available light (also often known as "available darkness"). That's why home video transfers are a nightmare. 6 hours of capture in one stretch is too much for most people to manage. If you expect to make a single capture that is perfect in most respects for all levels, color, and whatnot, you'll never finish capturing those tapes. You have to post-process VHS anyway, whether it's level-safe or not. Quote:
There are reasons why y=16-235 is known as "video safe" because of the difference in the way video is stored (YUV) and the way it's displayed (RGB). During display, y=16-235 is expanded to RGB 0-255. If your YUV video is already 0-255 or broader, the extremes at both ends are clipped (i.e., destroyed). If you want 0-235 RGB for computer display only, there are methods in Avisynth for going with that using the "PC609" matrix. On encoding to YUV, most out of spec values are ignored or clipped. Websites like YouTube will clamp YUV at 16-235 for 0-255 RGB display on the 'net. If you submit RGB 0-255 to a TV station or closed circuit entity for telecast, it would be rejected. Out-of-spec luma and chroma values seen on TV will be clipped and can cause visual disturbances that your TV can't handle -- or If not "disturbed" it will otherwise look like crap. |
Quote:
Quote:
Quote:
Quote:
I know that levels which were clipped during capture are gone forever. This post was entirely about using the histogram as a tool to prevent that from happening. The idea is to play the entire tape and record the histogram's values throughout. This could be data logged as text in tabular or CSV format, or it could be a video file that has been screen-capturing the histogram while the tape was being played. This initial playback would not be used for video capture, but rather to prepare for capture on the next run. This would provide a temporal "blueprint" of the tape's levels, which could be reviewed/scrubbed through in order to know where to properly set levels for capturing on the second playback. I do not expect a "perfect" capture of a 6 hour tape. The goal is to set the levels at an ideal spot; avoiding being overly conservative with an unnecessarily low contrast. Quote:
|
Thanks for your additional notes. I thought I'd read through all your posts but I apologize if I seem to have missed anything. Overall the good captures are very workable. Eventually you might encounter a tape that's so difficult to play properly it would require a full-fledged external frame tbc more powerful than anything you have now. But in that case it would be a tape that requires greater effort in other areas as well.
|
Was this thread split from another one? The quoting is confusing since only some of the posts are in this thread.
To answer the main question: Yes, it is possible to do a 1st-pass capture with the proc amp intentionally set to washed-out values allowing for extremes, and to then run an Avisynth script that creates a CSV which can be analyzed to check the distribution of dark/bright scenes. Once upon a time, I graphed all the frames of a Blu-ray (the image links are long-dead). I tend to agree with Sanlyn that this is less useful than you would think for a 6-hour tape with different extremes. |
Quote:
Thank you for the legit response to my thread! While I have to suspect you are both right regarding its usefulness over a 6-hour capture, I still believe this could be helpful for planning logical points for breaking a tape down into smaller optimized captures. I know I'm probably going way overboard here, but now that I know my idea isn't falling on deaf ears entirely... imagine how far you could take this:
While this may sound like overkill... if achieved, it could potentially eliminate the need for manual scene capturing altogether. Plus, additional time would be saved if/when doing additional captures from multiple playback devices, as they should be able to take advantage of the 1st pass data as well. |
People have been trying to do this automation trick for years, and failing. Good luck.
|
Site design, images and content © 2002-2024 The Digital FAQ, www.digitalFAQ.com
Forum Software by vBulletin · Copyright © 2024 Jelsoft Enterprises Ltd.