+ Reply to Thread
Results 91 to 108 of 108
Thread: Are my capture settings ok ?
I'm learning nothing here but thanks for trying to school me
Looks as if you did a nice job of denoising.
Your MPEG has a max bitrate of 12180 kb/s. You gain nothing from VHS with bitrates this high, but it doesn't hurt. Needs a little color work; highlights are blown away, faces are green, but you can always fix color later. The capture below is frame 15. I resized it a little horizontally to 640 pixels wide, and added a little red to the player's face.
Your 5800 isn't tracking that well. In frame 15 look at the bent metal bar ( I overlaid a box around it), and the one to its left. Mount the video in VirtualDub and cycle a frame at a time, back and forth between frames 0 and 16. Watch the player's face and the metal bars change shape. A line-level TBC could have prevented much of this.
Good work. You made a vast improvement over the original. I learned a lot from this thread.
Last edited by sanlyn; 27th Jun 2011 at 14:16.Our inventions are wont to be pretty toys, which distract our attention from serious things. They are but improved means to an unimproved end. -- Henry David Thoreau
Our inventions are wont to be pretty toys, which distract our attention from serious things. They are but improved means to an unimproved end. -- Henry David Thoreau
I wouldn't necessarily use that script filters or identical values for everything. All these suggestions people are making are meant to be a starting point for a specific source. You should learn what the filters do and customize them for each source. For example, in this sample, you might have used different values or filters - there are areas that are blown out (have a look at the details on logos on the ice - or maybe you're that ashamed of the Habs ). It probably didn't have the exact same characteristics as the other video, so there is no reason to blindly apply the same filters or values. If things look posterized to you , there is a good chance things were overprocessed
i think this was a dub so the error was imbeded onto the tape
thx a lot for your help , it's really generous from you guys
and i still have a lot to learn
and this was a segment from the same capture , i won't live old enough to filter each scene differently ( i maybe have 500h of tapes) so i try to filter for each capture when i get some very precious tapes i'll take more time into it
Well on ice exposure is definitely going to be different lighting setup than inside a studio .
It's not that difficult to filter in segments; you can use Trim() to divide up sections . Presumably it doesn't go back & forth that often
I would argue getting the levels correct is probably the most important thing to do - you can do it either in hardware or software
A good filter to start learning about basic levels is histogram() . You make adjustments with smoothlevels, then preview using f5 in avspmod. Your goal is to put everything within the brown lines, that corresponds to Y= 16-235 , which is "legal range" . Data outside that range won't be visulized on your TV. For example, like the logos the on the ice
You can try some "auto" filters as well, like autolevels() , but they tend not to work as well as manual adjustments
(I was joking about the Habs btw, they are awesome)
Bitrate Viewer shows it's about 20,000 kbps CBR:
MediaInfo says 9800. GSpot says 9830 avg. A quick calculation based on file size gives about 9800 (27,000 kbytes * 8 / 22 sec). I usually find Bitrate Viewer to be most accurate but something about this file has it confused. For some reason it thinks the frame rate is 59.94 fps, and hence shows the running time is only 11 seconds.
Last edited by jagabo; 27th Jun 2011 at 19:23.
the capture was 20M (the first one) I redid it with huffy but once converted to mpeg i did a variable 9500/9800 bitrate
if you add the audio it's even more was 224 audio i think
isn't the max bitrate 10080 video+ audio total ? (9800 being the max for video)
looking at your graph i may have pushed it a little too much
Last edited by smartel; 27th Jun 2011 at 23:14.
GSpot, which listed a target of ~9000 VBR, a peak at ~12000.
To get 90 minutes of video on a disc at 720x480 or 704x480 with a 4:3 VHS capture, I set both my encoders and my DVD recorder to 6200 VBR target, 9000 max. I use two methods to see what I believe is a display of the actual bitrate being "used" on playback (I don't know how great a test these displays are). In PowerDVD those videos play back at bitrates that hover between 4500 and 6700, usually in the upper 5000's almost all the time, depending on action, detail, etc. My Toshiba recorder's playback display shows the same figures, so I guess both playback devices are looking at the same thing. I haven't watched every frame of those videos, but I keep waiting for the displays to rise over 7000 but haven't seen it yet.
I don't really know how "accurate" that kind of display would be. Is there a utility that gives the actual bitrate or data count in a frame, much the same way computers show file sizes?Our inventions are wont to be pretty toys, which distract our attention from serious things. They are but improved means to an unimproved end. -- Henry David Thoreau
so for me i think it's better to find an overall setting for the capture . One capture should have about the same caracteristics ( same recording environnement) i rather have a little noise left and keep everything smooth
i know i should start with the black and white levels and move my way up to color balance etc etc ..
One capture should have about the same caracteristics ( same recording environnement)
Levels is the easiest thing to screw up and the most dangerous, especially in hardware. Hard blow outs mean that information is gone, permanently. Not recoverable in software or any method, unless you capture it properly. If you capture and the levels are ok, then you can make fine adjustments in software, but this will vary by scene
I would reduce both settings for RemoveDirtMC and LSFMod if you want less "plastiky" look. It's entirely possible these are the wrong filters to use for this example.
but .. but the 1st results look very promising i'm not aiming for perfection (not yet)
i posted this capture as an example for a bad noisy source , so i could learn a little something
they are so many filter combinaisons that someone could easily get really confused
i know that trying and see the results maybe i could identify filters needed just by looking at the source
but for now it isn't my case , i'm still at the test stages and i tend to filter less