althor1138 said:

What I mean is when you entered your captures into the script, was it cap1 or cap2 that was the darkest of the 2 captures.

The posted clip did not use your overlay script.  Some of the scenes were from the bright capture, some from the dark capture.

I sent you a PM with what is hopefully a better explanation of what i was trying to explain lol.  BTW, the clip you posted looks pretty good already I think. Nice work!

althor, I'm curious, the script in xvid4psp says "directshowsource"... is there a reason that it doesn't say "avisource"?  Does it matter?

lol I'm not really sure.  I'd say if the video looks correct then it doesn't matter.  What is xvid4psp anyway? Does it directly support avs files?

Puggo - Jar Jar's Yoda said:

althor, I'm curious, the script in xvid4psp says "directshowsource"... is there a reason that it doesn't say "avisource"?  Does it matter?

Generally avisource is considered more frame-accurate, so use directshowsource only as a backup option if avisource is giving problems.
But there is probably very little difference.

Puggo - Jar Jar's Yoda said:

 If there were a way to have it choose pixels based on which pixel is less extreme (that is, pick the pixel closer to the midpoint, which should use the lighter source in the dark areas, and the darker source in the light areas - rather than averaging them), then it could probably be used everywhere.  Hmm, I wonder if anyone has ever written a script to do that?

I actually think I've found a way to do exactly this in avisynth using only the mask,showalpha, and invert filters.  Here are some pictures to demonstrate.  It is an extreme comparison because I brightened one image considerably just to demonstrate what is happening.

Normal frame

Masked bright frame overlaid on top of normal

Bright frame

The image doesn't look great because I way overbrightened the bright frame to intentionally blow out the whites to show that it successfully only brightens the dark areas from the normal pic.

If you are interested let me know.  For now though,  I must do some housework or the wife will kill me when she comes home.

That's a promising idea.  I was wondering if there was a way of taking the bright clip and chopping out all information above 128 luma, taking the dark clip and chopping out all information above 128, and then simply adding them together.

Or maybe something slightly more sophisticated would be to do the above on the extremes (say, above 192 and below 64), and blending the center range.

Of course, there might be some strange effects I haven't considered.  If there is a way of "chopping" above/below a certain luma with some small degree of fuzziness, that would probably reduce the possibility of artifacts.

Puggo - Jar Jar's Yoda said:

That's a promising idea.  I was wondering if there was a way of taking the bright clip and chopping out all information above 128 luma, taking the dark clip and chopping out all information above 128, and then simply adding them together.

Or maybe something slightly more sophisticated would be to do the above on the extremes (say, above 192 and below 64), and blending the center range.

Of course, there might be some strange effects I haven't considered.  If there is a way of "chopping" above/below a certain luma with some small degree of fuzziness, that would probably reduce the possibility of artifacts.

I'm not positive but I believe this script is doing that.  I create a mask of the brighter clip and then invert it so that only the darker areas get overlaid onto the darker clip.  This leaves the bright areas of the dark clip untouched while passing the dark areas of the bright clip onto the corresponding areas of the dark clip. If i understand correctly grey pixels in a mask allow the pixels from the image to pass through with an opacity based on how close they are to white. This should mean that the middle ranges are a blend of both.  I've been checking this method by using the gout and a bright version of the gout and i've not seen any artifacting spatially or temporally.

I definitely would like to try it - can you post your new script too?  (when it's safe, of course)  :)  thanks!

ok I've edited the script in post 139 to incorporate the mask/overlay method.  You should be able to just open the second part of the script and try it directly.  The overlay in the variable avgcap1cap2 is set to mode=blend.  I'm not sure if that is best so you can try to change it to luma as well.  I don't think it will make that much of a difference though.

*EDIT*

If it's mixing the wrong parts of the images just let me know and I'll switch things around in the script.

Harmy said:

I would recommend using 23.976fps instead of 24fps, because 24fps can cause playback problems. Ady made his EpIV AVCHD 24fps and I know there were people who had to remux it to 23,976 in order to play it.

I've also read that playback of true 24fps can be problematic. I believe that the majority of Blu-ray Discs are encoded at 23.976fps?

Since a 0.1% difference in speed is negligible, and Puggo will be re-syncing the audio anyway, it would be wisser to go with 23.976fps. The AviSynth command is then:

AssumeFPS("ntsc_film")

Moth3r said:

Since a 0.1% difference in speed is negligible, and Puggo will be re-syncing the audio anyway, it would be wisser to go with 23.976fps. The AviSynth command is then:

AssumeFPS("ntsc_film")

Ooh, thanks, I was wondering about that.

thorr said:

d020 said:

Why would you like to add black bars? As said before, it´s not necessary in MKV. The test video looked great on my PC aside from the pulldown. In my opinion it only would blow up the file size and make it more complicated to edit the video or using anamorphic lenses on beamers etc.

But if there´s a good reason to do so maybe you could do a version without.

 

The reason is so you can play it on a blu-ray player and remove the requirement to use a PC or other device.  Adding black bars is negligable in terms of file size (it's all black and never changes), and also, it does not make it more complicated with anamorphic lenses at all.  It will be the same as playing any blu-ray.  On the other hand, if it was encoded without the black bars, it would need to be re- encoded to be blu-ray compatible (extra work for Puggo, or video loss for us if he doesn't do it to begin with and leaves it up to us to re-encode).  I am not 100% sure if what I am saying is true about the 720 requirement  because I haven't tried to create a blu-ray (actually I mean an AVCHD DVD) from Puggo's file yet to test it.  Maybe it is not needed.

My 2p's worth - if you're going to add the black bars to make it suitable for authoring to AVCHD or BD9, you may as well go the whole hog and release in that format. I don't know how much effect adding the black borders would have on the bitrate available to the video - maybe it is negligible. But those with BD players or PS3s can simply burn and play.

On the other hand, those who have alternative playback arrangements (Popcorn Hour, WDTV, HTPC, jailbroken Apple TV/XBMC, or even just hooking a laptop up to  the TV) would be happy with a MKV without the black bars.

Well right now he's just releasing sample clips, so I don't think it's going to be a concern for a while ;)

I hope there is a DVD i cannot watch HD on my PC and i don't have a PS3 or a blu ray player and have no intention of buying one.

If there is not I can also understand since this is not my project and i have no say.

If need be I can dvd-ize it.

I bought a blu-ray player a couple of years ago so I could see the restored version of 2001.  If that were the only movie to ever come out on blu-ray, I'd still think the purchase was worth it.  :)  And it plays DVDs better than my DVD player.  Today they're a third of the price than when I bought mine.  Even as much of a luddite as I am, I can't think of a reason not to get a blu-ray player these days.

That said, I'm sure it will be easy to make a DVD version of it.

Perhaps I should make a thread on it in the technical section (i did a search but came up empty) but is there a way to put out a an HD version that can be burned to disc and easily watched on any standalone bluray player without having to actually burn a bluray (to my knowledge blu ray burners still aren't all that common, or cheap)?  I've seen the term avchd or something mentioned here before & get the impression that might be what it is, but I really don't know.

canofhumdingers said:

Perhaps I should make a thread on it in the technical section (i did a search but came up empty) but is there a way to put out a an HD version that can be burned to disc and easily watched on any standalone bluray player without having to actually burn a bluray (to my knowledge blu ray burners still aren't all that common, or cheap)?  I've seen the term avchd or something mentioned here before & get the impression that might be what it is, but I really don't know.

Yes. It's called an AVCHD, and most blu-ray players will play them. Typically a dual-layer DVD is used.

Time for an update...

althor1138's script for combining captures has proven to be a superb tool, and we've tweaked it to the point where I am likely to use it for many, perhaps most, scenes.

Here is a new (silent) sample, this one about 50 seconds with a variety of scenes (about 49 Mb):

  http://rapidshare.com/files/455048968/demo2d.mkv

It was made using the following steps, in this order:

1. Vegas: very slight color correction, mainly to get the clips to match,
2. Avisynth: althor's two scripts to combine captures,
3. Vegas: addition of border,
4. Xvid2PSP:
     changed CRF to 18,
     Lanczos resizing to 1440/480;
     soft pulldown using assumeFPS("ntsc_film");
     very slight denoising using FluxSmoothST;
     very slight sharpening using LimitedSharpenFaster.

There was also a very slight extra sharpening on the two scenes from reel2 (Dagobah and Bespin), during step 1 above, which I probably won't do in the future. I set the horizontal size to 1440 because although the stills had a width of 654, the video has a width of 720 (i.e., I think the aspect ratio was incorrect in previously posted clips).

If anyone would like to suggest a better alternative for noise reduction and sharpening for this application, I'm open to suggestions.  Keep in mind that one of the main reasons for doing smoothing is not to eliminate grain, but to reduce encoding size.  Setting CRF to 18 made a noticeable improvement, but doubled the encode size. FluxSmooth brought it back down by half that amount with imperceptible effect on the video.