Laserdisc capture workflow. — Page 4

msycamore said:

Darth Mallwalker said:

Seems you have a different mastering than Arnie.d's X9 cap

Interesting...  I can understand that there was re-pressings made but you have to wonder why a new mastering took place of these releases. My JSC ESB proved to be a different mastering compared to Arnie.d's ESB cap as well.

It's stuff like this that makes me wary of TooT'ing three captures from three different pressings of the same LD release; e.g., TooT'ing captures from three different JSC sets, like what althor1138 wants to do.

Without a bunch of safeguards in place, you have to perfectly spatially- and temporally-align each capture. This usually isn't difficult, but how would you know if one of the three sets has a quirk (like the different mastering above) without doing a frame-by-frame scan?

I wish we had a good method for dropout/rot detection. I'd really prefer to do one excellent capture and fix dropouts/rot manually in PFClean.

I've uploaded a clip of side 1 here.  PW=OT.  I've cropped the picture, resized, and tried to do a sensible levels/tweak adjustment based on moth3r's advice. I've also sync'ed some sound to it.  I'd appreciate any feedback on it.  If everything seems in order I won't be far away from being finished.  I just have to sync the other 4 sides with audio and encode to blu-ray.  Soon I will start a thread in the preservation's forum.  I think we've gone over the capture workflow and this thread is turning into more of a Japanese Special Collection thread.  I still could use some advice on making a blu-ray.  My plan was to just use multiavchd. It seems like the best free choice.

AntcuFaalb said:

It's stuff like this that makes me wary of TooT'ing three captures from three different pressings of the same LD release; e.g., TooT'ing captures from three different JSC sets, like what althor1138 wants to do.

Without a bunch of safeguards in place, you have to perfectly spatially- and temporally-align each capture. This usually isn't difficult, but how would you know if one of the three sets has a quirk (like the different mastering above) without doing a frame-by-frame scan?

I wish we had a good method for dropout/rot detection. I'd really prefer to do one excellent capture and fix dropouts/rot manually in PFClean.

I am beginning to agree with you on this.  I'm not sure exactly what method toot uses for averaging the 2 "best" captures either.  Sometimes it seems to me like it gives the film a posterized look.

Wasn't there a scripted solution to replace the toot plugin?  I'd rather use something like that because you can at least tweak a script to fit your needs sometimes.  The toot plugin has no settings other than input clips.

Didn't realise that TooT actually averages the two closest captures. When I did some tests a while ago, I found that while a median function is best at removing dropouts, an average function is better at smooth over dot crawl.

Here is g-force's "pure" median function for three clips :

Function Median1(clip input_1, clip input_2, clip input_3, string "chroma")
{# median of 3 clips from Helpers.avs by G-force

chroma = default(chroma,"process") #default is "process". Alternates: "copy first" or "copy second"

Interleave(input_1,input_2,input_3)
chroma == "process" ? Clense(reduceflicker=false) : Clense(reduceflicker=false,grey=true)
SelectEvery(3,1)

chroma == "copy first" ? last.MergeChroma(input_1) : chroma == "copy second" ? last.MergeChroma(input_2) : last

Return(last)
}

I figured out why it was looking posterized sometimes.  One of my captures inexplicably shifts left or right by 1 pixel once in awhile.

I'm just going to use the first capture because the difference between one capture and an average of all 3 seems to be rather negligible in this case.  I'm sure it helps out in some situations but in this case it just adds layers of complication with no real benefit.  Of course I'll play with it as an afterthought but I think it's better that the preservation is as original as possible anyway.

althor1138 said:

I'm sure it helps out in some situations but in this case it just adds layers of complication with no real benefit.

I agree. There are two options for doing capture averaging/median without artifacts or resolution loss:

1  Process the captures shot-by-shot (or even frame-by-frame)

2  Get an amazing image registration program to do the work for you

I think the best solution, for now, is to do one excellent capture and work on fixing drop-outs and speckles (that aren't on the source) manually.

AntcuFaalb said:

althor1138 said:

I'm sure it helps out in some situations but in this case it just adds layers of complication with no real benefit.

I agree. There are two options for doing capture averaging/median without artifacts or resolution loss:

1  Process the captures shot-by-shot (or even frame-by-frame)

2  Get an amazing image registration program to do the work for you

I think the best solution, for now, is to do one excellent capture and work on fixing drop-outs and speckles (that aren't on the source) manually.

I think imagej using bunwarpj or something newer would actually do a great job at registration and could be easily automated and incorporated into an avisynth script via the call command.  I think the downside to that approach is that you have to do at least 2 colorspace conversions before encoding and it would take anywhere from 10 seconds to 10 minutes per frame depending on your accuracy settings.

I think I'm about to start playing around with my cheaper cld-d505 tonight and see what results it produces.  It looks less noisy but I'm pretty sure that's because the composite is filtered and recombined.

Still waiting on the Video Essentials to arrive.  The guy pretty much lied to me about mailing it and only now is shipping it out.  If it ever does arrive I'll be posting some stuff in the Laserdisc screenshot comparison thread to hopefully get some more definitive feedback on which of the 3 players I should be using for capture. I do think the HF9G will probably produce the best results but as Moth3r said earlier the real proof will come from a captured test signal pattern or some colorbars.

Here's some comparison between cards with a capture of the CLD-D505 vs. the CLD-HF9G included this time. I think the HF9G still looks better?

So I'm using multiavchd to burn this to a blu-ray with x264 as the encoding format. I use the little calculator to determine that I need around a 22mb/s bitrate to fill up a BD-25 but after the encode is finished the whole folder including menu and such is only a bit over 10gb. Am I doing something wrong or is that all that I'm going to get out of it? 

I'm almost certain that i put it in correctly. If I'd put in 2200 instead of 22000 by accident the file would have been much smaller than 10 gb I think. The resolution is 720x480p. Is there some cap on bitrate because of the resolution?

Thanks.  That link was like the one stop shop for all I needed to know about encoding to blu-ray.  I too believe that 22000 is overkill but I thought I'd give it a try. 

What I'm leaning towards is putting the raw version and a cleaned version both on the same disc so people can choose.  They'd both fit on the disc with room to spare even if they were encoded at very high quality.

I played with my Diamond ATI Theater HD 750 USB capture card earlier. I connected my Panasonic DVD-S29 (an excellent DVD player that outputs full chroma resolution via its analog outputs) and captured some patterns from my Digital Video Essentials DVD over composite.

I captured the pluge pattern, so I'll be able to determine if the AGC issue is indeed present sometime tomorrow evening.

I also captured the Snell & Wilcox Zone Plate to examine the 3D comb filter. It's not anywhere near as good as the 3D comb filter in the Panasonic DMR-ES10, but unlike the DMR-ES10, it switches from 3D to 2D smoothly, so that's a plus.

Unfortunately, the 5-line 2D comb filter cuts a lot of diagonal resolution from the "/" square, but manages to retain the diagonal resolution of the "\" square, which I find weird. What's also odd is that there's almost no rainbowing in the "\" square, but a lot of rainbowing in the "/" square and that manages to hide visibility of the loss of diagonal resolution from that square.

I was unable to see any difference resulting from modifying the comb filter settings. Do these even do anything? Have any of you managed to see a difference?

I also noticed an irregular "jump" happen every once in a while. In fact, it happened too often to not notice. Have any of you noticed something similar?

I'm convinced that the settings for the comb filter don't do anything.  If you figure out how to make it work please let me know.  Can you explain the jump that you are talking about? Do you mean like a single frame jumping up or down once in awhile or a series of frames?  Have you posted any screenshots yet?

althor1138 said:

I'm convinced that the settings for the comb filter don't do anything.  If you figure out how to make it work please let me know.  Can you explain the jump that you are talking about? Do you mean like a single frame jumping up or down once in awhile or a series of frames?  Have you posted any screenshots yet?

No screenshots yet, but I will post some when I get home tonight.

The jump is weird. It looks like the odd field moves up several lines for 1 frame.

Is it something like this? 

althor1138 said:

Is it something like this? 

Possibly. I'd have to see it in motion.

Can you post a 1-2 second clip containing that frame?

Actually, I kinda think what I'm seeing is inherent to this laserdisc.  I went and found a couple of these instances and played with a few settings and it seems like they are just cadence changes that last only a couple frames but the strange part is that the first frame or two is shifted down a couple lines and then it pops back up. I think since the frame jumps up and down you have to be careful not to throw away the wrong field or it looks like this. After a correct trim and pulldown applied it plays out progressively without a problem, other than the fact that the frame instantly jumps up and down by a few pixels for a couple of frames. I checked this against all 3 captures of this title and they all exhibit the same thing on the same frames.

Here is a clip. It happens around the 4 second mark.  This one jumps down for 3 frames and then back up. I've corrected it with the proper pulldown so it plays progressively. What do you think? I hope it's not the capture card lol.

PW=OT

althor1138 said:

Here is a clip. It happens around the 4 second mark.  This one jumps down for 3 frames and then back up. I've corrected it with the proper pulldown so it plays progressively. What do you think? I hope it's not the capture card lol.

Don't worry, the same jumps are present in my LD and Arnie.d's captures as well. Like your samples, keep up the good work!

I think I'm aware of the disadvantages of capturing in RGB colorspace(rounding errors, bigger filesize) but are there any advantages over capturing in yuy2 colorspace? 

Jetrell Fo said:

althor1138 said:

I really wish I had access to two more sets of the JSC at this point.  I could do a toot between 3 sets for a total of 9 captures and probably eliminate all of the rot and dropouts.

EDIT:  Actually anybody that owns the set and wants to participate can send me their set and I'll send them back to you with your raw capture saved on a stick or something.

I actually have a set coming in the mail.  Where are you located?

He's in Sweden IIRC.