AntcuFaalb

Puggo - Jar Jar's Yoda said:

AntcuFaalb said:
Question: What is the correct size? Someone mentioned in an earlier post that all I need to do is double the width. Is this correct? If so, please explain further, as I'd like to know!

Double-the-width seems to work for 16mm/Scope film.  I don't know what the right correction would be for your capture, though... if it's a capture of a 35mm film, I would expect the correction could be different.

Here's what TServo2049 said:

Also, a word about aspect ratio correction - as I said before, this is a raw telecine of an anamorphic print with no aspect ratio correction, so it is both squeezed by 50% (approx. 1.20:1) and cropped by the telecine operator to fit the 1.33:1 aspect ratio of the video realm. Thus, the image needs to be horizontally stretched 2x, to an aspect ratio of approx. 2.66:1. Unfortunately, VLC doesn't have such an option.

This sounds correct to me. What do you think?

^g-force

about the luma variations, sounds like brightness AGC. If so, deflicker won't touch it.

Fizick's DeFlicker page claims that it can fix AGC blinding if you set the Lag to a negative value. Also, I've verified that these brightness variations are on the source tape.

about the duplicate frames, IF 4 of the five captures always have the correct frame at any one time, the median-of-five script will take care of it for you automatically. The method you describe sounds like a LOT of work (you will get false positives that you will have to weed through, or you will miss some due to noise and other fluctuations)

There's no way I can be sure without watching all five captures very closely. I'd like to see how many duplicates MultiDecimate identifies before deciding on any single method.

about the left-hand curve, does it change based on overall brightness (i.e. change on scene changes)? Just thinking of power supply issues old CRTs had, maybe something similar in the capture equipment.

The lefthand-side curve seems unrelated to the brightness; its size and shape is random and non-periodic. For example, it'll be one size and shape for 5 minutes here, another for 7 minutes there, etc.

If you post some raw video that exemplify the problems you are describing, SOMEONE here will have an answer for you.

I will. I passed-out last night, so I didn't get a chance to.

OK, I went all-out and did captures 2-5 last night. I'm exhausted, but I think it was worth it.

I will do one more capture tonight, as VirtualVCR reported 0 "Dropped2" frames for all captures except capture #2.

Observation: The Leitch DPS-290 sometimes (rarely) sends out duplicate frames to catch up to the genlock'd signal. This would be fixable in hardware if the Panasonic AG-1980P had an EXT SYNC-IN, but it doesn't. I believe the duplicates are exact duplicates, so I should be able to use pass 1 of MultiDecimate to identify them and manually replace them with non-duplicates from one of the other five captures.

Observation: The temporal luma variations mostly happen at scene changes. Furthermore, they're most disturbing when moving from a bright to a dark scene; e.g., Tatooine scene -> Sandcrawler scene. I believe that splitting the film by groups-of-scenes-that-should-have-similar-brightness and applying DeFlicker to each group will fix this issue.

Question: Does anyone have any idea as to why the big ugly lefthand-side curve changes in size and shape throughout the entire film?

Question: What is the correct size? Someone mentioned in an earlier post that all I need to do is double the width. Is this correct? If so, please explain further, as I'd like to know!

Darth Mallwalker said:

^Spaced Ranger in the scenario you describe, would it look something like this?
So far I'm not seeing that in PS78v0.1. Fields look pretty-well blended there
but I suppose it could be the result of DivX encode, as AntcuFaalb said.

AntcuFaalb said:

PM sent. I think DivX may have screwed with the picture in v0.1, so don't trust what you see.

If you can handle it, I can arrange to send you my 60GB Huffyuv AVI of Capture #1. PM me if you're interested.

How about posting up a 60MB sample?
Maybe the first few seconds of receding STAR WARS logo? (Or any scene with motion and contrast)
That way we could all have a go at it...

Did you plan to IVTC the five captures first (if possible) or median first, then IVTC the result ?

 

I'll post a good representative sample tonight. Thanks for the suggestion.

I planned to do the median first, then IVTC. Which approach is better? I don't mind putting the time in if it's better to IVTC each of the five first...

msycamore said:

With all my talk of incorrect aspect ratio... when I did the correction on the Technidisc I noticed that it almost perfectly match the 2004 DVD for the most part when I resize that transfer to 704x480, I also noticed that the GOUT-print is pretty badly warped, the distortion is worst in the left part of the frame, anyway what's interesting is that I also discovered that the lightsaber shot in Ben's hut in the GOUT transfer is vertically stretched or horizontally squished, it seems to me that they spliced in that part from a different source, probably because of the "split-saber syndrome" in the '93 transfer we talked about earlier in the thread.

Can someone please post this frame:

from the '95 "Faces" transfer, I'm just curious to know if the GOUT source actually match the "Faces" release rather than DC.

I noticed this a few weeks ago while trying to come up with the minimal cropping needed for GOUT. Is this the only scene with this defect?

Darth Mallwalker said:

Just seen TServo's reply above.
IVTC sounds like fun . . . if it's possible.

I haven't seen v0.0 nor v0.1. Are those links still up?

PM sent. I think DivX may have screwed with the picture in v0.1, so don't trust what you see.

If you can handle it, I can arrange to send you my 60GB Huffyuv AVI of Capture #1. PM me if you're interested.

Darth Mallwalker said:

Distributed to theaters on six 2000-foot reels.
I believe the IPs, or whatever the film source used to make official home video releases,
comprise twelve 1000-foot reels (ROTJ thirteen).

IVTC sounds like fun...

OK, thanks for the information. Do you know anything about what TServo2049 said above?

Excellent work, russs15!

Capture #1 is done. I'll be starting #2 tonight.

Is anyone here (with experience in manual audio syncing) interested in helping out after the video captures are done?

Also, is anyone interested in helping IVTC the median-of-five video? I'd like to GOUT-sync it for Release v1.1 (non-raw). Do Darth Mallwalker's Reel Lengths apply here? I thought ANH came on six reels... ::confused::

I don't have anything useful to add, but I just want to say that I think this project is a great idea!

I'd be interested in seeing a US side-project.

Capture #1 is going well. The quality is pretty damn good; as good as it's going to get.

The chroma noise is still annoying (albeit easily fixable in software), but the color is vibrant in the non-blown-out scenes!

My week of experimentation is finally complete.

What I've learned:

• The temporal luma variations (blown-out blacks and whites at scene changes) are definitely a part of the VHS. I must have failed to notice this defect on my 10" CRT TV.
• The Panasonic DMR-ES10 stabilizes the horizontal jitter very well and slightly improves the big ugly curve on the lefthand side of the picture.
• Using the Panasonic DMR-ES10 in passthrough mode is preferable to using it in record mode, as the latter suffers from too many MPEG-2 artifacts.

Plan of action:

• I'll be doing five captures, at the rate of one capture per night, starting tomorrow evening.
• I plan to use g-force's median-of-five (a.k.a., Median2) function on the five captures.
• I'll have to capture the audio separately and sync it manually.

Goodnight!

SilverWook said:

The mind boggles! How many people went into a porno shop looking for a copy of Star Wars?

"Hey kids! I'm off to the Smut Shoppe, you want me to pick you up something?"

It does look interesting...

 Can I include this as a quote on the cover I design for PS78?

Here's my new setup!

russs15 said:

AntcuFaalb said:

Is there a GOUT-sync'ing how-to somewhere on OT? I would like to GOUT-sync my '92 Letterbox captures after I'm done with PS78.

I read this as a "how do I GOUT sync a random video (or audio) to the GOUT to which the answer is that this is not the droid you are looking for. I have not got the foggiest how to do this.

If it is as per the last response, stripping (demuxing) a known GOUT source and remuxing with another GOUT source, the answer is that it is pathetically easy (on a Windows PC with free software) and I will gladly  help anyone out...

I was asking about the former; i.e., GOUT-sync'ing a random video. I think I just need to IVTC, split at the reel-changeovers, add-in blanks for lost frames (if any), and then reassemble.

I've ordered a brand new/sealed 1995 (THX; Faces) P/S Fullscreen set. It should arrive soon.

The Panasonic DMR-ES10 should arrive tonight. I'm excited!

A persistent (archiving) live chat would be nice. There are plenty of batteries-included Java applets available in the FOSS world for this.

It'd sure beat the coordinating needed for IRC discussions...

negative1 said:

AntcuFaalb said:

Unfortunately, the ColourLike filter, even when applied on a frame-by-frame basis using ScriptClip, does a poor job at histogram matching.

I'm looking into histogram matching plugins for ImageJ. If I can't find anything that works well, then I may have to settle on the commercial program PureImage NR for color matching.

pureimage might work well for the occasional image,

but for automating, or doing several scenes. it is very

clunky, and overall not worth it. 

tried it here:

----------------

http://originaltrilogy.com/forum/topic.cfm/Methods-for-colour-matching-colour-grading/topic/9430/

 

more threads here:

------------------------

 http://originaltrilogy.com/forum/topic.cfm/Colour-matching-for-fan-edits/topic/7257/

http://originaltrilogy.com/forum/topic.cfm/Auto-correction-from-SE-colours-to-GOUT-colours/topic/6732/

 

you need to use professional filters and plugins in

premiere, avid, or other pro video editting software.

 

later

-1

I think I might just end up programming a histogram matching algorithm in C for AviSynth.

Thanks for the links!

I might just buy this book. The first edition contains the analog information that was removed for the second.

russs15: Can you list the sources (e.g., GOUT, '93 LD, etc.) for each video entry?

Also, I've just learned that Darth Mallwalker's IVTC script for The Starkiller's 1982 Video Rental Library VHS preservation does GOUT-sync'ing as well.

Unfortunately, I haven't yet received confirmation from my coworker as to whether or not he has a TESB bootleg.

I just realized: PS78 will be going through three different kinds of TBC. There's a field-based TBC in the Panasonic AG-1980; a frame-based TBC in the Leitch DPS-290; and a line-based TBC in the Panasonic DMR-ES10. Crazy!

frank678 said:

I was curious if you had had chance to try using the colormatch thingy for a whole scene from just one still yet? That way you could mix and match sources or even use one of the 70mm scans frames/screening photos/ production stills/other source etc etc.

Also whether you've compared your 1992 FS set with a 1982 vhs preservation, they are understood to be all from the same master between those years but it might be possible one version has a slight edge. Maybe once you've captured your vhs releases you could experiment with them via colormatch?  

 

Unfortunately, the ColourLike filter, even when applied on a frame-by-frame basis using ScriptClip, does a poor job at histogram matching.

I'm looking into histogram matching plugins for ImageJ. If I can't find anything that works well, then I may have to settle on the commercial program PureImage NR for color matching.

Thanks for the information, everyone!

Mavimao, you're right: my understanding of interlacing (at the time of the first post in this thread) was poor, at best. Now it's much better thanks to the information that both you and Moth3r provided.

My captures are raw (excluding TBCs, of course). I double-check the settings on all of my equipment and software prior to doing one to ensure that nothing fancy, e.g., noise reduction, is turned on. With each capture, I intend to take pictures and screenshots of the respective hardware and software settings involved so that someone with more experience can correct me if I do something in a suboptimal way.

In Film->NTSC_VHS conversions, is the film telecined (to get 29.970fps) and then interlaced? If so, would you deinterlace before or after IVTC'ing?

Wikipedia seems to indicate that all VHSs are interlaced, by design:

Each helical track contains a single field ('even' or 'odd' field, equivalent to half a frame) encoded as an analog raster scan, similar to analog TV broadcasts. The horizontal resolution is 170 lines per scanline, and the vertical resolution (the number of scanlines) is the same as the respective analog TV standard (576 for PAL or 486 for NTSC). In modern-day digital terminology, NTSC VHS is roughly equivalent to 333x480 pixels luma and 40x480 chroma resolutions (333x480 pixels=159,840 pixels or 0.16MP (1/6 of a MegaPixel)).[18], while PAL VHS offers the equivalent of about 335x576 pixels luma and 40x240 chroma (the vertical chroma resolution of PAL is limited by the PAL color delay line mechanism).