logo Sign In

PS78: Pre-ANH Star Wars Bootleg VHS from 1978 ***"RAW" DVD RELEASED*** — Page 12

Author
Time
Post link

I can make the sample smaller tonight if the size is a problem for anyone...

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time
Post link

That sample looks good! I watched it with VLC in it's original aspect ratio, and I'm impressed. Keep up the good work!

Screen grooming.  A most underrated activity. - hairy_hen

 

The mic was in picture!

^_^

Author
Time
Post link

Comments are between the Sample2 and initial capture, PS78_DivX.avi.  (i've got no clue what the colors should be)  The side black bars are differently proportioned but seem to be the same general size.  Sample2 is generally greener, see Tantive getting sucked into Star Destroyer shot, or the Rebel Soldiers uniforms.  The flickers are gone and the image is more stabilized.  Detailing is improved, in an early semi-close up of 3po the eye socket got kinda washed but in Sample2 it's clearer and better defined.  The close up of the older Rebel Soldiers, the face is paler/less red then the first capture.  How did the tape glitch which happens right before this sample come through these new transfers?

Author
Time
Post link

AntcuFaalb said:

The Leitch DPS-290 must have caused more issues than just the occasional frame-freeze. I did a short capture using the Panasonic AG-1980P -> Panasonic DMR-ES10 and it does seem to have a pattern.

It's a 1600-frame Huffyuv-compressed capture, so it's a little large. I'd greatly appreciate any help with determining the pattern(s). Thanks!

Here it is!

There is no pattern I can see - it is most definitely field-blended, with some frames completely unrecoverable (much like the PAL VHS tapes of Droids and Ewoks).

Guidelines for post content and general behaviour: read announcement here

Max. allowable image sizes in signatures: reminder here

Author
Time
 (Edited)
Post link

Moth3r said:

AntcuFaalb said:

The Leitch DPS-290 must have caused more issues than just the occasional frame-freeze. I did a short capture using the Panasonic AG-1980P -> Panasonic DMR-ES10 and it does seem to have a pattern.

It's a 1600-frame Huffyuv-compressed capture, so it's a little large. I'd greatly appreciate any help with determining the pattern(s). Thanks!

Here it is!

There is no pattern I can see - it is most definitely field-blended, with some frames completely unrecoverable (much like the PAL VHS tapes of Droids and Ewoks).

Could this be caused by the Panasonic DMR-ES10? Probably not, as it's just a Line-TBC.

I'm currently looking into AviSynth filters for field-blending correction. Can you suggest any?

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time
Post link

Srestore is good for dealing with the most common field-blending issues, but this source is unfortunately too far gone to be restorable.

Guidelines for post content and general behaviour: read announcement here

Max. allowable image sizes in signatures: reminder here

Author
Time
Post link

Moth3r said:

Srestore is good for dealing with the most common field-blending issues, but this source is unfortunately too far gone to be restorable.

That sucks, but do you think Srestore might be able to do anything? I'm thinking something like: srestore(frate=23.976, omode=6, speed=-25, mode=2)

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time
 (Edited)
Post link

I used srestore for a VHS transfer I did of the old Legend of Sleepy Hollow starring Jeff Goldblum (available at cinemageddon) which had many of the same problems your VHS does.

You'll have to deinterlace the raw 29.97i footage to produce 59.94 fps progressive footage so that srestore can have more frames to work with. You lose half the resolution, but using some more advanced deinterlace methods can produce good results.

Anyway, a basic re-purposing of my old settings with your proposed settings looks to work pretty well on your 1600-frame Huffyuv-compressed capture. EDIT - I've updated the EZKeepGrain setting from 0.6 to 1.4 to reduce the amount of denoising going on in QTGMC.

AssumeTFF()
ConverttoYV12(interlaced=true)
QTGMC( "slow", TR2=1, SourceMatch=2, Lossless=2, Sharpness=0, EZKeepGrain=1.4 )
srestore(frate=23.976, omode=6, speed=-25, mode=2)

QTGMC can be extremely slow and endlessly tweakable but many in the doom9 avisynth usage forum consider it to be the best deinterlacer out there.  The main limitation is that many of the settings require a YV12 color space but that should not be a problem for most.

None of these methods are perfect and some blending will remain, but I hope this helps!

Author
Time
Post link

I've used QTGMC with SRestore for some tricky stuff.

"Right now the coffees are doing their final work." (Airi, Masked Rider Den-o episode 1)

Author
Time
Post link

The thing about Srestore is that it works by substituting a blended field with an unblended neighbour. In sources such as this one, where some frames do not exist in unblended form, it can never restore all the original frames.

Guidelines for post content and general behaviour: read announcement here

Max. allowable image sizes in signatures: reminder here

Author
Time
Post link

Take my advice and if using QTGMC, always use "Placebo".

In mine and DJ's project, using any other setting for it came out looking slightly worse. Also, using Edimode="EEDI2" made it look smoother.

I don't know all technical details about the plugin but this is the experience I had with it.

Author
Time
 (Edited)
Post link

To resolve the issues with the Leitch DPS-290, I did five back-to-back (Panasonic AG-1980P -> Panasonic DMR-ES10) captures this past Friday night. These new captures are excellent replacements with no visible loss in quality and no randomly-frozen frames. Furthermore, the auto-tracking never changed since I did them back-to-back, so it's far easier to align these five captures than the previous five.

I looked at a scene from Interleave(c1.SeparateFields().SelectEven(), c1.SeparateFields().SelectOdd()).srestore(frate=23.976, omode=6, speed=-25, mode=2) and compared it with the same scene from GOUT. Amazingly, both are exactly 1063 frames. Some of the shot-transitions within the scene are off by one frame, but that can be repaired with AviSynth's FreezeFrame.

Should I still look at QTGMC?

Edit: Also, I was able to get rid of the ugly lefthand-side curve using DeFish(fov=48)!

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time
 (Edited)
Post link

IMO, you shouldn't stress yourself on this one. It IS a 35 year old tape, y'know.

Just taking out only the very worst flaws should be fine. The colours are the most important part here (in other words, it's probably better as a reference than for actually watching as a movie).

Author
Time
Post link

Agree, the less that's done, the better.

We want you to be aware that we have no plans—now or in the future—to restore the earlier versions. 

Sincerely, Lynne Hale publicity@lucasfilm.com

Author
Time
Post link

msycamore said:

Agree, the less that's done, the better.

Agreed. This is why I intend to do two separate releases. Release v1.0 will be the median-of-five'd capture and Release v1.1 will be something that's comfortably viewable.

I also intend to upload each of the five raw and unfiltered captures individually to a.b.starwars. Many won't have the resources or desire to play with these, but you never know...

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time
Post link

That's great. What you're doing is very much appreciated.

We want you to be aware that we have no plans—now or in the future—to restore the earlier versions. 

Sincerely, Lynne Hale publicity@lucasfilm.com

Author
Time
 (Edited)
Post link

Despite taking the Leitch DPS-290 out of the mix, the capture-card seems to have duplicated <10 frames (yes, frames -- not fields!) in each capture. I'm working on manually identifying and removing them, as they interfere with my ability to line-up each of the captures to one another.

The following algorithm outputs ~100 frames for each capture, many of which are false-positives.

c.WriteFileIf("current_frame != 0 && YDifferenceFromPrevious() < 1.0",  "current_frame", append=true, flush=true)

I have a Perl script that takes the list of frame numbers and outputs AviSynth code to display: frame-1, frame, frame+1, red frame, .... Identifying the duplicated frames is then simply a matter of visually inspecting each segment and writing down the result. I plan on double-checking these results against something like MultiDecimate before moving on.

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time
Post link

I have not got a clue what any of that last post actually means but I am convinced that you are doing an amazing job. Keep it up!!!!

4 - 5 - 3 - 1 - 6 - 2

Discuss…

Author
Time
Post link

Good News: For Capture #1 (the only one I've tested so far), MultiDecimate identifies the same duplicate frames as those I've manually identified!

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time
Post link

All that work you're doing is appreciated. Can't wait to see the final version of this project!

Author
Time
 (Edited)
Post link

More Good News: It's the same story for Captures #2-4. Furthermore, the numbers of duplicated frames are making sense; that is, when you subtract the number of duplicate frames from the total number of frames for each capture, you end up with each capture having the same total number of frames. Yay!

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time
Post link

Capture #5's being a pain-in-the-neck. Argh!

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time
Post link

Wouldn't 4 captures be enough? Or does it have to be 5 for that median thing to work?

Author
Time
Post link

You_Too said:

Wouldn't 4 captures be enough? Or does it have to be 5 for that median thing to work?

It has to be five for g-force's Median2 function to work. It's fine. I'm going to stick with it until I've solved the problem.

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time
Post link

might want to compare a TOOT or median1 with median2 on a short section, just to make sure that median-of-five is really needed before you kill yourself over the 5th capture.

-G