No idiot. I am not saying that. A correct screenshot captured on a 1920*1080 screen will look the same as any 4:3 dvd played on such a screen.
And, again, I never said what you put in to my mouth in your second point. How was I to even know that the dvd/vob was released by the record label. In fact the label you attach to the file suggests otherwise.
Third point. I will agree as far as the definition of master is concerned.
A lot of original commercial material ends up in private hands either by fair means or foul. I need not say any more.
I am not changing the subject. I am merely 'interested' in the source. The artist is not to my taste although I do own a dvd of 8-Mile which included a promo as an extra. I also looked at youtube earlier and found an official release of this clip. It is only 360p but color/definition-wise is close to your 'master'. Now that has been around for 10 years. Of course any collector will seek out the best - I am no different in that regard as already mentioned in other topics.
Try StreamFab Downloader and download from Netflix, Amazon, Youtube! Or Try DVDFab and copy Blu-rays!
+ Reply to Thread
Results 31 to 60 of 81
Thread
-
-
I was going to pitch in and try to help, but the OP appears to be too intent on engaging in personalities and also not answering legitimate questions about important video parameters.
It is impossible to help when the responses are so vague, and it is not pleasant to do so when the responses have such an edge of hostility. I'm too old for that sort of thing.
Several years ago I worked with "StainlessS" over in doom9.org to help create techniques for undoing interlacing artifacts that result from not deinterlacing prior to resizing. However, I don't think I'd get very far trying to suggest any of those fixes here. -
That's a loaded statement.
"direct encode"? (completely meaningless words; everything is a "direct encode" in digital video, source>output)
"analog transfer"? (with what hardware, from what master media, to what res/format/etc)
And you would be surprised to know that UMG and there team are more amateurish than you would think.
than how would some rando on the internet get a higher quality source to encode from
because it still is of better quality than the VOB.
Better quality? Not at all. It's screwed, probably irreparably.
Oh, I don't know ... some us would read it, even if the OP didn't.Want my help? Ask here! (not via PM!)
FAQs: Best Blank Discs • Best TBCs • Best VCRs for capture • Restore VHS -
I've answered every question that was asked of me I'm not sure what your talking about. Jagabo already provided me with a fix that seems good enough.
I really didn't think I was the bad guy here but perhaps I'm just ignorant. My eyes were deceiving me and people would most definitely prefer to watch the VOB rip over the master is that right? Ugh, beats me.
I was not disrespect or vulgar until DB83 started questioning the legitimacy of the source which after all isn't what this thread is even about.
But just as a finisher for this thread, I'm a 16 year old from Texas who just has a ton of free time and is a big fan of Eminem. I thought I'd restore old videos of his which other fans would appreciate (and they have). I'm no pro, I'm don't make money from this work. But I'd like to think I know quite a bit about video media. Enough to know that:
1. The video is of higher resolution, bitrate, detail, and actually has less interlacing artifacts, than the VOB dvd rip. I would take minor blending/combing over more detail any day of the week. I also no that if it was resized I wasn't the one who resized it whoever encoded it did.
You can check many VEVO rips of old music videos and you'll see a bunch of combing in them.
2. The video was encoded by someone who access to either the original film reels or the analog transfer, which would explain the rain-bowing. It's impossible to get this much better quality any other way. -
Well I do not think it inappropriate to question the legitimacy of the 'source'. Maybe you should take a little of your 'ton of free time' and read the forum rules.
And, yes, I can not prove it's legitimacy and neither can you. You, wherever you got this from, have blindly accepted the speel. And as I already wrote the vids on yt have the same color-balance etc. as this.
And you do not know as much about video media as you think you do. You keep ranting on about 'analog transfer'. What makes you think the real master - not the encoded one you possess - was analog ? Methinks you have put two and two together and made five. Or 'It was interlaced so it must have been analog'
Anyway. Just to keep you interested I have ordered my own 'master'. It comes on official dvd and should be different to the dvd you have and should land on my doormat in a few days. And since it is Christmas I may even share it with you. -
OK, sir smurf, you asked for it:
repair bad deinterlacing
I will warn that, since the original video was "fubar," to use that Army expression someone used earlier, the result is still pretty awful. However, we did actually manage to almost completely eliminate the massive "teeth" (resizing artifacts are much bigger than normal combing). I've always meant to go back and create a proper function out of that work because it has to be tuned for each case, depending on the height (how many scan lines) of the teeth. Also, my method for identifying each row was grotesquely crude.
It would also be interesting to study the mathematics of how those teeth get formed and why you end up with offsets that are multiple scan lines high.
Despite the crudeness of the result, I was actually pretty proud of figuring out that the solution to the problem was to apply motion estimation separately to each "tooth" in order to move it back to where it should be.
BTW, to the OP, I'm 67 years old, and there are a few others here who are a little beyond their teenage years. Our culture has changed a lot, but back in the 1950s and 60s, I would call anyone older than 21 "sir" or "mister." It's not your fault that the language is now coarser and the generational respect has disappeared, but you would do well to remember the old saying, "You catch more flies with honey than you do with vinegar."
If you want to address me in future posts, you can call me "grandpa." -
What? You bought a dvd of Like Toy Soldiers? That wouldn't be a master at all, would it?
I never got the master I have from a DVD. I bought it with money.
I guess we'll see, but I know one thing is certain:
The DVD you get won't look as good as the "master": incorrect aspect ratio, letterboxed, interlaced, lower resolution, lower bitrate, worse color
Can you agree on that at least?
Perhaps I'm not clear on what your trying to prove anymore.
But I will own up to the fact that I don't know for sure if it is a master, but I do know the person I bought it from said it was, and so did the person he/she bought it from.
I also know that the quality strongly resembles exactly what it says is in the metadata and looks better than any other VOB I've seen (besides the interlacing artifacts).
That's putting 1 and 1 together to make to 2: it is a master.
Perhaps someone resized it: maybe the person I bought it from or the person they bought it from, or maybe the label did it themselves because afterall, there is evidence they don't really care about these artifacts (like the fact they appear in a bunch of old videos).
But like my point before: Where would this encode come from if wasn't encoded by the record label or at least some other parent company that contains raw digital/analog captures? (My bets are on analog because of the rainbowing that appears on other videos of his at this time period).
Also, sorry that I got upset about you questioning the source. I think I understand now the reason for your doubt. -
That's pretty easy. It happens when using simple resizing algorithms (nearest neighbor, point, simple bilinear, simple bicubic) with a new size that's close to 1/2 the size of the original. Using all integers:
Code:input_y = (output_y * input_height) / output_height; // input_y is the scanline in the input video which we will copy from // output_y is the scanline in the output (downscaled) video which we will copy to // input_height is the height of the input video // output_height is the height of the output (downscaled) video
Code:input_y = (output_y * 480) / 260;
Code:input_y = 123 * 480 / 260 = 227
Code:0, 1, 3, 5, 7, 9, 11, 12, 14, 16, 18, 20, 22, 24, 25, 27, 29, 31, 33, 35, 36, 38, 40, 42, 44, 46, 48, 49, 51, 53, 55, 57, 59, 60, 62, 64, 66, 68, 70, 72, 73, 75...
You can see this by resizing in VirtualDub with the Nearest Neighbor algorithm. You can also see it with the Bilinear (interpolation only) and Bicubic (interpolation only) algorithms -- though those end up with less sharp bands. Many programs used to use those algorithms.
A simple C program to generate those line numbers:
Code:#include <stdlib.h> #include <stdio.h> #define INPUT_HEIGHT 480 #define OUTPUT_HEIGHT 260 main() { int y; for (int y=0; y<OUTPUT_HEIGHT; y++) { printf("%d, ", (y * INPUT_HEIGHT) / OUTPUT_HEIGHT); } }
-
That is a very interesting and useful analysis. I guess what you're saying is that, in your example, you have seven even fields that are used and then six odd fields (I think they should balance out, but that didn't seem to be the case in your example ... I'm not sure how that could be). Since even and odd are from different moments in time you get the characteristic multiple scan line horizontal shifts, back and forth as you look at a still shot of the resulting re-sized progressive frame.
This suggests that the algorithm I created in the doom9.org thread I linked to could be "tuned" by entering the original and final resolutions. That would let you get the exact height of the resulting blocks that need to be shifted horizontally (I just guessed, in the video we were restoring, based on what I saw visually). In many cases, even if the person possessing the video has no direct way of knowing what prior violence was done to that video by someone else, that person could make an educated guess about the original resolution and then see if the results prove that guess correct.
So thanks for that insight. I have no time right now to do anything about this, but because this is such a common problem, it might be useful to generate a generic function similar to the much-used blended frame recovery script, SRestore. The interlaced re-sizing problem we're talking about here is probably almost as common as the field blend problem that SRestore attempts to fix. -
The total number of lines coming from each field does not have to balance out. Consider when resizing by exactly 1/2, 480 to 240. All output scan lines will all come from one field (0,2,4,6...). Note that my algorithm used truncated integers. Other algorithms may round rather than truncate. And they may not start at zero. They may work in floating point rather than integer...
The full list from my simple program (modified to output with leading zeros):
Code:000, 001, 003, 005, 007, 009, 011, 012, 014, 016, 018, 020, 022, 024, 025, 027, 029, 031, 033, 035, 036, 038, 040, 042, 044, 046, 048, 049, 051, 053, 055, 057, 059, 060, 062, 064, 066, 068, 070, 072, 073, 075, 077, 079, 081, 083, 084, 086, 088, 090, 092, 094, 096, 097, 099, 101, 103, 105, 107, 108, 110, 112, 114, 116, 118, 120, 121, 123, 125, 127, 129, 131, 132, 134, 136, 138, 140, 142, 144, 145, 147, 149, 151, 153, 155, 156, 158, 160, 162, 164, 166, 168, 169, 171, 173, 175, 177, 179, 180, 182, 184, 186, 188, 190, 192, 193, 195, 197, 199, 201, 203, 204, 206, 208, 210, 212, 214, 216, 217, 219, 221, 223, 225, 227, 228, 230, 232, 234, 236, 238, 240, 241, 243, 245, 247, 249, 251, 252, 254, 256, 258, 260, 262, 264, 265, 267, 269, 271, 273, 275, 276, 278, 280, 282, 284, 286, 288, 289, 291, 293, 295, 297, 299, 300, 302, 304, 306, 308, 310, 312, 313, 315, 317, 319, 321, 323, 324, 326, 328, 330, 332, 334, 336, 337, 339, 341, 343, 345, 347, 348, 350, 352, 354, 356, 358, 360, 361, 363, 365, 367, 369, 371, 372, 374, 376, 378, 380, 382, 384, 385, 387, 389, 391, 393, 395, 396, 398, 400, 402, 404, 406, 408, 409, 411, 413, 415, 417, 419, 420, 422, 424, 426, 428, 430, 432, 433, 435, 437, 439, 441, 443, 444, 446, 448, 450, 452, 454, 456, 457, 459, 461, 463, 465, 467, 468, 470, 472, 474, 476, 478
-
Some days ago I wrote in this thread
"If you possess an original dvd I would equally argue that you can get as good a quality from that. "
Well my dvd has arrived. Silly of me to think that an original dvd would be as good as what you have.
Here is a comparison of your 'studio produced' dvd to the one I received (this is just the main video - I did not rip the BTS clip)
There are some noticeable differences between the two certainly on bitrate and audio. I have prepared a sample to show one other difference namely that this dvd does not include the logo/title. But it does contain a final sub-title/credit missing from your 'master' so length is slightly longer.
I leave it to you to judge how this official dvd compares to your 'master'. I have no interest in deinterlacing/cropping/resizing/conversion to NTSC etc. So if you want the whole thing you get the PAL format vob as is. And it cost me the equivalent of two bucks. You probably paid more than that.
For the record I am a few months younger than johnmeyer. But you can call me SirLast edited by DB83; 11th Dec 2019 at 09:51.
-
While that is one of best VOB's I've seen, it is still mathematically and perceptively of worse "quality", but maybe my definition of that word is different than yours.
It's encoded in MPEG2 which is less efficient than AVC, which I'm sure you know. But your 9000kbps MPEG is only equivalent to about 4,000kbps AVC, which means that is of lower bitrate and quality than my master, which is 9,000kbps AVC which is about equivalent to 18000 MPEG.
Not only that, but there is a weird distortion where the pixels the black bars and I can conclude that those pixels are intact in my master.
Also, of course, it is of higher resolution and isn't an upscale. You can confirm this yourself by comparing the detail levels in screenshots.
So this concludes that my master is of better quality, which in this case I'm defining as fidelity, than the VOB.
However, I think it looks good enough to the point where I may be able to substitute frames from into my master.Last edited by embis2003; 12th Dec 2019 at 15:32.
-
Yes, that is a key component of any of these discussions, and the important conclusion is that the only thing that matters is what looks good to you (unless you're doing it for a client).
I haven't looked closely at the various samples posted in this thread, but I did see that there is a difference in resolution, color, and gamma. From what you are saying it sounds like there may also be a difference in encoding artifacts. There is no doubt that the JPEG (DCT) compression that is at the heart of the old MPEG-2 standard does produce some pretty easy-to-see artifacts, whereas AVC encoding is more clever in disguising compression degradation.
Aslo, some people are more sensitive to loss of visual detail; some to color wokiness (don't mention DV video to various people in this forum who absolutely despise the 4:1:1 colorspace); some to compression artifacts; some to GOP glitches (slight jerks when each keyframe gets you back to a "real" frame rather than a delta); and so on.
The somewhat obvious advice is: perform your own tests and then do what looks good to you. -
I am not going to get in to a discussion about the relative merits of bitrate for AVC and mpeg2 etc. etc.
Shame you could not be a little more accommodating. I did say 'may'.
Plenty of copies available on fleabay or you could always borrow some frames from the youtube version. -
-
I could equally argue that you made your own mind up about the dvd even before you saw it.
If you bothered to check all available sources BEFORE you blurted out you would have noticed that those lines both at the top and bottom of the dvd are not there in your wonderful copy. And they are not there because they have been cropped away and the frame has been resized - no call that stretched - to keep the AR. In the youtube version did you ever wonder why there some small black bars both at top and bottom ? Because those lines were cropped away and the bars added to retain the AR without stretching.
They may even be there, the black bars I mean, in the other dvd source since the top and bottom of the video have a different tone to the rest of the letter-boxing.
I have my own opinion of where your 'master' actually originated from. You would never in a millennium accept that.
And had you kept your big mouth shut and just thanked me for my 'uninterested' effort I would happily have uploaded the entire vob for you to do whatever you felt with it. -
All of them have problems, but if you composite - mix and match with masking/rotowork some manual work colormatching and reframing/resizing to match, maybe some frame interpolation you might be able to fix some of the problems better
That PAL version has field blending - it's not that great either . By the time you deinterlace/srestore , and reframe/resize to match - it's going to be significantly worse
The "real" master from that video would likely be a betacam sp/sx or digibeta tape variant, given the year. It's possible the AVC version was encoded directly from that, but even ignoring the field resizing issues - it's not a great AVC encode either. Some parts are better. But there are banding and posterization artifacts that even the "lower" quality versions do not have in some areas -
I could equally argue that you made your own mind up about the dvd even before you saw it.
If you bothered to check all available sources BEFORE you blurted out you would have noticed that those lines both at the top and bottom of the dvd are not there in your wonderful copy. And they are not there because they have been cropped away and the frame has been resized - no call that stretched - to keep the AR. In the youtube version did you ever wonder why there some small black bars both at top and bottom ? Because those lines were cropped away and the bars added to retain the AR without stretching.
I have my own opinion of where your 'master' actually originated from. You would never in a millennium accept that.
For me, the fact that the "master" looked of better quality than any VOB I've seen was confirmation for me.
I will conduct an experiment with the VOB and try to get it to match the "master". If, in the end, I can get it look as good as it, I will accept that isn't a true master. -
-
I was gonna upload a short sample of my feeble attempt to convert the vob to AVC with a higher bitrate. Resampled fps back to 29.97. Even cropped etc to remove the issues as the top and bottom of the frame.
Given PDR's comments which I full respect I might not bother now. The biggest issue for me is the text in the book at the start. That, I will concede, is not as sharp.
There are always two ways to read something. Just for a moment go back to that mkv and read the mediainfo report. Here is the relevant line "Eminem - Like Toy Soldiers (FLAC-Master-480p-H264-ddnt32)" The was I interprete that is it the sound that is mastered NOT the video.
But if you want this just to see if it is visually any better than the vob, in the spirit of Christmas, just say 'please'. -
Last edited by embis2003; 27th Mar 2021 at 07:14.
-
Your AVC version is 100% definitely not sourced from his VOB, or your other VOB
And if you have access to other versions, it's entirely possible you can mix and match to get a better result. Each of the ones here have slightly different problems on different frames. They probably all have been through different steps and butchered in different ways. And I'm not just talking about the combing. There are other issues you can "fix" too, such as bad deinterlacing artifacts, aliasing, line buzzing -
Wow um. Isn't that the one you made?
I noticed some immediate problems first of like the motion isn't very smooth, some artifacts, and most of all that the text on the book is barley readable in that. In my AVC version it is. Perhaps it was your deinterlacing method?
What I wanted was the original VOB you ripped. But Its mostly irrelevant now though since we basically confirmed the AVC isn't source from any VOB posted in this thread. -
Well you were not readin' wot I rote >> "I was gonna upload a short sample of my feeble attempt to convert the vob to AVC with a higher bitrate........"
And I already mentioned the text issue which is on the dvd. That could well be a result of the encoding method since run times are identical whereas a PAL dvd would typically be sped up.
These conversions are not my forte. Just like getting the dvd, this was just an experiment.
Not gonna upload the complete vob. You already failed that exam. Like I already stated, plenty of copies available. -
-
First of all, I'd like to apologize for how hostile this thread got. I think its fair to say I got frustrated.
But I would like to inform that this thread did aid in me creating what I consider to be a nearly perfect repair of this video.
I used manual frame interpolation for frames which had combing. I found it worked better than using jagabo's automatic method (although I think the actual interpolation in that script was higher quality) because instead of editing out the false positives, I could just manually select everything that was combed.
For some frames, I elected to resize to half height (original resolution was 864,480, I Lanzcos4Resized down to 864,240, and then doubled using nnedi3, and then used Lanzcos again to get it back to its original ratio) to remove combing as opposed to interpolating, because they would either be too close to a scene change to interpolate or interpolating just wouldn't work (???).
I then scaled up the de-combed result using Topaz Video Enhance, which dramatically improved detail and removed the banding issues.
For those curious, here is my final result: https://drive.google.com/file/d/1FWpsEDGpUTvCwfWzNTVv4ciJiqAOvArB/view -
Very respectful comment.
nearly perfect repair of this video.
For those curious, here is my final result:Want my help? Ask here! (not via PM!)
FAQs: Best Blank Discs • Best TBCs • Best VCRs for capture • Restore VHS
Similar Threads
-
Deinterlacing an online video with comb artifacts?
By Master Tape in forum RestorationReplies: 4Last Post: 21st Jul 2018, 03:48 -
Remove interlaced artifacts on progressive video.
By TempUser_ in forum Video ConversionReplies: 4Last Post: 18th Jul 2018, 12:38 -
Why are interlace/combing artifacts only an issue on progressive displays?
By 90sTV in forum Newbie / General discussionsReplies: 12Last Post: 28th Jan 2017, 08:22 -
Artifacts In Progressive SD Animation Source
By LouieChuckyMerry in forum DVD RippingReplies: 48Last Post: 13th May 2016, 21:27 -
Dot Crawl Artifacts from Composite Source?
By Ish Kabibble in forum CapturingReplies: 9Last Post: 20th Mar 2015, 06:48