Combing

User avatar
x265
Joined: Tue Dec 25, 2012 1:52 pm
Org Profile

Re: Combing

Post by x265 » Fri Jun 07, 2013 7:18 am

“The only true wisdom is knowing that you know nothing.”

User avatar
l33tmeatwad
Joined: Wed Feb 16, 2005 3:22 pm
Location: Christiansburg, VA
Contact:
Org Profile

Re: Combing

Post by l33tmeatwad » Fri Jun 07, 2013 10:09 am

I actually have worked with this footage on multiple occasions and this sitting on my hard drive right now. I will say the original releases did not have these issues and it's new to the "remastered" edition releases.
Mister Hatt wrote:That doesn't sound right at all, seeing as the OP has declared it to be mostly ok when forced FILM, which implies most of it is soft-telecine and pattern detection is useless.
The opening sequence is NOT mostly okay with forced film, the DVDs were not soft telecined.
Mister Hatt wrote:Additionally, the distortion is clearly combing, NOT aliasing
You are right, it's not aliasing, I used the wrong term as I was thinking of anti-aliasing filters to cover that up since it's not fixable in this particular instance, I corrected my original post to avoid reader confusion. As for this particular footage, it appears as if they may have removed the interlacing the footage then reinterlaced the footage leaving behind some wonderful artifacts throughout the entire footage. He could selectively use an anti-aliasing filter to cover up just the frames needed, but that would be basically the first few and last few frames of almost every sequence throughout the entire show...

The picture of shinji staring at his arm is from the first frame of that sequence, here is the first five frames of the interlaced source, the first TWO are the exact same fame and the third frame is the shift as the footage drops down slightly before zooming in. I'm not aware of any way to fix an isolated 2 frame sequence where the frame is duplicated in the state that it is in like that aside from using an anti-aliasing filter like DAA().
Spoiler :
Image Removed
Mister Hatt wrote:and the ghosting effect looks like what is very typical of an incorrect field order setting in forced deinterlacing, a common issue on mostly soft-telecined content with hybrid or VFR sections.
Okay, now I know that section well, it is the only spot in the series that is like that, and only the ONE character int he bottom part of the screen is like that (for the entire sequence, even when she is not moving). She displays like that no matter how she is decoded (or played back for that matter...she shows up like that on TV). She is actually "ghosted" before you deinterlace as well, that part is definitely something wrong with the source material and not his method of handling the footage.
Spoiler :
Image Removed
Additionally, I HIGHLY doubt an entirely cel animated shot would be "hybrid VFR"...
Mister Hatt wrote:I think your suggestion will definitely fix some stuff in a damaging way, but for most people it will be enough to make it look ok. There is a better way to do it for quality and speed's sake which I've outlined.
I will agree with you there, it is damaging to the footage, but unfortunately the footage was already messed up pretty badly.
Mister Hatt wrote:Making assumptions about the mastering of a DVD, especially something relatively high profile, is not the right way to help anybody.
Oh I'm sorry, I guess you weren't aware of the Evangelion movie release that were HARD LETTERBOXED and poorly encoded...
Spoiler :
Image Removed
I could make a list of titles (Dragon Ball being one) that have had botched releases. Just because something is high profile does not mean it won't get messed up in the process of it getting mastered for release, ESPECIALLY with anime.
Mister Hatt wrote:Incidentally, still cels inside a larger frame was among the first digital composition effects, long before CG and computer rendering.
I haven't researched every bit of how the show was animated, so you may be right on this, but it doesn't seem to be a factor with how the footage was mastered for this "remastered" release.
Mister Hatt wrote:(PS I've seen the actual masters, you're wrong)
I'm thinking you may have worked with the older DVD release if that is the case, else it would be nice if you actually shared an actual method for cleaning up this footage... :roll:

Edit: Images Removed
Last edited by l33tmeatwad on Mon Aug 05, 2013 10:13 pm, edited 1 time in total.
Software & Guides: AMVpack | AMV 101 | AviSynth 101 | VapourSynth 101
PixelBlended Studios: Website | Twitter | YouTube

Mister Hatt
Joined: Tue Dec 25, 2007 8:26 am
Status: better than you
Contact:
Org Profile

Re: Combing

Post by Mister Hatt » Sat Jun 08, 2013 2:42 am

I.. what? How can you say it's not soft-telecined when the DGIndex screenshot from the OP literally shows it?! From the look of things, there is a small section which isn't but I don't think that screenshot is taken at the end of the indexing process as the percentages don't match up. There looks to be maybe a 40 frame sequence which is pure interlacing according to that screenshot, which is 0.6% shy of what I expected it to be.

As far as the masters go, yes I thought this was the original, not a remastering. I've actually seen a lot of hybrid cel stuff tho, because re-animated content is done at a different framerate and then composited with the original. It's quite common to mix 12fps and 18fps content like that especially.

Mister Hatt
Joined: Tue Dec 25, 2007 8:26 am
Status: better than you
Contact:
Org Profile

Re: Combing

Post by Mister Hatt » Sat Jun 08, 2013 2:48 am

x265, could you retake that index screenshot when it's FINISHED indexing and all the boxes are filled? Some data is missing but we can decide on the best way to handle pulldown with the completed picture. Incidentally, any reason to use dgdecodenv on a dvd? I think from what l33tmeatwad said there is some field blending but it also looks like some pure interlacing as well, so I'd be interested to see how that's coded to possibly remove said blending even though it's part of individual fields.

User avatar
l33tmeatwad
Joined: Wed Feb 16, 2005 3:22 pm
Location: Christiansburg, VA
Contact:
Org Profile

Re: Combing

Post by l33tmeatwad » Sat Jun 08, 2013 8:29 am

Mister Hatt wrote:I.. what? How can you say it's not soft-telecined when the DGIndex screenshot from the OP literally shows it?!
I actually missed his screenshot... :oops: ...I can see his source video was encoded differently from the Platinum Edition US release...Here's the video info from the first episode of that DVD release:
Spoiler :
Stream Type: MPEG2 Program
Profile: main@main
Frame Size: 720x480
Display Size: [not specified]
Aspect Ratio: 4:3 [2]
Frame Rate: 29.970030 fps
Video Type: NTSC
Frame Type: Interlaced
Coding Type: P
Colorimetry: BT.470-2 B,G*
Frame Structure: Frame
Field Order:
Coded Number: 42013
Playback Number: 2
Frame Repeats: 0
Field Repeats: 0
Considering his were encoded differently and both have the exact same artifacts (at least from the screenshots given), I afraid the combing artifacts are in the original source these two were encoded from.
Mister Hatt wrote:As far as the masters go, yes I thought this was the original, not a remastering. I've actually seen a lot of hybrid cel stuff tho, because re-animated content is done at a different framerate and then composited with the original. It's quite common to mix 12fps and 18fps content like that especially.
Personally I haven't seen many shows with animation added later, however I will agree that many scenes can possible be mixed with content of varying framerates when it comes to animation, it shouldn't really affect much unless you start throwing in really odd framerates, ones above 24fps, or adding things on top of telecined material.
Software & Guides: AMVpack | AMV 101 | AviSynth 101 | VapourSynth 101
PixelBlended Studios: Website | Twitter | YouTube

User avatar
x265
Joined: Tue Dec 25, 2012 1:52 pm
Org Profile

Re: Combing

Post by x265 » Sat Jun 08, 2013 12:44 pm

Image

Image
“The only true wisdom is knowing that you know nothing.”

User avatar
l33tmeatwad
Joined: Wed Feb 16, 2005 3:22 pm
Location: Christiansburg, VA
Contact:
Org Profile

Re: Combing

Post by l33tmeatwad » Sat Jun 08, 2013 1:10 pm

I took a look at your samples and have some suggestions and apologize for overlooking them when posting my previous reply. First off, as it's already been said, since your source is soft telecine you SHOULD be using force film. To really figure out what's going on, running ignore pulldown flags will give you the raw video (NOT RECOMMENDED FOR REGULAR ENCODING!), and based on the clips you uploaded those combing artifacts are in the raw video. Feel free to use check one of the full episodes with this method to see if there are any leftover interlaced frames in the footage, however I get the feeling those artifacts were a result of the footage originally being interlaced, then inverse telecined improperly before being encoded for DVD. As I stated before, the condition of your clips are the exact same as the hard interlaced version on the release I have (same frames have the exact same combing artifacts).

Also, as Mister Hatt mentioned, you should probably be using the latest version of DMPGDec.
Software & Guides: AMVpack | AMV 101 | AviSynth 101 | VapourSynth 101
PixelBlended Studios: Website | Twitter | YouTube

User avatar
x265
Joined: Tue Dec 25, 2012 1:52 pm
Org Profile

Re: Combing

Post by x265 » Sat Jun 08, 2013 11:00 pm

frames 47015-47412 are blended even after viewing in ignore pulldown mode.
“The only true wisdom is knowing that you know nothing.”

Mister Hatt
Joined: Tue Dec 25, 2007 8:26 am
Status: better than you
Contact:
Org Profile

Re: Combing

Post by Mister Hatt » Sun Jun 09, 2013 8:26 pm

NO NO NO, never EVER use Force FILM for soft-telecine unless it is 100% FILM! That is NOT the case here. You need to use Ignore Pulldown and then figure out which parts are not telecined. You have about 10,000 frames which are interlaced (but possibly coded progressive) that you need to find. Use trim() in avs to filter those as necessary (depends on progressive vs interlace coding how you do it) but the rest of your content should be relatively simple.

For the frame range above (47015-47412), is that frame blending or field blending? Check it with separatefields().selecteven() or something. Field blending is the worst possible situation at this point, frame blending is relatively ok to fix.

PS: I have a tool which actually outputs the ranges of pulldown/interlacism, but I just rewrote it to be platform agnostic so you'll need to send me a d2v so I can test a bit if you want.

User avatar
l33tmeatwad
Joined: Wed Feb 16, 2005 3:22 pm
Location: Christiansburg, VA
Contact:
Org Profile

Re: Combing

Post by l33tmeatwad » Mon Jun 10, 2013 2:19 am

Mister Hatt wrote:NO NO NO, never EVER use Force FILM for soft-telecine unless it is 100% FILM!
It really depends on the source as thing such as post production added studio logos before or after a movie can throw off the percentage from being 100% film, or errors in the mastering. Really it comes down to evaluating the source when it comes to how you are going to handle it, however sometimes a source with 99.9% or higher works with force film. At the end of the day it comes down to properly and carefully analyzing your source and deciding which method works best for handling it properly.
Software & Guides: AMVpack | AMV 101 | AviSynth 101 | VapourSynth 101
PixelBlended Studios: Website | Twitter | YouTube

Locked

Return to “Video & Audio Help”