View Full Version : MeGUI: General Questions and Troubleshooting Thread
MrVideo
8th June 2011, 20:50
The source is a "semi" professional DVD ;)
The "semi" professional kinda worries me, because LigH's question was my next question. What about 2:3 pulldown? After taking Groucho2004's advice and posting a sample, please provide a link here to that posting.
This I gotta see.
rapscallion
8th June 2011, 21:05
If you have NTSC, always check first if it is 3:2 Pulldown = Telecine instead of interlacing; in that case, IVTC can make it progressive (more or less) losslessly. PAL does not use 3:2 Pulldown.
May I ask, how do you check that ?
MrVideo
8th June 2011, 21:50
May I ask, how do you check that ?
Visually. No really, that is how I find the cadence locations when editing.
When using a video editor, like VideoReDo, you step through the frames in any area that has action. If the video has 2:3 (technically it is 2:3, not 3:2) pulldown, you'll see three frames that have no interlacing artifacts and then two frames that do (a double image in the single frame).
If that pattern is there, it has 2:3 pulldown. If you see anything else, i.e., every frame within the action is double imaged and/or blended (25 fps material converted to 29.97 fps or film transferred to 29.97 that was not 2:3 converted), then it doesn't have 2:3 material.
magnetite
8th June 2011, 23:23
I have a question about MeGUI's quality settings. What settings determine the quality of your video aside from a low CRF value (eg. I'm using CRF 14)?
pereant
8th June 2011, 23:35
After taking Groucho2004's advice and posting a sample, please provide a link here to that posting.
This I gotta see.
I've started a new thread as suggested: http://forum.doom9.org/showthread.php?p=1506819
Thanks for all ! You're great helpful people :)
@ rapscallion:
The AVS Script Creator has an automatic interlacing tester (Filters: Deinterlacing – Analyze...); but no automatic detection will ever be as smart as your brain.
@ magnetite:
In general, there are two groups of factors which are related to the video quality (or, more exactly: the amount of quality loss compared to the original video source):
a (major): Quantization, bitrate, filesize. CRF 14 could already be considered "archive grade accuracy", many people would already be satisfied with CRF 18.
b (minor): Encoding efficiency, redundancy removal efforts. May reduce the size by permilles, maybe even percents, when comparing videos with a constant quantization. Almost only interesting if the resulting filesize is limited, and the quantization is a result of the bitrate distribution, not the other way round. If reliably low quality loss is more important than limited file sizes, don't tweak your encoder too much; especially "psychovisual enhancements" are made for reducing the annoyance of already perceptive loss (tweaking "subjective quality"), but non-perceptive loss does not need that (already has good "objective quality" = small measurable difference).
It's like racing cars: Spoilers (b) are no substitute for more cylinder capacity (a).
Music Fan
9th June 2011, 09:07
Concerning the de-interlacing of Ntsc dvd movies, we can add another solution : when it's soft-pulldown (like most of movies on Ntsc dvd), we can remove the pulldown with TSMuxer (check "remove pulldown").
You'll get a progressive mpeg-2 file in 23.976 fps rid of 3:2 pulldown, without re-encoding of course.
Then you open it in Avisynth or MeGui which won't have to de-interlace it.
We can see if the pulldown is soft or hard with GSpot. Two examples gathered in same picture ;
http://img139.imageshack.us/img139/4346/gspotht9.jpg
Above : hard pulldown or video source
Below : movie source with soft pulldown.
MrVideo
9th June 2011, 09:18
when it's soft-pulldown (like most of movies on Ntsc dvd), we can remove the pulldown with TSMuxer (check "remove pulldown").
I forgot all about that. Good point. The DVD player is doing the 2:3 pulldown. The original encoding is a lot nicer as well, as they could actually encode at a higher bitrate, since there are less frames, or they can add more content per layer.
Music Fan
9th June 2011, 09:41
Actually, if we remove pulldown, I don't know what happens if we wanna keep the subtitles, because I wonder if their timecode is in 23.976 fps or 29.97 fps.
Anyway, if it's in 29.97 fps, it's not a huge problem because we can convert the subtitles timecode from 29.97 to 23.976 fps with several programs, but it's important to do to keep the sync between video and subtitles when we encode the video in 23.976 fps (whatever the de-interlacing method).
flebber
9th June 2011, 12:17
Is there an option in megui oneclick to force it to use ffms instead of dgi? I am using mkv input so I can use ffms.
pereant
9th June 2011, 12:48
Is there an option in megui oneclick to force it to use ffms instead of dgi? I am using mkv input so I can use ffms.
Doesn't it use ffmsindex on mkv files automatically?
If you load a mkv file manually with the file indexer GUI it chooses ffmsindex and dgindex is greyed out.
I'm a clueless individual though - just trying to provoke minds ;)
flebber
9th June 2011, 13:11
yeah the file indexer gets it right and uses ffms.
raj1402
9th June 2011, 13:50
for dvdrip whats the correct frame rate
29.970 or 25fps or 23.976fps
Music Fan
9th June 2011, 13:55
It depends on the dvd. You can analyze files with MediaInfo or GSpot.
Zathor
9th June 2011, 14:52
Is there an option in megui oneclick to force it to use ffms instead of dgi? I am using mkv input so I can use ffms. The only way to force using ffms when dgi is available is renaming/delecting your license.txt file.
magnetite
9th June 2011, 20:33
@ LigH
Oh okay. Another thing I wanted to mention is that this video was 24 FPS but it's been run through Interframe so it's 60 FPS. With more information in the file I would assume that CRF 14 would be good for a 24 FPS source, but since it's 60 FPS that it might need a lower value to be similar to the source.
LigH
10th June 2011, 07:20
Current x264 versions already respect the frame rate in their calculation of acceptable loss (to indirectly support variable frame rates too): The longer you can see each frame, the less loss is acceptable for a given rate factor.
Inspector.Gadget
10th June 2011, 20:30
Actually, if we remove pulldown, I don't know what happens if we wanna keep the subtitles, because I wonder if their timecode is in 23.976 fps or 29.97 fps.
Anyway, if it's in 29.97 fps, it's not a huge problem because we can convert the subtitles timecode from 29.97 to 23.976 fps with several programs, but it's important to do to keep the sync between video and subtitles when we encode the video in 23.976 fps (whatever the de-interlacing method).
It shouldn't matter. Unlike PAL speedup, NTSC pulldown doesn't change the play time of the video, so subs should always be in sync.
Edit: Also for pulldown removal DGPulldown works well. The caveat of course when removing pulldown is to always make sure DGIndex reports 100% FILM so you don't create sync problems. GSpot will display 3:2 pulldown if some number of flags are found, not if the content is all only film, so it's not a reliable guide.
bamsebo
10th June 2011, 22:25
I have an h264 1920x1080i25 broadcast ts that I am trying to re-encode using MeGui and x264.
however I am seeing some odd artefacts when playing back the avs script even before encoding. I have tried various deinterlacing options without any luck.
http://img221.imageshack.us/img221/4919/vlcsnap00006r.png (http://imageshack.us/photo/my-images/221/vlcsnap00006r.png/)
Uploaded with ImageShack.us (http://imageshack.us)
any help would be appreciated.
thanks Barry
Playing the original ts file doesnt show any artefacts
rapscallion
10th June 2011, 23:04
Concerning the de-interlacing of Ntsc dvd movies, we can add another solution : when it's soft-pulldown (like most of movies on Ntsc dvd), we can remove the pulldown with TSMuxer (check "remove pulldown"). You'll get a progressive mpeg-2 file in 23.976 fps rid of 3:2 pulldown, without re-encoding of course.
Then you open it in Avisynth or MeGui which won't have to de-interlace it.
We can see if the pulldown is soft or hard with GSpot. Two examples gathered in same picture ;
.
Is it correct that if we load a 1080i *.m2ts file into Tsmuxer and the "remove pulldown" box is greyed out, that we can then use Megui to encode progressive, rather than interlaced ?
I have to admit that this interlaced stuff really makes my head hurt.
flebber
10th June 2011, 23:06
The only way to force using ffms when dgi is available is renaming/delecting your license.txt file.
That did the trick, just what I wanted FFMS with oneclick. Thank you.
Music Fan
11th June 2011, 06:43
It shouldn't matter. Unlike PAL speedup, NTSC pulldown doesn't change the play time of the video, so subs should always be in sync.
Right. But I guess a player could be lost with two different timecodes ; one in 23.976 fps for the video and the other in 29.97 fps for the subtitles.
Also for pulldown removal DGPulldown works well. The caveat of course when removing pulldown is to always make sure DGIndex reports 100% FILM so you don't create sync problems.
I believe DGPulldown only works with avisynth, which re-encode the video.
TSMuxer can open a mpeg file, remove its pulldown and export it in mpeg, without re-encoding.
Then you can re-encode it (without thinking to the de-interlacing method) or play it in 23.976 fps, in a TS or MKV container. It's useful for people whose player can't make reverse pulldown but can send 23.976 fps when there is no pulldown.
GSpot will display 3:2 pulldown if some number of flags are found, not if the content is all only film, so it's not a reliable guide.
Yeah, for some exotic files, but for a movie on Ntsc dvd, it's generally entirely in 23.976 fps with a pulldown flag.
But of course, we should always verify it with Virtual Dub (for example) with a short extract of the movie before re-encoding it (before and after using TSMuxer).
Music Fan
11th June 2011, 07:00
Is it correct that if we load a 1080i *.m2ts file into Tsmuxer and the "remove pulldown" box is greyed out, that we can then use Megui to encode progressive, rather than interlaced ?
No, the pulldown flag in mpeg is used on dvd, which can't be in 1080 lines.
Unless the 1080i mpeg file comes from US HDTV, but I don't know if then can send soft pulldowned video or only hard pulldowned.
And on Blu-ray, I believe there is no 3:2 pulldown flag, or there is a different kind of flag.
If it's a hard pulldowned movie source, you can't just encode it in 60 hz without de-interlacing it. You have to de-interlace it in 23.976 fps.
If it's a video source, there is no pulldown, so you keep it interlaced or not, as you want.
And if it's a mix of movie and video sources, that won't be simple.:D
When you wanna re-encode a file, the first thing to do is to know where it comes from. It's half the work and will help you to encode it correctly. But of course, you also have to look at it carefully and analyze it with Media info and Gspot.
rapscallion
11th June 2011, 15:27
Thanks Music Fan! I had a feeling it wasn't going to be simple.
The video is a retail BD of a concert from broadcast tv, called David Foster- Hitman Returns.
Media Info :
Video
ID : 4113 (0x1011)
Menu ID : 1 (0x1)
Format : AVC
Format/Info : Advanced Video Codec
Format profile : High@L4.1
Format settings, CABAC : Yes
Format settings, ReFrames : 2 frames
Duration : 2h 22mn
Bit rate mode : Variable
Width : 1 920 pixels
Height : 1 080 pixels
Display aspect ratio : 16:9
Frame rate : 29.970 fps
Resolution : 8 bits
Colorimetry : 4:2:0
Scan type : Interlaced
Scan order : Top Field First
Single stream @ 37gb, and as you can see, MI shows "interlaced, TFF, @29.70fps". However, when I load it into Virtual Dub and advance frame by frame, I don't see any interlacing whether I set Preview Mode to "progressive" or "Bob TFF".
Music Fan
11th June 2011, 15:56
It's weird, I guess this show has been filmed in interlaced.
In progressive mode in Virtual Dub, you should see blended frames in the moving parts.
rapscallion
11th June 2011, 16:25
It's weird, I guess this show has been filmed in interlaced.
In progressive mode in Virtual Dub, you should see blended frames in the moving parts.
....and, I don't.
Zathor
11th June 2011, 16:48
Can someone help me with this ImageSource issue? Whenever I use ImageSource in an AVS file, no MeGUI version opens it after 1911! Can someone help me? I have Avisynth 2.58 installed and I use Windows 7 64-bit.
Post some example image files. I have tested this with a simple script and it does work in MeGUI 2031.
ImageSource("d:\megui_icon%d.png", 1, 2, 25)
ConvertToYV12()
MrVideo
11th June 2011, 16:53
It's weird, I guess this show has been filmed in interlaced.
In progressive mode in Virtual Dub, you should see blended frames in the moving parts.
First off, you can't film with video. I have no clue how you put film into a video camera. :D
Nothing stops them from recording @ 29.97 progressive, but dumping it out at 59.94 fields/sec interlaced for Blu-ray, which they have to, because 1080p29.97 doesn't exist as a Blu-ray format.
Same thing with Doctor Who. They shoot it 1080p25, but because 1080p25 is not a Blu-ray format, it is converted to 1080i25, so you do not see any interlacing artifacts.
Music Fan
11th June 2011, 17:03
First off, you can't film with video. I have no clue how you put film into a video camera. :D.
What word should we use in place of "film" ? In french we say "filmer" for both interlaced and progressive (cinema) cameras.
Nothing stops them from recording @ 29.97 progressive, but dumping it out at 59.94 fields/sec interlaced for Blu-ray, which they have to, because 1080p29.97 doesn't exist as a Blu-ray format.
Same thing with Doctor Who. They shoot it 1080p25, but because 1080p25 is not a Blu-ray format, it is converted to 1080i25, so you do not see any interlacing artifacts.
Yes, and it can be encoded with the "fake interlaced" AVC's option.
MrVideo
11th June 2011, 17:13
What word should we use in place of "film" ? In french we say "filmer" for both interlaced and progressive (cinema) cameras.
Two words, for starters, captured and recorded, with recorded being the better one. You can't really use "taped" these days either, as a lot of stuff is recorded onto hard drives, or other removable media.
Yes, and it can be encoded with the "fake interlaced" AVC's option.
There is that.
Inspector.Gadget
12th June 2011, 01:45
I believe DGPulldown only works with avisynth, which re-encode the video.
No, it's a standalone app that doesn't have any connection to DGIndex. All it does is take in MPEG-2 with 3:2 flags and spit out the same stream without the flags, no re-encoding.
Yeah, for some exotic files, but for a movie on Ntsc dvd, it's generally entirely in 23.976 fps with a pulldown flag.
If only this were the case :) . Studio intros and similar encoded in the same title are often hard-telecined or true interlaced or even 30p encoded as interlaced. You're right, though, in that newer DVDs (last 5 years or so) seem to be eliminating that pathological behavior.
any help would be appreciated.
thanks Barry
Playing the original ts file doesnt show any artefacts
That looks like DGAVCIndex failing on an interlaced H.264 stream because it's built on an old version of FFMPEG and doesn't handle all types of H.264 interlaced content. Try re-muxing to MKV and using FFMSIndex instead. Alternately, that may be a transmission error in your stream and the decoder in your player is helpful enough to conceal errors by dropping/duplicating frames.
Music Fan
12th June 2011, 07:11
No, it's a standalone app that doesn't have any connection to DGIndex.
Ok, I was thinking to DGIndex.
Are you sure that DGPulldown can remove pulldown and not only add it ?
Because I don't see "remove pulldown" option on this picture ;
http://neuron2.net/dgpulldown/dgpulldown.html
Inspector.Gadget
12th June 2011, 18:16
Music Fan: See the DGPulldown.txt file that's packed with the program:
Note that if the custom rate conversion is selected, and if the source rate is
specified as equal to the destination rate, then all pulldown is removed and the
stream is flagged as having a rate equal to the specified destination rate.
So set both to be 23.976 and get an output file with all pulldown flags removed.
Music Fan
13th June 2011, 08:15
Ok, thanks.
rapscallion
13th June 2011, 17:45
@Music Fan
Re my previous interlacing posts. Could you please download this 1 min vid clip (250 MB) and tell me if it's truly interlaced or not :
http://www.megaupload.com/?d=AIEUM1DS
I can't get any info via Gspot and, as I mentioned, I don't see any blending in VD. Maybe I just don't know what to look for, but would sure appreciate your taking a look at it for me. Thanks
Music Fan
14th June 2011, 09:29
It's interlaced ! We can see it clearly in Virtual Dub on the pianist and singer's hands.
So you can keep it interlaced or de-interlace it if you want.
bamsebo
14th June 2011, 10:34
@Music Fan
Re my previous interlacing posts. Could you please download this 1 min vid clip (250 MB) and tell me if it's truly interlaced or not :
http://www.megaupload.com/?d=AIEUM1DS
I can't get any info via Gspot and, as I mentioned, I don't see any blending in VD. Maybe I just don't know what to look for, but would sure appreciate your taking a look at it for me. Thanks
mediainfo:
Video
ID : 4113 (0x1011)
Menu ID : 1 (0x1)
Format : AVC
Format/Info : Advanced Video Codec
Format profile : High@L4.1
Format settings, CABAC : Yes
Format settings, ReFrames : 2 frames
Codec ID : 27
Duration : 59s 993ms
Bit rate mode : Variable
Bit rate : 28.7 Mbps
Width : 1 920 pixels
Height : 1 080 pixels
Display aspect ratio : 16:9
Frame rate : 29.970 fps
Color space : YUV
Chroma subsampling : 4:2:0
Bit depth : 8 bits
Scan type : Interlaced
Scan order : Top Field First
Bits/(Pixel*Frame) : 0.461
Stream size : 205 MiB (81%)
Music Fan
14th June 2011, 10:43
@ bamsebo : as said above, mediainfo is not enough to be sure that the video is really interlaced.
LigH
14th June 2011, 14:06
"Interlaced encoding" is not equal to "interlaced content".
Even though a video may be encoded in interlaced mode (both fields more or less independently), it may or may not be combed (the fields having a regular timing difference).
The only way to tell for sure is to watch the video field-wise and check if the motion is progressing constantly from field to field without interruptions or blending ghost.
rapscallion
14th June 2011, 15:16
Thanks Music Fan !!
@bamsebo: also, see my post on the previous page, # 6226
@LigH -exactly why I needed some help, and I thank all for the responses.
GG-Xtreme
16th June 2011, 04:24
Sorry, didn't see this thread when I posted. This is my question:
http://forum.doom9.org/showthread.php?t=161630
namus
17th June 2011, 21:40
I am not sure whether this is a DGIndex or meGUI problem. But I am asking anyway. I can upload samples if needed.
I have a very strange source (DVD9). DGIndex preview informs me it is NTSC interlaced (fps 29.97) and last line of my d2v file says 100% VIDEO. But while going through frame by frame, it does not look like a interlaced video at all, i.e. there is no field line visible even during motion. Rather, I find a duplicate frame in every 6th frame. So it must be some kind of 3:2 pull down, but with all progressive frames. It is not even usual cycle of 5, but 6. But then why does DGindex say it is interlaced?!
What is even more strange that meGUI automatic detection of source type fails with the d2v file. It just gives me the error that it "cannot detect the source type"! This is the first time I have encountered such a message from meGUI.
So what is the right approach? I can always use decimate(6) to manually remove the duplicate 6th frame and forget about interlacing, which is actually not there. But is that really correct, or am I missing something. Is there a possibility of making a mistake by manual decimate?
Thanks in advance! :)
MrVideo
17th June 2011, 23:03
I have a very strange source (DVD9). DGIndex preview informs me it is NTSC interlaced (fps 29.97) and last line of my d2v file says 100% VIDEO. But while going through frame by frame, it does not look like a interlaced video at all, i.e. there is no field line visible even during motion. Rather, I find a duplicate frame in every 6th frame. So it must be some kind of 3:2 pull down, but with all progressive frames. It is not even usual cycle of 5, but 6. But then why does DGindex say it is interlaced?!
What are you using to view the video and step thru it frame-by-frame? Lots of programs only show you a field and not the complete frame. As a result, it looks progressive. Plus, in that mode, it should be a duplicate frame every 5th frame, not 6th. 2:3 pulldown results in the pattern starting over every 5 frames, hence 2:3 pulldown.
Using a program, like the VideoReDo editor, will show you frames, not fields.
namus
18th June 2011, 00:56
What are you using to view the video and step thru it frame-by-frame? Lots of programs only show you a field and not the complete frame. As a result, it looks progressive. Plus, in that mode, it should be a duplicate frame every 5th frame, not 6th. 2:3 pulldown results in the pattern starting over every 5 frames, hence 2:3 pulldown.
I used two methods:
1. Loaded the VOB files in DGIndex and moved frame by frame. There was no visible interlacing in any frame.
2. Then I saved the d2v file (using Honor pulldown flag in DGIndex) and loaded in a simple avisynth script and viewed in Virtualdub and moved frame by frame.
In both cases I did not see any interlacing (field lines). But I found duplicate frames at every 6th frame (NOT 5th).
Is it possible that I am encountering some PAL to NTSC converted DVD with progressive frames? This thread discusses such a possibility: http://forum.doom9.org/showthread.php?t=42692 :
If it was telecined from a PAL (25fps) master, as you seem to be saying, then add this to your script:
Telecide(Guide=3)
Decimate(6)
If this is the case for me, I have three questions:
1. Do I use Telecide even if I don't "see" any interlacing? Or just Decimate(6) should be fine?
2. If there is no interlacing, why DGIndex thinks it to be interlaced?
3. Why MeGUI fails to detect the source (may be relevant question for this thread)?
MrVideo
18th June 2011, 02:09
1. Loaded the VOB files in DGIndex and moved frame by frame. There was no visible interlacing in any frame.[quote]
DGIndex has a viewer? I wouldn't know, since I do all viewing and editing with VideoReDo.
[quote]2. Then I saved the d2v file (using Honor pulldown flag in DGIndex) and loaded in a simple avisynth script and viewed in Virtualdub and moved frame by frame.
While I've used Virtualdub, I've only done so with inverse telecine files.
In both cases I did not see any interlacing (field lines). But I found duplicate frames at every 6th frame (NOT 5th).
Just because you don't "see" separate fields, doesn't mean that it isn't interlaced. By definition, both NTSC and PAL are interlaced "analog" formats. If the frame rate is 29.97, it is interlaced. Keep in mind, that you are viewing complete frames. If the complete video frame is from a single frame source, it will "look" as if it is progressive. PAL is the same way, especially when film is transferred to video. A single film frame = a single PAL frame, making it look progressive, when it really isn't. The every 6th frame is worrysome.
Is it possible that I am encountering some PAL to NTSC converted DVD with progressive frames? I've never heard of PAL being converted that way to NTSC, as most standards conversion does blending to avoid that frame jerkiness. It will still be interlaced.
2. If there is no interlacing, why DGIndex thinks it to be interlaced?
Because it is interlaced, as explained above.
Music Fan
18th June 2011, 10:19
The easiest way to see if your video is progressive or not is to open your mpeg or vob file directly in Virtual Dub Mod (without avisynth). Choose "progressive" in the preview field mode to blend the fields without de-interlacing.
You can also do that with Virtual Dub (not Mod) which needs a mpeg filter ("MPEG2.vdplugin" file to put in "C:\Program Files\VirtualDub\plugins32" directory).
MrVideo
18th June 2011, 11:25
The easiest way to see if your video is progressive or not is to open your mpeg or vob file directly in Virtual Dub Mod (without avisynth). Choose "progressive" in the preview field mode to blend the fields without de-interlacing.
I must be missing something here. And just how is blending the fields supposed to give you info as to if it is interlaced or not?
As mentioned above, NTSC analog output is interlaced, it can't be progressive. If the frame rate was 23.976, then the video on the DVD would be progressive, with the DVD player doing to 2:3 into interlaced mode.
Since it is a 29.97 frames/sec video, and because the program said it was interlaced, why is it so hard to be believe that the video isn't interlaced? This is a DVD we are talking about. It is meant for playback on NTSC analog display devices, which are only interlaced.
If you were to watch the current series of Doctor Who, starting with series 1, you'd swear that it was progressive. Yet, the DVD, and OTA broadcast, is interlaced. It is shot progressive, but converted to interlaced for DVD/Blu-ray/OTA. Yet, you can look at each frame, and swear the frames were progressive.
Work with this DVD as is. Don't deinterlace it, as you'll accomplish nothing, other than potentially reducing vertical resolution.
Why does everyone seem to hate interlacing? Don't destroy vertical resolution and spatial information needlessly. If the device you are going to play it on (a handheld of some sort, etc), then you may have no choice, otherwise, leave well enough alone.
Music Fan
18th June 2011, 12:29
I must be missing something here. And just how is blending the fields supposed to give you info as to if it is interlaced or not?
Because you see it of course. Something like that ;
http://www.windowsmoviemakers.net/PapaJohn/54/Interlaced-Versus-Progressive_files/image010.jpg
Two different frames are blended by Virtual Dub. You won't see that if the source is progressive because the fields come from the same frame.
As mentioned above, NTSC analog output is interlaced, it can't be progressive.
As said above, interlaced encoding doesn't mean interlaced content (even if it was originally interlaced, because it could have been de-interlaced, I already saw that). And as dvd doesn't support 29.97p, it has to be encoded in 59.94i.
http://en.wikipedia.org/wiki/Telecine#2:2_pulldown
2:2 pulldown is also used to transfer shows and movies, photographed at 30 frames per second, like Friends and Oklahoma! (1955),[3] to NTSC video, which has 60 Hz scanning rate.
Interlaced encoding of a progressive source doesn't create new frames, it only divides 30 frames in 60 half frames (fields).
Work with this DVD as is. Don't deinterlace it, as you'll accomplish nothing, other than potentially reducing vertical resolution.
Why does everyone seem to hate interlacing? Don't destroy vertical resolution and spatial information needlessly. If the device you are going to play it on (a handheld of some sort, etc), then you may have no choice, otherwise, leave well enough alone.
Encoding in 29.97p won't reduce quality if the file is not "really" interlaced and contain actually 29.97 different frames and not 59.94.
You don't even have to make a simple de-interlacing, you can encode directly your interlaced mpeg file in 29.97p, the encoder will blend the top and bottom fields and it won't be a problem as they come from the same frame.
But this situation is rare, it's generally really interlaced (59.94 different frames).
When it's progressive, we can consider it as a progressive encoding with a flag for the player (like in the soft pulldown system but simpler).
It's like movies on Pal dvd : I don't know if mpeg-2 encoders divide the 25 frames in 50 half frames or if they encode 25 frames with a flag, but it doesn't matter, it won't change the result if you encode it in 25p (in h264 for example).
MrVideo
18th June 2011, 17:26
Two different frames are blended by Virtual Dub. You won't see that if the source is progressive because the fields come from the same frame.
Two different FIELDS, not frames are blended. Doing that would make sense, IFF the program you are using to display the video doesn't show both fields.
As said above, interlaced encoding doesn't mean interlaced content (even if it was originally interlaced, because it could have been de-interlaced, I already saw that). And as dvd doesn't support 29.97p, it has to be encoded in 59.94i.
Technically, interlaced video is interlaced video, period. There is no such thing as "interlaced content." You could have a video of still images, whereas each image stays on the screen for a number of seconds and then rapidly fades from one slide to the next. You'd swear that the video was progressive, because both fields looked the same, until you got to the rapid fade where two fields could look different. Yet, the video is 100% interlaced. As for your note about DVD and 29.97p, I thought that is what I was trying to say.
Also, while not as common as it should be, the value given should be in frames/sec + the frame encoding, in order to avoid confusion as to whether the video is frames or fields per sec. So, instead of 59.94i, it should be 29.97i. While 480i59.94 doesn't exist, 1080i59.94 can (not on Blu-ray though).
Interlaced encoding of a progressive source doesn't create new frames, it only divides 30 frames in 60 half frames (fields).
I never said that it didn't. That is what I was trying to say. I know how interlaced video works, as I've been working with video for over 50 years.
Encoding in 29.97p won't reduce quality if the file is not "really" interlaced and contain actually 29.97 different frames and not 59.94.
Actually, even video of stills has two different fields for each frame, as one field is of 1/2 of the image and the other field the other half. There will be differences in the two fields. When you deinterlace, blending of some sort is taking place, which reduces the vertical resolution of the image. If the video was such that the even lines were black and the odd lines were white, deinterlacing would probably result in an image that was gray. You'd swear that the video was interlaced and not progressive because of the alternating black and white lines, yet it could be progressive because the source could be of an image that has alternating black and white lines.
Deinterlacing for the sake if deinterlacing is not always wise.
You don't even have to make a simple de-interlacing, you can encode directly your interlaced mpeg file in 29.97p, the encoder will blend the top and bottom fields and it won't be a problem as they come from the same frame.
You don't want any blending, as that will reduce vertical resolution. You want to just join the fields into a progressive frame.
But this situation is rare, it's generally really interlaced (59.94 different frames).
59.94 FIELDS, not frames. As mentioned above, it is always interlaced.
It's like movies on Pal dvd : I don't know if mpeg-2 encoders divide the 25 frames in 50 half frames or if they encode 25 frames with a flag, but it doesn't matter, it won't change the result if you encode it in 25p (in h264 for example).
Depends on how you set up the AVISynth script, i.e., to deinterlace the fields or to join the fields.
Music Fan
18th June 2011, 17:51
I'm not gonna answer everything, too long. And we apparently both know how it works, we say the same things (but english is not my mother tongue).
But at least one thing ;
59.94 FIELDS, not frames.
Fields = frames for the camera when it's interlaced. But it keeps one line on two.
59.94 fields = 59.94 different frames wich can be reconstructed with a good de-interlacer to get 59.94p on a recent screen.
When your interlaced file contains actually 29.97 different frames, you don't have to keep it interlaced, it's not useful and you loose bitrate for nothing.
You don't want any blending, as that will reduce vertical resolution. You want to just join the fields into a progressive frame.
blend = join when the source is progressive, it's a simple 2:2 pulldown.
vBulletin® v3.8.11, Copyright ©2000-2025, vBulletin Solutions Inc.