HQ Conversion Software

[Pinstripes23]

the test sample Fulci provided is PAL and i have finally found the answer as to why ffmpeg decimated frames as it converted it to 23.97 fps. the reason is that all PAL DVD's are created by taking the 24 fps film and applying what is called 2:2 pulldown,

I don't doubt that that technique is used but ALL? No what they usually do is take the 24fps film and speed the movie up, thus reducing the running time of a PAL encoded movie. And the audio has to be changed in length also. Don't you notice a pitch difference in voices of a PAL DVD compared to when you watched it in the theater. Or compare it to the NTSC version. Compare Fulci's pal2ntsc2 clip to his original PAL clip and tell me if you notice the pitch difference in the audio, and also compare the video lengths. He slowed the movie down to 23.976fps(no decimating), thus the length will be longer. If you want to know THIS is the proper way for NTSC DVD's. If you were to get the retail NTSC version of this movie you will see it will basically have the same characteristics as this conversion he did.

I and manono already explained why ffmpeg decimated the frames - in order to keep the video length the same and convert to 23.976fps, the encoder must remove 1 frame/sec. I mean think about it - 25fps and you're trying to go to 23.976fps - (encoder)solution = remove 1 frame within the 25 frames displayed per second.

as you can see the process of applying 3:2 pulldown does indeed cause the video to be interlaced and this also explains why tmpgenc also deinterlaces the video when you choose to have it inverse telecine the video.

which means that your supposition that by definition all NTSC DVD's that run at 29.97 fps are interlaced is in fact true. to confirm i ripped a number of different NTSC DVD's that i have and sure enough every single one of them was interlaced.

When viewing the movie yes you see the interlacing, but that's not how it was originally stored on the dvd. It was originally encoded progressively to 23.976fps, then pulldown flags are inserted into the video stream so that playback will be 29.97fps.

if you analyse each of the submitted videos you will note that mine is the only one made up entirely of unique frames, ffmpeg didn't remove a single unique frame, it's just that both the original and Fulci's has duplicate frames.

No because the original pal clip is sped up from 24fps, like I said up above, so all the frames that were in the 24fps film is used and unique. Nothing is duplicated.



Edit: here is from your link

2-2 Pulldown

PAL (and SECAM) use a 50 field per second scan rate. The most common method of putting 24 fps film on PAL video is called 2-2 pulldown.and each film frame is scanned twice. The film movement is sped up slightly to be 25 fps rather than 24 fps. This causes the film to run in less time and also causes music in the audio track(s) to be played back at a slightly higher pitch. However the process is referred to as 2-2 pulldown (referring to the pulldown claw film advance) rather than 2-2 pullup, the term "pullup" refers to the speeding up in time of a program.

A less common version of 2-2 pulldown keeps the 24 fps film playback speed, but has scanning of every twelfth film frame 3 times. De-interlacers with 2-2 pulldown sensing for PAL need to take this into account, otherwise the threesomes will be treated as a bad edit every 12 frames and/or the de-interlacer may fail to lock onto the 2-2 pulldown cadence.

Look at what I bolded.

[deadrats]

one more thing, i just authored a DVD using DVDAuthorGUI, a free open source dvd authoring app, and it authored the dvd within 14 seconds.

i then burned it onto a DVD+R and the resulting DVD played beautifully on both windows and linux using a HP DVD-ROM and on a standalone DVd player (Mintek) hooked up to a standard tv.

anyone doesn't believe me, feel free to try it yourselves.

here's the complete process i used:

use gui4ffmpeg to convert the PAL sample to NTSC sample, 16:9, ac3, 23.97, cbr, took under 1 minute

used dvdauthorgui to author dvd, took 14 seconds

burned dvd, took less than 20 seconds.

so inside of 2 minutes i had transfered the PAL test clip to a NTSC DVD.

[jagabo]

I was wondering what software would do the conversion with the least amount of quality loss, i dont care how long it takes.

He didn't ask for the fastest method that gave the jerkiest results.

Try burning Fulcilives' NTSC MPG files to DVD. Look closely at the sequence right after the hatch on the BMW closes, as the people walk up the path (starting at 57 seconds in the original PAL MPG file). Your video shows an obvious jerk every second. Fulcilives' two videos play much more smoothly.

[FulciLives]

I started a new job today so forgive me for "missing" so much "action" LOL

I have good news and I have bad news.

I know why my PAL2NTSC1.MPG file has a few interlaced frames at the very end (the last 1 -2 seconds).

Follow my process here ... it makes sense but I don't know how else to tell it without going over the steps I took one by one.

The first time around I had used the multiplexer in MPEG2VCR to create my MPEG file. I loaded it into VirtualDubMod and realized it was OUT-OF-SYNC and I thought it was the fault of the multiplexer in MPEG2VCR because I did not yet realize that VirtualDubMod does not like DGPulldown files (at least not the 25fps ---> 29.970fps files).

So thinking it was the multiplexer in MPEG2VCR to fault I decided to create an MPEG file by authoring the separate M2V and AC-3 using TMPGEnc DVD Author (I recently upgraded to the 2.x version from the 1.6x version).

I then opened the VOB file into VideoReDo and did quickstream fix on it to get a MPEG file.

That was how the first PAL2NTSC1.MPG clip was made.

Well I didn't keep any of my intermediate files SO I just re-did it all using the same exact steps and the same exact AviSynth script and tools. I did update DGMPGDec, AviSynth and DGPulldown to the latest versions.

The M2V as encoded had no interlaced frames (I eyeballed it with VirtualDubMod). The M2V that DGPulldown spit out (dong 25fps ---> 29.970fps) had no interlaced frames (again I eyeballed it with VirtualDubMod). I was happy!

I then went about making the MPEG file the same way ... authored with TMPGEnc DVD Author 2.x but this time I loaded the VOB file directly into VirtualDubMod before trying the VideoReDo step of "extracting" to an MPEG file.

Well what do you know ... interlaced frames just as before! I went ahead and converted the VOB to MPEG with VideoReDo anyway and as I expected the interlaced frames where still there in the MPEG file.

So now I'm thinking "TMPGEnc DVD Author 2.x is doing something bad" so I took the M2V and the AC-3 (this time I kept all the intermediate files) and then used DVDauthorGUI to create a DVD. I loaded the VOB file into VirtualDubMod.

Well what do you know ... NO interlaced frames!

So then I decided what if I just use MPEG2VCR to multiplex and not run the files first through a DVD author program. One apparently screws it up (TMPGEnc DVD Author 2.x) whereas the other doesn't screw it up (DVDauthorGUI). So I went ahead and multiplexed with MPEG2VCR and it does play back correctly with PowerDVD but not in VirtualDubMod but again NO interlaced frames. So yeah I should have just done that to begin with (multiplexed with MPEG2VCR) and we never would have had an interlaced frame issue making deadrats think he "won" somehow. What a complete idiot he is ... anyways ...

We have discovered something important ... TMPGEnc DVD Author 2.x apparently has issues with the DGPulldown PAL 2 NTSC method.

I am hoping that this is simply a "end-of-stream" issue because I would like to continue using TMPGEnc DVD Author 2.x and I can live with a second or two of interlaced frames at the very end of the video since that would be in the credits of a TV or MOVIE etc. and I can deal with that.

So again it is the authoring stage in TMPGEnc DVD Author that introduces the interlaced frames ... not sure how that happens ... but that's the facts folks.

Now of course the question is ... did older version of TMPGEnc DVD Author have this issue or was this something new to version 2.x ... I have a bunch of PAL2NTSC DGPulldown conversions that were made with TMPGEnc DVD Author 1.5/1.6 so I will have to get one of those and RIP and check the very end of the last VOB file and take a look see.

I wonder ... is there a tool (VirtualDubMod?) that you can use to "search" for interlaced frames without having to "eyeball" it?

Here I was all happy to finally upgrade to TMPGEnc DVD Author 2.x so when I do my PAL2NTSC conversions I can include a 2nd audio track (for the commentary) since the older version only allowed 1 audio track. I also liked being able to make a 16x9 WS enhanced menu instead of being stuck with the 4:3 menus of the older version.

Something just dawned on me ... I know some MPEG-2 encoders (HCenc can do this as can CCE) have an option where you can force the last encoded frame to be an "I" frame. Hmmm ... me wonders if that will have an effect at all!

OK I think I am going to re-encode this one more time with that option and see how TMPGEnc DVD Author 2.x handles it.

BTW the new "corrected" PAL2NTSC using DGPulldown is uploaded. Again this file was created the exact same way (same programs and same AviSynth script etc.) as pal2ntsc1.mpg but this time I did not run the file through TMPGEnc DVD Author 2.x

Here is the link: http://rapidshare.com/files/6258961/pal2ntsc1fixed.mpg
File Name: pal2ntsc1fixed.mpg
File Size: 87.0 MB
The file size ended up .2 MB bigger even though I used the same bitrate in HCenc but we all know how that can vary a tiny bit from one encode to the other.

- John "FulciLives" Coleman

[FulciLives]

I just did another re-encode with HCenc but this time I selected the option that forces the last frame to be an "I" frame.

No joy though as TMPGEnc DVD Author 2.x somehow manages to screw up a few frames again at the very end!

Yes I double checked the file at all stages ... the original M2V from HCenc output ... the M2V from DGPulldown (doing 25fps ---> 29.970fps option) ... both with no interlaced frames.

Again when you multiplex with MPEG2VCR it works with no interlaced frames.

But author with TMPGEnc DVD Author 2.x and load the VOB directly into VirtualDubMod and those damn interlaced frames are back again.

Argh !!!

- John "FulciLives" Coleman

[vhelp]

fulci,

Try another vdub. If you use the older version like virtualDubMod v1.5.1.1a,
you shouldn't get any Interlace (or other) forms of real-time 3:2 interpretations
( or other IVTC-like interpretation -- how about, IVTC.I )

For instance, if you have Film source that you Encoding to 24p w/ 3:2 pulldown,
vdubMOD vXX will real-time process it and give you WYSIWYG 3:2 frames. IOW,
you get 3 Progressive frames, and 2 Interlaced frames. Thus..

That's what these latest vdubMOD's do these days, and this is the (probably)
reason why you are seeing stranded interlace frames, cause it's hick-up'ing
on the tail end of the IVTC.I process and it seems like it can't find enough
pattern frames for what it consideres de-IVTC'able.. weather 3:2, or 2:2, or
whatever. Actually, I thought that these methods were only specifically ment
for restoring back to Film 24p, hence 3:2 pattern.. and not PAL stuff.
( I think there is a way to turn this off, though -- I foget, however )

Hope that made sense to ya.

-vhelp 4150

[jagabo]

I'm using VirtualDubMod 1.5.10 and it doesn't perform the pulldown. The interlaced frames appear in VirtualDubMPEG2 which has the option of ignoring pulldown flags. It also happens with DGIndex with ignore pulldown flags selected. The old MPG file is simply messed up. The new MPG file is perfect though.

[FulciLives]

I appreciate the comments by vhelp but I think Jagabo is correct ... the original MPEG file (pal2ntsc1.mpg) does have interlaced frames at the tail end and again this is happening ONLY when I author with TMPGEnc DVD Author 2.x

Again if I take the same files and author with DVDauthorGUI the result is pure progressive as it should be.

Again if I multiplex the video and audio with the multiplexer found in MPEG2VCR the result is pure progressive as it should be.

I don't know the why or the how but TMPGEnc DVD Author 2.x is the culprit here.

I HAVE A THOUGHT ...

Is there a way with AviSynth that I can add like 2 seconds or so of just plain BLACK video at the end of the video I am encoding?

If it is a corruption that only happens within the last second or two then I can ... to be safe ... add say 3 seconds of black to the end of the clip so the corruption happens there and then who gives a shit ... right?

Although will that throw audio sync off or will the audio start as it should and then just end as the video continues for those extra 3 seconds or so?

- John "FulciLives" Coleman

[Pinstripes23]

try adding this after your source line:

+blankclip(height=576,width=720,fps=25,length=75,p ixel_type="YV12").killaudio

so it's like this:

mpeg2source("e:\testclip.d2v")+blankclip(height=576,width=720,fps=25,length=75,pixel_type="YV12").killaudio

It adds 3 seconds of blackscreen at the end of the clip

[FulciLives]

try adding this after your source line:

+blankclip(height=576,width=720,fps=25,length=75,p ixel_type="YV12").killaudio

so it's like this:

mpeg2source("e:\testclip.d2v")+blankclip(height=576,width=720,fps=25,length=75,pixel_type="YV12").killaudio

It adds 3 seconds of blackscreen at the end of the clip

Well I'm getting ready to leave for work so I can't try this until tonight (and I might not even have time this particular evening) but this might just be a good work-around for those of us (me!) wanting to use TMPGEnc DVD Author 2.x when doing PAL2NTSC the DGPulldown way.

- John "FulciLives" Coleman

[deadrats]

I appreciate the comments by vhelp but I think Jagabo is correct ... the original MPEG file (pal2ntsc1.mpg) does have interlaced frames at the tail end and again this is happening ONLY when I author with TMPGEnc DVD Author 2.x

Again if I take the same files and author with DVDauthorGUI the result is pure progressive as it should be.

Again if I multiplex the video and audio with the multiplexer found in MPEG2VCR the result is pure progressive as it should be.

I don't know the why or the how but TMPGEnc DVD Author 2.x is the culprit here.

I HAVE A THOUGHT ...

Is there a way with AviSynth that I can add like 2 seconds or so of just plain BLACK video at the end of the video I am encoding?

If it is a corruption that only happens within the last second or two then I can ... to be safe ... add say 3 seconds of black to the end of the clip so the corruption happens there and then who gives a shit ... right?

Although will that throw audio sync off or will the audio start as it should and then just end as the video continues for those extra 3 seconds or so?

- John "FulciLives" Coleman

want to know why you are seeing interlaced frames? the answer can be found here:

http://en.wikipedia.org/wiki/Telecine#2:2:2:2:2:2:2:2:2:2:2:3_pulldown

http://en.wikipedia.org/wiki/Telecine

http://en.wikipedia.org/wiki/PAL

in particular pay attention to this part:

quote: "When PAL is used to transmit content which was originally composed of 25 progressive full frames per second, the odd field of the frame is transmitted first. This is opposite to NTSC. Systems which recover progressive frames, or transcode video should ensure that this 'Field Order' is obeyed, otherwise the recovered frame will consist of a field from one frame and a field from an adjacent frame, resulting in 'comb' interlacing artifacts."

and then go back to gspot and look at the field order for the original file and the file you created.

i know you consider me "an idiot", "clueless", and "dangerous to this forum" but at least i know how to read and understand what i read. the PAL DVD you used to create the sample is obviously created by applying 2:2 pulldown to the original 24 fps source and this is confirmed by the fact that gspot says it's interlaced with top field first while the file you created is bottom field first which coincides with the NTSC standard.

as for who gives a shit, you should. you placed yourself on a high horse by providing the guides (you are the author of the guides, are you not?), you attacked me when i tried to help the original poster and you made an ass out of yourself by proclaiming that you methods were the only correct methods for converting PAL to NTSC.

your little chearleading section has repeatedly said that i have embarrassed myself in this thread, exposing myself as incompetent. congratulations, i have done the same to you.

welcome to the idiots club.

-deadrats, for when you absolutely must be humiliated by an idiot.

[FulciLives]

@deadrats
First thing first. I am a god among men. Period.
Now as usual you haven't a clue what you are talking about. As I already very clearly pointed out the interlacing issue (that only appears at the very end of the clip) is the result of TMGPEnc DVD Author 2.x doing something ... funky. It doesn't happen when the clip is not first run through TMPGEnc DVD Author 2.x which is something I have already clearly pointed out and verified even by jagabo with a clean clip for all to download and see.

Anyways let me get on with what I wanted to say ...

I went ahead and did what Pinstripes23 suggested and used the following AviSynth script:

LoadPlugin("C:\COPY_DVD\dgmpgdec148\DGDecode.dll")
LoadPlugin("C:\Program Files\AutoGK\filters\RemoveGrainSSE3.dll")
MPEG2Source("C:\test2\testclip.d2v")+blankclip(height=576,width=720,fps=25,length=75,pixel_type="YV12").killaudio
RemoveGrain(mode=1)
LanczosResize(720,480)
Crop(8,60,-8,-60)
AddBorders(8,60,8,60)

This is exactly the same script that I've been using except that +blankclip part which adds 75 frames of black audio-less video to the end of the clip.

I encoded as usual (using HCenc) with all the same settings (this time I did not use the option to force an "I" frame for the last frame).

I then imported the M2V into TMPGEnc DVD Author 2.x (after DGPulldown of course to do 25fps ---> 29.970fps) and the 5.1 AC-3 and created a DVD on the HDD then used VideoRedo to extract the VOB to an MPEG file.

Same as I've been doing.

I am happy to report that here are NO visible interlaced frames. I say NO visible interlaced frames because I am guessing there are a few in the last 1 - 2 seconds but who can tell since it is black video.

I can now use TMPGEnc DVD Author 2.x without worrying about interlaced frames showing.

I am very happy!

If anyone wants to see the clip here is the info:
Link ---> http://rapidshare.com/files/6392304/pal2ntsc3.mpg
File Name: pal2ntsc3.mpg
File Size: 88.6 MB

- John "FulciLives" Coleman

P.S.
Here is the log from HCenc:


-----------------------------------------
| HCenc - MPEG2 encoder - rel. 0.19.0.0 |
-----------------------------------------

input: c:\test3\hcenc3.avs
output: C:\TEST3\HCenc3.m2v

--------------------
| encoder settings |
--------------------

profile: BEST
frames: 0 2339
framerate: 25.00
aspect ratio: 16:9
bitrate Kb/s: 7500
max. bitrate Kb/s: 8000
pulldown: no
closed gops: yes
VBV check: yes
scene change det.: yes
interlaced: no
goplen,B-pic: 12 2
dc_precision: 10
scan method: ZIGZAG
bias: 99
chapter frames: 0
time code: 0 0 0 0
CPU: SSE3
matrix: MPEG

--------------------
| source stats |
--------------------

nr. of frames in source: 2340
width*height: 720*480
fps: 25.00
nr. of frames to encode: 2340
frames to encode: 0 - 2339

---------------------
| encoding - pass 1 |
---------------------

pass 1 encoding time: 0:01:14 (74 s)
average fps: 31.5

--------------------------------
| encoding - intermediate pass |
--------------------------------

bitrate set to: 7500000 b/s
est. outfile length: 85693 kB
intermediate encoding time: 0.4 s

---------------------
| encoding - pass 2 |
---------------------

pass 2 encoding time: 0:07:25 (445 s)
average fps: 5.3

------------------
| encoding stats |
------------------

total encoding time: 0:08:40 (520 s)

intra matrix used
8 16 19 22 26 27 29 34
16 16 22 24 27 29 34 37
19 22 26 27 29 34 34 38
22 22 26 27 29 34 37 40
22 26 27 29 32 35 40 48
26 27 29 32 35 40 48 58
26 27 29 34 38 46 56 69
27 29 35 38 46 56 69 83

non-intra matrix used
16 17 18 19 20 21 22 23
17 18 19 20 21 22 23 24
18 19 20 21 22 23 24 25
19 20 21 22 23 24 26 27
20 21 22 23 25 26 27 28
21 22 23 24 26 27 28 30
22 23 24 26 27 28 30 31
23 24 25 27 28 30 31 33

nr. of gops: 206
nr. of frames: 2340
nr. of I-frames: 206
nr. of P-frames: 779
nr. of B-frames: 1355
average quant (non linear): 2.491
VBV underflows detected: 1
VBV underflows fixed: 1
minimum bitrate: 142
maximum bitrate: 7988
average bitrate: 7378

[Pinstripes23]

Glad you got it sorted out Fulci and able to use YOUR favorite authoring program. Damn, this deadrats has a hard time of understanding the concept. He may be able to read but the understanding part....he is really clueless. When I read about the interlacing in the last few seconds I decided to convert the original PAL clip and see what would happen using muxman for authoring. No interlacing was seen running it through vdubmod. So it's definitely a TMPGenc DVDAuthor problem.


@deadrats

Fulci may be the author of the guide but the whole method is based on the dgpulldown program. And it is a tested and tried true method that works for many people. I have converted pal-to-ntsc many times using this method and I am glad this program came out cuz I used to convert to NTSC the long way(stretching the video/audio). So you keep using your method of duping/dropping frames method. Anyone who wants to keep ALL the frames will use the dgpulldown method and convert as efficiently as possible.

[deadrats]

@deadrats
First thing first. I am a god among men. Period.
Now as usual you haven't a clue what you are talking about. As I already very clearly pointed out the interlacing issue (that only appears at the very end of the clip) is the result of TMGPEnc DVD Author 2.x doing something ... funky. It doesn't happen when the clip is not first run through TMPGEnc DVD Author 2.x which is something I have already clearly pointed out and verified even by jagabo with a clean clip for all to download and see.

you sure you don't mean "a chimp among men"?

answer me one question, oh "god among men", why is it the original clip you provided, you know, the one from the PAL DVD is also interlaced and can be verified by every piece of software that is capable of distinguishing between interlaced and progressive?

i say it's because it was authored by applying 2:2 pulldown to the original 24 fps source, which by definition means it's interlaced.

i'm dying to hear your explaination...

[FulciLives]

@deadrats
First thing first. I am a god among men. Period.
Now as usual you haven't a clue what you are talking about. As I already very clearly pointed out the interlacing issue (that only appears at the very end of the clip) is the result of TMGPEnc DVD Author 2.x doing something ... funky. It doesn't happen when the clip is not first run through TMPGEnc DVD Author 2.x which is something I have already clearly pointed out and verified even by jagabo with a clean clip for all to download and see.

you sure you don't mean "a chimp among men"?

answer me one question, oh "god among men", why is it the original clip you provided, you know, the one from the PAL DVD is also interlaced and can be verified by every piece of software that is capable of distinguishing between interlaced and progressive?

i say it's because it was authored by applying 2:2 pulldown to the original 24 fps source, which by definition means it's interlaced.

i'm dying to hear your explaination...

This was already covered by manono in an earlier post in this very thread.

The PAL DVD is PROGRESSIVE. End of story.

I admit I don't know much about PAL 2:2 Pulldown but according to the Wikipedia entry that you referrenced ... and I quote ...

2:2:2:2:2:2:2:2:2:2:2:3 pulldown

This pulldown method is sometimes used in order to convert 24 frame/s material to 25 frame/s. Usually, this involves a film to PAL transfer without the aforementioned 4% speedup. For film at 24 frame/s, there are 24 frames of film for every 25 frames of PAL video. In order to accommodate this mismatch in frame rate, 24 frames of film have to be distributed over 50 PAL fields. This can be accomplished by inserting a pulldown field every 12 frames, thus effectively spreading 12 frames of film over 25 fields (or "12.5 frames") of PAL video.

So if PAL 2:2 Pulldown was used on THE DESCENT the running time would match that of the NTSC version as well as the film version (since both run at 24fps and yes I know NTSC is 23.976fps but close enough and nobody argues that).

However we find from the IMDB that the film running time is 99 minutes yet the PAL DVD has a running time of 95 minutes. Oh wait that is a 4 minute difference? I wonder why?

Let me quote your Wikipedia source yet again ...

PAL speed-up

Motion pictures are typically shot on film at 24 frames per second. When telecined and played back at PAL's standard of 25 frames per second, films run 4.2% faster. Unlike NTSC's telecine system, which uses 3:2 pulldown to convert the 24 frames per second to the NTSC frame rate, PAL results in the telecined video running 4.2% shorter than the original film as well as the equivalent NTSC telecined video. It also increases the pitch of the soundtrack by 70.67 cent — ⅔ of a semitone, which people with absolute pitch will notice immediately.

However, some movie enthusiasts prefer PAL speed-up over NTSC's 3:2 pulldown, because the latter results in telecine judder, a visual distortion not present in PAL sped-up video.

So do the math.

99 x 4.2% = 4.16 or approximately 4 minutes
99 - 4 = 95

You jack ass

- John "FulciLives" Coleman

[jagabo]

deadrats, you seem to misunderstand the difference between encoding a video as interlaced and the individual frames of the video being interlaced.

True interlaced video contains two separate pictures in each frame. One field contains every other scanline of one picture from one point in time, the other field contains the complimentary scanlines of a picture from a different point in time (1/50 second later with PAL, ~1/60 second later with NTSC). To prevent artifacts from cross contanimating the two fields an MPEG encoder needs to be told that the frames it is compressing contain two separate pictures. Internally it splits the frame into two images and encodes them separately. When the codec decodes the video it weaves the two fields back together into a frame. When this frame is displayed on an interlaced TV the display device needs to know which of the two fields to send to the TV first. Hence you have top-field-first and bottom-field-first flags included in the MPEG data.

A progressive video can be encoded as if it's interlaced. This simply means that the encoder handles the source frames as two interlaced fields. Again, the codec splits the frame into two separate fields, compresses them separately and on decompression weaves them back together into a frame. But since the frame was progressive before compression, it is progressive again after decompression.

Field order flags are immaterial for progressive video compressed as interlaced. Both fields come from the same point in time so it doesn't matter which you display first on an interlaced display.

No doubt you are familiar with the 3:2 pulldown that normal NTSC television uses to display 24 fps progressive frames at 60 fields per second. On DVDs The video is usually compressed as 24 progressive frames per second and pulldown flags are added to instruct the DVD player on how to perform the pulldown to 60 fields per second. This is the best compromise for displaying 24 frame per second film on a 60 field per second TV. It's also possible to perform the pulldown before encoding as MPEG. This results in 2 of every frames being interlaced so the entire video has to be encoded as interlaced. The DVD player then performs 2:2 pulldown and needs to know the field order.

The DGPulldown method applies 3:2:3:2:2 pulldown flags to display 25 progressive frames per second as 60 fields per second. The results are smoother than duplicating one frame out of every 5 and performing 2:2 pulldown (what your first sample did) or dropping 1 frame every second and then performing 3:2 pulldown (what your second sample did). Simply slowing the video down from 25 fps to 24 fps and performing 3:2 pulldown may (arguably) give the smoothest results but requires that audio be adjusted to match the longer running time.

[vhelp]

Just what is your problem with that statement? Perhaps 90% or more of the PAL DVDs
of movies are encoded as interlaced. These are movies that show in the movie theaters at 24fps
(progressive) and are speeded up to 25fps (still progressive) to be made into PAL DVDs. They are
then encoded as interlaced. Doing that is as simple as not turning on progressive encoding in the
encoder. Therefore the source *IS* progressive, but they are encoded as interlaced. Not so hard
to understand, is it? Such programs as DGIndex, Bitrate Viewer, or ReStream will show it as
having been encoded as interlaced, but there's no interlacing to be found. The fields of each
frame are taken from the same point in time, unlike true interlaced video. You can scoff and
curse all you want, but none of it helps your case.

Exactly. The encoding *as interlace* is just that.. the encoder inserts "flags"
into the stream to tell gspot ( i mean, the decode/player ) that the video
source type is Interlace. Then, the next job of the (this time) decoder/player
is to play the "flagged" stream as interlaced. This has an special effect for
this type of source output. So, even though the video is PROGRESSIVE, the
*player* will think its Interlace and make the necessary adjustments to output
the video as Interlace to your TV, even though the source video who is PROGRESSIVE
in nature but flagged as Interlace, is still PROGRESSIVE.

Then, the next job is the TV's.

I don't know why others find this hard to belive. If they *still* don't,
then perhaps you can *MAKE* them belive, by showing them how to do it
with an actual test source (of there own) and let them decide if they
still don't belive.

In the example below, I use my favorite movie, "The Fifth Element" for
this scenario because it is TRUE PRGRESSIVE 24p on my dvd's.. ripped with
smartripper. Note, 24p meaning 23.976 fps.

Ok. So lets take our 24p (23.976 fps) PROGRESSIVE dvd, rip it to hdd, and
encode it as Interlace. Note, this does not have to be a PAL or PAL->NTSC.
This is just a demonstration of how to encode PROGRESSIVE video as if it is
(to look like it is, by the player) Interlace.

In TMPGenc, do the following setup:

A - open a PROGRESSIVE video source (ie, 'The Fifth Element')
B - [settings] \ Advanced \ Video Source Type:[Non-Interlace (progressive) ]
C - [settings] \ video \ Encode Mode:[Interlace ]
D - Encode the video to either an .MPG or .M2V

Then, to review what the source is now, (is it Interlace or Progressive) ..

E - using your favorite dvd2avi or variant, open it up and pres F5 key,
--- and observe the Frame Type field. It *should* say:
---
--- ** Frame Type:[Interlace ]

..notice though, it is *STILL* Progressive, by evidense of reviewing the
video frames (see below) inside a timeline, like virtualDubMOD, where there
is no evidense of Interlace (stair-stepping, or saw-tooth) in the frames.

Ok, still not convinced, then try this last step..

F - open video file inside virtualDubMOD
G - scan through manualy for any interlace frames -- hint: you shouldn't find any.

Regarding the above, this is what the DGIndexpulldown is doing.. faking
that it is interlace (using flags that most dvd players understand to mean
to deal with fields in the video and treat it not like Progressive, but
like Interlace -- and illusion)

That's it

Now, you may ask the question, (which is probably what should have been
asked, instead of arguing) -- why do they do that ??

..And, wait for your answer

-vhelp 4152

[manono]

the PAL DVD you used to create the sample is obviously created by applying 2:2 pulldown to the original 24 fps source and this is confirmed by the fact that gspot says it's interlaced with top field first while the file you created is bottom field first which coincides with the NTSC standard.

Pure gibberish. Makes no sense at all. As FulciLives said, it was created using the common 4% speedup from 24->25fps.

i have already admitted i was wrong about Procoder and don't plan on recommending it to anyone anymore.

Why not? It does the conversion, as do all the MPEG-2 encoders with which I'm familiar. It's how it's set up by the encoder (the person behind the keyboard) that determines whether or not the PAL->NTSC conversion is done correctly. My problem is with the Procoder enthusiasts that seem to think that using some sort of template or preset for this sort of thing is the way to go, when all it does is screw up the conversion. It may be easy, but it's inferior. Here's the testclip properly converted (given the limitations of the testclip source, and the way it was cut) to NTSC using Procoder:

http://rapidshare.com/files/6414206/manonoProcoder.VOB.html

28 MB.

[FulciLives]

That Procoder clip looks damn good for such a low bitrate.

- John "FulciLives" Coleman