PDA

View Full Version : Megui - Interlaced or Progressive? Noise


skaleton
20th August 2010, 22:10
Hi,
I am trying to encode a PAL dvd into XVID with Megui.
DgIndex says that the video is Interlaced
The analysis in Megui says that it is Progressive.
I applied this script on a sampleAssumeTFF()
SeparateFields()The sequence is, as far as I can tell, ...aabbcc...
Therefore I applied what Megui suggestedLoadPlugin("C:\Program Files (x86)\MeGUI\tools\dgindex\DGDecode.dll")
DGDecode_mpeg2source("C:\xxx\VTS_01_1.d2v")
#deinterlace
crop( 16, 2, -10, 0)
Spline36Resize(704,528) # Spline36 (Neutral)The result was not satisfactory.
It seems to me that there is a lot of "noise" (you may find a more appropriate term), I tried the denoise options and Deblock but did not see much improvement.

Here is a link to a sample (http://www.megaupload.com/?d=Q6HD0B67)
I would appreciate if someone would take the time to have a look at it and advise me how to best handle it.

Thanks

neuron2
20th August 2010, 22:32
It's progressive *content*.

DGIndex said it was *encoded* as interlaced, which is true.

That's very noisy. Where did the source come from?

skaleton
21st August 2010, 02:04
Thank you neuron2 for your reply.
I have to confess that I don't quite understand. I made a search with "progressive, content, interlaced, encoded" and found every combination of the 4 terms without any clear explanation, even things like "progressive encoded in interlaced mode".
So I would be grateful if you would explain it to me or help me find the information.
Is the "progressive content" as analysed by Megui the source or the result of the encoding as enterlaced as analysed by DgIndex?
In my attempt to encode it to XVID, should I consider it progressive or interlaced?

The source is a retail DVD from the UK of a series from the 70s.
So is there something I can do about that noise?

neuron2
21st August 2010, 12:02
If there is motion at the field rate, i.e., in a motion scene you have a new picture (temporal moment of sampled video) at every field, then the *content is interlaced*. If there is motion at the frame rate, then the *content is progressive*. The clip you posted has progressive content so you should process it as such.

When you compress video with an encoder, you can compress it by fields, in which case you have an *interlaced encoding*. If you compress it by frames, you have a *progressive encoding*.

Interlaced content may be compressed with interlaced or progressive encoding. Also, progressive content may be compressed with interlaced or progressive encoding. It's not always guaranteend that the content and encoding will match.

I cannot advise you about noise filters because I rarely work with such poor material and so have no experience to draw upon. Hopefully some more experienced members will drop by to advise you about it.

skaleton
21st August 2010, 13:25
Thank you neuron2 for your explanations. It is very kind of you but I am afraid that it is all a bit complicated to me.
If it is at all possible to simplify, is the content that determines what filters to use?
In the present case, processing as progressive means no de-interlacing, (and since it is PAL, not IVTC), just cropping and resizing, doesn't it?

As for the noise, I did some tests with Undot(), FluxSmoothST(7,7) and Convolution3D("movielq") but they don't seem to make much difference ... at best.
Google is not very helpful on the subject. I might try Denoise for VirtualDub.
I have a few TV movies and old series that are just as bad and as you say "Hopefully some more experienced members will drop by to advise me about it".

neuron2
21st August 2010, 14:35
If it is at all possible to simplify, is the content that determines what filters to use? Yes, of course. Also the problems that the video content has will determine the needed filtering.

In the present case, processing as progressive means no de-interlacing, (and since it is PAL, not IVTC), just cropping and resizing, doesn't it? That's right, you do not need to deinterlace progressive video content.

As for the noise, I did some tests with Undot(), FluxSmoothST(7,7) and Convolution3D("movielq") but they don't seem to make much difference ... at best. Your source is so bad that I doubt any filter is going to be very helpful. But as I said I have little experience dealing with such poor sources.

skaleton
21st August 2010, 21:18
Thank you neuron2.

IIRC, should I increase the bitrate and size with a source with noise than I would normally aim at with a cleaner source?

What kind of resize filter would you recommend? Would a sharp one tend to enhance the noise? Or is it not related?

Still in connection with the noise, does the final size matter?
The source is 720x576 4:3 and I am planning to re-size it to 640x480 with a video bitrate of around 1600 kbps