what is the right way to create an Anamorphic KDVD?
 

Hi everybody.

I don't know how many minutes (widescreen material) you guys fit on one DVD-R. Currently I'm trying to cram 3 movies (all in all 327 min.) onto one, with a 384kb AC3 for each movie. Quality is poor at 720x576. I had a CQ of 56 as I recall. Is 327 min. widescreen just too much to put on 1 DVD-R at that resolution? :roll:
One movie is pretty high action, the other two are medium to low action. :roll:

Also what is the right way to create an Anamorphic KDVD? I know it was discussed before, but whenever I set Input=16:9 and Output=16:9 PowerDVD squishes it down (faces are not egg-headed but squished together horizontally) except when I UNcheck keep AR. Will that look right in my Standalone?

The anamorph input and anamorph output handling will be done in your avisynth script only!

So you still end up with a "center" setting in TmpgEnc just a encoding at "16:9" added in the encoder options of TmpgEnc (which only sets the flag to the encoded stream). Source is just 1:1 as its a avs input.

Thats all.

According to Quality:

- Try to lower your AC3s to 256kbit for example - just test it.

- Change to 704x576

- A CQ of 56 at full resolution (704/720x576) is not that bad (ok, not good enough ;-) ) as it will be not horizont. resized that much like 528x576 or even 480x576.

- Low action as you see it as a whole in the movie doesn't have to mean something. Bitrate peaks also will be happen at scenes where for your "interpretation" is no action :arrow: Diamond sparcles, ... also a romantic "interpreted" calm scene where raindrops fall onto a watersurface :arrow: the bitrate will rise up very much.



:)
Inc.

I wil say more "encoding a 16:9 source into a 16:9 target" is _exactly the same_ to do "4:3 source in 4:3 format".

Remember what is the definition of "anamorphic" : a 16:9 image stetched to fit in a 4:3 container.

You can verify that under movie stacker : load a source that is anamorphic, check the "anamorphic" box for the destination and look at the resizing parameters Moviestacker gives. Then uncheck BOTH anamorphic boxes (source and target). See the resizing parameters again... they are identical !

Yep!

Thats what I meant, no size "maipulation" done in TmpgEnc.
All these in and iff output parameters do have their sense primary when TmpgEnc does the Resizing. But we don't do that, thats also why we love avisynth.

I'll give it a try, thanks for your help!

How would you guys do predicition for 3 movies on one DVD-R? Right now I'm putting all 3 d2v files into one AVS script and let it predict. The problem is that those sources are of very different quality. One is very clean and high quality, the other one is quiet noisy. So it's not a good idea to treat all three sources with one AVS script. What way is there that you can specify the filters/resizing parameters etc. for each source seperately but in ONE AVS script? :roll:

Another questions I have is to use IfoEdit to get the correct GOP structure of the original DVD because I'd like to have exact chapterpoints since DVDLab does support ChapterXtractor. However, TMPGEnc doesn't seem to recognise the .txt file IfoEdit gives me. When I load it in TMPGEnc's Force Picture Type setting, no GOP structure appears, so i guess I have done something wrong. I recall there are several different presets you can save your .txt file as. Which one is the right one for TMPGEnc? :roll:

All depend on what you are using for encoding.

With TMPGENC, you surely use CQMatic for predicting and there is a setting where you specify the number of stream on the media. You just have to indicate 3 there, choose DVD are target medium, and CQMatic will do a prediction for a target size of 1/3 of DVD.

If you use VMesquite method and CCE, just divide 4.37/3 and use this calue as targeted size :-)

Hi all:
Jellygoose wrote:
Well, sure there are better methods out there. This is the method I use:
You can apply two different set of filters, according to your sources. But, the CQ will be the same for both films.
This works for me.
Tells us if does work for you.
Cheers

Jell,

you should determine clips in your avs so every source gets its own filtering, like

a=mpeg2source("xxxxx.d2v")
a=a.Denoiseheavy here ()
a=a.Resize here ()
#
b=mpeg2source("yyyyy.d2v")
b=b.Denoiseheavy light ()
b=b.Resize here ()
#
c=mpeg2source("zzzzzz.d2v")
c=c.Denoiseheavy medium ()
c=c.Resize here ()
#
return a+b+c

ok, but all streams have to have same...

... colorspace (I think you do just deal with mpeg sources and so they all do have 4:2:0 YV12)
... FPS (... here the same, as your sources could all be 25 FPS)
... Endsize after resizing! (ok, all should be 704x576 not imortand if encoding 1 anamorph afterwards and 2 not)

EDIT: Our Digitall.Doc was más rapido! :)

PS: I use this also to apply at the beginning of all may captures one of my Movie-Company Intros ... as I finally got a nice collection. Cause many broadcasts do got that animated logo, when the logo flys in or something (like Pro7 = Interlaced! *würg* horizontal animation)

Second:
If using IfoEdits txt file, you only need to let it give you the "I"Frame Timestamps! When loading that txt into TmpgEncs Force Picture Type settings, it "could" be that it takes a long time! So your machine did NOT bomb!

Hey incredible:
I was faster (first time!) :lol: , but maybe you're far accurate.
Don't think that makes a difference, but there's no need to use a=a. ... each time (look my code). But in the end, I think we said both the same.
Yep, you're right, and I forgot to say that :oops: .
:lol:

But please don't use that limiter() as it first treads the luminance at its ends very bad and second if using standard KVCD Template settings, this will be done before encoding via tmpgEnc! :wink:
I got bad experiences with limiter :(

Ok, one more question concerning Anamorphic Streams... Please correct me if I'm wrong somewhere.

When I have a 16:9 Display DVD (as most DVDs are right?), and I load it in Moviestacker "Anamorphic" box is automatically checked, meaning that the film IS coded with egg-heads, and that there's a flag in the stream that tells the DVD player to squish the video horizontally, so that it is displayed as 16:9 (no more egg-heads) on TV right?
So when I copy that DVD to KDVD I have two options.

First one is Input=Output AR, meaning that I encode it too as an Anamorphic stream. (no resizing or add borders, Input= 1:1 (avs) Output=16:9 Video Arrange=center) With that in Tmpgenc's preview screen, when resized to 352x240 (standard) the picture would look horizontally stretched (egg-heads), as the stream is actually encoded just so, and only the 16:9 flag in it tells the DVD player to display it correctly.

Second Option, to encode as Non-Anamorphic and 4:3 display flag. That would give the correct AR with DVD Players and 4:3 TVs. It would also save LOTS of space on the DVD, since a lot less film-pixel is encoded.

Now how would such a stream would look on a 16:9 TV?
What are the advantages encoding Anamorphic? Why would you code a movie with egg-heads, and tell the player to squish it down? It doesn't make sense to me. Are there more details present?
Can someone enlighten me?

1. 352x240 = always non-anamorph! as its not supported at that size

2. http://www.widescreen.org/widescreen.shtml

Also a nice flash animation where they show how does 4:3 look at 16:9 tv sets.

Means

a) a little 4:3 moviearea within the 16:9 and borders left & right added.

b) that 4:3 movie zoomed so it fits in its effectively movietreatment the total of 16:9

;-)

with 352x240, I was talking about the size of the preview window in TMPGEnc. I know that Anamorphic is only supported by DVD resolutions, such as 720x576/704x576...

Okay, I get the flash, totally...

However, it still does not make sense to me. so once more, sorry for being stupid. Why would I encode as Anamorphic at all? Why wouldn't I just resize to 16:9 image with 4:3 display? It makes no sense to me that DVDs that are really 16:9 are encoded as Anamorphic (stretched to egg-heads). Why is that so?

Because of a full anamorph vertical 576/480 resolution on a 16:9 tv set!
The quality is better than doing just a zoom to fit a 4:3 video to a WHOLE 16:9 Tv set width Proportionaly.

Because anamorphic videos will be stretched in their WIDTH to 1024x576 on a 16:9 Tv set. Means only be scaled in their width but not in their height. And at non-anamoprph 4:3 Streams you have to scale the width AND the hight (to fit proportional the complete width of a 16:9 Tv set)! And as the height resolution of a stream is most responsable for sharpness on a TV this makes sense.

But that advantage stands very much in relation to the more bitrate you need. So in case of anamorph encodings we do need more bitrate to enjoy that advantage ;-)

That's exactly what I mean. Do you think it's worth to encode Anamorphic at all?
Would a 4:3 stream be stretched to fit in width and height automatically when displayed on a 16:9 tv?

A 4:3 stream is treated the same way as a 4:3 TV brodcast program. You have to use a zoom mode to fit the 16:9 entirely.

Ok, i decided to delete this post - - too much info not needed :roll:
-vhelp

Update.. no longer available :roll:
-vhelp

Well its all a thing of space/bitrate that is needed, so IF you see an anamorph stream on a 1&:9 TVset you will recognise the quality.Its a matter of settings, as almost every 16:9 TVset gots a zoom option to compensate this.

That's why I do all my KDVD in anamorphic and my KVCD in non-anamorphic. On a KDVD we have all the space wanted to "waste" a little the bitrate with an anamorphic image. As my TV is 16:9 compliant, the qualitty is far better.

I did a miniKDVD with 352x288 res and 16:9 flags.
Just to fit the surround audio line on a 80min CDR

It was a backup of my Linkin Park - Live in Texas btw, so video quality didn't matter that much to me.
Bitrate was about 850, and video looked kinda creepy. =)

But it was fun.

???

As I know, 352x288 is not supported in anamorph said in 16:9 flagged mode.

Hmm... I got no probs playing either in PowerDVD or two different standalones tested (don't remember models though).

Try it yourself ;)

On DVDMaestro for example you can set as anamorphic ANY DVD compliant source, even MPG1 ! It worked, I've done it.

I encoded everything in 4:3 stretched, not as anamorphic, and just put the anamorphic flag when authoring.

You are all right.

Anamorphic flag is (as pulldown ;-)) a MPEG2 feature only. But... there is also such a flag in DVD header.

Thus you can do 4:3 DVD with a 16:9 video stream or the opposite (for sure, the A/R will be completely wrong).
I just did it on my first attempt with DVDMaestro (imported a 16:9 m2v but didn't cxhange the project to 16:9 :-().

Alright, just wanted to warm this up again...

what filter combinations is everyone using currently to make KDVDs from commercial DVDs... I'm talking about average quality DVDs, with not much noise. The goal would be fitting 2-3 movies on one DVD Anamorphic Output. The script I'm currently using is:

I'm actually quite satisfied with the results, although sometimes low detail areas tend to look a bit plain...
Please everybody post your favorite script, and I can do some benchmarks... hehe! :lol:

This is mine :
Note that the "72_72" means that this lien has to be used in case of movie with 72 pixels wide top/bottom borders. These are actually 2.35:1 movies.
16_16 is for 1.85:1 movies. You can check the width of the borders in DVD2AVI.
[/code]

hy all !

i use this script for 2 movies with ~4 hours in cce (Q=20),
muxed with the ac3-5.1 or 2.0 channels .
if the final size is too big, i re encode only the audio with softencode.
the good is that i don't need to do prediction!
Brasil is too hot and if i use more time, the media can damage( and maybe the player)
remember what i posted here:
http://www.kvcd.net/forum/viewtopic.php?t=9242

this script is called "mixed emotions" :

mpeg2source("D:\kdvd\x.D2V")

Undot()
Asharp(2,4) # do a sample, if needed change to (1,4)
Levels(0,1.15,255,0,255) # all vobs encoded loose some bright

GripCrop(720,480,overscan=1,source_anamorphic=true ,dest_anamorphic=false)
# atention with true or false,depend of the source(moviestacker help a lot)

BicubicResize(GripFit_resize_width,GripFit_resize_ height,0,0.6) # big thanks to fabrice
# - it don't loose (cut) the scenes on the sides like crop
# and/or GripSize(resizer="BiCubicResize") that give strange anamorphics sometimes

deen("a3d",3,4,1,4) # light cleaning

DCTFilter(1,1,1,1,1,1,0.5,0)
GripBorders()

ConvertToYUY2() # if needed change to (Interlaced=true)

Trim(0,-1).FadeOut(150) # use for better transition between the movies and soft end !

that's all ! :wink:

0.15 in gamma is more than "a little". But if you like it this way...

You should go to 704*480 and use overscan=2. You won't see the difference but your file will bit a lot smaller.

New or old version of deen ? If you used old one, try the new :
http://atlas2.tgv.net/~media-video/f...c9c1b3be9b4655

Use it withou any parameters first as I didn't try with the ones you like so much.

Note: why don't you don't dest in anamorphic ? You will probably have a widescreen TV a day and you will regret to have done all your KDVD in 4:3.

Phil wrote:
"0.15 in gamma is more than "a little"
i know Phil but just a little,the recommendation is 1.3 !

"You should go to 704*480 and use overscan=2. You won't see the difference but your file will bit a lot smaller."
i knew too...like i posted, size isn't a problem for my target (~ 4 hours)!

"New or old version of deen ?"
is the old version.
i got the new in the link that you posted in another thread
(seems the same link here) but don't did tests !

"You will probably have a widescreen TV a day ..."
like i posted..."you're a joker",
i don't have money for that!
:lol: :lol:

thanks for all recomendations!
:wink:

Hey Phil!

Thanks for the script I love it!! :D
I thought it would be a little too much temporal filtering, but it's really great. Thanks... :wink:

Phil and Jell,
please explain and help me:

no matter if using the script that Phil posted
(i know is for PAL, i change the values for NTSC)
or any other script using crop and letterbox:

:arrow: i loose part of the scene in the picture!

i use the script posted and loose part of the scene in
top,bottom,left and right.

if i am not clear, "copy source to the clipboard" in vdub,
using any script with crop/letterbox (all possibles and combinations)
paste that picture in paint!

do the same with another script without crop/letterbox
(using any other resize like bicubic,lanczos...)
and paste that picture i (open another Paint)
and compare the sizes and the scenes from the first image and the second.

please, answer if only i see it or if i did something wrong!
thanks in advance!
:)

Jorel ? To crop in english actually means "remove / cut / erase".
So all scripts that have a Crop line remove some parts of the picture !

Moreover, with letterbox you paint black border ON the picture, so you also lose the image that is normally under the border.

That is different with an addborders which adds a border ARROUND a picture (arround is not "onto", do you understand the difference ?)

So all what you describe is normal.

Phil wrote:
"That is different with an addborders which adds a border ARROUND a picture
(arround is not "onto", do you understand the difference ?)

very very clear my friend :!:

then.....why use crop if you
:arrow: remove some parts of the picture !?!?!
:?

and why use letterbox if you
:arrow: paint black border ON the picture?!?!
:?

the 2 situations loose part of the pictures!
i only want "resize" the picture without "loose" anything :!:

then.......why use crop/letterbox?!?!?
:arrow: this is unclear for me :!:
:?

Because :
1/ it's faster
2/ it doesnt hurt the rest of the image. A resize does affect the WHOLE image and if the resizer is bad, all the picture is screwed up.
3/ I crop 8 pixels. That is under the normal overscan of my TV that means that the part I crop, I do not see it when I whatch the original DVD. So it's not a problem if I loose it also on my KDVD backup ;-)

See reason 1 and 3 above. Just replace the word "crop" by "paint".

I never said you won't loose anything by using this :!: But as I just told, what you lose with this method is what you already lose when you watch your original DVD and you don't cry about it, do you ? :-)

By the way you can also ask to Moviestacker to give you correct parameters for resizing but for the reason #2 I enumerated above, I find this is a bad idea.

@Jelly
You're welcome. In fact I wonder recently why I put 3 filters but I think that there are too much spatial filtering, not temporal. In other words I would try without undot.
Now if you rather to have a little less temporal, you can change temporalcleaner for temporalsoften. It is lighter.

Many times I was read in this forum that speed makes no sense.
In many other scripts you trust in resize, why you don't in this one?
I think if you resize and then addborders, you don't loose any part of the image, only it will loose just these borders. In other words, you will see more than in the original DVD.

yes Phil,
i had understood it all! :wink:

this is what i mean too (Prodater64 post):
"I think if you resize and then addborders, you don't loose any part of the image,
only it will loose just these borders.

:arrow: In other words, you will see more than in the original DVD. "

true, all my encodes and samples show that :!:

you found the right words Prodater64 :!:
:D

Because it doesn''t. But here speed comes in common with both other 2 reasons. And you are right, the order wasn't correct : the first reason in word of importance should have been the #2. The #1 comes in last position.

You are talking about scripts to make KVCD where i MUST resize !
Here I'm doing DVD. Did you see any script of mine for DVD where I use resize ? ;-)

But I what I will see would have been screwed up by the resizer. I'd rather to lose 1% of the image than to look at 100% of it distorted by a mathematical interpolation. FYI "bilinear" and "bicubic" stand for "bilinear interpolation" and "bicubic interpolation". Interpolation means rounding/averaging/guessing of pixel values.

Think about that... You too Jorel :-D

Note: Prodater, if you want to fight with me ask yourself why I didn't post yet my results about MCE... :wink:

1 - I understand that like a joke (for the smile)
2 - Anyway, I don't want to fight with nobody at all!
3 - I'm here learning (¡you are a professor!), i'm an "asker", sorry.
4 - You can see I learn (see point 1 and this one too) :)
5 - BTW, since you mention it, I cannot stay waiting for, Phil!!!! 8O

Edited: I forget it, finally I agree with you, I prefer too, croping than resizing, as you explained.