KVCD: would others post their procedures?
 

I have some thoughts going through my head, and am trying to pin down the procedure most people use for encoding (the manual way, rather than acp or DVD2SVCD). Personally, I do it like this:

1. Obtain source files, either VOBs or AVIs
2. Separate AC3/PCM audio
3. Adjust audio length, if necessary
4. Encode audio with HeadAC3he
5. Load source file in VirtualDub and get border sizes
6. Enter information into FitCD to obtain cropping/resizing parameters
7. Create an appropriate Avisynth script including Sampler()
8. Make sample strips to determine correct CQ or CQ_VBR
9. Comment Sampler() line
10. Encode
11. Mux audio and video in bbMPEG
12. Author with VCDEasy
13. Burn with CDRWIN

If others could post their procedures I'd be grateful, as I'm trying to find points of similarity for a potential project :).

Re: Procedure
 

:bigooh: On boy, that sounds like 2003 is going to be interesting :jawdrop: :wink:

Your procedure is almost identical to the one I use, except I haven't used Vdub to get border sizes. Is there that much difference to doing it with Vdub than to let FitCD calculate them :?:
I also let VCDEasy do the burn.


-kwag

I'm the same as Kwag almost identical to my method apart from the vdub part.

Re: Procedure
 

FitCD won't calculate the size of the borders on the existing source material. That's the part for which I use VDub.

I had a problem getting cdrdao to work with my burner, but I've not tried it with the latest version of VCDEasy yet.

Re: Procedure
 

:imstupid: I didn't get that :eeks:, maybe too much alcohol from New Year's eve on my system :roll:

@SansGrip,

4. Encode audio with HeadAC3he

In this step, I wait until after creating video file (.m1v) and then
create the audio with HeadAC3he. Sometimes audio will need
adjusting if video comes out too large or too small :)

5. Load source file in VirtualDub and get border sizes

Cropping with VirtualDub or Tmpgenc is easier than with FitCD,
because you can visually see what your doing. SansGrip and
Mauddib are working on a way to call VirtualDub via FitCD and
everything could be done within FitCD. :D

7. Create an appropriate Avisynth script including Sampler()

Here's the part where your decisions will either create a great video
movie or a failure. How you decide which resolution to use, which
filters have the best results for your video type, what will fit on 1 or 2
CD's for the movie lenght, widescreen vs fullscreen encodes, and
more. I feel the most common video type and settings should be
discussed as samples (e.g. KVCDx3 CQ or CQ_VBR).

-black prince

@SansGrip,

I forgot, plenty of screenshots, e.g. bbMPEG, FitCD, avi sample scripts,
etc. 8)

-black prince

I have an idea for guessing a good starting value for automated sample encoding... Theoretically if one knew the frame count, resolution and how much of the frame was actual information (as opposed to borders) it should be possible to arrive at a pretty good estimate of the quality value to use. I've tried this out manually and had promising results, but it's too long-winded to do every time unless it's automated.

In order to do this in a program one would need to enter the cropping and resize values too, so I figure if I'm going to implement that in a program I might as well make it do the resize part of FitCD's job as well.

Essentially I'm thinking of putting together an "all-in-one" tool for KVCD encoding. At first it would incorporate the functionality of both FitCD and KVCDP, but it could in theory be expanded to include automatic audio extraction from AVIs (properly handling both PCM and AC3), hands-free encoding with HeadAC3he and TMPGEnc, and could even duplicate the important parts of VCDEasy via vcdimager and cdrdao.

I think before I start planning this in too much detail I should check out acp, because it sounds like the two might be doing the same job ;).

Anyway, that's why I was asking the question. Feedback appreciated.

Correct borders with Vdub
 

Hi SansGrip,

Please correct me if I'm wrong. If I understand it correctly, this is the way I would get my "correct" borders with Vdub.
I open my clean .avs with Vdub. That is, no resizing or addborders on the script. I add the "null transform" filter and crop the top and bottom. For example, I got 56 on top and 59 on bottom. So if my targer resolution is 528x480, it would subtract 480 - 57 -59 = 366. So my resizing would be "LanczosResize( 528, 366 )" and my borders would be "AddBorders( 0, 56, 0, 59 )"

Actually if I want one block overscan, then lanczos would read "LanczosResize( 512, 250 )" :?: :?: ( 528 -16 ) ( 266 -16 )

Is this the correct procedure, or did I miss something :roll:

Edit: On add borders would be: "AddBorders( 16, 56, 16, 59 )" because of the one block overscan. Forgot to add the 16's.

-kwag

Re: Correct borders with Vdub
 

That's almost the way I do it, with the difference that I note the border sizes and new dimensions I obtain through VDub, then enter those into FitCD. It'll then work out all the cropping and resizing for me.

I don't bother using AddBorders in Avisynth, I just set TMPGEnc to use "center" arrangement and it handles that for me :).

Re: Correct borders with Vdub
 

But then what are you doing, cropping in AviSynth?
Also, for what I see in the way I just did it above, there's no need for FitCD 8O. My aspect is correct with the manual calculations :!:

-kwag

Re: Correct borders with Vdub
 

Yep, using the values from FitCD.

I'm not sure, but I think that way will fail with anamorphic source. Also it won't be rounded correctly compared to the values FitCD suggests.

Basically you're using VDub to find out how big the borders are in the source material. You then transfer those values over to the "film pixel" textboxes in FitCD and proceed as normal, except now you're cropping those borders in Avisynth before doing anything else (except stuff like Telecide). This makes your script run faster because the filters have to process fewer pixels.

Re: Procedure
 

Your procedure is almost identical to the one I use, except I let VCDEasy do the burn, and extract the DVD chapters with ChapterXtractor.

Re: Correct borders with Vdub
 

so KWAG... was it you or the calculator that had too much eggnog??? heheh sorry man, had to be done. :wink:

anyway, after you AddBorders and get your correct output resolution, what are you selecting as your Source Aspect Ratio and how do you determine that?

here is my procedure:

01. Obtain source files, either VOBs or AVIs
02. Separate AC3/PCM audio
03. Adjust audio length, if necessary
05. Encode audio with HeadAC3he
06. Create an appropriate Avisynth script (with cropping and resizing) including Sampler()
07. Make sample strips to determine correct CQ
08. Encode
09. Mux audio and video in TMPGenc
10. Author and burn with VCDEasy

Re: Correct borders with Vdub
 

No, my stomach had too much beer :bugeyes:
I'm doing similar to SansGrip, but I found it faster and "clearer" to see the edges with TMPEG's internal crop. It gives me the same information as Vdub. Then, as SansGrip said, I take parameters and enter them in "Film Pixels" of FitCD and set left border and righr border. Everything else then is done by FitCD. In TMPEG, I'm using "Center" and 1:1 VGA as source aspect, as it seems to make no difference if I set to 4:3 or 16:9. So 1:1 is actually "Whatever comes resized via AviSynth" :wink:
Aspect looks correct on TMPEG and played back on WinDVD too.

-kwag

ahhh, you are a beer man. me too, but my unhealthy diet of around 30/week makes it feel like new years every day! hahahaha...

try 16:9 with full screen (keep aspect) and let me know what your results are. i hope its the same. thanks for the help!!!

:Drunk: I have to keep it "Center" because I removed the "addborders" from the script, and the .avs has the correct resize. If I set "Full Screen (keep aspect)" the output is wrong. :wink:

-kwag

For DVD rips

01. Rip VOBs with SmartRipper
02. Separate AC3 audio(s) and .d2v project (DVD2AVI) (Sometimes it works better to let SmartRipper extract the audio stream(s))
03. Get DVD chapters with Chapter Extractor or from SmartRipper log
04. Get subtitles (bmps + sub) with SubRipper
05. Adjust audio(s) length(s), if necessary
06. Encode audio(s) with HeadAC3he
07. Load .d2v in TMPGEnc and find the borders (film pixels)
08. Create an appropriate Avisynth script (with cropping and resizing) including Sampler()
09. Make sample strips to determine correct CQ
10. Encode
11. Mux audio(s) and video in TMPGenc (I have had better luck with it for SVCD muxing, my player doesn't like 480 lines mpeg-1)
12. Mux Subtitles as CVD with WinSubMux GUI
13. Author and burn with VCDEasy
14. Print covers and labels for the CD's :)

If I'm doing the movie in two CD's I treat each CD as a separate project so that I can make sure the movie is split at a chapter transition and each CD is fully filled.

For AVI captures

01. Capture with AVI_IO, HuffYUV or MainConcept MJPEG
02. Load .avi in TMPGEnc and set the borders (use TMPGENC to crop) and Ghostbuster parameters (if necessary)
03. Create an appropriate Avisynth script including Sampler(), MJPEGcorrect (if necessary), Decomb, Ghostbuster, etc.
04. Make sample strips to determine correct CQ
05. Load .avs in TMPGEnc and do Cut editing to eliminate comercials (makes little difference in prediction doing this at this point)
06. Encode video+audio (external audio coder set to toolame)
07. Author and burn with VCDEasy (may set chapters with the built-in player of the new version)
08. Print covers and labels for the CD's :)

Hi,

How do you do that ?? :eeks:
Is it really necessary ?

thx

--
apoc

It can be necessary, as a last resource, if you have synch problems. The most friendly way to do this with an audio editor like CoolEdit, sometimes a video editing tool like Ulead's helps.

@Kwag,

Here the instructions SansGrip gave me to Crop using FitCD.
I created a avs script with no filters as input to VDub->video->add->
null transform->OK->cropping. Then I following the steps below.
It gets tricky when top-bottom and/or left-right borders are different.
The advantage of doing it this way is faster encodes. Albeit FitCD
can give the crop settings, it would be nice to make this a more
user friendly process :)


-black prince

I generally only need to adjust the audio length if I'm converting a PAL movie into NTSC. To do this I use AssumeFPS(23.976), which means the video now runs about 4% longer and I need to "time stretch" the audio an equal amount for it to remain in sync.

Edit: Oops, forgot to say how I do it. I either use a utility called WLA (available from Doom9's download page, I think) or I use WaveLab. The latter is easier since I just enter the new running time instead of having to calculate the new total-samples value.

A quick clarification to what I wrote: when I say "uncheck anamorphic", I mean "uncheck destination anamorphic", not source.

Also I've since stopped putting AddBorders in my Avisynth script. I now set TMPGEnc's arrange method to "center" and it adds whatever borders are necessary.

@All,

It turns out that LanczosResize(496,352,0,8,720,464) in this
format will crop and resize. LanczosResize(496,352) and
Crop(0,8,720,464) is the same thing. There is debate as to whether
a resize should include a crop command. Having a separate crop
command before resize will speedup the encode process.
Choosing LanczosResize in FitCD will give you the both :D

-black prince

@SansGrip,

I wanted to add some more ideas to enhance documentation.
I'd like dividing each step into separate How to Guides. This makes
it easier to update or add more information to each step.

For example:

1) Ripping DVD

2) Converting VOB to .d2v

3) Converting Audio

4) Cropping and Resizing

5) Predicting File Size

6) Encoding

7) Multiplexing

8 ) Authoring and Burning


-black prince

That's a good idea. Of course, all of these guides can be written right now, except number 5 ;).

Here's what I do:
1) Ripping DVD with smartripper

2) Converting VOB to .d2v with dvd2avi

3) Converting Audio with headache

4) Cropping and Resizing with fitcd

5) Encoding with tmpg and kwag's templates

6) Multiplexing with tmpg

7) Burning with nero

I didn't add Predicting File Size because I haven't tried it yet,but I plan on trying it soon. Most of the time my movies fits on a cd so I burn it, and if it's over I adjust so the movie will fit. But this is the way I do it.

@SansGrip,

The manual prediction method is available. It's need to include a section
for CQ prediction, where Kwag takes the current CQ and adds the new
CQ then divides by 2 for the corrected new CQ. CQ_VBR is what he
has available. 8)

-black prince

@SansGrip,

Here's the manual file prediction guide:

http://www.kvcd.net/forum/viewtopic.php?t=2102

-black prince

What is the advantage of doing it this way? Does it help speed up the encode process? Between this and doing a Crop() separate from the LanczosResize() I might finally get my encodes down to a more sane amount of time :)

The advantage of letting TMPGEnc add the borders, for me, is that if I change my mind about what resolution to encode at there's fewer things to change in the script ;).

I'm not sure how it affects encode time. Probably not a lot.

I think with exception of FitCD, file size prediction and kwag templates, we can refer to Doom9 guides, they're very detailed and reliable. No need to reinvent the wheel :)

@acidfire,

Some of us use DVD Decrypter instead of Smartripper. Some use
BeSweet instead of Headac3he. Some multiplex with bbMPEG instead
of Tmpgenc and some use VCDEasy instead of Nero. Maybe these
steps should be without specific software so more than one way
can be described in the explaination of How To step. :?

@GFR

Maybe we should recreate all the steps, since some of Doom9's
guides have not been updated and are not specific to KVCD settings.
I never thought anyone would write an automated program like
acp, because we were using DVD2SVCD. Now it makes sense
because we can make changes to this program when new developments
arise more quickly and the same would be true for our own guides. :)
Also, putting updated dates on a step, let's everyone know the latest
changes for that step. :)

For example:

4) Cropping and Resize

5) File Prediction (updated: 01/07/03) :?

Step 4) has no changes, but step 5) was changed on 01/07/03.


-black prince

@ black prince,
Yes I understand what you are saying :lol: , but SansGrip wanted to know how everyone makes their kvcd. And in his first post he had his method and the software he used. So that is why I added the software I used. But I do agree with you when you say we should not limit it to certain software. I'm not sure if it would be hard or not but maybe we can have each step then inside each step can be different software and how to use each software :roll: . But first we should get the basic steps down and go from there.

@acidfire,

acidfire wrote:
Yep! your right. I was thinking ahead about designing a guide that new
visitors and experienced users could refer to for starting with KVCD.
I'm sure the steps within steps can describe other methods to accomplish
the same outcome. Guides change and with this forum where change is
rapid we need a source to keep up to date. My hope is that only stable
processes will be posted for update and only certain forum members will
be able to make updates. 8)

-black prince

I follow the steps and software suggestions put forward by Kwag and SansGrip. I simply don't have the time to do the extensive amount of testing that they've done, so if they say something works I tend to believe it and use it. This was especially true of bbMPEG. I loathed trying to use that software first time I tried, because it never worked. I asked Kwag what was wrong and he whipped up a screenshot of his settings and I haven't had a problem since!

To the point: KVCD has become much more than just a TMPGEnc template, it is a whole process from the ripping through to the burning. While I agree several software solutions will do the job, for any guides that are created for this site I think as much detail as possible should be included _based on the software being used by the authors_. Hence, if Kwag/SansGrip use HeadAc3he over BeSweet, for example, then document the process for HeadAc3he! Trying to document the process for multiple apps is very time-consuming, and will lead to either no guides getting completed or 'rush-jobs' that forget a few critical steps causing a flood of new posts here in the forums. If others want to contribute process documents for alternate software, by all means, but overall there should be a "Complete Guide to KVCD" which documents a single path from start to finish.

(NOTE: I mention Kwag/SansGrip above without any intention of slighting anyone else that has contributed significantly to the KVCD process).

@ black prince,
Sounds Great! :lol: I don't know much about making guides but maybe SansGrip, Kwag, and you can get someothers to help with making guides. And maybe assign certen sections to one perston. It might make it easier for all involved in making the guides.:wink: Example:One person workes on just audio or if you can get enough people say one person just dose heada3he in the audio section. And that person keeps their section updated. Hope this help or gives you some ideas! :lol:

This would be my preference not because kwag and myself use it, but because it's better :mrgreen:.

I agree. I think the first step is to get the "most common case" documented, whether that be HeadAC3he or BeSweet, acp or DVD2SVCD, a calculator or KVCD Predictor, etc.. This was one of the reasons I started the thread.

KVCD is by nature open source, and the best thing about open source is that anyone can contribute. I think it would be a very good idea if we set up some "teams" to document certain aspects of the process, then a few individuals could go through the results and edit and integrate into a whole. I suggest a few individuals because I find things tend to be more consistent that way.

Personally I much prefer coding and testing to writing documentation (as you can tell from my readmes ;)), and I think am more useful to the cause in that role, but would be more than willing to help out with the final integration.

An additional thought struck me while playing with the 3-disc version of Minority Report: a useful section of the documentation would deal specifically with AC3 audio from AVI sources:

1) Use VirtualDub to extract
2) Use BeSplit to repair the resulting file
3) Use BeSplit to join separate AC3 files if necessary

Then one could refer to the generic HeadAC3he guide for actually encoding the AC3.

@All,

Here's a link to Divx Guides, that has a structure we could discuss.

http://www.everwicked.com/modules.ph...des&file=index

The guide uses links within the guide for steps needed to create
the Divx video. I'm interested in the structure and how easy it
would be to follow along. You may have a better one or want
to comment on this one :D

-black prince