Batch Intermediate Creation Utilities

[Eugenia Loli-Queru]

wrong url

[Eugenia Loli-Queru]

So, I tried it with the latest ffmpeg version http://ffmpeg.arrozcru.org/builds/bi...latest.tar.bz2 and it works well, thanks. Just one feature request:
Is it possible to be able to change the size of the MJPEG proxy? You see, there's no reason to have a full sized proxy that takes more space and time to decode. It would be nice for the proxy to either have its own avisynth script, but I think that even better it would be for an input field, to enter e.g. a 640x360 size.

[Khaver]

Corrected url.

http://www.westernwoodstructures.com...reator-v1b.zip

I'll add a resize input box for proxy files in the next release.

[racer-x]

One of the few times I use an Intermediate codec, is when I wnat to do a Slo-Mo effect. This tool helps things go a little more quickly.

While editing HDV footage in Vegas, I select the region I want to slow down and export to HDV.m2t. The smart rendering works well and quickly produces the clip. I then use your ffmpeg-bic to load the clip, select Mjpeg and load a custom AVS script I keep in the folder. This produces very clean 50% slow-mo with audio to match.

Here's the script I used if anyone is interested:

### Mencoder Batch Intermediate Template
LoadCPlugin("C:\Program Files (x86)\AviSynth 2.5\Filters\yadif.dll")
Directshowsource("__vid__",pixel_type="auto")
Yadif(1,1)
Assumefps(29.97)
TimeStretch(tempo=50)#(audio speed %)
ConvertToRGB(matrix="PC.709", interlaced=false)

[Eugenia Loli-Queru]

MJPEG is extremely lossy, I wouldn't recommend it for anything other than proxies.

[racer-x]

While I agree that for heavy filtering and constant re-compressing, it would be much better to use a lossless codec like lagarith.

However for my purpose Mjpeg is a very good choice. Vegas loves it and with compression set to min, its 125 mbps and the quality is very good. I checked frame grabs of source and destination and could not see any visible differences. Plus the overhead is very light and I can easily playback at full speed on my system.

[Khaver]

Racer-x,

Have you tried to see if there's a speed difference using the yadif deinterlacer in ffdshow (if you have the latest one) as compared to using the plugin in avisynth?

[racer-x]

Well I never thought about using the built-in de-interlace filters in ffdshow, so I gave it a shot. The ffdshow/yadif combo (11 fps) is indeed a little quicker than the Avisynth/yadif method (10 fps).

I'm glad you mentioned this because I did some experimenting with the de-interlace filters in ffdshow. It's giving me lot's of Ideas about HD to SD resizing ect. ffdshow has some very good de-interlace filters.

[Khaver]

New version of FFMpeg-bic.

I've added an input box so you can specify the resolution of the proxy files if you don't want it the same as the intermediate files. Enter as WxH. Leave blank for same res as intermediates. This value will be saved between sessions.

You can also specify if you only want proxy files to be created.

http://www.westernwoodstructures.com...reator-v1b.zip

[Eugenia Loli-Queru]

Thank you, it works wonderfully! BTW, one last addition would be nice to remember the last opened directory in the "add files" explorer window. I usually put my files on the same drive/folder (but just in a different subfolder).

[Khaver]

Per Eugenia's request, revised FFMpeg-bic.exe will now remember the last add directory between sessions.

http://www.westernwoodstructures.com...reator-v1b.zip

I'm thinking about dropping Mencoder-bic unless there are requests to leave it in.

[improser]

I see in the AVS template to 24p are a 24.00 framerate? They would be 23.976 or im wrong...?

Anyway, great program! please go ahead to implement more things!

[Eugenia Loli-Queru]

24000/1001=23.976, so it's fine.

BTW Khaver, I updated my HDV and AVCHD pulldown removal tutorials on my blog and included your method.

[Khaver]

The assumefps function can use the format: AssumeFPS(clip,int numerator,int denominator)
This is how the template is set up. 24000/1001=23.976

I'm glad you like the program. I'm not too sure what else needs to be added for creating intermediates and proxies. If you can think of something, let me know. Maybe just adding more formats as the FFmpeg developers add them.

[Khaver]

Thanks Eugenia, it's an honor.

I checked out your blog. You might make mention that with the AVS templates provided you need to verify that ffdshow is doing the deinterlacing.

[Eugenia Loli-Queru]

Np! Thanks for the great utility. I tweeted about it btw, get the links to check the method out here http://twitter.com/EugeniaLoli hopefully I didn't forget anything in the method.

[Eugenia Loli-Queru]

>you need to verify that ffdshow is doing the deinterlacing.

For the 50i/60i files. But for the PF24 files this is done by the TIVTC, no?

[Khaver]

I think the tdecimate function requires progressive frames. This tempate doesn't use the tdeint function that I believe the HV20Pulldown template uses.

[Khaver]

Eugenia,

I guess the original HV20Pulldown avs file used TFM to deinterlace and generate the progressive frames that tdecimate requires. My script relies on ffdshow to feed tdecimate with progressive frames. It seems either way works. What I think TFM does is actually find the duplicate frames and replace one with a copy of the other one. That way the duplicate frames will actually be exact duplicates so tdecimate will work better. Maybe I'm wrong, but just deinterlacing the 24p in 60i will give you the same progressive frames, only the duplicates are what the camera actually outputs. I think they're close enough for tdecimate to work. Plus, just deinterlacing with ffdshow is a little faster than using TFM to do the deinterlacing.

[Eugenia Loli-Queru]

Ok, let me get this straight, cause you are confusing me. Does your 24p avs script with the rest of the utils used remove pulldown properly, or not?

[Khaver]

Yes it does. You just have to make sure deinterlacing is being done before being fed to the AVS script. That's why I have the ffdshow configure button on my utility so you can easily check and adjust things as needed. And as racer-x found out, you also have to make sure ffdshow IS the default decoder for your source files.

In the IVTC-24p.avs script you could add TFM() right before tdecimate() and in ffdshow turn off deinterlacing. This gives the same 24p output file as my default method.

I tested this on a few 24pin60i clips I have. You might want to test it yourself.

[Khaver]

This revision saves the last used intermediate compression format between sessions.

http://www.westernwoodstructures.com...reator-v1b.zip

[racer-x]

Khaver wrote:

Yes it does. You just have to make sure deinterlacing is being done before being fed to the AVS script. That's why I have the ffdshow configure button on my utility so you can easily check and adjust things as needed.

Just so you or anyone running a 64-bit version of Windows knows, that the ffdshow configuration button won't work on 64-bit versions of Windows, because it is installed in C:\WINDOWS\SysWOW64.

To be honest, I can't remember if I installed the 64-bit version of ffdshow only, or both 32-bit and 64-bit versions on my XPx64 install............

[Eugenia Loli-Queru]

Hmm. If you are using ffmpeg, why not use ffmpeg to decode too? Why add ffdshow as another layer? This way, the user doesn't have to worry about such details.

[Khaver]

I wanted the flexabilty to be able to use avisynth plugins and/or the filters built into ffdshow. FFmpeg has only 1 deinterlacer. FFdshow has several good ones (Tom's MoComp, Yadif and the one ffmpeg uses). My tool is meant to be a sort of pre-processor for the created intermediates, not just a pulldown remover. Using just the ffdshow interface, I can manipulate the video in a variety of ways (noise reduction, grain addition, sharpening, etc.) while I create the intermediates and I can use an avisynth plugin at the same time if needed.

There are already many GUI front ends for ffmpeg and mencoder.

Flexabilty is a good thing.