Encoding lagarith from the commandline

[Foofaraw]

Could somebody suggest a 64bit compatible way of encoding a file to lagarith from the command line?

Like ffmpeg would be good, but it doesn't encode, it only decodes.
Preferably an exe/bat file (and not scripts or avisynth)

[Atak_Snajpera]

why do you need lagarith?
ffv1 offers similar compression efficiency and it is available in ffmpeg

Maybe because Lagarith is faster to decode than FFV1?

[Foofaraw]

Originally Posted by Atak_Snajpera

why do you need lagarith?
ffv1 offers similar compression efficiency and it is available in ffmpeg

Does it compress as good? I've heard lagarith is a far better.

However I also need something with a directshow codec (that works on 64 bit Windows 7) - lagarith has that, does ffv1?

[Foofaraw]

Originally Posted by SassBot

Maybe because Lagarith is faster to decode than FFV1?

That too

[spawnbsd]

Originally Posted by Foofaraw

Does it compress as good? I've heard lagarith is a far better.

No, using FFV1 with arithmetic coding (-coder ac) is much better compression than Lagarith; though as mentioned the decode speed is far worse.

[mastrboy]

One way i can think of is saving virtual dub processing settings and loading them via the command line for vdub.exe, but can't say for sure this will work with 64bit though... And virtualdub can't save/encode to any other container than avi, so a remux would be needed later if that is not your wanted container.

[Foofaraw]

Originally Posted by spawnbsd

No, using FFV1 with arithmetic coding (-coder ac) is much better compression than Lagarith; though as mentioned the decode speed is far worse.

Ok, might try it if there is a directshow interface.

Originally Posted by mastrboy

One way i can think of is saving virtual dub processing settings and loading them via the command line for vdub.exe, but can't say for sure this will work with 64bit though... And virtualdub can't save/encode to any other container than avi, so a remux would be needed later if that is not your wanted container.

Yeah, that seems to work. And avi is fine for now

[Foofaraw]

Btw, for those who know: Can FFV1 be destructive depending on arguments, or is it always lossless ?

[Atak_Snajpera]

ffv1 is lossless obviously. it is also available via ffdshow (encoder and decoder)

[Foofaraw]

Originally Posted by Atak_Snajpera

ffv1 is lossless obviously.

Huffy is obviously lossless, but has a setting that can make it lossy. So there is nothing obvious about anything. Especially since there seems to be 50 different settings

Originally Posted by Atak_Snajpera

it is also available via ffdshow (encoder and decoder)

Indeed, and its output crashes Media Player Classic.
But the files do seem small

[Atak_Snajpera]

So there is nothing obvious about anything. Especially since there seems to be 50 different settings

where do you see those 50 different settings??? Lossless is lossless. Period.

Indeed, and its output crashes Media Player Classic.

it does not crash on my machine. Tested on 1920x1080 24fps source

Originally Posted by Atak_Snajpera

Lossless is lossless. Period.

That's easily falsified. Choose a colorspace that involves color subsampling the input and it's no longer lossless. Either way, his question had to do with if there were settings to make the codec lossy to improve compression. All you had to say was "no".

[Atak_Snajpera]

That's easily falsified. Choose a colorspace that involves color subsampling the input and it's no longer lossless.

most sources is already 4:2:0 so default colorspace YV12 won't hurt quality. Unless he is pro and he has access to 4:2:2 footage.

[Foofaraw]

Originally Posted by SassBot

That's easily falsified. Choose a colorspace that involves color subsampling the input and it's no longer lossless. Either way, his question had to do with if there were settings to make the codec lossy to improve compression. All you had to say was "no".

Could you suggest to me what would be a good colorspace to use?

It supports YV12,444P,422P,411P,410P,RGB32, but I take it from your reply that none of those colorspaces use subsampling. What is the difference?

Also, for "Coder Type" spawnbsd suggested AC (VLC seems grainy) ?
"Context Model" small or large?

Also, are there resolution limits?

[Atak_Snajpera]

Could you suggest to me what would be a good colorspace to use?

the same as source

Also, for "Coder Type" spawnbsd suggested AC (VLC seems grainy) ?

AC offers slightly higher compression but it will be slower obviously

"Context Model" small or large?

small is faster but large may offer higher compression. Your choice.

[Taurus]

Originally Posted by Foofaraw

Indeed, and its output crashes Media Player Classic.

Just tested: ffdshow ffvh and ffv1 are working stable.
And no crashes in MPC HC.
Somethings burked on your side.

Originally Posted by Atak_Snajpera

most sources is already 4:2:0 so default colorspace YV12 won't hurt quality. Unless he is pro and he has access to 4:2:2 footage.

Not if he's using this for lossless of encoding screen capture footage or the output of a capture card that can be in YUY2. You presume his source is YV12 to begin with but that may not be the case.

Originally Posted by Foofaraw

Could you suggest to me what would be a good colorspace to use?

It supports YV12,444P,422P,411P,410P,RGB32, but I take it from your reply that none of those colorspaces use subsampling. What is the difference?

What is the source of your input file?

[Foofaraw]

Originally Posted by Taurus

Just tested: ffdshow ffvh and ffv1 are working stable.
And no crashes in MPC HC.
Somethings burked on your side.

Plays fine in Windows Media Player, VLC, VirtualDub

Originally Posted by SassBot

What is the source of your input file?

It varies. It can be 3d rendered (no settings for color space), it can be camcorder, and sometimes external sources (don't know what space they might be in)

I might add that ultra high commercial quality is not required - as long as it looks sensible on a TV via HDMI its probably fine

I tried running something through vdub and it says input was YUV420 and output became RGB888 - i couldn't really tell the difference.

[Bloax]

Originally Posted by Foofaraw

I tried running something through vdub and it says input was YUV420 and output became RGB888 - i couldn't really tell the difference.

That's because RGB24 is a fully precise colorspace (if we're talking 8-bit anyways).

The input format (YUV420/YV12) has a halved color resolution, for example. You won't see a difference going from YV12->RGB24, but you might be able to spot the difference in RGB24->YV12.