Is there any benefit to converting lossy to lossless.

[jpvoodoo]

I have a ton of low quality 3g2 files from an old cell phone that I want to convert to h.264.
Is there any benefit to going from a lossy format like 3g2 to huffyuv before going to h.264?

I might be encoding them to something else later if that matters. I was concerned about converting from lossy to lossy and later to lossy again.

[poisondeathray]

I have a ton of low quality 3g2 files from an old cell phone that I want to convert to h.264.
Is there any benefit to going from a lossy format like 3g2 to huffyuv before going to h.264?

I might be encoding them to something else later if that matters. I was concerned about converting from lossy to lossy and later to lossy again.

No benefit for that usage case.


If you convert from the cell phone video directly to h.264, you'd get the same results

Using a lossless intermediate is all negatives in that case - huffyuv will be much larger in filesize, less compatible, take more time, and sometimes not even lossy (the original huffyuv does not support 4:2:0, so you introduce some loss by going to 4:2:2, then back to 4:2:0, because the method it uses is not nearest neighbor)

[jpvoodoo]

I tried one and the difference is negligible.

But maybe you can also answer the question generally as well.

Are there cases where someone would get better results? For example, transcoding a reasonable quality mpeg to huffyouv before h.264? In what instances might it be recommended? How do you decide?

[jagabo]

The process of converting a lossy codec video to another lossy codec video involves first decompressing the source to uncompressed video, then compressing that uncompressed video to the new lossy codec.

lossy source ---> uncompressed video ---> new lossy compression

The process of converting a lossy codec video to another lossy codec via a lossless intermediate involves first decompressing the lossy source to uncompressed video, saving that uncompressed video with a lossless codec, then opening that lossless intermediate, decompressing it, and finally compressing with the new lossy cocec.

lossy source ---> uncompressed video ---> lossless intermediate file ---> uncompresseed video ---> new lossy compression

As you can see, all you are doing is wasting time (compressing and decompressing the lossless intermediate) and space (for the large intermediate file). So, in general, a loss intermediate accomplishes nothing useful.

There are some exceptions:

1) If your editor can't read the source video (can't handle the container or codec) converting to a lossless intermediate can give you a file your editor can read.

2) Long GOP, out-of-order codecs, can be slow to seek and hard to random seek (often seeking to the wrong frame) -- for both editors and encoders. If you're having trouble with this using a lossless (all I-frame) intermediate will help. Such a lossless file is quick to seek and always seeks to the right frame.