How to compare video (codecs)?

[Flynnstone]

What's the (best) way to compare codecs?
I'm no sure of the source material yet, but lets say DV.
So I envision encoding the DV to another form, say H264.
Then converting it back to DV.
Then it would be good to run both image streams side by side. Quicktime?
Is there a visual diff for video?
Is DV a good format? what would be better?

I ran into some comments about MPEG-2 having better image quality than MPEG-4. This seems counter intuitive to me. I would like to get the source and compare.

Any help is appreciated.

Tim

 

[-DH]

It sounds like you want to compare apples to oranges; they're both fruit, but what is the point in comparing them? DV is compressed 5:1 as its recorded to tape (or HDD). All MPEG formats are usually compressed much, MUCH more than that. Typically the higher the compression, the lower the quality. That's why its best to start with the highest quality possible and why most commercial production companies use still use film or very high end video formats where one camera/lens could cost more than your house.

MPEG formats are generally used as a delivery format - not really intended for editing (although they can be). The point is to NOT compress any more than you need to ... and when you do need to, do so using the codec that best fits your delivery method.

MPEG-2 can be higher or lower quality than MPEG-4. It all depends on the frame rate, pixel resolution and bit rate used during the encoding process. MPEG-4's main advantage is that it is much more scalable than MPEG-2.

-DH

 

[Flynnstone]

I chose DV because the source is likely to be SD (NTSC).
And DV is lightly compressed.
And I'm not sure what I can get.

It all depends on the frame rate, pixel resolution and bit rate

But for the same frame rate, pixel resolution and bit rate, H.264 should be better, correct?

How do you test?

 

[LethalWolfe]

The best way to compare codecs is to start w/a higher quality image than the codecs you are comparing. For example, if you want to compare DV and MPEG4 you wouldn't want to start w/DV footage you shot on a camera because then you have to factor in the quality of the camera as well (and unless you are shooting a w/a hi end camera you'll never push the codec to its potential).

So maybe a good place to start would be to create an animation in Motion using the Broadcast NTSC preset then export it using the animation codec. This should give you a very high quality NTSC QT to work with. Then you can import that into Compressor and export as many variants as you like. Once you have all your different QTs you can play them side by side to compare how they look. If you have PS or PS elements you can export a frame from each QT and compare the stills as well in there as well.

As -DH says, you have to look at the whole picture when doing a comparison. For example, HDCAM (the de-facto HD b'cast standard) is compressed more than DV (up to 10:1 vs 5:1) but HDCAM is way, way higher quality than DV.

Also, even though MPEG-2 and MPEG-4 are mostly known for their lower quality, more compressed delivery variants there are very high quality versions of the codecs as well. For example, HDCAM is based on MPEG-2 and HDCAM SR (a newer, higher quality version of HDCAM) is based on MPEG-4. The biggest different between between these high quality acquisition codecs and their lower quality delivery versions is HDCAM and HDCAM SR have a higher bit rate and are I-frame only (meaning they use intra-frame, not inter-frame, compression). Quick Wiki link.


Lethal

 

[Flynnstone]

The best way to compare codecs is to start w/a higher quality image than the codecs you are comparing. For example, if you want to compare DV and MPEG4 you wouldn't want to start w/DV footage you shot on a camera because then you have to factor in the quality of the camera as well (and unless you are shooting a w/a hi end camera you'll never push the codec to its potential).

Agree. Any source I can get would come from a NTSC SD camera.
I don't think I can get uncompressed, but ...
DV might be easier to come by and decent quality.

The source is under water video. The complaint with MPEG-4, is that, say you're taking a video of a pipeline. A fish swims into the picture. MPEG-4 tends to sharpen the fish and blur the background. In this instance, the pipeline is the interesting bit and the fish is a "foreground noise".

So I want to see if this is the case.

For testing, it would be nice if I could use Quicktime with 2 synchronized windows. That way I could step frame by frame and compare.
Am I off base with this?

Also what about MJPEG. There seems to be a setting of 1 to 20 quality. Is this standard?
What is a reasonable setting? and compression ratio?

 

[bigbossbmb]

The complaint with MPEG-4, is that, say you're taking a video of a pipeline. A fish swims into the picture. MPEG-4 tends to sharpen the fish and blur the background. In this instance, the pipeline is the interesting bit and the fish is a "foreground noise"

That sounds like the camera is on auto-focus and has nothing to do with the codec.