Hardware MPEG-2 Encoders Better Than CCE-SP?

[Don R.]

Is it acknowledged that hardware encoders such as those marketed by Sonic, OptiBase or Vitec make MPEG-2 files that have better color fidelity, contrast and resolution than Cinema Craft CCE-SP?

The reason I ask this question is the other day I saw the results of a comparison test between a hardware encoder and CCE-SP. I keep seeing the hype written about software encoders being better than hardware encoders so this test was very interesting and has pretty well settled the question for me.

The same Quick Time Reference file was encoded to a MPEG-2 file by a $1,500 hardware encoder and then used to author a DVD. The DVD authored from the hardware encoded MPEG-2 "looked" better on a TV scrren than a DVD authored from the same Quick Time Reference file but encoded by CCE-SP at the highest bit rate and quality settings since size of file was not important in this test. Sonic's DVD Producer was used to author and burn both DVD's.

With all they hype that people have written about CCE-SP, I was surprised to see Cinema Craft come in behind a cheaper hardware encoder.

Has anyone else seen the results of a head-to-head comparison of CCE-SP and a hardware encoder? If so, what was your opinion?

[Don R.]

The local Optibase rep encoded a 3 1/2 minute test tape that I put together and had encoded using SP, MainConcept 1.4, Sorenson Squeeze 3.5, BASIC, and TMPGENc.

The Optibase encoded MPEG-2 files made a better looking DVD than did the software encoders. The Sorenson software was the best of the software encoders.

Have the software encoders caught up with SP?



Originally posted by paceri
where did you see those results... link please :)

[DDogg]

made a better looking DVD than did the software encoders Don. that's pretty subjective. Can you flesh out why it looked better? More smoothing? Saturation? etc.,etc. Do you have the bandwidth or webspace to allow us to download a snip of the two clips plus a corresponding snip of the original? If so I could do some tests using some of the newer mathmatical comparison methods.

[Don R.]

Morning, Dog,

What about me sending you a MiniDV tape with the 3 minute test on it?

Would send it FedEx if you like.

[Don R.]

Forgot to say why the hardware encoder MPEG-2 file looked better.

Image was much more brilliant. There was none of the haze look. More contrasty.

The hardware encoder reminded me of prints made from negatives from a camera with a high contrast lens versus prints from a camera with a low contrast lens.

I do not have a website but I do have a DSL line so downloading is not too much of a problem.

Sending you the MiniDV that contains the tape that Optibase ran through the hardware encoder might be the best apples-to-apples comparison method.

No problem at all doing that.

Don R.

[DDogg]

Appreciate the offer but it would not help as my encoding setup is quite different from yours. To be objective we would have to see the original, the hardware encoded snip and the CCE encoded snip from your system and particular setup, especially since you use QT and report a "haze" from your encoding setup. Frankly I would not be surprised if a vertical hardware solution using the latest technology/chips, etc. would outperform a software based solution like CCE. One would hope so at least.

Don, slight OT, how do you capture your raw DV from your camera? In other words do you use a proprietary DV Codec with Avid? Where I am going is to suggest you cap a small snip with some cap util without Avid being involved so that the MS DV or one of the standard VFW codecs like the Sony or Panasonic codec is used for playback as a benchmark. Load that directly into CCE and encode a snip and see if you get the same "haze". May not be worth your trouble, but I was trying to isolate Avid and especially QT out of the equation.

but I do have a DSL line so downloading is not too much of a problem. I have a good downstream so that might work if you have a FTP setup. If so, let me know via PM yout instructions.

[Don R.]

Just finished observing the first two hardware encoder tests and there was NO contest.

Back to the drawing boards with SP.

[sh0dan]

The no-contrast-hazy look could be because of luma scaling, done wrongly by a DV-codec. Some codecs excibit this unfortunate behaviour.

If you posted some screenshots or two short m2v sections we could have a look.

[Don R.]

Sh0dan, that is an excellent comment. As a matter of fact, the USA retailer has asked for a copy of the DVd codec on my PC plus copies of the tape and encoded results so he may have a firt hand look at the problem.

However, I am certainly not the first to note this problem.


Hopefully, CCE will be able to handle the problem and work out a solution.

Thank you again for your comments.