Feature request: three different encoding modes for HC

[Boulder]

As the latest HC release has an automatic interlacing detection mode (i.e. hybrid encoding), it would be great if DVD-RB also supported it in a proper way.

Currently, when you set a VTS progressive, HC actually encodes it using the hybrid mode because DVD-RB omits the *PROGRESSIVE call from the ini file and thus the hybrid encoding mode is used.

So, to take advantage of the three encoding modes, DVD-RB would have to either call *PROGRESSIVE, *INTERLACED or leave both of these out of the ini file.

[jdobbs]

Ok. I changed it for the next release. DVD-RB will specifically use either *PROGRESSIVE or *INTERLACED.

[Boulder]

What about the hybrid encoding mode, would it be possible to utilize that?

[jdobbs]

It really wouldn't apply to DVD-RB. It does its own hybrid determination based upon the source and overwrites the flags in the M2V during REBUILD.

[Boulder]

There are situations when the source is flagged as interlaced but the contents are actually mixed progressive and interlaced material. Many PAL documentaries are like this for example :)

[jdobbs]

Even more NTSC. Often they switch back-and-forth between interlaced and telecined. DVD-RB looks for those and handles them. It looks at each frame individually.

I'd have to redesign DVD-RB entirely to allow the encoder to make those kinds of calls itself. Right now they'd be overruled anyway.

[Boulder]

Is there any way to test how it works in my PAL cases?

[Boulder]

Even more NTSC. Often they switch back-and-forth between interlaced and telecined. DVD-RB looks for those and handles them. It looks at each frame individually.

I'd have to redesign DVD-RB entirely to allow the encoder to make those kinds of calls itself. Right now they'd be overruled anyway.But what does it do to the encoding mode - encode as interlaced? In PAL, you might be better off encoding in hybrid mode and flagging as interlaced.

[jdobbs]

I guess I don't understand why. What does HC use to determine whether it is interlaced or progressive that is different from DVD-RB? Does it have an algorithm employed, or does it also look at individual frames?

[Boulder]

Here's one description by hank:

"It now analyses fields instead of frames so interlaced can now be detected. That's done by looking at the difference of the variance & luma of the upper/lower field.

If a frame is really progressive it will be encoded progressive, otherwise it's done interlaced with real field DCT encoding for the MB's which need it. The scanning method, zigzag/alt, is also automated."

So where DVD-RB encodes as interlaced (?), HC encodes the frame either as progressive or interlaced depending on its analysis. This provides better quality because progressive content should be encoded as progressive with zigzag scan enabled.