Magix Mainconcept AVC render quality problem

Comments

nepomuk wrote on 5/26/2019, 1:32 PM

Conversion of progressive material to true interlaced is not simple, because the missing field has basically to be generated in some way. There exist various approaches, like
- separate the progressive source frames into odd an even (virtual) fields, which in this case have no temporal difference, then encode and flag as interlaced. This will satisfy authoring tools and players which expect an interlaced format, but of course the temporal resolution is the same as the source. It is sometimes referred to as "fake interlaced" as the underlying material is progressive.
- Generate the missing field by blending adjacent frames and use the blends as "in between new frames" (or fields). This seems to be what MAGIX MC now does in VMS 16.
- Synthesize the missing field by temporal interpolating rather than just use blending. I have some doubts that VMS 16 does this properly. It should not produce blends or shadows, but is prone to other artifacts.

Unfortunately the documentation of VMS is not clear what happens under the hood, but what happens is definitely different when we render with Draft/Preview.... or Best quality.

Maybe I am missing something and someone else can shed some light on it.

EricLNZ wrote on 5/28/2019, 3:03 AM

I've now rendered my videos, that were unsatisfactory with the Magix MainConcept Blu-ray codec, with the original MainConcept Blu-ray codec. Using the same VBR 1 pass max bitrate 35 Mbps, Average 25 MBPS I find:

The files produced by the original codec are usually larger by up to 17%. Unfortunately MediaInfo only tells me the bitrate is max 35 and variable. It doesn't indicate the average so I cannot compare.

The Magix MC files correctly indicate they are Standard PAL. The original MC files say they are Standard NTSC even though they are correctly 50i Top field first. Presumably this won't cause me problems with DVDA. It is accepting them without recompression although I haven't created a disk image with them as yet.

nepomuk wrote on 5/29/2019, 2:11 PM

I am confused now. Are your videos frame rate converted by any of the encoders, like from 50i (PAL) to 29.97i (NTSC) or vice versa? Or do the old MC and MAGIX MC just report the same source differently?

EricLNZ wrote on 5/29/2019, 7:13 PM

I am confused now. Are your videos frame rate converted by any of the encoders, like from 50i (PAL) to 29.97i (NTSC) or vice versa? Or do the old MC and MAGIX MC just report the same source differently?

No, nothing is changed. Attached is MediaInfo for the two files and as you can see they are otherwise identical apart from a difference in file size. The original MC being larger. Unfortunately MediaInfo doesn't report the average bitrate to compare the difference They should be 25 Mbps average. Both correctly show Max 35.00 Mbps which is what I set them to as the template 40 gives problems with DVDA. This was the subject of a thread here.

The video is 5 Mins 19 Secs long i.e. 319 Secs. That gives average bitrates of 28.14 Mbps for the original MC and 24.29 Mbps for MagixMC. So the Magix MC is closer to the requested 25 Mbps average. As a test I also did a 1 minute render using 20 Mbps CBR with each. The file sizes were close (147,233 KB for original MC and 147,374 KB for Magix MC).