Question with AVCHD, 1080i, Deinterlacing...

[xLusi0n]

OK, so I did lot of research on the net before posting this.

My AVCHD recordings are 1080i at 1440X1080i X 60

When I play it in DivX, it plays at 1440X1080i but looks squeezed together. When I play it in all other media players, it stretches it out to 1920X1080 like it's supposed to...so I just blame this part on DivX Player.

I believe this is called anamorphic widescreen and the new AVCHD recorders record in full 1920X1080i/p.

Question #1: Do I lose quality when encode it to H.264/MKV and it makes it 1920X1080 for me (from 1440), should I not resize it?

I know that I should deinterlace video for computer playback and definitely before resizing video...but,

Question #2: The quality of the deinterlaced product can vary by the deinterlacer filter used?

Question #3: Before deinterlacing it was 30 frames / 60 fields per second...after deinterlacing it will stay 30 frames, but when displayed back on to a TV...how does it recreate 60 fields if its not interlaced anymore?

Question #4: On some of the video converters, there's an option to double the frame rate after deinterlacing...why would I do that? Would it be higher quality? I know it would increase the file size...

Sorry for such long posts and dumb newb questions. I really appreciate everyone's help!

[J_Darnley]

1 - Yes, you always lose quality when you use lossy encoding. The two sides to resizing are that if you do it before encoding you can spend time with a fancy resizer that does "hard" things which are too slow to use in real-time. But doing this also requires more bits to keep the same quality if you don't resize. It also increases the encoding time if you resize before encoding because the encoder has to process more video.

2 - Yes, there are a wide variety of deinterlace filters which produce varying subjective qualities at varying speeds. From simple quick ones such as Avisynth's Bob() through yadif to really slow ones which use motion compensation.

3 - The even and odd fields no longer show different points in time.

4 - Yes, same-rate deinterlacing effectively removes half the temporal resolution. Whereas double-rate deinterlacing retains the temporal resolution. People like this for sports and other fast-motion, fast-action videos

[xLusi0n]

Thanks for the reply,

So, to summarize it, when I convert my AVCHD files to MKV, I should deinterlace, and double the frame rate if it's fast action type stuff but leave it at 30 otherwise.

I archive my AVCHD originals anyway and just want something to use to send people (smaller file) at a reasonable quality.

Back to question #1, perhaps I'm not understand or didn't word my question correctly. I know I'm recording in anamorphic 1440x1080 resolution. Disregarding the loss from using lossy encoding and other factors, does the picture degrade by going from 1440X1080 to 1920X1080? What is filled into the videos to make up for the 480 lines missing?

If I don't resize my 1440X1080 to 1920X1080, the same players should still automatically stretch my video out to play widescreen 16:9 right? (It does that already with the original M2TS file).

Thanks.

[scharfis_brain]

if you just want to resize in the horizontal direction (eg. 1440 -> 1920) you don NOT need to deinterlace.

deinterlacing only is required for vertical resizing.

[xLusi0n]

if you just want to resize in the horizontal direction (eg. 1440 -> 1920) you don NOT need to deinterlace.

deinterlacing only is required for vertical resizing.

Ah, thanks for that tip. Makes sense since only those lines are interlaced.

Ultimately, the recorded resolution is only 1440 (at 1.33 PAR) so when I convert it to 1920 (at 1.0 PAR), where do the extra lines come from? Is the quality difference noticeable?

[scharfis_brain]

ultra crisp detail may be blurred slightly.

just leave it as it is unless you need to conform to a strict standard that asks for a certain resolution.

[J_Darnley]

Ultimately, the recorded resolution is only 1440 (at 1.33 PAR) so when I convert it to 1920 (at 1.0 PAR), where do the extra lines come from? Is the quality difference noticeable?
Where do they come from? They are created, but since it would be noticeable (in a bad way) if only the "missing" lines were created, so all the output lines are created based on the input. How are they created? By computer magic.

I would agree with scharfis_brain, I leave all my DVDs in their anamorphic size and I would probably do the same with any other video format except if it is too big to be shown on my screen (which 1440x1080 is). In this case I would resize a width of 1280 and the appropriate height including cropping and adjusting for the pixel aspect ratio.

[xLusi0n]

I found a great explanation of anamorphic size: http://seansense.net/2008/04/06/final-cut-express-40-1920x1080-vs-1440x1080