GK not recognizing interlaced films

[dap_werewolf]

I've been using AutoGk now for a couple of days on some of my mpeg2 captures because I've had to answer some questions on it, I usually use Gordian Knot. The few files I put in were determined by AutoGK as being progressive, which was a rare sight, being tv captures. So I even asked a friend if tv caps CAN be progressive, he said yes, so I didn't worry about it, though after about 250 captures I had never run across one yet. After about the third one in a row I decided to open the source file in GKnot, and when I ran the stats in DGIndex, sure enough it said the source was interlaced. Now I'm wondering if my recommendations to newbies that they capture to high res mpeg2 and encode to avi in AutoGK were all that good. I capture to mpeg2 at a bitrate of 12000, source is PAL. Yes, I know I could tell someone to force deinterlacing in hidden options, but how does one determine that the source is interlaced if it says during the analysis that it's progressive? Or rather, which one is right, the analysis in AutoGk or the stats in DGIndex through GK? Using GK I can check for artifacts, but how do I tell one to do that in Auto?

[dap_werewolf]

Actually I'm not so sure I named this thread right, as now I went one step further and put one of the mpegs that AutoGk said was progressive into GK and created a .d2v file (DGIndex, as I said above, said this same file was interlaced). I opened the .d2v and enlarged it, and I scanned it carefully but couldn't find any artifacts, so now I'm thinking that it's not AutoGk but GK (or rather DGIndex in GK) that's making the wrong determination. I'm using GK 0.35.

I've been reading on interlaced sources for hours now and it seems like for every statement there's a contradiction, as to whether TV satellite output (mine is from the UK so it would be PAL) can be progressive and most say that it arrives to the tv interlaced, even if it was progressive at the studio. But I'm beginning to think that many of the sources which GKnot told me were interlaced really weren't, because I can't see any artifacts whatsoever on the last two that I checked. I guess that the only way is to scan the file carefully for artifacts, but it's still interesting to me why AutoGK says one thing and GK says another about the same file. They both use DGIndex don't they?

[niamh]

You could start by reading this thread (http://forum.doom9.org/showthread.php?s=&threadid=91335&highlight=hollywood+why+bother) about DGindex reporting things interlaced, and particularly manono's link on flags :)

[dap_werewolf]

Originally posted by niamh
You could start by reading this thread (http://forum.doom9.org/showthread.php?s=&threadid=91335&highlight=hollywood+why+bother) about DGindex reporting things interlaced, and particularly manono's link on flags :)

Thanks for the thread, it was informative, and of course I could be wrong, but I don't believe it applies here. The sources I am referring to are PAL TV captures. All the material I've been reading is, as I said, contradictory, so far I've basically come to the conclusion that Sky Digital Sat broadcasts interlaced material for the time being, although plans for progressive hi-def tv sources are in the works. I'm also beginning to think my capture device deinterlaces it as it captures, even though I can't find any material to support this or negate it. I think I will have to stick to analyzing by eye each time for now, but I'm still interested to know why AutoGK and GK analyze the same file differently.

[len0x]

- GK doesn't analyse source at all
- DGIndex just displays flags from mpeg2 stream and often wrong because source is mastered incorrectly
- AutoGK does actual frame analysis and is never wrong about the source

[dap_werewolf]

Thanks, that's the answer I was looking for :)