[RESOLVED] Show=true display progressive_frame indication from NVCUVID

[TheThing_89]

Hi there.

I'm using DGDecNV/DGIndexNV to attempt to deinterlace or work with MBAFF source video. According to the docs, PureVideo Deinterlacer can use progressive_frame indicator from the NVCUVID. Since I'm dealing with MBAFF source and I'm trying to debug it, I was wondering if it was possible to have the show=true flag display the progressive_frame indicator so I can verify it is correct or false.

Best regards:
TT

Update:
I wanna emphasise that this thread is for talking about progressive_frame indicator, especially in relation from what NVCuvid reports when dealing with for example MBAFF content. It is not the thread to talk about how to deal with MBAFF de-interlacing. If people are interested, I'm willing to start a new thread on tackling that problem and getting to the bottom of it.

With love (nohomo): TT

[Rocky]

Shouldn't be too hard. I'll see what I can do. Thank you for the suggestion and welcome to the forum!

[Guest 2]

Shouldn't be too hard. I'll see what I can do. Thank you for the suggestion and welcome to the forum!

I quote the request, joining to my other thread about use_pf.

It could be nice to have a flag beside every frame in DGI to see what NVCUVID reports about it, so we can manually go to it and investigate.

[TheThing_89]

Thank man

Cause it seems from initial inspection that purevideo deinterlacer is deinterlacing frames it shouldn't be. As an example, here's a frame comparison from the BD that I found: http://www.framecompare.com/image-compa ... n/9JJMCNNU
Direct image link to DGSource(".."): https://i.imgur.com/Xdq61r1.png
Direct image link to DGSource("..",deinterlace=1, use_pf=true): https://i.imgur.com/9aFmMj1.png

Zoomed in version: http://www.framecompare.com/image-compa ... n/7DDPLNNX

Unfiltered raw, looks fine normal progressive frame


With purevideo deinterlacer:


As you can see, with purevideo deinterlacer, it seems to add interlacing effect strokes on the frame and I wanted to know debug why it's deinterlacing this frame cause it looks progressive to me.

Hence the debug information

And thank you I was pleasently surprised to see this program and forum so active considering all the dead links I encountered while doing a re-install and re-update on all encoding programs, filters and such.

[TheThing_89]

I can upload a sample if that would help onto an HTTP server but that's a completely unrelated issue to be honest and probably shouldn't be discussed here.

Thanks again for a wonderful tool

[Rocky]

Sure go ahead with that. If you need to, send me a PM about matters you don't want public.

[Bullwinkle]

Hey Rock, DGIndexNV already shows the progressive_frame flag in the Info dialog, but only for MPEG2 (Frame Coding). Why don't you add it for all the video types? Much easier than mucking about with DGSource().

Would that work for you, TheThing_89?

[Rocky]

Seems reasonable. Thanks, Bullwinkle!

TheThing_89, be aware that it is common to encode progressive content as interlaced. In such a case where the flag is not accurate, it could be better to use a separate adaptive deinterlacer that tries to deinterlace only interlaced areas. FieldDeinterlace? TDeint?

[Curly]

use a separate adaptive deinterlacer that tries to deinterlace only interlaced areas

Why don't we have a nice fast HBD CUDA deinterlacer like that? I'm sure Sherman could whip it up in a jiffy. Not!

[Sherman]

If it's macroblock adaptive theoretically we could use the macroblock type to determine whether to deinterlace a block or not. That is, do things at block-level. However, the deinterlaced area detection method may already approximate that sufficiently.

[TheThing_89]

Hey Rock, DGIndexNV already shows the progressive_frame flag in the Info dialog, but only for MPEG2 (Frame Coding). Why don't you add it for all the video types? Much easier than mucking about with DGSource().

Would that work for you, TheThing_89?

Uhh, are you talking about the info dialogue for the full process? I just wanna be able to view individual frames if they have progressive_frame flag on them, not on the full video itself if that makes sense.

ressive content as interlaced. In such a case where the flag is not accurate, it could be better to use a separate adaptive deinterlacer that tries to deinterlace only interlaced areas. FieldDeinterlace? TDeint?

I know but I've never found a good deinterlacer that can deinterlace MBAFF correctly. There are definite full frames that I've seen that are fully interlaced but with MBAFF it can be done on a macroblock level and most de-interlacer just fiddle with frames they shouldn't be fiddling with. Never found the options to make TDeint or FieldDeinterlace only deinterlace the frames they should be deinterlacing. As you saw in my screenshot, purevideo deinterlacer in nvidia was definitely fiddling with progressive frames.

Right now I just wanna know what the progressive_frame indicator is showing cause I wanna know if the bluray is just derped or the deinterlacer is derped or nvidia is not flagging frames correctly. It will set me on the correct path.

[TheThing_89]

I wanna emphasise that this thread is for talking about progressive_frame indicator, especially in relation from what NVCuvid reports when dealing with for example MBAFF content. It is not the thread to talk about how to deal with MBAFF de-interlacing. If people are interested, I'm willing to start a new thread on tackling that problem and getting to the bottom dealing with MBAFF.

With (nohomo) love <3
TT

[admin]

Hey, we don' tolerate bullies telling people what they can post. And this "nohomo" stuff is way over the line. So user TheThing_89 is deleted for unfriendliness. If you want to participate here you can make a new account (using same email if you like) and then do not try to bully our members or make stupid remarks about sexual orientations. And BTW Rocky added the capability to see the progressive_frame flag for AVC in slipstream 207. You don't have to Save Project (the full process); you can just do a Preview at any point on the timeline and then stop.