I thought so too Rory, that is, it being interesting.

Grazie

deleted by me . . appears I got a double IE post . . . maybe IE needs deinterlacing too?

Grazie

I thought it was a parent/child thing

Lol - g

"Grazie:I've been playing with the i to p options. What would you think would happen, coming from a PAL 50i?"

I think that it depends entirely on the algorithm used by the program. Thus, you can convert 50i to 25p or "upconvert" from 50i to 50p.

The television video world has preferred to "upconvert" from 50i to 50p because it is easy to do and because it reduces large area flicker considerably.

For so called filmlike results you could go from 50i to 25p with similar interpolating algorithms but updated per frame and not per field as above.

If you are referring only to Vegas way of doing it, it bears investigation. Good question since it is applicable to the 60i world also.

50p = 2 seconds of PAL.

I only know Vegas. So, yes.

And yes again, it was this sense of having all things P from an I source and thought of the double-PAL option - kinda got me excited as to what could be wrung-out or extracted from the fields. And yes again with the 60i World.


Grazie

Well the best way to deinterlace is either with the Mike Crash smart deinterlacer or the new one that comes with Boris BCC7. PAL deinterlace and uprez to 720p is better looking than if you try to do the same thing with NTSC by quite a bit, but it's still not like shooting HD.

" What would you think would happen, coming from a PAL 50i?"

You gain nothing and loose almost nothing.

Assume you shot 1080i50, just because the numbers are easier to work with. Exactly the same if you shot 576i50.
Each field contains 540 lines and were taken 20mS apart. Converting to 50p with de-interlace = interpolate you get frames of 10290 x 1080 that playout every 20mS. Each one of those frames though at best can only have 540 lines of vertical resolution. End result is you have the same spatial and temporal resolution. Nothing gained. As pointed out above this is probably done quite a lot in LCD and plasma TVs and in processing by Vegas etc.

You are however handling more data than you strictly need to. The 1080 lines only contain data from 540 lines simply interpolated to 1080 lines. Possibly you are wasting storage space although I'm inclinded to doubt it. Interpolation does involves smoothing / averaging to create pixels that didn't exit. Potentially if you kept repeating the process of going from 50i to 50p to 50i image quality may degrade.
I'm not certain what smart de-interlacing would achieve in this process. Potentially the parts of two fields that had no motion could be simply merged to give full raster resolution and the parts that do have motion interpolated. It could yield a better outcome but if you tried going back from 50p to 50i you might get a bit of a mess.

As far as I can see the only reason the option exists in Vegas is for compatility with 3rd party apps. Staying within Vegas the only use I can think of is for checking field order without using a proper interlaced display.

Bob.

Hi,

I always wanted to go from 50p to 100i, repeat what camcorder is dooing. But don't know how to do it in Vegas. It doesn't make sense if there is the same content in UF and LF. Fields should be interpolated. Knowing that, slomo would be be easy to achieve.

Best regards,

Sorry but I'm not understanding what you're trying to achieve.
If the camera is shooting 50p, say the EX1 at 720p50, then splitting the frames into fields gains you nothing making it 100i.

Bob.

Sure Bob, and thanks. But is there an advantage when it then comes to editing and rendering? I've been tempted to explore this option when I've been hearing about the advantages of using P media and as my media is interlaced I wished to explore the anything gained or lost imperative.

Interesting. Thanks Bob!

Grazie

"But is there an advantage when it then comes to editing and rendering?"

No, as I said Vegas would almost certainly be doing this internally anyway. From my experience I and P are the same to edit and the same to render with Vegas. P footage is easier to key and motion track and in general to composite however I'm far from certain that holds true if it's a real apples to apples comparison. Interlaced was usually from the cheaper cameras and progressive from film or very expensive video cameras. Be aware old news dies a very slow death in this game.

" I've been hearing about the advantages of using P media and as my media is interlaced I wished to explore the anything gained or lost imperative."

There are certain advantages to P media, no denying that however most start when the media was shot. Kind of the cake is baked and you cannot get the eggs back out of it. I should also warn there are disadvantages to shooting P that should not be taken lightly. I once shot a something 720p in a venue with fine textured bricks and trying to stop the jitters was a major pain.
If you want to explore the realm of P media then you can:

1) Create some in Vegas from the media generators.
2) Almost all stock footage is P and there's plenty of it to be had for free.
3) You'd do better converting your 50i to 25p using a smart de-interlacer.

I still mostly shoot interlaced because that's mostly what I have to deliver.I also deliver to YouTube and use Mike Crash's free smart de-interlacer to get 720p25. Takes a bit of fiddling around to get it working right but it does a pretty good job.

Bob.

"farss: Each one of those frames though at best can only have 540 lines of vertical resolution. End result is you have the same spatial and temporal resolution. Nothing gained"

This is only true for an archaic and abandoned method of interpolation at the dawn of digital video. Modern methods of interpolation, thanks to cheap and fast memories, provide considerably better vertical resolution closer to the theoretical limits, 1080 in your example.

Borrowing from MPEG methods, it is possible to interpolate not only from 2 contiguous fields but also from adjacent frames (additional fields), again if you want to use more temporary memory and/or processing speed. The interpolation may be more accurate but you cannot get something for nothing, so some artifacts may show up .

A good deinterlacer is very useful when capturing a still from video and if there is motion. Otherwise you get geartooth edges.

That is why investigating Vegas de-interlacing method is appropiate, whether 50i or 60i. Another interesting task on the list of things to do ...

Hi Bob,

I mean about :

real_time_environment -> IN -> camcorder processing -> OUT -> 50i

... I can deinterlace it to 50p.

But can also record file as 50p. I would like to have tool which can treat 50p as real_time_environment and produce 100i, like this :

50p -> IN -> plugin processing -> OUT -> 100i

... I can deinterlace it to 100p.

100p is beter for slomo then 50p. But can't find INTERLACER plugin with reasonable algorithm of building fields.

Best regards,

"That is why investigating Vegas de-interlacing method is appropiate, whether 50i or 60i. Another interesting task on the list of things to do ..."

At Grazie's request the discussion is limited to what can be done with Vegas out of the box. You may say " Modern methods of interpolation, thanks to cheap and fast memories, provide considerably better vertical resolution closer to the theoretical limits, 1080 in your example. "[/i]

The simple fact is Vegas only provides two mechanisms for de-interlacing, Interpolate or Merge.

I am well aware of advanced image processing technologies. They take time to do their thing and the results are quite variable. For anything in motion there is also the issue of motion blur.

I'd also point out that by design each field in interlaced video has nothing like 540 lines resolution by design. Due to line pair averaging used to reduce jitter and improve S/N the actual figure is closer to 400 lines or less.

Bob.

You're getting yourself confussed by the numbers and how cameras work.

In traditional 50i the camera scans every second line 50 times per second. You can treat the 50 fields per second as frames to get 50p with reduced vertical resolution. The temporal resolution equates to 50fps regardless

When a camera is set to 25p it may split each frame into two fields to produce 25PsF50. In this case the temporal resolution is 25fps NOT 50fps.

Set the camera to say 720p50. The temporal resolution is 50fps. Split into two fields if you like to give 50PsF100. If you de-interlace that to 100p you get a sequence of two identical frames and the temporal resolution remains 50fps. I know of no camera that outputs 100i. None of the standard camera interfaces support 100i. SDI doesn't and I'm pretty certain HDMI doesn't either.


Using software such as After Effects you can use time remapping with pixel by pixel tracking. This takes a LOT of CPU power. It can work really well or produce some quite amazing artifacts. If you cannot afford a real high speed camera that is your best solution. Much the same technology is also available for free in AviSynth or Virtual Dub.

Bob.

"farss: The simple fact is Vegas only provides two mechanisms for de-interlacing, Interpolate or Merge.

Due to line pair averaging used to reduce jitter and improve S/N the actual figure is closer to 400 lines or less."

I do not know of any other mechanisms other than Interpolate or Merge to convert from interlaced to progressive. The real issue, in addressing Grazie's question, is what kind of Interpolate and what kind of Merge Vegas uses. Depending on the algorithms (process) the results can be quite acceptable or pretty bad. I hope to find the actual answer because it helps in many situations. So far, grabbing a still in Vegas 9 (albeit 60i) is not very satisfactory if there is any motion at all.

As to the line pairing averaging, only some cameras use it. Better quality camcorders do not line average, specially for sports and outdoor scenes. For stationary, high detail and high contrast scenes, such as fine text, cloth grain and similar it is better to filter in post.

Grazie, what is your source of 50i video material ? What camera brand and model ? This a crucial starting point to address your interesting question.

"I do not know of any other mechanisms other than Interpolate or Merge to convert from interlaced to progressive"

There's Mike Crash's Smart de-interlacer, there's better systems than that which use motion vector compensation. Most standards converters use the latter. I use Mike's filter when extracting stills from interlaced video or producing 720p from 1080i


"As to the line pairing averaging, only some cameras use it. Better quality camcorders do not line average, specially for sports and outdoor scenes. For stationary, high detail and high contrast scenes, such as fine text, cloth grain and similar it is better to filter in post. "

Maybe CHEAP cameras don't use it but certainly the expensive one do. Without it fine details create line twitter with interlaced video. Certainly the EX1 and every other camera that shoots P and I reveals the use of it when tested. In any interlace mode vertical resolution drops and signal to noise increases. When you're broadcasting live there is no option to fix it in post.

Bob.

[Apologies to Grazie for this off topic]

"There's Mike Crash's Smart de-interlacer, there's better systems than that which use motion vector compensation. Most standards converters use the latter. I use Mike's filter when extracting stills from interlaced video or producing 720p from 1080i"

Mike Crash's plugin is somewhat outdated. Donald Graft issued a beta version adding edge-directed interpolation. It works fine with virtualdub and may provide better results than cubic interpolation. In order to get scale 1 screenshots, click on the last four thumbnails at the end of this post (blah blah is in french, sorry).