What is the best de-interlacing method
 

I am looking for a de-interlacing method that will not lose me 50% of my vertical resolution? Also, if you have a method, please tell me which software or utility is used to achieve it.

Thanks

A popular method that can be done in most any NLE (without the aid of a plugin) is the following:
-Import your clip twice
-Place the clips on separate layers, one on top of the other
-Set the top clip to odd fields
-Set the bottom clip to even fields
-Set the transparency of the top clip to 50%

With this method, you'll be loosing like 25% instead of 50%. This is a lot better than the 'duplicate fields' method of dienterlacing.

If you are willing to spend some money, then look at some of the available plugins. I use Re:Vision Effects' FieldsKit Deinterlacer and it works very well. It's not one of those things that works wonders right out of the box, though - you have to tweak the settings a bit.


You may be interested in these links:
http://www.creativecow.net/articles/...deinterlacing/
http://www.creativecow.net/articles/...wixtor_review/

DVfilm Maker is one reasonably priced option that runs on the Mac and PC: http://www.dvfilm.com/maker

If you're using FCP on the Mac you should also check out the "Joe's Filters" package which has a nice deinterlacer plus a lot of other great plug-ins: http://www.joesfilters.com

Thanks guys, but again, is there no way I can do this de-interlacing without losing that much resolution (25% is still a lot).

You lose resolution when you simply discard one of the video fields, but the programs I mentioned are adaptive deinterlacers, which means they treat the static areas of the image differently from the areas with movement. They will give you the maximum possible resolution.

Of course all methods are starting out with interlaced video shot at 60 fields per second. Unfortunately this is not the same as true progressive video which captures 30 complete frames each second.

best thing to do (and i know this might sound silly) is to shoot in progressive.

no seriously.. to save teh time and hassle of al these tricks and methods, best to start at the source..

I used to do what your wanting to do, then i thought, well, i could save alot of processor time and get teh look i need jsut by shooting in this method.

Another way is to shoo tin Frame mode, which obviously isn as good, but pretty much similar to, if not better (and faster as theres no processing onvolved) than the Software solution.

my 2cents

The reason I shoot in i mode is that I want to convert to 24 from the 30p mode of my hd10u. This means I am already giving up my hi def, so you can imagine that the remaining resolution is important to keep.

Dwight: Converting from 30p to 24p may not be the only way to go. In one of those links I posted, it was recommended to originate in 29.97 interlaced (60 fields) and then deinterlace that, using some adaptive/interpolation method and rendering out to 60p. Then you take that footage and convert it to 24p. You'd be giving the program more footage to work with and it would look better.

Dwight, When converting from 30p to 24p, you have to completely drop every 5th frame of video creating stutter over the entire project. When transferring 60i to 24p they are working with fields that they can blend together tocreate a smooth(er) motion where the gaps would have been. 30p to 24p is so hard to work with, that most post houses won't even touch it.

Shoot in 60i, or trade in your JVC-HD cam toward a DVX or XL2.

My Film Effects has a great adaptive de-interlacer in it. It treats static areas differently from moving areas, and even in moving areas tries to preserve as much rez as possible. It also does 24p from 29.97fps 60i video.

That means that it does a great 24p from the FX1 footage that's been posted around, but poorly from 30p footage from the HD10.

For 30p footage you need to invest in something like twixtor or boris optical flow, or other super-slowmo program that can take many many times longer to render to produce the end result as it will calculate inbetween frames for you.

30p is pretty much the worst frame rate as to convert it to PAL or 24p you have to do expensive processing as above. Both 24p and 60i convert to PAL much, much cheaper, and with better results, IMHO.

Graeme

Trading in my hd cam is at the moment not a reasonable option (and adding the dvx100 or the xl2 will happen in the coming year "I hope"). I will try the plugins, and I believe that twixtor has a demo version that I can try to see if it works with my footage. As for the de-interlacing options, there does not seem to be a perect solution, so I will just have to try them all (well that what coffee is for)