Never use 2 field capture?

[Hulk]

http://people.freenet.de/codecpage/

The page listed above says never to use 2 field capture when compressing for a number of what seem like logical reasons. Is this true?

What is the difference between two fields and interlace capture as opposed to one field capture? What are the advantages/disadvantages of each?

[Dr Mordrid]

For doing VCD's or half-height video sizes for display on a computer screen I tend to agree, with a few exceptions.

Capturing in 2 fields and then using most software to "delace" the video to VCD height is silly. 90+% of sofware "delaces" by just dropping a field, so capturing both fields is redundent. Both Premiere and MSPro do this.

ON THE OTHER HAND if you use a program like FilmRender, which has a smart delacing module, resizing an interlaced clip to VCD size (single field) can actually improve the quality of the result IF its set up properly. However this does depend somewhat on the source material. Experiment....

For full height video destined for viewing on a TV 2 fields are essential.

Dr. Mordrid

[Walrus]

Hulk: For lossy codecs, the case can certainly be made ESPECIALLY if the video doesn't move too much that you may as well set your motion-JPEG codec to encode the full resolution frame as a progressive frame and not an interlaced frame. Of course you could always just lower the compression level which would negate this artifacting due to lower effective vertical resolution for each field. That said, interlaced MPEG-2 (which is a far more lossy codec than MJPEG) can look quite good in interlaced mode even with the same issues. (just look at any decent DVD of a TV program) The downside to using the technique described on the site is that heavy motion causes very heavy interlacing artifacts, and these do not compress well. They end up being very blurry, and this could take motion that is otherwise fluid and make it very messy. The real answer to this (if you can) is to capture with a lossless codec like HuffYUV and avoid all of these compression trade-offs until your final rendering stage. (you get the maximum quality this way)

[kefoo]

Hey Doc! Where can I find FilmRender? It's not showing up in any search engines. Thanks.

Kefoo

[Walrus]

kefoo: You may also want to take a look at VirtualDub and Donald Graft's Smart Deinterlace filter add-on for it. Both are free, and incredibly powerful. (Virtualdub also now has a good inverse telecine feature as well for recovering the 24fps progressive stream from a NTSC signal)
Virtualdub is at http://virtualdub.com.. if you dig around, there's a link to Donald Graft's filter page in there

[Dr Mordrid]

FilmRender can be had at;

http://www.nurkware.com

It's not a freebie, but the trial version is way useful. Besides delacing it also does other very neat stuff like user defineable mono & color film grains, frame blend, frame clip and color channel control.

Dr. Mordrid

[Flying dutchman]

Doc,

My own experienc is that it's better to capture full-frame and do a "smart deinterlace" followed by bilinear or bicubic resize for three reasons:

1) you'll improve your s/n ratio quite a bit.
The resize process will "average out" most of the (VHS?) noise.

2) fine horizontal lines (such as in a wire fence, football goal, brick wall etc) tend to magically disappear and reappear if you simply discard one field. I find this rather annoying. Again, the resizing process will take these lines into account (50% of the info remains!)

3) the same applies for thin vertical lines, because the Marvel simply discards the unwanted pixels.

I guess that internally it always captures at full resolution, but does a kind of "nearest neighbour" resize if you want it to capture a smaller resolution...