Lower field first

[Bill Ravens]

There seems to be a lot of confusion about this setting, Rob. Vegas has two options, either progressive or bottom/top field first. Those are the ONLY choices.

Now, if you have interlaced footage, or you want interlaced footage out, you select bottom/top field first and voila, you get interlaced out.

If you want de-interlaced, OR, if you're starting with progressive footage and you want progressive (aka de-interlaced) out, you select "progressive".

Now, in the special case of frame mode, as in the XL1s, where you get "simulated" progressive, but really, it's still interlaced footage, it depends what you want out of Vegas. If you want de-interlaced(same same if I call it progressive) then you select "progressive". If you're starting with progressive in the input stream, fine, NP. BUT, if you're starting with interlaced footage(and the XL1 is INTERLACED footage, even in the frame mode) then you will, by default, de-interlace it when you select progressive. If it's progressive in, it doesn't matter what de-interlacing mode you pick, interpolation or blend. If it's interlaced footage in and you select progressive, then it matters what de-interlacing method you pick. Since the XL1s frame mode captures both fields at the same instant, pixel shift not withstanding, then selecting interpolate will give a higher rez output, IMHO, than blend.



Does this help?

[Andre De Clercq]

If ïnterpolate is being set, the system interprets one field and interpolates the lines for the next field (in order to get a "progressive" frame) ending up in vertical resolution loss as compared to the original progressive vertical resolution. Back to the field order point... Lower/upper field, top/bottom field, even/uneven field is a terminology which very "popular", but confusing (lower field in NTSC is upper field in PAL...) The only correct way to define the field order is "field order A" or "B" and "field 1 first"or "field 2 first". Because of all this confusion, even by "professionals, the best way is to test both and see what is right for the actual system.

[Michael Best]

Where's 'Interpolate Fields'?

[Bill Ravens]

Michael...

The choices INTERPOLATE, BLEND, or NONE appear in the options dropdown for the "Interpolate Method".

Andre...

"Interpolate", in the mathematical definition, is to calculate an intermediate point based on the definitions of the endpoints, usually done in a linear fashion. True, when the fileds are separated by a temporal difference, "interpolate" will reduce the rez. However, in the Canon frame mode, since no temporal difference exists between the captured fields(the endpoints), one can argue the following:

if one lets A=beginning field(endpoint)
B=ending field(endpoint)
C= an intermediate point(equal to half the difference for this example)
then, the following can be shown:
1-A=B
2-A+B = C
3-letting A=B, then substituting A+A=C
4-or 2A=C
5-In the definition of "mathematical interpolation, then the mid-point is 1/2 A
6-or A=B=C

So, you see, an interpolated result of two equal endpoints is itself equal to an endpoint. There is no rez loss when interpolating between two equal fields.

[Andre De Clercq]

Bill, indeed there is no temporal difference, but there is spatial difference even in the pseudo framemodes. Spatiial difference is what counts and that's what you also get when there is no motion in an interlaced picture. Temporal difference is only a subset of temporal difference.

[Bill Ravens]

Andre...

If both fields are captured at the same instant in time (ie there is no temporal difference) how can there be a spatial difference? The only possibility I can conceive is that the 3 sensor arrays are separated by a physical difference, in which case you'd get a "3-D" effect (parallax)

[Andre De Clercq]

Bill, each of the 480 active NTSC lines can have a different content in progressive and pseudo progressive modes. One can theoretically get a scene containing 240 black horizontal lines spatially interleaved with 240 white lines together forming a frame. This hypothetical image would, if recorded in interlace mode, end up in a black field followed by a white one. Interpollation would then result in a fully black or a fully white sequence, depending whether field one or field two was used for interpollation. Also oblique image structures generate 480 different line contents, in vertical direction and horizontal direction as well

[Bill Ravens]

Ah! Yes, I see, Andre. The scan lines are displaced, vertically, yes, of course, duh. Thanx for your patience. So, blur then becomes the favored de-interlacing method?

[Andre De Clercq]

I'm happy we got agreed Bill. For the hypothical image content, indeed blur would be best, for other contents interpollate would be better... " intelligent " deinterlacers adapt their algorithme permanently. B.t.w. Bill you live in a splendid area...I have been there in May this year. My wife and I travel quite a lot...We stayed in the Santa Fe El Rey Inn. I remember the beautifull
Plaza, the Loretto chapel and St Francis Cathedral, Canyon road, Bandelier, the unique pueblo's and much more.

[Bill Ravens]

Yes, it is quite beautiful here. We've been under a drought for about 5 years, it's affected many of the trees as they're quite dead. It's sad the scale of it. The rains have returned, this year, however. Things are quite green and the wildflowers are blooming.