Progressive Scan Question

Greetings all. I am using a DVX100A PAL and shooting for web streaming and downloading so my ultimate output format is Real Media and Windows Media Player. I want to encode Progressive footage as I have read it gives the best results. I am shooting in Progressive mode.
OK here are my questions: Do I need to use my DVX100A to capture from (into Premier Pro 1.5) or can I just use my Sony deck and just use Matrox Tools as normal or can I use the DVX100A and Matrox Tools to capture Progressive footage (this would mean, I assume, that I could then open a new Premier Pro 1.5 project using my Matrox RTX100 and not have to render the captured footage)?
Once I have edited the footage I want to archive it onto MiniDV for future use as Progressive footage. Do I have to use the DVX100A to record it onto to keep it as Progressive or can I use my Sony deck?
Any help with this would be very much appreciated.
Karl

It depends on where in the chain is/are the scalers. If the monitor has a good scaler, or you have an outboard scaler, then the picture will be made progressive by this device and progressive capture is unecessary. If not, or you are not sure, you can arrive at progressive scan video by progressive capture (with the dvx100 anyway) or deinterlacing on the computer via the NLE software.

When you’re working in PAL, it doesn’t matter; you only need bring it into a PAL timeline. (Your NLE should recognize progressive material. But count yourself lucky that you don’t need to deal with pulldown issues!) Then, just use the appropriate PAL settings to export to WM or RM. You’ll be set and you won’t need to worry about scaling.

As for which to use, the camera or the deck, that, too doesn’t matter. As long it reads and records PAL DV, you’re set. DV is inherently interlaced, so you can’t record “progressive” to tape in a true sense no matter what you use – but your deck and NLE will take care of that for you. No worries.

The real difference between progressive and interlaced comes at the time of acquisition. In progressive, you’re shooting an entire frame all at once; in interlaced, you’re shooting only half of it at any given time, by odd lines or even lines. The practical difference is that in progressive, each frame is a whole picture; in interlaced, a “frame” is made up of two pictures, one comprised of the even lines and the other comprised of the odd lines. If something is in motion, you will see in each “frame” that it’s split into two, scan line by scan line, so the edge of a moving car, for example, will look something like a ladder.

In progressive, you won’t have those interlacing artifacts. That’s why it’s better, especially when adding effects or doing chroma keying or other things which really need pixel-accurate workspace.

And, if you “deinterlace” using field interpolation, the fields interpolated will be the two halves of the same actual progressive frame, leaving you with a perfect progressive frame, and not an approximation as it would be with deinterlacing something captured at 50i.

If you acquire your footage in 25p, you’re fine. You don’t have to worry about scalers.

That is where the division/multiplication factor of 2 comes into play...25p or 50i, same thing.

All Footage is recorded onto the tape as an interlaced signal, regardless if the CCD's captured progressive.

The image once captured gets broken into fields and written to tape.

It doesn't matter what you capture with, as the signal will not change, but it depends on how the footage is displayed.

I am not a Premiere user, but it should work similarily to FCP, AVID etc.. in that it will either display 1 field when playing back during editing. Or it will combine the fields, in the latter case, you'll be seeing the progressive signal, if the signal was originally captured in progressive mode.

make sure to check that your footage is being displayed at full res. I don't know if Premiere works the same, but in Final Cut there is an option to set Field Dominance to None, so that they are both equally displayed, the same option may exist for you.

anyways I am rambling, to answer the question, capturing with camera or deck will produce the same results. I dunno what a scaler would have to do with capturing..

EDIT: De-interlacing your progressive footage effectively throws away half your resolution.

De-interlacing, and field interpolation are quite different and used interchangeably when they shouldn't be.

That is confusing....why would you deinterlace (make progressive) progressive scan video? Can it be done if you wanted to?

hmmm.. it is indeed confusing..

You wouldn't de-interlace progressive footage, but I was trying to state that a de-interlace on progressive footage would throw away half the res..

Quicktime, only displays DV footage using 1 field(effectively the same as de-interlaced video) and I have seen people having problems with their footage because only one field is being displayed...

ok nevermind.. I should go to bed.

Read this: ftp://ftp.panasonic.com/pub/Panasonic/Drivers/PBTS/papers/Progressive-WP.pdf

http://videosystems.primediabusiness.com/ar/video_progressive_need_know/index.htm

"But what caught my attention the most was that for all the passion, there was an amazing lack of knowledge about progressive video."

These are some of the best technical articles I've read on the subject of interlaced, progressive, and frame mode CCD capture. They address NTSC but apply equally to PAL (other than the vertical resolution figures) What many proponents of interlaced mode fail to acknowledge is that interlace scanning must sacrifice vertical resolution to execute an interline twitter reduction filter. Although the camera will store progressive or interlace footage in the same alternating (lower and upper) fields, the content of the fields is quite different. The technology is well over 50 years old and was implemented to compensate for the inability of the broadcast components to display a full picture scan in 1/30s.

Progressive NTSC fields have 240 lines of res, to give you the full NTSC 480 line complement on display. To contrast, the same temporaly displaced interlace fields will carry at best 180 lines for a frame equivalent of 360 lines. If you start with interlace, (50i) in your case, you will also lose more resolution when deinterlacing. Even the best deinterlacers/scalers must interpolate moving areas in the interlace fields, meaning lost detail. Remember they are starting with 120 less lines of resolution.

There is no de-interlace necessary for progressive footage on a PC. The display adapter weaves the interlace fields together without any blending, interpolation and thus no loss of information.

I have to admit, you lose me with that explanation. can you simplify?

Sure...although I suspect you don't really need the simplification.

Take two film cameras both filming a slow pan over a newspaper. Both are taking 60 pictures every second. To simplify, let's assume both can reproduce 480 lines of resolution vertically. The cameras are panned over the paper at the same speed, and 480 lines of text are showing.

We unload the first camera and toss out every second picture so now we have 30pics per second. We shred the 30 pics into 480 lines, number them and set them aside. This is our progressive footage.

The 2nd cameras shots are shredded in the same way. But now we alternately discard the odd stripes of the first pic, even numbered stripes of the second pic shot etc leaving 240 stripes per shot. We revisit what's left of each picture and randomly remove another 60 or so stripes. That's our interlaced footage.

So now we project our pictures from each camera onto a wall. We can't really see any flicker differences. Which footage would you rather use to read the paper? It's the loss of 60 stripes in the interlace footage that makes all the difference. How much of the paper would be readable on our "interlaced" projection?

Sure...although I suspect you don't really need the simplification.

Take two film cameras both filming a slow pan over a newspaper. Both are taking 60 pictures every second. To simplify, let's assume both can reproduce 480 lines of resolution vertically. The cameras are panned over the paper at the same speed, and 480 lines of text are showing.

We unload the first camera and toss out every second picture so now we have 30pics per second. We shred the 30 pics into 480 lines, number them and set them aside. This is our progressive footage.

The 2nd cameras shots are shredded in the same way. But now we alternately discard the odd stripes of the first pic, even numbered stripes of the second pic shot etc leaving 240 stripes per shot. We revisit what's left of each picture and randomly remove another 60 or so stripes. That's our interlaced footage.

So now we project our pictures from each camera onto a wall. We can't really see any flicker differences. Which footage would you rather use to read the paper? It's the loss of 60 stripes in the interlace footage that makes all the difference. How much of the paper would be readable on our "interlaced" projection?

I assume that in this example you're envisioning the equivalent of a 1/60 shutter speed in video acquisition?

yep, 60 frames per second, 1/60s shutter...that's a departure from the 1/48s used by most film cams.

To read a paper (or look at a still photo) yes, the later example is the preference. But this is video, at 60 fps versus 30 fps, my vision and processing will blend the half picture fields (2 of them) into one image per 1/30th of a second giving exactly the same perception.

And what will your brain do with the missing 60 lines of text?

You are correct....IF IT IS A STILL PICTURE ONLY. If it is video/motion, the missing lines are conveniently located in the next 1/60th of a second, too fast for our brains to know it was missing (for 1/60th of a second)

30p and 60i are distinctly different. Try shooting some interlaced and progressive footage with the DVX and then compare them. The motion signature is quite different proving that we can identify 60 samples of motion (if not more)

To say 30p is no different than 60i is to not have compared the two :)

CC, the 60 lines are effectively missing from the interlace line boundaries as a result of filtering to reduce interline flicker. They don't show up in the next 1/60s. Remember my example is assuming 480 lines of text. Try filming a 480 line pattern of alternating horizontal black and white lines with interlaced, then progressive, and let me know if you still think interlace footage looks the same.

These are extreme examples, however, they illustrate a point.