MovieStuff PAL transfer question

[blackangus1]

<BLOCKQUOTE><font size=2 face="Verdana, Helvetica, sans-serif">quote:</font><table border="0" width="90%" bgcolor="#333333" cellspacing="1" cellpadding="0"><tr><td width="100%"><table border="0" width="100%" cellspacing="0" cellpadding="2" bgcolor="#FF9900"><tr><td width="100%" bgcolor="#DDDDDD"><font size=2 face="Verdana, Helvetica, sans-serif">Originally posted by MovieStuff:

Okay, so let me get this straight: If I capture from my Hitachi camera via component, frame by frame (which is how the DV8 works) and then export that footage from the time line as a numbered bitmap sequence, the PAL end user can import that numbered sequence into his/her PAL NLE system and render it out as a 25fps PAL video clip. </font></td></tr></table></td></tr></table></BLOCKQUOTE>
Sure.

<BLOCKQUOTE><font size=2 face="Verdana, Helvetica, sans-serif">quote:</font><table border="0" width="90%" bgcolor="#333333" cellspacing="1" cellpadding="0"><tr><td width="100%"><table border="0" width="100%" cellspacing="0" cellpadding="2" bgcolor="#FF9900"><tr><td width="100%" bgcolor="#DDDDDD"><font size=2 face="Verdana, Helvetica, sans-serif">Furthermore, if I snug up the super 8 image to the top and bottom of the NTSC frame during transfer, that frame will end up in the middle of the PAL frame with some black top and bottom, assuming that they enable the "maintain aspect ratio" function of their NLE system and the loss in resolution would be minimal.
</font></td></tr></table></td></tr></table></BLOCKQUOTE>
Nope. They will have to stretch the image to maintain the proper aspect ratio.

If you want to try an NTSC-&gt;PAL DV conversion, you can test it, for free. Canopus' converter is freely downloadable as a demo, but it's limited in that it only works with Canopus-created DV files. BUT, you can get around that: you can go to:
http://www.baobab.net/dvconv.htm

This company has written a program that makes DV files freely interchangeable from one codec to another, including to/from Canopus. Both are free downloads. So you can download both programs, and give it a test. Import the projects back into Premiere and examine the quality on the computer monitor. You can compensate for the aspect ratio by resizing your computer monitor and see if the quality is adequate.

<BLOCKQUOTE><font size=2 face="Verdana, Helvetica, sans-serif">quote:</font><table border="0" width="90%" bgcolor="#333333" cellspacing="1" cellpadding="0"><tr><td width="100%"><table border="0" width="100%" cellspacing="0" cellpadding="2" bgcolor="#FF9900"><tr><td width="100%" bgcolor="#DDDDDD"><font size=2 face="Verdana, Helvetica, sans-serif">And finally, since I am avoiding the entire miniDV compression issue, the color of the numbered bitmap sequence should pretty much look like it does over here from the original transfer.</font></td></tr></table></td></tr></table></BLOCKQUOTE>

Sampling at 4:2:2, you'll avoid the DV color decimation issue. As for how accurate the color translation is, that's still to be seen. After all, NTSC is notorious for having finicky (well, lousy) color control and color reproduction, whereas PAL is far superior.

<BLOCKQUOTE><font size=2 face="Verdana, Helvetica, sans-serif">quote:</font><table border="0" width="90%" bgcolor="#333333" cellspacing="1" cellpadding="0"><tr><td width="100%"><table border="0" width="100%" cellspacing="0" cellpadding="2" bgcolor="#FF9900"><tr><td width="100%" bgcolor="#DDDDDD"><font size=2 face="Verdana, Helvetica, sans-serif">Is there any way to save this information on a miniDV tape? Not as a video signal but just the raw digital information? </font></td></tr></table></td></tr></table></BLOCKQUOTE>
There are people who have attempted to use DV decks as data recorders. I'm sure they sort of work, AS LONG AS THERE ARE NO DROPOUTS. Dropouts are an annoyance in video data, but they would be catastrophic in a data stream. But, hey -- you could always try it.

<BLOCKQUOTE><font size=2 face="Verdana, Helvetica, sans-serif">quote:</font><table border="0" width="90%" bgcolor="#333333" cellspacing="1" cellpadding="0"><tr><td width="100%"><table border="0" width="100%" cellspacing="0" cellpadding="2" bgcolor="#FF9900"><tr><td width="100%" bgcolor="#DDDDDD"><font size=2 face="Verdana, Helvetica, sans-serif">
I suppose if they wanted to send me a precious 1 gig micro drive, I could save it to that! An expensive proposition but should work, right?
</font></td></tr></table></td></tr></table></BLOCKQUOTE>
You can get 50-packs of 700MB CD's for free if you shop during rebate season, so a 1-gig micro drive doesn't seem to be a better solution. As crimsonson said, you can burn a full CD in 4 minutes.

You could consider turning the still sequence into DVD-quality MPEG-2. You could fit about 20 minutes of DVD material on a standard CD, and DVD MPEG-2 is 720x480 4:2:2. Then the client would need to find some way to convert that to PAL... hmmm... doesn't solve the format issue, but it at least makes it a cheap way to deliver the data.

[blackangus1]

<BLOCKQUOTE><font size=2 face="Verdana, Helvetica, sans-serif">quote:</font><table border="0" width="90%" bgcolor="#333333" cellspacing="1" cellpadding="0"><tr><td width="100%"><table border="0" width="100%" cellspacing="0" cellpadding="2" bgcolor="#FF9900"><tr><td width="100%" bgcolor="#DDDDDD"><font size=2 face="Verdana, Helvetica, sans-serif">Originally posted by MovieStuff:
But if an NTSC frame is saved as a JPEG or bitmap file, doesn't the original pixel shape become academic? The PAL NLE system won't know whether the pixel came from a digital camera or a video camera or a photo scanned on a flatbed scanner, will it?

Roger</font></td></tr></table></td></tr></table></BLOCKQUOTE>

??? Of course it matters -- just as the aspect ratio of the display device must be taken into account. TV pixels are not square!

Whatever the original acquisition device is will establish the shape of the pixel. For example, an NTSC camera uses oblong-shaped pixels, slightly undersquare. A flatbed scanner typically uses square pixels. If you scan in a picture on a flatbed (using square pixels) and then display it on a TV monitor, you'll see that the image is stretched tall on the TV. Try scanning in something circular, you'll see that on the TV it's noticeably oblong. To get a round image on a TV from a scanned image, you have to resize. Typically for NTSC you would scan in at 720x540 and then resize in Photoshop to 720x480.

NTSC pixels are undersquare, taller than they are wide. PAL pixels are slightly oversquare, wider than they are tall.

[Scottness]

Hey Roger why don't you just buy PAL gear?

Scot M!

[crimsonson]

<font color="#a62a2a" size="1">[ July 07, 2002 06:39 PM: Message edited by: crimsonson ]</font>

[MovieStuff]

<BLOCKQUOTE><font size=2 face="Verdana, Helvetica, sans-serif">quote:</font><table border="0" width="90%" bgcolor="#333333" cellspacing="1" cellpadding="0"><tr><td width="100%"><table border="0" width="100%" cellspacing="0" cellpadding="2" bgcolor="#FF9900"><tr><td width="100%" bgcolor="#DDDDDD"><font size=2 face="Verdana, Helvetica, sans-serif">Originally posted by blackangus1:

??? Of course it matters -- just as the aspect ratio of the display device must be taken into account. TV pixels are not square!
</font></td></tr></table></td></tr></table></BLOCKQUOTE>

Okay, but then explain to my why I can export a still frame (jpeg) and then open it in Photoshop and it will be 720x486, which when I print out on to paper will measure exactly the same size as a canvased image I create from scratch that is ALSO 720x486. One comes from a video image, the other comes from Photoshop as an original image. The Photoshop pixels aren't "TV" pixels yet both images will measure exactly 10 inches by 6.75 inches when printed on paper.

Therefore, it seems to me if I send that Photoshop file to someone in Australia, it will measure exactly the same there, as well. They should not be able to tell which file came from a video frame and which was created as a Photoshop original, since they both measure 720x486 at 72dpi. Again, if they put that image on their non linear system, how (or why) will it see things differently? More to the point, if THEY create something that is 720x486 and send it to me, won't it measure the same as my original image if I put it on the timeline here?

I would understand if I was sending them a video file but I'm not. I'm essentially sending them a Photoshop file. I thought once something was made into a JPEG, it no longer exhibited the original pixel characteristics. I don't get it.

Roger

[MovieStuff]

<BLOCKQUOTE><font size=2 face="Verdana, Helvetica, sans-serif">quote:</font><table border="0" width="90%" bgcolor="#333333" cellspacing="1" cellpadding="0"><tr><td width="100%"><table border="0" width="100%" cellspacing="0" cellpadding="2" bgcolor="#FF9900"><tr><td width="100%" bgcolor="#DDDDDD"><font size=2 face="Verdana, Helvetica, sans-serif">Originally posted by Scottness:
Hey Roger why don't you just buy PAL gear?

Scot M!</font></td></tr></table></td></tr></table></BLOCKQUOTE>

Not that simple, Scott. The DV8 isn't a film chain where you simply point one type of camera or another at it. The entire system is NTSC, from the camera that is built into the DV8 to the NLE system I use to capture each frame for playback. Unfortunately, my system will not allow for playback at the PAL conversion rate. It is NTSC only. So I'm trying to see what is possible with the tools I have.

Roger

[blackangus1]

<BLOCKQUOTE><font size=2 face="Verdana, Helvetica, sans-serif">quote:</font><table border="0" width="90%" bgcolor="#333333" cellspacing="1" cellpadding="0"><tr><td width="100%"><table border="0" width="100%" cellspacing="0" cellpadding="2" bgcolor="#FF9900"><tr><td width="100%" bgcolor="#DDDDDD"><font size=2 face="Verdana, Helvetica, sans-serif">Originally posted by MovieStuff:


Okay, but then explain to my why I can export a still frame (jpeg) and then open it in Photoshop and it will be 720x486, which when I print out on to paper will measure exactly the same size as a canvased image I create from scratch that is ALSO 720x486. One comes from a video image, the other comes from Photoshop as an original image. The Photoshop pixels aren't "TV" pixels yet both images will measure exactly 10 inches by 6.75 inches when printed on paper.
/[QB]</font></td></tr></table></td></tr></table></BLOCKQUOTE>
Of course what you're saying is accurate: 720x486 will be 720x486 in PhotoShop, regardless of the originating format. The difference is, is the video-originated item displayed and printed accurately, aspect-ratio-wise? And the answer must be, "no", because the printer uses square pixels and the TV uses tall/skinny pixels. But the PhotoShop one will be displayed accurately because the computer monitor uses square pixels.

Think about it for a second: we all know that NTSC TV is 720x480, yet we also know that NTSC TV is 1.33:1. 720x480 would be 1.5:1 if the pixels were square. But they're not square! That's the whole point. The pixels are not square, so you can't treat them as if they were. If you do, you'll end up with a distorted aspect ratio.

Computer monitors use square pixels, for the most part. Think of the screen sizes you're used to seeing: 640x480, 800x600, 1024x768: each of those are 1.33:1, the same as the physical dimensions of the monitor. But NTSC uses 720x480, not 640x480, yet that image all fits in the same physical 1.33:1 space. The reason? The pixels are skinnier and crammed tighter together, because they are not square.

Scan a photo of a circular object in at 720x480, then put it on the timeline and look at it on the TV, and what do you get? An oval. Export a still from video of a circular object, and print it out on your printer. What do you get? An oval. One will be taller than wide, the other wider than tall.

This is important to remember when looking at the PAL&lt;-&gt;NTSC conversion, because neither TV standard uses square pixels (although PAL pixels are closer to square). So if you put an NTSC 720x480 image in the middle of a PAL 720x576 frame, you will end up with a substantially vertically squished display.

<BLOCKQUOTE><font size=2 face="Verdana, Helvetica, sans-serif">quote:</font><table border="0" width="90%" bgcolor="#333333" cellspacing="1" cellpadding="0"><tr><td width="100%"><table border="0" width="100%" cellspacing="0" cellpadding="2" bgcolor="#FF9900"><tr><td width="100%" bgcolor="#DDDDDD"><font size=2 face="Verdana, Helvetica, sans-serif">[QB]They should not be able to tell which file came from a video frame and which was created as a Photoshop original, since they both measure 720x486 at 72dpi.</font></td></tr></table></td></tr></table></BLOCKQUOTE>
Aha, there you have it. They will (or should) easily be able to tell which one originated from video and which originated as a square-ratio photoshop original, because the video frame will not be "true" -- it'll be stretched horizontally.

Try this simple test: make a 640x480 image in Photoshop. Then duplicate it, but stretch it to 720x480. Output the 720x480 image onto your TV, and you'll see that it looks like the 640x480 image on the computer screen. The 720x480 image will look squashed-down on the computer screen but it will look correct on the TV.

<BLOCKQUOTE><font size=2 face="Verdana, Helvetica, sans-serif">quote:</font><table border="0" width="90%" bgcolor="#333333" cellspacing="1" cellpadding="0"><tr><td width="100%"><table border="0" width="100%" cellspacing="0" cellpadding="2" bgcolor="#FF9900"><tr><td width="100%" bgcolor="#DDDDDD"><font size=2 face="Verdana, Helvetica, sans-serif">Again, if they put that image on their non linear system, how (or why) will it see things differently? More to the point, if THEY create something that is 720x486 and send it to me, won't it measure the same as my original image if I put it on the timeline here?
</font></td></tr></table></td></tr></table></BLOCKQUOTE>
No, because the aspect ratio of PAL pixels is different than the aspect ratio of NTSC pixels. What you send to them, if they tried to display it on their TV, will be squashed down vertically. Circles will not be circles, they'll be flat ovals. And if you tried to display something that they sent to you, the opposite would be true.

Keep in mind this is ONLY FOR VIDEO IMAGES. If they scan something on a flatbed and send it to you and you display it on your computer monitor, it will look exactly like you would expect, because their scanner and your monitor both use square pixels.

<BLOCKQUOTE><font size=2 face="Verdana, Helvetica, sans-serif">quote:</font><table border="0" width="90%" bgcolor="#333333" cellspacing="1" cellpadding="0"><tr><td width="100%"><table border="0" width="100%" cellspacing="0" cellpadding="2" bgcolor="#FF9900"><tr><td width="100%" bgcolor="#DDDDDD"><font size=2 face="Verdana, Helvetica, sans-serif">I would understand if I was sending them a video file but I'm not. I'm essentially sending them a Photoshop file.
</font></td></tr></table></td></tr></table></BLOCKQUOTE>
Yes, but one that was created from a video source, and that video source uses nonsquare pixels. Your photoshop file does not display square pixel aspect ratio, it's a bucket of pixels that were designed, intended, and executed as tall skinny pixels. Playback on a square-pixel monitor will reveal squishing, and playback on PAL (which is even oversquare) will reveal more significant squishing.

Again, check the pictures at the Canopus site, they clearly demonstrate what happens.
<BLOCKQUOTE><font size=2 face="Verdana, Helvetica, sans-serif">quote:</font><table border="0" width="90%" bgcolor="#333333" cellspacing="1" cellpadding="0"><tr><td width="100%"><table border="0" width="100%" cellspacing="0" cellpadding="2" bgcolor="#FF9900"><tr><td width="100%" bgcolor="#DDDDDD"><font size=2 face="Verdana, Helvetica, sans-serif"> I thought once something was made into a JPEG, it no longer exhibited the original pixel characteristics. I don't get it. </font></td></tr></table></td></tr></table></BLOCKQUOTE>
JPEG just compresses the existing pixels into a different compression format, but it doesn't change the aspect ratio of each individual pixel. The issue you're dealing with is that the video imaging system creates the original pixels in a tall undersquare shape. And the video imaging system you want to go to manipulates pixels in a squat, oversquare shape.

[MovieStuff]

<BLOCKQUOTE><font size=2 face="Verdana, Helvetica, sans-serif">quote:</font><table border="0" width="90%" bgcolor="#333333" cellspacing="1" cellpadding="0"><tr><td width="100%"><table border="0" width="100%" cellspacing="0" cellpadding="2" bgcolor="#FF9900"><tr><td width="100%" bgcolor="#DDDDDD"><font size=2 face="Verdana, Helvetica, sans-serif">Originally posted by blackangus1:
Of course what you're saying is accurate: 720x486 will be 720x486 in PhotoShop, regardless of the originating format. The difference is, is the video-originated item displayed and printed accurately, aspect-ratio-wise? And the answer must be, "no", because the printer uses square pixels and the TV uses tall/skinny pixels.</font></td></tr></table></td></tr></table></BLOCKQUOTE>

Well I guess that's my confusion, here, because when I do print them out, the video images don't look distorted at all to me, which is why I brought it up. Hmmmm. I understand what you are saying but I have yet to see the effect you are talking about and I actually do print out a LOT of video frames to paper for roto work and animation. I will have to experiment with the whole circle thing and see how pronounced it is. Perhaps it is doing it and I just have never noticed.

Roger

[Scottness]

&gt;&gt;Not that simple, Scott. The DV8 isn't a film chain where you simply point one type of camera or another at it. The entire system is NTSC, from the camera that is built into the DV8 to the NLE system I use to capture each frame for playback. Unfortunately, my system will not allow for playback at the PAL conversion rate. It is NTSC only. So I'm trying to see what is possible with the tools I have.

No I thought so Roger - I was just asking for fun

[img]wink.gif[/img] I'm sure there's a way too - if it's all just a mathematical structure -- assuming of course you can get your mind around the posting in simulcast (and what people are describing here [img]confused.gif[/img] )

Scot M

[mattias]

<BLOCKQUOTE><font size=2 face="Verdana, Helvetica, sans-serif">quote:</font><table border="0" width="90%" bgcolor="#333333" cellspacing="1" cellpadding="0"><tr><td width="100%"><table border="0" width="100%" cellspacing="0" cellpadding="2" bgcolor="#FF9900"><tr><td width="100%" bgcolor="#DDDDDD"><font size=2 face="Verdana, Helvetica, sans-serif">
DV is field based.
</font></td></tr></table></td></tr></table></BLOCKQUOTE>

no. unlike mjpeg for example, dv is actually frame based. not that this matters to anyone anyway, since that's far below the abstraction level we're working at. the video out from a dv deck is obviously field based...

(that's the only one that was easy to answer. man, are you making this complicated?) ;-)

/matt