24p questions

[Mike Minor]

Thanks for the response AJ your response made a lot of sense. Unfortunatly sounds like theres not much to be done at this stage.

also, i recorded in 2:3 not 2:3:3:2.

[A. J. deLange]

Alvaro,

No, the ISO rating stays the same as long as you don't touch the gain control. Just as the ISO rating of a film stays constant as you vary shutter speed and aperture so does the sensitivity of the CCD.

There is nothing inherently wrong in speaking about the ISO sensitivity of a CCD sensor - it is done all the time in the still camera world. The main difference is that you don't have a control labeled "gain" in a still digital camera. It is labeled ISO but it is exactly the same thing - a gain control.

If you expose an 18% gray card with an XL2 such that the light meter in the viewfinder reads at the center of its range (so that all the pixels fall in the middle of a luminance histogram (i.e. around 128 counts) then the ISO value which makes an INCIDENT light meter read the same combination of shutter speed and aperture (or any combination of equivalent exposure) is the "speed" of the camera. If I do this with tungsten lighting, 30p, 16:9, gain 0, tungsten balance and a shutter speed of 1/30th then it takes an f/4 on a 105 mm Nikkor to center the lightmeter in the camera. A Sekonic in front of the card set to ISO 200 gives me 1/30th at an f/4 so my measurement shows the speed of the camera as ISO 200 (and running around outside with the Sekonic confirms that this seems to be about right). Perhaps the biggest danger is that different people will measure it different ways! BTW I'm not saying that my reading is better than whoever posted 400. It was done pretty quickly and my method may not be what was used previously. This is why the ISO standard was written.

There is no reason that I can think of that someone couldn't take accurate measurements of the speed of the CCD (using whatever modifications to the ISO standard apply to CCDs) and publish this number. We could then subsequently take all our light readings with an incident meter using the published ISO number. We'd all figure out pretty quickly what our "working" ISO was exactly as still photographers do. I'm not advocating this as a way to go, just pointing out that it should be doable.

[Kevin Chao]

it either looks like compression artifacts... or a dirty head...

[Mike Minor]

Thanks Kevin. If they're compression artifacts, how can I get rid of them?

[Charles Papert]

The issue is that the curve of video is different than film, so that readings taken in the shadows will not expose the same as one would "expect"...i.e. a virtually non-existent reading may be quite visible on screen. Thus a light meter, or classic definition of ISO is not as transferable to digital as it might seem. Thus, what appears to make sense as a 400 ASA rating for the camera may act more like a 640 or 800 ASA rating when examining values in the toe.

[A. J. deLange]

That's quite true and the shoulder is different too - no grace with respect to over exposure in digital. And it's also true of still digital cameras which use ISO ratings as a matter of course. An exposure meter doesn't know about gamma or latitude. All it knows is how to get Zone V correctly exposed given the amount of light available regardless of whether the "film" is reversal or negative or CCD. The skilled wet photographer learns how to tweak the ISO setting for his combination of film, camera developer and paper. The skilled still digital photographer does the same and there is no reason why a videographer can't do it too. Whether we are aware of it or not the camera's meter has an "ISO" setting - so many watt seconds per square cm gives 50 IRE and the camera will, in automatic mode, adjust iris and/or shutter to get that many and if the subject is an 18% gray card or the scen is average, all is well. If it isn't, that's what the AE shift control is for and why those of us who use manual don't always center the "needle".

[A. J. deLange]

While surfing about looking for details on the ISO standard I found http://www.spectracine.com/determini...o.htm#plancha1 which is a short article describing the means for determination of the ISO rating of a professional video camera. They suggest the same method I described except that they want the "crossover chip" which is, I guess, one on the test target used by pro videographers set to 55 IRE. I don't know whether this chip is at 19% but I'll bet it's close. They then go on at great length explaining why using an incident meter is a far better way to set exposure than using a monitor. Now I've got nothing to gain by pushing this technique except that maybe I'll discover that these guys are right and get better video (there certainly are some advantages to incident light metering) but these guys sell light meters to Hollywood so you may conclude that they aren't as dispassionate as I.

[Steve Brady]

No, AJ's right, it's a pulldown issue.

Like AJ said, every four 24fps frames get distributed among 5 30i video frames like so: AA BB BC CD DD... The first, second and fifth video frames are "Whole" frames, because both fields contain the same 24fps frame, and the third and fourth video frames are "Split" frames, because each field contains a different 24fps frame. So the video sequence is WWSSW. However, you can start capturing the video at any frame. If you start capturing at the third video frame in the repeating sequence, for example, then your app needs to know that the pulldown on the clip is SSWWW. I don't use FCP, so I can't tell you how to set that up, I'm afraid.

Also, a word about 2:3 vs 2:3:3:2. It's fairly widely put about that this is a marketing gimmick, but the real point of it is that if your final material is going to be delivered at 24fps (not just film, but progressive DVD like you seem to be doing here, or web video, or whatever) then you'll get better picture quality, because rather than having to reconstruct two of the 24p frames from separate video frames, each 24p frame is stored in it's own video frame (you just throw out the split frame in the middle of each sequence of five). The only reason to shoot 2:3 rather than 2:3:3:2 is if there's a possibility that the footage will need to be viewed as 60i video without undergoing the conversion to 2:3.

[A. J. deLange]

Mike,

There is something that can be done but you may not like the results. If you "deinterleave" i.e. throw away all the high or all the low fields the motion blur will go away. To see what this looks like view the footage in QT with out setting quality to High. What you may not like is that half your vertical resolution goes out the window with the motion lines. If this is acceptable then there is probably some way to do the deinterleaving in FCP but I don't know what it is.

[Ed Bicker]

Hello AJ,

FIrst, I have two questions. I have an XL2, 20X
When you refer to QT, what do you mean??

Second, I shot a nature scene last week and I can not remember, what frame rate I was in. It was either 24p or 60i, but the problem that I saw was that I had horizontal rectangles forming as I moved the camera across the landscape. It was not there earlier in the video, but I was going through the process of shooting at various frame rates and all of a sudden, as I was playing this back these rectangles began to form and , of course, they destroyed the quality of the image. ANy idea why they would have formed when I panned to the right or left??? I do not remember them being there in the view finder.

[Matthew Nayman]

Sound like you have a dirty head in there... send a cleaning tape through it..

Matt

[A. J. deLange]

QT refers to QuickTime - Apple's way of handling video. All Apple computers come with Quick Time Player and it is available free for Windows platforms as well. Many other programs are based on QT such as Final Cut Pro.

Rectangles appearing erratically throughout DV video are an indication that bits have been dropped. The most likely cause is indeed a dirty head.

[Mike Minor]

How do you alter the quicktime quality setting? I poked around in preferences and couldnt find anything.

Also, anyone how to Deinterleave footage on a mac thats ultimately going to dvd?

[A. J. deLange]

In QT: Movie > Get Movie Properties will open a window with two pull down menus. Use the left one to select the video track. Now use the right one to select Quality. Check the "High Quality Enabled" box and you are set.

In the FCP browser window: Effects > Video Filters > Video > Deinterlace brings up a filter which allows you to select the field you want to retain.

[Mike Minor]

you're the man AJ! Thanks!