Frame Mode Question for Charles Papert
 

I keep reading posts extoling the virtues of originating your project in XL1s frame mode versus deinterlacing in post. Pros and cons abound concerning resolution loss in frame mode and resolution loss by deinterlacing in post (keeping in mind that frame mode isn't really anything close to a true progresive scan). I've even read your article about the taping of the Seinfeld American Express commercial and the apparent "mixup" halfway through the production concerning the frame mode controversy. You are a working DoP and have seen your own results. What is your definitive answer on using frame mode and it's effect on resolution loss versus not using frame mode and deinterlacing in post? I hope this question isn't too nebulous for a definitive answer.

I would like to insert one thing: de-interlacing in post will also
always result in resolution loss (one method more so than
another). I'll leave the rest up to mister Papert.

I continue to be chagrined at this whole arguement for/against frame mode. On the one hand, there have been viewpoints, ad infinitum, extolling the virtues of "the film look", which includes grain and a somewhat degraded image quality relative to video. Video, they say, is too lifelike, too real. Now, video cameras have the "frame mode" capability and people decry the loss of rez. As usual, I'm completely awed by the chronic inability of image-makers to understand what it is they really want.

I think individual filmmakers know what they prefer themselves. It's just that everyone hasn't come to a consensus. When I had my GL2, all I did was use the Frame mode. I wasn't planning on blowing up to film, and it all worked out great... for me. I was full aware that 30fps wasn't about to translate to film's 24fps.

Using Frame Mode on a PAL version would be the only option in Frame Mode if you do plan on going to film. That's what they did with 28 Days Later... Sorta like 25p.

So basically it's just about what your purpose is, and what you prefer within the realm of your purpose. Or something like that...

Agreed Imran (although I think you stole that last line from "Kung Fu" or something)

I can't keep up with the latest developments in de-interlacing technology, nor have I shot anything on the XL1 that was intended to be output to film (including those Amex spots). The post production guys on that project do keep up on this stuff, and they still recommeded frame mode.

For me, the slight resolution loss of frame mode acts similarly to a light diffusion filter; I find it a pleasing image. I have yet to see an audience start counting lines of resolution when they watch a film. The perceived image is all that counts, and it works for me. If I was to embark on a large scale project for myself, I probably would do some tests with Magic Bullet etc. and make a comparison of the results, including variable speed pans and the like. Haven't needed to yet.

This then is my definitive answer for myself...can't speak for anyone else out there! (I think the definitive answer for many other people at this point is just shoot it with a DVX-100 on 24p, to be honest)

In spite of my ranting on this subject, I agree with you guys.

*Wax on*
*Wax off*

Also, Bill, how do you pronounce your production company's name?!

CHAL-CHEE-WEE'-TUHL

it means "Blue Stone" (aka turquoise) in Toltec....
a much revered spot near my home.

Here's my Cliff's Notes method for determining whether you should be shooting in Frame Mode or not. (Applies to NTSC XL1(S) and GL1/GL2. Numbers differ for PAL.)

A. What will be the frame rate of my final output?
- If 30 frames per second, go on to question B.
- If 24 frames per second (as with a transfer to film), or other non-multiple/non-factor of 30, go to question C.
- If unsure, go to question C.

B. Do you want your final output to be interlaced, or progressive scan?
- Interlaced. Will provide fastest frame refresh on a cathode ray tube NTSC television. Shoot Normal Mode.
- Progressive. Fast pans and fast-moving objects like cars may appear choppy. Shoot Frame Movie Mode. (Always choose this option if shooting solely for the web.)
- Don't know yet. Shoot Normal Mode. (See below.)

C. Will you be dropping/doubling or temporally interpolating frames in post?
- Just dropping/doubling frames. Shoot Frame Movie Mode. (Choose this when using Vegas without any special plugins.)
- Temporally interpolating frames. Shoot Normal Mode. (Choose this when using Magic Bullet, Re:Timer, or one of the expensive professional services. Always select this when shooting for eventual distribution on film.)
- I'm not certain. Shoot Normal Mode. (While the spatial resolution of Frame Movie Mode cannot be recovered in post, neither can the temporal resolution of Normal Mode, and retaining the latter is the safer bet.)

This line of reasoning is based on the fact that Frame Movie Mode will always yield higher spatial resolution than interlaced frames interpolated. (The caveat being that temporal interpolators appreciate having finer-spaced data points to operate on.)

cool! thanx for sharing

Thank you all for your input. I really need to s**t or get off the pot. Or, more appropriately, shoot some serious tests and judge the outcome. I hear so much about how interlaced or deinterlaced does not matter so much as careful lighting, frame composition, acting and story. GOD how I hope my projects never look like soap operas.

Testing will basically determine how it will look. Because without
it you won't know how it will look or what influences the look and
in what way. Testing is good!

Test it is!
 

In the spirit of the great inventors of our time... I shall test extensively. Thanks for taking the time to answer.

I shot a simple scene with myself, nothing moving but me and Magic bullet looked horrible compared to frame mode.

Maybe I screwed something up but I don't see how.

In After Effects, I simply chose magic bullet and it does a little auto thing, and it chose 30 fps and I checked de-interlace. Maybe that was my mistake, I don't know. It takes a long time, I know that much.

It obviously takes out frames of the NTSC moving image and made it very jerky and there was a loss of image.

The same footage was used in Vegas, doing their 24fps and it looked far better.

I also lost audio in After effects for some reason making it useless.

In After Effects(Magic B.) I outputed to AVI using the DV codec provided, setting it to high quality output as well....

Either Magic Bullet sucks, or I did something really wrong. The test takes so long that it's hard to test over and over again quickly.

My scene was only 46 seconds long too.

Is there a good Magic Bullet tutorial explained on here so I might try to most optimum settings next time I try a test?

Anyhow.....loss in resolution or not, the GL2's Movie Mode looks great right out of the camera in my opinion and saves alot of time as well in post. It also looks more movie like RIGHT as I am monitoring what I am shooting, which helps.

Your Magic Bullet settings are incorrect. Be sure to look at the manual and be sure to setup the interpret footage and composition frame rates exactly. Then be sure to set the render options out correctly, including pulldown. Otherwise it will look bad. I also thought you could just dive in and go, but you have to RTM with this one.