updated 20 Jun 2005

 

Jon Fordham's FX1 Review, Part Two
an article by Jon Fordham, Director of Photography, D.I.T.

Jon is a contributing member of our DV Info Net Community.

While the news of the script change sunk in, I took some time to set up the FX1 and the DVX100A in front of a Putora 7A9 chart. I did a series of three different comparison shots. DVX in 60i and 30p at three different focal lengths. And the FX1 at 60i and CineFrame30 at three different focal lengths. I'll take the time to post the settings and my thoughts on this test separately and Heath will be supplementing with still frames of the results.

After I finished up the chart test, I read the script and then went over the scene with Heath. The good news was that the script only required two setups. The bad news was that the talent had a time limit. We only had a few hours to shoot the 5 page script before the actors had to go!

The script was about three sisters having a conversation on the living room couch. While I dig Heath's living room, it was a bit drab in color tones and the walls are white. I suggested that we move the conversation into another room of the house in hopes of finding brighter colors and walls any color other than white! Heath and I finally settled on having the girls hang out in a bedroom with some nice blue walls. I breathed a sigh of relief upon the success of getting away from the white walls! The scene was INT Day. And thankfully the room had a nice large window with plenty of light coming in. I decided to use the light from the window as my Key, but at a full side light position to the talent. I then supplemented with a 1K fresnel gelled with Full CTB bounced in to fill the shadows. After switching the table lamp's bulb with a bulb of a lower wattage and gelling it with 1/2 CTB, I was ready to roll.

The scene started with a girl at the window who then crossed to the bed. Knowing the sunlight pouring in would be too much for any format to handle, I made the decision to get a separate cutaway of her and set up the master with the girl just on the edge of frame.

We used a Sony 8" Standard Definition CRT monitor (practically an industry standard for field monitoring) to evaluate our image during the shoot. Certainly not as good as having an HD monitor to view the image in full resolution. But a hell of a lot better than going off an LCD. And considering that 90% (that's a conservative number) of your audience will probably never see the finished project in its full resolution, using an SD monitor is just as valid if not a better way of working.

With the FX1 set to output a live 480i downconverted image in anamorphic 16/9, we plugged it into the Sony 8" via Y/C, set the monitor for 16/9 and everything worked exactly as it was supposed to. It's always nice when things work right the first time. Using the cameras SMPTE bars, calibrating the monitor was quick, easy, and accurate. And thanks to the user assignable buttons, bringing up the color bars was as simple as pressing a single button on the side of the camera. SMPTE split bars and access to them without navigating a menu is a welcomed change from the VX2K and PD150.

The image looked very nice on the monitor. Very clean, clear, and color accurate with just the right amount of chroma saturation for my taste. However, the sensitivity of the CCD's and latitude capability were noticeably less than what I had hoped for. Even with sunlight pouring in from the window, and a 1K fresnel, I had to open the aperture all the way to get the shadow side of the talent (which was the direction we were looking for most of shots) at an acceptable level. And the few spots of sunlight that were hitting the wall or furniture were above 100 IRE. Like I said, sunlight pouring in would be beyond the range of most any format. But that's when we're talking about an actor standing at the window. There was sufficient fall off by the time the sunlight hit the wall and furniture to be in range. But alas, the FX1's latitude handling seems to fall a little short.

Having tested a variety of settings in the days before, I decided to set up the camera right down the middle. Color 0, phase 0, WB shift 0, skin tone DTL OFF, and the sharpness at the default setting of 11. I did not use the CinemaTone gamma setting. But I did use the CineFrame24 setting.

I decided not to use the CinemaTone gamma setting because I felt the resulting contrast was too stark for this film. I'm at a bit of a loss as to what Sony's intention was with the CinemaTone gamma setting. To my eye, the CinemaTone gamma does clamp down on the highlights which should give you a bit more play in the upper end of your exposure. But in the process it also forces the shadows down to a much darker level. The result is an image that is more contrasty and possibly an even narrower range of latitude. This is the complete opposite of what the DVX100A's CineLike gamma settings do. The DVX100A's CineLike gamma functions actually produce less contrast delivering a more even tone and actually increasing the range of latitude.

The only thing I can think of as to what Sony means by "CinemaTone" is a picture that seems richer because of the higher contrast. As a DIT, I often get asked the question, "Can you make it look like film?" To which my reply is, "You tell me what film looks like, and I'll make it look that way." I'm not looking to get into a debate of what film looks like. Yes, I know what people are referring to when they discuss the look of film. But the reality is that Film only looks a certain way because of how it's handled. The way it's lit, exposed, processed, printed, tinted, timed, etc. And that doesn't take into consideration the many different film stocks available or today's new DI timing process. So, what does film look like? Ask ten different people that same question and you'll get ten different answers. I suspect that Sony's CinemaTone gamma setting is what Sony thinks film looks like.

CineFrame...

There has been a lot of discussion about Sony's CineFrame option. I've read plenty of explanations of what CineFrame is and how it works. The explanations range from simple one line answers like, "it's an in camera de-interlacer" .‘ or the camera does a magic bullet process in camera" to complex explanations complete with graphs and very technical sounding theories. The problem is, I have yet to find white papers that support any of these theories. And I haven't read any official explanation from Sony on how it works. If anyone has white papers that verify any of the theories, claims etc, I would love to read them. Until then, I'm going to throw out my own theory based on what my eyes saw. To my eyes, Sony's CineFrame30 appears to be using the exact same method as Canon's Frame Movie Mode of getting non interlaced frames. As I understand it, Canon's Frame Movie Mode takes one field from the Red and Blue CCD and interpolates it with one field from the Green CCD creating a single non interlaced frame. However, this method cuts vertical resolution by about 25%. Sony's CineFrame24 looks as if it's taking those 30 non interlaced frames and just drops frames to get to 24. As everyone knows, getting to 24 frames from 30p or 30 Frame Mode produces less than ideal motion. Add to that the 2:3 pulldown of those remaining frames, and it's no wonder why CF24 produces such stuttered movement.

Now, I've read multiple declarations that CF24 is "unusable" or that it is a "look filter for amateurs". And while I understand why such declarations would be made, I disagree. First of all, any option at your disposal is a tool that can be used to create the images you desire, regardless of how expensive or sophisticated it may or may not be. Prior to the DVX100, Canon's Frame Movie Mode was the only option available to acquire non interlaced 3 chip DV footage. And whether you were shooting for the web, sports, or just desired a motion signature that took the edge off normal interlaced video, Frame Movie Mode was used by a hell of a lot of productions. Even though the XL1 lacked enough pixels for a 1 to 1 NTSC pixel ratio and even though Frame Movie Mode was noticeably lower in quality, productions still used it to great effect.

I chose to use CF24 for two reasons. One, because during my testing I found that the stutter was not an issue when operating the camera as I would any other 24 frame capture system. Whether shooting film or digital, 24 frames per second is a frame rate that requires practice to get smooth motion. Yes, CF24 has an unnatural stutter to it. No, it's not as good as 24p. And NO, CF24 won't be appropriate for every 24 frame project. But when CF24 is shot with care and discipline, the results are nice. Two, I always make every effort I can to do as much in camera as possible. Why? It's the best insurance I've got that what ends up on the screen in the final cut is representative of the work I've done. While I don't doubt that Heath would've followed through with some sort of smart de-interlacer, or Magic Bullet type of processing, I'd much prefer to work with what I've already tested and seen the results firsthand. Since this project was an effort amongst friends, time and money to test and test and re-test multiple ways of achieving different frame rates and motion signatures didn't exist. Perhaps a smart de-interlacer or Magic Bullet process could achieve smoother motion and preserve more resolution. But without having tested the process, I was not comfortable with relying on that possibility.

During the few days of testing the camera on my own, I determined that the FX1's CF24 produced satisfactory results for the motion signature I desired. And thus was the best choice among the options I had tested to achieve a motion signature that removed it from interlaced video. CF30, which produces smoother motion and likely higher quality results in terms of… resolution was an option that I considered. But even though I'm a big fan of 30p both on the DVX and other Progressive Scan cameras, 30p's smoother motion retains a bit of the video look. For those that aren't keen to the motion signature of 30p, I would describe it as a hybrid type look. Somewhere between 24p and 60i. For commercial, industrial, sports, or digitally screened narrative shorts, 30p is a great choice. However, for this project I decided that I'd prefer a motion signature that was another step removed from 30fps. Even if it meant a few moments of noticeable stutter. An added benefit of doing as much as possible in camera is the time and money it saves in post. If I can get the frame rate, color, contrast, tone, balance etc. that I am aiming for, then post doesn't have to spend time rendering the project through some de-interlacer, I don't have to spend a lot of time color correcting, and the production ultimately saves money by cutting down on the amount time it takes to get the job done. And for anyone who's done a supervised transfer or DI, you know that time is money.

Being pressed for time and given that the film was a pretty straight forward narrative that didn't lend itself to unusual angles and frames, we set about covering the scene with the standard wides, meds, and singles. With the exception of one or two brief moments when the action and corresponding camera movement abruptly changed, the CF24 motion signature worked very well. During the course of the five hours we spent shooting, no major problems or anything unusual popped up that were caused by the performance or design of the FX1.

Go back to Part One of Jon's FX1 review.
Move on to Part Three of Jon's FX1 review.
Written by Jon Fordham, Director of Photography.
Thrown together by Chris Hurd.