60p mode on EX3

[Charles Papert]

Hi chaps.

I am doing a shoot this weekend that is meant to suggest a classic TV talk show. We are shooting on EX3's; obviously not a match for the old tube cameras! but I'd like to shoot in 60p mode to keep the live look. We are also doing a fair amount of greenscreen on this shoot. I had read that the EX3 keys nicely, possibly even better than the HVX200 due to the increased resolution and other factors.

What I'm wondering is:

a) I understand that it will be easier to pull the keys if we shoot 60p rather than 60i, which will necessitate shooting at 720. Might this however hamper the keying process due to lower resolution than 1080p? Would we be any better off shooting 1080 60i?

b) This seems like an easy one, but it's throwing me off a bit--what would be considered the "standard" shutter speed at 60p--1/60th? or "off"? I so rarely shoot any flavor of 60 except for slo-mo purposes. And will there be any exposure compensation required from shooting 720p60 vs, say, 1080p24 which is what I have shot with this camera in the past?

thanks all--appreciate it in advance.

[David C. Williams]

EX1/3 use 4:2:0 color space, which means a square block of 4 averaged together. For keying this means you can get nasty saw tooth color when in interlaced. Blurs the edges somewhat. Progressive gives excellent results though, even in 720. If you can, score a HD-SDI recorder for better 10 bit 4:2:2 color, it's not necessary for good results.

720P is 500 ASA and 1080P is 400 ASA, so 20% more light needed for 24P. 1080i incidentally is 800 ASA due to dual row summation to reduce twitter. I generally use 180 deg shutter by default, and just change it depending on light and effect needed. I just did an office comedy with Kino-Flos at 720/60P 180 deg, and it looked very sharp and butter smooth.

[Charles Papert]

Thanks David.

Am I wrong in thinking that "standard" 60 frame video mode (i.e. before we had choices like 24p) was effectively shot with a 360 degree shutter, i.e. none? As the capture of 60 images per second automatically assumed a 1/60th of a second exposure. So I'm trying to understand exactly what a 180 degree shutter, which I would normally think of as 1/48th of a second when shooting 24p, delivers when shooting 60p...there is a quantifiable difference between this and shutter off, and does it require exposure compensation?

[David C. Williams]

180 deg basically gives you a shutter opening duration of 50% of whatever your frame rate is. So yes, shutter off equals 360 deg, 90 deg is 25% opening duration per frame. If your frame rate is 60, 180 deg = 1/120th, 90 deg = 1/240th, etc. If your frame rate is 24, 180 deg = 1/48th, 90 deg = 1/96th, etc.

The EX lets you set the shutter to act as angle like film, or regular time based. Film can't do 360 deg though of course. Angle will automatically scale the shutter to your frame rate, while time stays to whatever you set it regardless of frame rate.

The exposure compensation for shutter off and 180 deg is 1 stop, 90 deg is 2 stops, etc. Just like film really, half the light, 1 stop open. Going from shutter off to 180 deg exposes the CMOS for 50% of the total frame duration, so open 1 stop for the same levels in your image.

I've never really heard of shutter off being "the" standard for 60P. I live in PAL land though, it may be a U.S. default? 180 default works for me, I just fiddle it depending on the shot and the effect I'm looking for.

[Bob Grant]

I recently shot a dance performance at 720p50 to deliver SD 50i. I had the shutter set to 180deg and looking at the final DVD I'm thinking I should have had the shutter off i.e. 1/50th which would match the normal shutter speed for 50i.

[Vincent Mesman]

I recently tested a few green screen recordings with an EX1 of myself in order to get the best settings.

It turned out that with the software I was using in combination with the lights setup and the 1080p25 format without shutter (meaning 360 deg angle) I was not able to recreate transparent motionblur. So every time I moved my arms fast in the final output the arms resized to something bigger. Very ugly.
I also lit the green screen too much, resulting in extreme spill and I was not able to balance correctly between talent and green background.
Also I preferred to use (close to) standard gamma to improve the skin detail, color and s/n ratio in post.

You can't use 4:2:0 with interlaced footage, certainly not for chromakey.
Always use progressive with chromakey.
Always uprez to 4:4:4 when dealing with high color constrasts, for cleaner results - even for backplates.

In my experience the 720p50 or in your case 720p60 is a very universal format. It is easy to convert to interlaced or 25/30 fps progressive formats. When shooting upper body only, you even might want to rotate the camera 90 deg to fill the frame to the max.

Run a few tests that matches the situation - not only 1 single frame. Test your software and stay away from lossy compression as much as your budget can handle.

You will be surprised how much you can learn in post from bad recordings.

[David C. Williams]

Originally Posted by Bob Grant

I recently shot a dance performance at 720p50 to deliver SD 50i. I had the shutter set to 180deg and looking at the final DVD I'm thinking I should have had the shutter off i.e. 1/50th which would match the normal shutter speed for 50i.

This is where someone should have suggested a quick test beforehand, when you have to deliver with a process you haven't used before :)

What was the specific issue?

[Ed Kukla]

Since the camera has no 60P mode, only 60i or 30P in the 1080 format, why not shoot 1080 30P instead of 720 60P??

[Charles Papert]

I have considered that, Ed, but 30p delivers a look that is much closer to 24p than 60i, which for once is desirable (again, this is meant to replicate a classic live television program). I don't believe the motion cadence of 30p will the right one for this project.

[Bob Grant]

I recently shot some green screen tests with the EX1.
Screen was a Scrim Jim 72" stretched fabric.
I used 2 4x55W fluro lights for the screen and the same for the talent (my wife).
Screen was lit one stop below key, iris at around f8. WB off white card, EX1 indicates 5200K. Camera was set to 1080HQ, 25p. I failed to turn Detail Off.
Screen reads within a few degs of pure green on scopes. I was pretty surprised at how noisy the screen is and this caused some difficulty in getting a clean key as I was testing the rather extreme idea of keying fast moving hair in an ECU shot. Once I got a good sample of screen in Keylight results seem OK, considering this is my first real attempt at green screen work anyway.
Based on my first tests I'd keep plenty of light on the screen, get the subject a good distance from the screen. Use sidelight / backlight to limit spill rather than under lighting the screen. Make certain the screen is out of focus unless the screen material is very smooth.

Results can be seen here HD Chromakey Test 1 on Vimeo

I tried the CK FX in Vegas however the need to preserve detail and the noise made balancing it impossible. Ultra worked quite well but Keylight seemed to do the best job. Any attempt at edge manipulation caused major problems in the moving strands of hair. Probably most peoples keying tasks are much simpler than what I'm trying to achieve.

I should mention that motion blur I'm told is very difficult to key, perhaps a faster shutter speed to remove/reduce it is the trick and then add MB after keying.

[David C. Williams]

F8 is well past the diffraction limit for green on the EX, you might get better results keeping above F4 if you can so you don't lose resolution.

Another trick is to light your green screen with green if possible. Screens are never perfect green, and reflect some across the spectrum. They can't reflect what they don't get though, and will cut down your noise.

Faster shutters will give you cleaner edges as you surmise. The EX detail circuit uses luminance to generate detail by default, you can change it to non additive mix of g and r, g, g+r. I've never tried it for keying, but it might give better (or worse :) results? Worth a quick look.