Rolling shutter improved with proper shutter speeds?

[Jon Fairhurst]

Tonight I tested pans and tilts, and I think they look better than before. I was running 1/60, 1/50 and 1/40. I then slowed to 1/30 and camera motion looked terrible.

I'm wondering if the 5D MkII didn't just get an upgrade in the handheld department. So often the old firmware would dial down to 1/30 that it wasn't funny. Or shoot outside, and you get 1/300. Surprise, surprise. Pans and tilts looked terrible.

I guess it's always possible that Canon sped up the sensor reads, but unlikely. (Mark, do you have your flash unit handy?) I'm thinking that the 5D MkII got a worse-than-deserved reputation for being bad handheld than merited by rolling shutter alone.

Give it a try. Dial in 1/60 and set a good exposure. Shoot handheld. Is it better than you remember it to be? I tried it with a 28mm and was able to get fairly brutal. With an 85, I was able to do handheld with a loupe but no rig, and it wasn't half bad. Sure, I could get lines to bend if I wanted to with the 85mm, but that's no surprise.

I'm thinking that a number of people moved the 5D with a 1/30 setting and simply wrote it off.

[James Miller]

I totally agree, it was the first thing I noticed but thought it was too good to be true.
It does look better with or with OIS. Also 50Htz lighting flicker is more control able for obvious reasons.

[Mark Hahn]

Originally Posted by Jon Fairhurst

Tonight I tested pans and tilts, and I think they look better than before. I was running 1/60, 1/50 and 1/40. I then slowed to 1/30 and camera motion looked terrible.

I'm wondering if the 5D MkII didn't just get an upgrade in the handheld department. So often the old firmware would dial down to 1/30 that it wasn't funny. Or shoot outside, and you get 1/300. Surprise, surprise. Pans and tilts looked terrible.

I guess it's always possible that Canon sped up the sensor reads, but unlikely. (Mark, do you have your flash unit handy?) I'm thinking that the 5D MkII got a worse-than-deserved reputation for being bad handheld than merited by rolling shutter alone.

Give it a try. Dial in 1/60 and set a good exposure. Shoot handheld. Is it better than you remember it to be? I tried it with a 28mm and was able to get fairly brutal. With an 85, I was able to do handheld with a loupe but no rig, and it wasn't half bad. Sure, I could get lines to bend if I wanted to with the 85mm, but that's no surprise.

I'm thinking that a number of people moved the 5D with a 1/30 setting and simply wrote it off.

I was hoping now that Canon fixed everything I wouldn't need to revisit the pain of analyzing the 5D2 behavior. I am happy to throw away the mylar and the flash data.

I can tell you this though. The two scanning lines, one resetting and one reading, should make the actual amount of Jello constant with regards to shutter speed, since it takes the same amount of time to scan the sensor at any speed.

However, if you have more or less motion blur it might make the subjective problem seem better or worse. I could argue both ways. More motion blur might hide the jello, or motion blur might add to the jello and make it worse.

I can't think of any experiments to measure such a thing since it is a subjective visual experience.

We got the firmware today. We need to play a few more days to make anything near a conclusive statement.

I myself am still hoping to shoot at 1/48.

[Jon Fairhurst]

Originally Posted by Mark Hahn

I myself am still hoping to shoot at 1/48.

1/50 should be pretty close - assuming that the numbers aren't lying. (I know. Bad assumption!)

Regarding the flash test, I'm just wondering if they sped up the sensor read-reset time in firmware. It's unlikely, but you never know.

[Mark Hahn]

Originally Posted by Jon Fairhurst

1/50 should be pretty close - assuming that the numbers aren't lying. (I know. Bad assumption!)

Regarding the flash test, I'm just wondering if they sped up the sensor read-reset time in firmware. It's unlikely, but you never know.

I'll redo the flash measurement when it's convenient, but I don't see why they would change that. I've been assuming all along they are running the CMOS at the max lines/sec. It is expensive to make it fast. 150 fps would be nice, but VERY expensive. 24fps would be a bit cheaper.

[Jon Fairhurst]

I did a quick test and found a 72.5% rolling shutter factor. Previously I measured about 75%. My flash is sloppier than yours (turns on fast, off slow), so the difference is likely due to measurement error. It's basically unchanged.

My result for a 1/60 indicated shutter speed is 1/55. That's pretty darn close, considering my measurement tolerance.

Anyway, it seems that my perceived improvement in handholdability is purely due to having a proper shutter speed. Too slow and it gets blurry and gross during movement. Too fast, and it's choppy and stuttered.

There's a reason that Hollywood went with a 180 degree shutter duration. If it didn't look good, they would have chosen something else...

[Chris Barcellos]

As with my FX1, 1/60 will be locked in most of time. IF Tramm can put ISO on the wheel, I would like that, but I am not about to bitch about it....

[Silton Buendia]

So what are the recommended best shutter points to use with this camera? I am here 1/60 but what else for faster shutter speeds?

[Jon Fairhurst]

Originally Posted by Silton Buendia

So what are the recommended best shutter points to use with this camera? I am here 1/60 but what else for faster shutter speeds?

Choosing faster shutter speeds is an art, rather than science. If you really want the image to look like frozen images, use the fastest shutter speed that you can. In general, you'll be limited by how much light you can throw at the subject.

Another use of high shutter is when you've mounted your ND filters, you want a certain DOF with your aperture, you've set ISO to 100, and the image is still overexposed. At that point, increase the shutter speed or decrease the aperture to find the right balance.

Really 1/60 (or so) is the right choice for natural motion. Higher and lower shutter speeds are either chosen for a special effect (which is a personal decision), or to compensate for exposure limitations.

[Bill Binder]

This is so true. I shoot a lot of indoor, dark, jazz stuff, and the 1/30 shutter was killing me. Now I can lock 1/50 or 1/60, and then trade off ISO with Aperture depending on how dark it is, how much noise I'm willing to live with, and how much or little DOF I want. A beautiful thing. Although it would have been great (for me personally) to have exposure lock and comp when using Auto ISO in manual video, but can't have everything, and this is a huge improvement overall.

Also, I think running a 1/50 shutter with shallow DOF at 30p (not 60i mind you) will seriously close the gap on the 24p crowd. Don't get me wrong here, it'd be great to have 24p, but just saying...

Finally, although I think IS is a good thing under certain conditions with video, I actually am fairly convinced that IS also has the potential to cause a lot of problems -- problems that can easily be mistaken for rolling shutter. On a handheld shot that's trained on something, it's great, but for any sort of panning/tilting/running/gunning, I actually turn it off and get better results. But that's just one man's opinion.

[Mark Hahn]

Originally Posted by Jon Fairhurst

I did a quick test and found a 72.5% rolling shutter factor. Previously I measured about 75%. My flash is sloppier than yours (turns on fast, off slow), so the difference is likely due to measurement error. It's basically unchanged.

My result for a 1/60 indicated shutter speed is 1/55. That's pretty darn close, considering my measurement tolerance.

Anyway, it seems that my perceived improvement in handholdability is purely due to having a proper shutter speed. Too slow and it gets blurry and gross during movement. Too fast, and it's choppy and stuttered.

There's a reason that Hollywood went with a 180 degree shutter duration. If it didn't look good, they would have chosen something else...

When Edison invented the thing, I'd bet dollars to donuts the only concern was to darken it long enough to move the film to the next frame. Also 1/2 open is too round a number to be scientifically designed.

I'm one of those that believe when all theaters go digital, frame rates will shoot up and they will market it as some new wonderful advance. I also don't believe tubes and vinyl make better sound.

[Jon Fairhurst]

Originally Posted by Mark Hahn

When Edison invented the thing, I'd bet dollars to donuts the only concern was to darken it long enough to move the film to the next frame. Also 1/2 open is too round a number to be scientifically designed.

I'd agree that's how it originated. And then people did detailed tests of other settings - and people preferred the 180 thing.

That said, I don't think 180 degrees is the magic perfect number. A bit slower or faster is fine. However, we know that 360 is too sloppy, and 10 degrees is too stuttery.

I'm one of those that believe when all theaters go digital, frame rates will shoot up and they will market it as some new wonderful advance. I also don't believe tubes and vinyl make better sound.

I think 24p is another happy accident. I saw the Batman Begins trailer recently with everything smoothed out to 120 Hz. Rather than a fantasy world, it looked like I was standing on a Hollywood set with pre-graded lighting.

BTW, SMPTE has standardized 24 and 48 fps for digital cinema. There is a proposal to add 25p and 30p (2k & 4k), as well as 50p and 60p (2k). In addition, 16, 18, and 20p are proposed as optional for showing early works (and the T1i? ;))

The main reason for the proposal is to show content at native rates and to avoid frame rate conversion.

Some believe that 3D needs higher rates as high as 120fps to achieve the "flicker fusion threshold".

It's not clear that the proposed frame rates could be changed on the fly.

Personally, I'd film fantasy stuff in 24p, and documentary stuff in 30p or 60p. It would be cool to be able to intercut the two, depending on the point of view.