Clarification for rolling shutter

[William Wedig]

I've seen what it looks like and know why it happens and why it sucks.

My thought is (and I've never seen this mentioned here) if you increase the shutter speed it should help with the problem. Correct? My thought is that it will be scanning for a shorter amount of time so difference between when it scans the top and when it scans the bottom shouldn't be that large.

I always assumed this to be true and then my friend started asking me questions which started to bug me out so I figured I'd post.

We're going to shooting with Lomo Anamorphics so I want to know how I can adjust for the problem as it's increased with Ana's. We also have normal spherical lens for anything that it gets too problematic with.

Does this work and if not, why not?

 

[Adrian T.]

Unfortunately, the rolling shutter speed is independent of the shutter speed. It's always the same. In fact, by increasing the shutter speed you get less motion blur and therefore you can see the skew much clearer.

Only way to get less skew is using a lower resolution (3K/2K).

 

[William Wedig]

Ok, well that's very unfortunate.

If the shutter is rolling at the same speed, then what does the shutter speed in camera do? Is it just how long each pixel exposes or does it have something else going on as well?

 

[J. Eric Camp]

My understanding is that the "roll" stays the same and it's just individual "acceptance time" that changes.

 

[Paris Remillard]

If you're concerned about skew, as opposed to the strobe/half frame illumination issues, and since you're using anamorphics, thus cropping from the center of the frame anyway, you might try mounting the lenses and camera sideways for the specific shots where there is the concern. You'd lose a bit of resolution and FOV since the 1.2:1 ratio would be swapped, then have to rotate each frame 90 degrees and crop the top and bottom in post, but you'd get compression/expansion instead of skew since the readout would be from side to side instead of top to bottom. I haven't tried this yet, still waiting for my birger mount, but it seems to me that it should be less noticeable than skew in small doses. Won't help with the jello effect caused by movement in both axes, but... I've got a shoot on Wednesday and someone else is paying for PL lenses : ) so I'll see if I can do a rough test to see if it might work at all. Worth a shot.

 

[William Wedig]

If you're concerned about skew, as opposed to the strobe/half frame illumination issues, and since you're using anamorphics, thus cropping from the center of the frame anyway, you might try mounting the lenses and camera sideways for the specific shots where there is the concern. You'd lose a bit of resolution and FOV since the 1.2:1 ratio would be swapped, then have to rotate each frame 90 degrees and crop the top and bottom in post, but you'd get compression/expansion instead of skew since the readout would be from side to side instead of top to bottom. I haven't tried this yet, still waiting for my birger mount, but it seems to me that it should be less noticeable than skew in small doses. Won't help with the jello effect caused by movement in both axes, but... I've got a shoot on Wednesday and someone else is paying for PL lenses : ) so I'll see if I can do a rough test to see if it might work at all. Worth a shot.

We do have round front anamorphics so that's not a bad idea. The other option is just using out regular (non anamorphic) primes for any shots that might have issues...

That really sucks about the rolling shutter. I thought it actually followed your shutter setting that you put in camera...

 

[Paris Remillard]

I did a quick test this afternoon to see the difference between quick panning with the camera sideways vs. normal/upright. It was rough, unscientific, handheld, but... The difference is huge, to my eyes at least. And to another director/DP who was with me. I don't think that I would ever notice the compression/expansion artifacts when masked by motion blur.

The reason that skew looks so objectionable to me is because it doesn't follow reality. When I whip my head to look at something, things blur as my eyes move past them, but the geometry of the world doesn't change. Vertical lines remain vertical. With compression/expansion, at least vertical lines also remain vertical, so the compression part doesn't jump out as being "off" and the motion blur masks most of the effect.

Now, I know that I also have to consider that the width of the frame is only 16mm/2K sized when the camera is on it's side, so the readout will be faster and subsequently rolling shutter artifacts will be less severe, when compared to the full 4k frame(if it were to be read horizontally) but I still think it would be worth it for camera manufacturers to consider trying a horizontal, side to side read of CMOS sensors with rolling shutters. Still doesn't get rid of any of the artifacts, but does seem to help mask them based on, again, very unscientific tests. I'll try to post footage soon so that people can judge for themselves.

 

[Mark Crabtree]

Less than 1/48th

Less than 1/48th

Would the slightly increased duration time of say 1/40th of a sec help mask the skew with a little more motion blur on those occassions you are doing fast pans?

 

[sander kamp]

The reason it looks worse when using anamorphics is that you have to stretch the frame horizontally to 200 percent, which makes the skew 2 x worse as well. When using the anamorphics the other way around you reduce the skew to 50 percent compared to normal, or 4 times less than using normal anamorphics.

There is also no need to turn the camera if you can mount the lens 90 degrees rotated.

 

[Chris Burket]

What exactly is this rolling shutter issue? What does it look like?

Did anyone see it in the anamorphic shots from the Argentine at the red booth? If its a major issue, I'm wondering how an entire film was shot with anamorphics?