Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Rolling Shutter

Dylan Macleod CSC

Dylan Macleod CSC

Well-known member
Joined
Sep 23, 2007
Messages
1,315
Reaction score
0
Points
0
Website
www.dylanmacleod.com
I haven't seen a whole lot of concern listed here regarding what appears to be a significant side effect of the rolling shutter.

I saw a recent RED test that showed fast panning left and right. The side effect was that the image appears to slant one way or the other. It is most notable in vertical lines.

This concerns me.

Has this been talked about anywhere?

Is it even possible to "fix" this?

I searched and only saw discussion of strobes and flashes.

Dylan Macleod, CSC
Toronto, Canada
www.dylanmacleod.com
 
Do you have a link to the footage online anywhere? I have been hawk-eyed for this type of artifact since the cameras started shipping and haven't seen a single example of rolling shutter problems; so if you've got one, I absolutely want to see it!
 
cailyoung said:
Do you have a link to the footage online anywhere? I have been hawk-eyed for this type of artifact since the cameras started shipping and haven't seen a single example of rolling shutter problems; so if you've got one, I absolutely want to see it!
Click to expand...

I do not. I was invited to see the footage (ultimately to be printed to film!). It was someone elses test, but made me wish I had tested for it when I had my hands on "Madrid".

I think anyone with a camera should perform this simple test. Shoot anything with prominent vertical lines (like a window) and do successively faster pans. The windows will appear to slant as you pan. Moreso the faster you get. Like "strobing" in film, this may become a "RED RULE"..."do not pan faster than...."

Really, shouldn't be an issue with 95 percent of what we might shoot.

Whip pans? Forget about it.

Extreme handheld? Who knows?

Dylan Macleod, CSC
Toronto, Canada
www.dylanmacleod.com
 
Evin Grant said:
My understanding and experience is that the rolling shutter scans fast enough to eliminate any "Shearing" problems. I'll try and test it soon if I have the time.
Click to expand...

Don't worry about it, Evin. We've got a very comprehensive rolling shutter test planned for mid Jan. I've already built the rig to test it.

Of course I think it might have been tested at LART, but I doubt they did what I'm doing.

I want to put this issue to bed once and for all and get on with life. To know a thing is to understand a thing, and therefore to do better work with the tools at hand.
 
dmacleod said:
I do not. I was invited to see the footage (ultimately to be printed to film!). It was someone elses test, but made me wish I had tested for it when I had my hands on "Madrid".

I think anyone with a camera should perform this simple test. Shoot anything with prominent vertical lines (like a window) and do successively faster pans. The windows will appear to slant as you pan. Moreso the faster you get. Like "strobing" in film, this may become a "RED RULE"..."do not pan faster than...."

Really, shouldn't be an issue with 95 percent of what we might shoot.

Whip pans? Forget about it.

Extreme handheld? Who knows?

Dylan Macleod, CSC
Toronto, Canada
www.dylanmacleod.com
Click to expand...


Forget about whip pans?

Well I can't speak for anyone else, but it strikes me as very silly to let this kind of technical niggle limit your creative freedom.

Of course if there was a system to record the camera tracking data realtime in 3d, and save it to metadata, it would be pretty easy to match this slanting in post.

Did I mention how awesome it would be to have such a system?

Well I'm gonna keep fucking saying it till someone listens.

R.
 
It's all Fear Uncertainty and Doubt (FUD) until it's comprehensively tested. The footage I've seen and experimented with is a long way from the HV20 style issues and seems quite trackable.

But, there needs to be tests. And tests we'll do.
 
Jim Arthurs said:
It's all Fear Uncertainty and Doubt (FUD) until it's comprehensively tested. The footage I've seen and experimented with is a long way from the HV20 style issues and seems quite trackable.

But, there needs to be tests. And tests we'll do.
Click to expand...

Fear, yes. Uncertainty and doubt, a little.

The issue does exist. I've seen it.

"Tracking it later" is not what I wanted to hear. :wacko:

I work as a cinematographer. I show up on set. I deliver the images. Production takes them home. If I am lucky I get to go to a color correct.

RED is a whole new ballgame which I can't wait to sink my teeth in to. But the more issues that have to be dealt with after the shoot day...the less desirable it will be to use this camera.

Dylan Macleod, CSC
Toronto, Canada
www.dylanmacleod.com
 
dmacleod said:
Fear, yes. Uncertainty and doubt, a little.

The issue does exist. I've seen it.

"Tracking it later" is not what I wanted to hear. :wacko:

I work as a cinematographer. I show up on set. I deliver the images. Production takes them home. If I am lucky I get to go to a color correct.

RED is a whole new ballgame which I can't wait to sink my teeth in to. But the more issues that have to be dealt with after the shoot day...the less desirable it will be to use this camera.

Dylan Macleod, CSC
Toronto, Canada
www.dylanmacleod.com
Click to expand...

Right, but it's really only an issue if you have to do vfx isn't it?

So if you have to do vfx, you'll have to track it ANYWAY.
 
Dylan if you're wanting to remove the slant in post you're completely out of luck. As soon as you go beyond pan and tilt camera moves of a static scene the amount of rolling shutter effect is completely pixel independent.

Someday it might be correctable with optical flow techniques but we're a long way off and will pretty pointless with the development of better global shutter sensors.

We did test rolling shutter effects at LART. It's not a global shutter, but imo it looks very very good. Hopefully they'll post the relevant footage on vuze soon. It involved vertical and horizontal lines getting moved about at various speeds.
 
Ruairi Robinson said:
Right, but it's really only an issue if you have to do vfx isn't it?

So if you have to do vfx, you'll have to track it ANYWAY.
Click to expand...

It is not exclusively a VFX issue. It is something that visually distorts the image. What I saw looked a little "cartoon" like. The whole image kinda slants in the opposite direction of the camera movement.

Something that someone might say..."whoa...what was that?"...like I did.

Dylan Macleod, CSC
Toronto, Canada
www.dylanmacleod.com
 
im.thatoneguy said:
We did test rolling shutter effects at LART. It's not a global shutter, but imo it looks very very good. Hopefully they'll post the relevant footage on vuze soon. It involved vertical and horizontal lines getting moved about at various speeds.
Click to expand...

I agree that it's virtually/practically impossible to compensate for the various vectors that would happen with a moving camera and moving objects. The only solution is simply a better shutter technique down the road with whatever next generation sensor RED uses.

I'd also agree from the two 3D tracks I've done that I haven't hit a problem with the subject matter that would be called a rolling shutter issue. But both of these tests were just for fun with 2nd generation clips with subject matter that had a bit of fudge room on a perfect track.

The test I intend to do will be pretty clinical and visual at the same time. It's EXTREMELY important for us VFX guys to know the exact amount and under what conditions this issue comes into play.

I'm sure someone at an FX house somewhere has done the due-diligence on these sort of tests, but they're not commonly available, or were done with in-house resources on a given project and not for public consumption.

Hell, I'm buying one of these cameras and I before I put all my money down I damn well want to know when to duck and cover, or when to shrug it off and keep shooting...
 
I posted on this topic several months ago (use the search). We use a rapid reset sensor system which is much faster than a typical CMOS rolling shutter. It is a bit slower than film so it is easier (not much) to induce skew with RED than with film. But it is MUCH better than a standard rolling shutter.

We have pointed this out to every DP (and VFX god) that was interested in using RED on features. So far, no one has seen it as a problem.

Jim
 
While I haven't seen skew in any of my shoots I have been more frustrated with the split screen artifcating by bright flashes of light etc...Is there any work around for that currently, or is that just an inherent problem we will learn to deal with due to CMOS technology?
 
LighthouseMEdia said:
While I haven't seen skew in any of my shoots I have been more frustrated with the split screen artifcating by bright flashes of light etc...Is there any work around for that currently, or is that just an inherent problem we will learn to deal with due to CMOS technology?
Click to expand...

That is what Build 12 is for... and the new cameras. Are you on Build 12?

Jim
 
Lots of things have changed in the chain from the sensor on back. That's why the need for the Black Shading calibration. I'll have to ask our guys what specifically and see if we can post what they were...

Jim
 
You must log in or register to reply here.
Back
Top