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 :)

Frame split defect with strobe light

Finner said:
If your SPFX guy gives you an explosion that lasts 1 frame you really should not pay him anything.
Click to expand...

My worry is not only for things that last only one frame, but things that APPEAR in one frame.
Granted, any pyro effect worth capturing is going to last several frames. But if the first frame of the explosion looks split, then it ruins the effect.

First I thought "we'll fix that in post". But after considering the many uses I have planned for the red camera (renting it first) in the TV show I work in, this splitting issue could become a major time consuming problem.

Not only for strobes and explosions, now I am worried about fast moving subjects.
One recurring effect in my show is cars running over people. If the car appears fast enough, will the first frame look split? Or even worse, the car lights will look split as it moves from frame to frame?
 
As Jarred indicated we will have true 360 degree shutter, it's not enabled yet
 
How might this split frame cause problems in greenscreen (or other matte) shots. Say if I was shooting a fast moving object and to reduce the amount of motion blur I would lower my shutter angle ( or increase speed),would I then start to see this split in my matte. Thanks for al the help with this issue. I am just trying to get my head around what this is and what I would need to change in my work flow to avoid it.
 
I don't think it would be possible for the split frame artefact to occur with a moving object, no matter how fast it was moving, as nothing can move instantaneously from one position to another, in the way some kinds of light can (effectively) instantaneously change from off to on and vice versa.

The potential issue with fast moving objects is a shearing artefact, which has been discussed at great length before. Has anybody with one of the production cameras done any testing to see what the shutter shear now looks like?
 
taylor said:
How might this split frame cause problems in greenscreen (or other matte) shots. Say if I was shooting a fast moving object and to reduce the amount of motion blur I would lower my shutter angle ( or increase speed),would I then start to see this split in my matte. Thanks for al the help with this issue. I am just trying to get my head around what this is and what I would need to change in my work flow to avoid it.
Click to expand...

For an object in motion (in relationship to the camera) to even approach what is seen in a strobe induced split, it would have to be moving at a rate that is in all likelihood, nearly impossible. The only 'charge' to your workflow would be identical to other methods of acquisition, namely, to increase the shutter angle (duration) and accept the motion blur which will result.
 
Hey Rob,

I'm just wondering if you can explain the rolling shutter a little better, my understanding is that in for example a 1/48th shutter, the top row of pixels begins collecting light first and then very quickly each successive row of pixels begin collecting light.

Then exactly 1/48 of a second later the top row of pixels stops collecting light and each successive row after that stops collecting light. As the data is read off the sensor.

Is that correct?

And if so then my question is: How long does the roll out take exactly?

And in addition is the rollout actually scanning each row together, or is it also rolling down each pixel like a non interlaced cathode ray tube would. Thus you might see a little one pixel notch in the strobe clipping.

Is their any chance that the shutter could work in a non-consecutive way in the future, creating other kinds of splits that might be more attactive in situations where strobes or even lightning are not controllable?

Maybe their is a better word for this period I'm calling the rollout.

Thanks,

IBloom
 
"I don't think it would be possible for the split frame artefact to occur with a moving object, no matter how fast it was moving, as nothing can move instantaneously from one position to another, in the way some kinds of light can (effectively) instantaneously change from off to on and vice versa."


Thanks for all the help on this question. I hope that the above quote is the case...but as we have been told in this thread that this is a "known issue" I was hopping for some definative answers. Though I hope it will not be a problem (yes I am on the list) I might have to wait and see. Thanks all
 
it is a known issue (thats been around for years before Red) that is not unique to our camera.. and we (as we have before) are finding new ways to reduce or eliminate.
 
Thanks Jarred. Never met you, but your posts always seem to make me feel better. :)
 
Rob Lohman said:
As Jarred indicated we will have true 360 degree shutter, it's not enabled yet
Click to expand...

Question?

Once 660 degree shutter is enabled will we be able to select other angles?

For instance if I wanted a 220 or a 15º shutter angle would this be possible?

Method
 
The rolling shutter only can give a split image when something appears or disappears instantly, as a flash.

An explosion takes a good couple of frames to develop, so the main problem remain flashes of light. Right now (as has been said, the readout speed has improved a lot) I don't think that even muzzles would be a major issue.

To minimize this effect I propose to use 2k windowed (if the readout is faster) at 72fps, so that you could always get a usable frame fully flashed.
 
Is it not possible to turn off the rolling shutter and sample the whole chip for 1/48th of a second?
 
SF Geek said:
Is it not possible to turn off the rolling shutter and sample the whole chip for 1/48th of a second?
Click to expand...

To directly address every photoreceiptor site instantaneously would require a pixel clock that was infinite in speed. As with all other things infinite, this too isn't possible. Frankly, I feel people are making way too much of this. In real world conditions, this isn't going to be much of an issue. Even strobe illuminated shots can avoid the problem if the strobe has a duration long enough or is properly synchronized to the camera.
 
I believe it is duration OR synch. Even with the unsync'd strobe in the one piece of footage posted so far showing this effect, if you look through it frame by frame, you can find some frames that are fully illuminated by the strobe. (There are no electronic strobes I'm aware of that actually last a full 1/48 sec.)
 
Yep, but these are flashes that start on the 1/48th of second before the frame starts and end in the 1/48th of second after the frame takes place.

So if the flash isn't synched and it lasts for a multiple of 1/48 of second then you will see either the start or the end, unless it is the case before amentioned.
 
Synchronization is obviously the best solution, but with a flash duration longer than the frame rate, then only the leading/trailing frame would be affected, the intermediate one would not be.

I have simulated lighting strikes without synchronization using a very large photo flash (Norman P2000) set to have a very long duration (near 50 ms). The total output was too high, so I just punched the strobe head through some ND to adjust the illumination level to match what I desired.
 
Just throwing out a wild idea here, so I have no clue if it's feasible or even wanted without introducing a new form of artefacting, but....

As far as I can figure out the sensor begins its sweep from top to bottom for a single frame. So in the case of a mismatched strobe, it comes on half way through the sweep, hence the line artefact.

Is it therefore possible to perhaps tell the sensor to begin its sweep in the middle of the sensor and split into two directions, up and down, at the same time? Or come in from the top and bottom at the same time? Could that maybe mean that although there might still be a sync issue, the actual image from it would appear more aesthetically pleasing than the current method?

Like I said, I have almost no knowledge of how these things work at the technical level, I'm just trying to think outside the square and perhaps fire some creative spark in others more versed in the field.

Cheers!

Paul
 
Double reading (up and down) is the same problem as instantaneously reading. It would require doubling the clock speed I believe.

I'm not sure it would look any better even if you could. In such a situation you would have a bright center bar through your image with darker letterboxing.

As far as explosions are concerned this is really only a problem for any effect which lasts less than than the duration of the shutter (aka 48th of a second). So like finner said. No problem at all. What would most likely happen is your explosion would look a bit squashed.
 
As much as we're talking about syncing the shutter, we're not sure how the Red would do that, correct? Considering all of the things they're working on, syncing would probably be something that wouldn't be addressed until far into the future, yes?
 
You must log in or register to reply here.
Back
Top