Rolling Shutter Suggestion

[Charles Angus]

But I believe the OVF got cancelled so that route is closed for RED.

You don't need an OVF if you want a mechanical shutter. You do need a mechanical shutter if you want an OVF, though (pellicles aside).

Mechanical shutter is the proven solution to this problem. It could be much thinner than a film shutter, too, because it wouldn't need to be angled for the mirror.

 

[Nir Shelter]

Would be great to see an in camera option to solve this problem which I have seen many times in high frequency flashing lights scenes, not sure about it's technical viability though.
If the charge from each photosite is dumped sequentially at given intervals it may be possible to sample luminance data from one roe of photosites calculate an average then measure it against it's adjacent roes. If a difference that exceeds a user defined threshold is measured then a user controllable feathering adjustment can, once again in theory, be applied to control the hard edge. Though arguably this could be done in post as well and could very well be impossible technically to implement in camera. I am only paying lip service to the idea.

 

[Phil Bates]

...I don't know the specifics of CMOS design but I don't think that can be done with how current sensors work. The sensor reads one line at a time, top to bottom. It's binary, not progressive: a line is either already read or not read...

I hear you, but if the wizards at RED have figured out how to extract HDR out of an MX sensor, then maybe we're not just talking about a simple binary on-off scenario any more. There must be some other kind of sensor trickery going on, and so maybe, just maybe, it can be used to simulate a feathered rolling shutter.

Graeme, are you out there?

Phil

 

[Peter Majtan]

Peter,

I used techniques to illustrate a real problem. There was no way to duplicate a feathered rolling shutter(the solution) without generating it, thus the need for an illustration. However, the effect is real, it happens all the time when shooting lightning. Here are actual examples (frame grabs) of this problem shot on RED three weeks ago:

...

Phil

Hi Phil, I've never questioned the existence of the problem. As a DP I've faced it too many times to remember. I really like Your idea and I hope RED will do something about it. I was merely pointing out that the kind of correction You have shown is not achievable on the current RED footage, which is why I think You have used Photoshop to better explain Your proposal...

Typed posts are so easy to misinterpret...

Cheers

Peter

 

[Daniel García]

You don't need an OVF if you want a mechanical shutter. You do need a mechanical shutter if you want an OVF, though (pellicles aside).

Of course. I was lumping them together because the OVF was the only known possibility on RED's roadmap of a mechanical shutter implementation.

BTW Do you know of any digital camera without an optical viewfinder that has a mechanical shutter?

 

[Phil Bates]

Typed posts are so easy to misinterpret...

Peter

Yes, they are. Sorry about that.

Phil

 

[Aaron Newsome]

My camera has a physical shutter (Viper). I've shot concert scenes and never noticed a problem with flash partly illuminating frames, although I know it is entirely possible that it could. To see how common the problem might be, I just conducted a semi-scientific test in my living room. I rolled camera and popped my Metz 45 CL-4 about 10 times and then inspected the footage.

Every frame that I popped was lit 100% even across the frame.

This may not be a fair test though. I do believe the Metz light dies a slow death.

 

[Jeff Kilgroe]

As has been pointed out, film cameras, or cameras with a mechanical shutter, do exhibit split-frame exposure and skew problems. But also as has been pointed out here, it's different. Mechanical shutter cameras have a much softer transitional edge due to the shutter blade being defocused. it also resides in front of the image plane physically and light leaks around it, so you don't see the split-frame exposure like you do in the with a scanline sensor. Or at least not ones with the hard horizontal edge.

There really isn't a way to feather an edge of the transition or the sensor reset. That would be something for post. And it's not always the ideal solution anyway. I think the ultimate solution would be a sensor that can adjust its read/reset timings and pattern. Would be really useful to be able to dial the reset timing of the sensor from 0 as a global shutter, up to about 5ms to be more in-line with many film cameras. To me, global shutter has that sterile, HD video look. But there are times when it is the best tool for the job. I also mention reset pattern... Could be useful to specify a scanning direction -- to invert it or move from horizontal to vertical. Or even give the system a reset mask, that details which photosites reset in which order, perhaps giving the ability to create diagonal reset wipes, as a simplistic example, that could better mimic a mechanical shutter.

 

[Rick Burnett]

So I have been thinking about this since last night, and slept on it before I answered because it is a fascinating topic to discuss, and I think I might have a hardware solution.

The biggest issue you face is that when the blade is moving across causing the soft edge, the incoming light that is pushed around that soft edge is being applied for that short amount of time to the film, exposing it. For VERY BRIGHT sources, this is why it can fade across so nicely and not be so harsh.

The problem with doing something automatic with the current mechanism of how lines are read or in post is you have have no way to separate out time based exposure to know what needs fixing and what doesn't. For instance, a light source could be recording through the whole frame that ends up being a certain brightness, then a flash happens half way through the read that ends up being the same brightness because it was brighter to start with so it took less time to absorb the same number of photons. This is why I think a lot of people are saying that this is hard to fix, and from this perspective, they are right!

However, let's take a different approach. Given that most people shoot at 180 degrees essentially, the camera has time where the sensor data isn't needed. This can be a good thing.

What I propose is this, a dual exposure system. It works like this:

1st exposure normal setting (like 1/48 for instance).

2nd exposure occurs IMMEDIATELY after 1st exposure and just cycles through the read/reset immediately without waiting.

(so the exposure of the second frame at best is 5ms or 1/200)

The second frame is processed mathmatically in terms of every scan line is diffused upwards towards the top of the frame and added together, then added to the first normal frame. Since the second exposure is SO fast, it will ONLY grab the powerful light source and help diffuse it through the frame. Anything lower will not affect anything in the final image.

You could even push the processing to post and just store the two images in say a 48fps container.

While not perfect, what this relies on is the time it takes for a line to read out and then that line starting again. Given I believe 5ms is for ALL the rows, this method would be VERY usefull and tuned properly, could really help alleviate this problem.

Now, that said, this may not be possible with the design of the sensor and the associated circuitry that drives it, but I really believe that this method would work, and well.