GH2 = No rolling shutter/jello ?
- Thread starter BWyldeck
- Start date
Nobody said they were. You're just too busy trying to nitpick and bust, without sufficient scientific proof, everything that people are saying to notice. However, they can be used somewhat interchangeably due to the fact that some of the terms define each other.
The effect of "skew" is a visual representation of the physical action of the generic term we are using for CMOS sensor readout called "rolling shutter". "Jello" is just skew in multiple directions in quick succession. Therefor the effect of rolling shutter can be described as skew or jello depending on the conditions.
It may just be semantics but that's never really stopped a forum troll before has it? :2vrolijk_08:
Emanuel
Cross-Examiner
The most funny is when today the 'video artists' are 'too good' for having the infamous jello on their work... Almost 100 years ago, Lartigue left the autochromes once he could have the color but not handle the motion like this:
Jacques Henri Lartigue [ LINK ]
Car Trip, Papa at 80 kilometers an hour, 1913
aka Grand Prix de l'A.C.F Automobile Delage - 26 juin 1912
Last edited:
Hey, Svart, a little "reading comprehension" lesson for you:
I see one mention of CCDs being more immune to wobble and skew, but the conversation has so far been about rolling shutter. So when you quote me saying that I don't believe a firmware update can "fix" rolling shutter, and then in your response talk about skew exclusively, then you put forth the strong indication that you believe that rolling shutter = skew, do you not?
What's your excuse?
That's right, nobody said they were the same. We didn't have to because everyone else knows that they are the same. You apparently do not. let me give you a second dose of "reading comrehension".
Skew, wobble and jello are just some of the optical effects attributed to the physical sensor readout, (aka rolling shutter) and how it interacts with the world through the lens.
Rick Burnett
Section Moderator
In my experience, skew relates to tilting images either to the left or right due to movement in the horizontal direction (panning).
Jello refers to compression and expansion of the image from moving in the vertical direction (tilting) and looks like jello expanding and contracting.
While they can both occur at the same time, they are fixed (if at all possible and mostly not) in different ways.
Jello refers to compression and expansion of the image from moving in the vertical direction (tilting) and looks like jello expanding and contracting.
While they can both occur at the same time, they are fixed (if at all possible and mostly not) in different ways.
First of all, everyone knowing something is a false statement in and of itself, because it's not possible. Second of all, you just got after me for pointing out that that rolling shutter and skew are not the same and now you're saying that everyone knows that they are the same.
So which is it? Does rolling shutter = skew or not?
Jello is just "skew" in very short motions in multiple directions. It's simply the change in pixel readout justification in comparison to those around it. Since CMOS reads pixel by pixel in line by line fashion similar to the way old tube cameras (and the inverse to how tube TVs display) then you can have the optical output of a line read look shortened or lengthened when the camera moves mid-line read. In short, that means that if the camera changes directions between pixels, the remaining pixel read on that line will look different. To the same effect, if your camera moves slow enough to avoid mid-line changes, then you could still get line to line changes in justification which will manifest as skew/wobble/jello.
Panning back and forth is slow enough that the pixel to pixel reads aren't completely messed up but you'll still get a small amount of extra blur. However you will still get line to line justification changes and your image will lean(skew)
Put the camera on a car and go down a gravel road and the camera will vibrate. Think of it is tiny pans in random directions which will interfere with the pixel to pixel reads in a line but not pan the image enough in any direction to manifest as a leaning image. This will end up as wobble/jello.
Panning back and forth is slow enough that the pixel to pixel reads aren't completely messed up but you'll still get a small amount of extra blur. However you will still get line to line justification changes and your image will lean(skew)
Put the camera on a car and go down a gravel road and the camera will vibrate. Think of it is tiny pans in random directions which will interfere with the pixel to pixel reads in a line but not pan the image enough in any direction to manifest as a leaning image. This will end up as wobble/jello.
There you go nitpicking again. You're not much one for allegory are you?
"Skew" IS the optical effect of the action of a "rolling shutter". For all purposes they are one and the same when the term "skew" is used to describe the optical effect of the "rolling shutter".
John Caballero
Veteran
Jeez! Why people that are always complaining, pixel peeping or finding faults from anything from the hack to anything they want to find fault with, don't just go out and shoot something? If the GH1 or any cmos camera is producing too much wobble/jello/skew or not enough sharpness just go out and buy/rent a 35mm film camera, buy real film and shoot something. You can't afford that? Then get a CCD camera or go into 3D animation. But then again they will find something wrong with that too!
Rick Burnett
Section Moderator
This is NOT my experience. EVERY CMOS sensor that I have studied reads out each full line in parallel. It is just each line that is read serially. I have tested my cameras quite a bit and can tell you that I do not get ANY compression of the image in the horizontal direction. This is easily testable by setting a very fast shutter speed and panning very fast. If you take one line while moving and compare it to the same line when you were not moving, you will see the distances between high contrast spacial points are the same. I have done these tests when I was doing VFX work so I knew what limitations I had in shots for rotoscoping and such.
When you tilt back and forth fast on a cheaper camera with VERY SLOW rolling shutter, even on my iphone I don't see horizontal compression or expansion (but haven't tested it thoroughly).
The rolling shutter, as I said, causes 2 different effects both because of the serial line reading.
1) Skew, which is the issue of horizontal movement and caused my successive line reads of the rolling shutter. Objects moving exactly horizontal will have a lean to them relative to the difference in time from when the lines were parallel-y read.
2) Expansion/Contraction, which is the issue of vertical movement and also caused by successive line reads of the rolling shutter. Objects moving exactly vertical will flatten or lengthen in the vertical direction based on whether they are moving up or down relative to the sensor.
Obviously you can get both phenomenon but they are dealt with in different ways. For instance, I use a stabilizer (black bird) to eliminated expansion and contraction and it works GREAT! Skew is a little more acceptable to me in small doses. I Also just pan slower
The only thing wrong with CCD now is that "no shallow depth of field derf!"
John Caballero
Veteran
You get a similar 16mm DOF with 2/3" CCD cameras though.
Rick Burnett
Section Moderator
I'm an electrical engineer and programmer, I agree I like to get excited about both the left AND right sides of my brain!
I like to get excited about both the left AND right sides of my brain! Barry_Green
Moderator
Not entirely true. CCDs also exhibit vertical smear on overbright highlights. And, when going for smaller chips, CCDs just can't deliver the same pixel count with the same sensitivity and the same heat management that CMOS can. Which is why we're seeing a proliferation of CMOS in little-chip cameras, and not so much infiltration in the 2/3" range.