GH2 = No rolling shutter/jello ?
- Thread starter BWyldeck
- Start date
Tracey Lee
Well-known member
yeah I am almost 100% certain that it's impossible to get the same rolling shutter skew with 35mm film cameras...not physically possible.
Blackout
Veteran
if the firmware in effect sped up the timing of the shutter's speed from it's first pixel to it's last, it could help reduce the skew....
Yes, but you have to intentionally do this. What's next? Are there going to be people claiming that you get aliasing and moire on 35mm film?
Barry_Green
Moderator
Please provide examples of 35mm motion picture camera film exhibiting jello or skew.
Barry_Green
Moderator
The speed of the Red's sensor's read/reset cycle was substantially increased (resulting in less perceivable skew) through firmware updates.
Whether improvement translates into "fixed" is debatable. The HPX370 represents a massive improvement over the original HPX300; I used the term "they fixed the skew problem" in my review of it and there were some who were displeased by that. And I can see their point. What I meant was, "they've removed the deficiency and now made it comparable to everyone else" but to some, "fixed" means that the issue would be entirely eliminated.
Which isn't ever going to happen on a rolling-shutter CMOS camera. Whether through firmware or whatever, if it has a rolling shutter, it will have all the rolling shutter artifacts.
What they can in some circumstances do is speed up the sensor to minimize the impact of the effects. And truthfully, there does come a point where you may just say "heck, it's good enough" and quit worrying about it. And that's an individual decision. For interviews, the HPX300's super-slow read-reset looked just fine, but for sports it was too much. The new HPX370's read-reset time is fine for any purpose for which a Red One or EX1 was appropriate. But even they still exhibit all the same rolling shutter issues. And all rolling shutter CMOS cameras will; you can't change that by type, you can only improve it by degree.
Barry_Green
Moderator
Well, let's specify -- with a 35mm cinema camera, I agree. With a 35mm still camera using a focal-plane shutter, it is possible, because a focal-plane shutter that travels a little slit across the frame, is basically the same concept as a CMOS rolling shutter.
But with a 35mm cinema camera, it isn't gonna happen. The 35mm cinema camera has a rotating shutter that results in much of the frame being globally exposed. About the only way you're going to get a 35mm movie camera to exhibit cmos-style skew would be to use a hyper-short shutter angle, such as 10 degrees or something crazy like that, thus eliminating the global exposure of the frame in exchange for emulating the scanning style of a rolling shutter.
Barry_Green
Moderator
Nope. Won't skew. It'll blur, yes, but it will not skew.
Ian-T
Veteran
No jello....just skew.
Barry_Green
Moderator
Read the article on CMOS rolling shutter in the "articles" section and it'll explain it pretty clearly.
You are correct that the rolling shutter could be described as that it exposes a line at a time, sequentially. It is that very nature of progressively exposing only a little bit of the frame, instead of the entire frame globally, that causes the skew.
Barry_Green
Moderator
Please provide an example.
I was reading "fixed" as: rolling shutter doesn't happen anymore. I agree that you can minimize the effects of it.
Which is why I love my HPX-170...I know it's there when I need a shot that sends CMOS into a rolling shutter frenzy.
Ian-T
Veteran
"Explicit Ills" 2008
Not sure if it was film or digital but I read it was shot on 35mm.
Edit: Saw the movie...witnessed the skew. I thought it was shot on a DSLR and looked it up (forgot the source) but it was supposedly shot 35mm anamorphic. But within the first ten minutes of the film I witnessed some skew and noisy images. That's what prompted me to look it up.
EDIT 2: This is what I found:
Film negative format (mm/video inches)
35 mm (Fuji Eterna 400T 8583)
Cinematographic process
Super 35
Printed film format
35 mm (anamorphic)
Aspect ratio
2.35 : 1
Last edited:
Well then you have a lot to learn.
skew is inherent in the architecture of both CMOS and CCD.
CCDs are interline sampled devices in which each line is sampled at a time, not per pixel line a typical CMOS sensor. This itself would lessen the amount of skew you would see, but then we haven't added the fact that CMOS sensors can read each color pixel at a time while CCD's have to interpolate via buffering, processing and reconstruction. I guess the easiest way to think about it is that CMOS images are more like what the sensor is seeing whereas what you see from CCD is a heavily processed image.
Anyway, another way you can affect skew is that you can clock each sensor accordingly based on how fast your sampling routines and what kind of architecture you have. Since CCDs are technically passive devices, you could read them as fast as you can run the shifting routines, which would in turn be regulated by your A/D clock speeds and/or bus speeds. On CMOS, the active nature of the device means that there was a designed native clock speed that the sensor was set to run at since it has the sampling and such on board already. Generally this is designed to work with whatever your DSP is and generally follows the rule of being as cheap as what you can live with.
If you had an agile sensor fast enough and you had enough available DSP, you could very well change inherent processing speed with a simple firmware update. As processing routines are optimized and shortcuts are figured out, it's very natural for electronics designers to release firmware updates to speed things up. I do it all the time in my business.