Tehnical info

[Gogu]

I wanted to know what is the tech stuff regarding shutter speeds and framerates. I don't understand how the sensor works when you set your cam to shoot at 60fps with 1/100 for example. It refreshes the sensor 60 or 100 times per second ?

Does the camera record truly 60 fps or does it process this internally somehow ? Do different cameras do this differently (HVX, EX1, Varicam even RED) Sorry if this has been talked about in another post, i searched the boards...

PS. And a personal curiosity about the EX1 body... The "SXS card" part of the body seems to be linked with the "custom buttons part" by only a small sort of cable if you look through the gap fomed by the 2 parts. It looks like the camera is made of modules. Is the data tranfsfer between those 2 modules donw solely with the help of that little cable ?
Don't know if you can understand my mumbo jumbo but i was really curios to how the camera was designed.

Thanks!

[Carl Marxx]

:violin: To answer that P.S. part of your question, the gap, I've been told, is to cool off the SxS memery cards by allowing air between the three parts, tha back and main frame and lenses. The wire is the umbilical chord that attaches the main and memory parts of the camera; this was confirmed by the Sony rep. I think the best way to find out about a camera of this type, is just is to pick it up and shoot with it! For you to fully understand the engineer's reasoning for a camera of this caliber is like a man's fully understanting of a woman, it will never happen!

[lawriejaffa]

Yeah - especially her mind! Whaaa

[Gogu]

Thanks for your answer. I was just curios about the design.

But doesn't anyone know about how does it record the number of framerates and how the sensor works ? :thumbsup:

[matthew77]

The sensor reads from top to bottom in 1/60 second (actually bottom to top - the image is inverted on the sensor)

So when the EX1's internal computer says "read the sensor" it takes 1/60 second to do so.

How often this happens is determined by the framerate. When set to 24p, for example, the sensor is read 24 times per second (actually 23.976).

Before the sensor is read for each frame, the sensor is reset. This discards the light information that has thus far accumulated (since the previous reset) and allows new, fresh light information to accumulate.

Exposure time, or "shutter speed" determines that interval. If the shutter is set at 1/48 second, the camera resets the sensor (a process which also takes 1/60 second from top to bottom) and 1/48 second later the camera reads the sensor. And this process repeats 24 times per second.

Since it's a rolling shutter camera, each line of the sensor is exposed at a slightly different time, but the effect of the above is that each given line of the sensor is exposed for 1/48 second in a cascading fashion.

Operation is very similar to the focal plane shutters in film and digital SLR still cameras. The first curtain opens (the reset) and then the second curtain closed (the read). For faster shutter speeds the second curtain begins to close before the first one finishes opening, effectively creating a slit that scans across the image.

Since the EX1's sensor can read or reset no faster than 1/60 second, the frame rate is limited to 60fps at the high end.

This is pretty much the same process for all non-mechanical digital motion picture cameras. The main difference among cameras is whether the whole sensor is reset/read at one time (global shutter) or in a cascading fashion (rolling shutter).

In your example, 60fps at 1/100 second, the camera initiates a reset, and then 1/100 second later initiates a read. This process is repeated 60 times every second. The read cycle begins before the reset is finished, creating the effect of an exposure window or slit moving down the sensor. The EX1 truly records 60 frames per second.

[Gogu]

Thank you very much for the answer matthew. Very nice explanation .

But your answer arises a few other "technical issues" let's say.

If i shoot at 1/10 in 60fps how can the camera do this 60 times a second when after the first 10 times that second is already up - not adding up the reset times that you said can be no shorter than 1/60 .

And in my example ( 1/100 at 60 fps ) isn't it possible that after the 1/60 reset is issued the 1/100 read may "catch" up to the reset and...then what happens ? Let's suppose a 1/200 shutter... :undecided

Am i making any sense or are the little gears inside my head spinning in the wrong directions ? :shocked:

PS. Are the global shutter cameras creating a "slit" of exposure on their sensors or they have to wait for the reset and then read again ?

Thanks again!

[matthew77]

The 1/100 is the delay between reset and read. But they both "travel" down the sensor at the same speed - 1/60 second to get across the sensor.

It's like two runners (Mr. Reset and Mr. Read) that run the same speed starting at different times. Mr. Read will never catch Mr. Reset. They start 1/100 second apart and they finish 1/100 second apart.

To continue the (weak) analogy, the next set of two runners, spaced 1/100 second apart, start 1/60 second after the first set of runners started. This means that runner 3 starts running before runner 2 has finished. But all runners are running the same speed, so they never catch each other.

Hope this makes sense.

(and it's not really possible to shoot 60fps at 1/10 second exposure [without frame accumulation that is a special effect]). The slowest shutter at 60fps is 1/60 second. In this case the reset for the next frame immediately follows the read from the previous frame.

[Gogu]

OH...now it makes sense. I didn't catch from your first post that 1/100 represesnts just the delay from read to write.
And do you have a link or somethin with all this info so i can read it in peace, cause 100 new questions just popped up in my head right now after reading your last post and I don't want to keep bugging you with these things.

Anyways here they are:

1) I thought that 1/100 shutter is a 1/100 shutter speed not a delay between two electronic commands. So i thought that 1/100 is the time the sensor is kept open to receive light (just as in classical film cameras). From what you are saying I just understand that the shutter speed is fixed at 1/60. How can you freeze motion better with faster shutter speeds if the camera is fixed at 1/60.

2) If I have 1/100 at 24fps and then at 60fps how can the camera record 24 and 60 times/second separately ? I mean it's math...1/100 + 1/60 + 1/100 and so on ...I don't know how many this you can add these numbers up until they give you one second but I surely know that the same formula cannot work for both 60 and 24 times per second.
In film cameras the whole mechanics inside it start to move faster as you crank the speed up but in the digital equivalent the same amounts of intervals are periodically rolling one after the other.

This is how i understand it and it just doesn't seem logical to me. sorry! And thanks again for your time trying to make me understand this. :costumed-smiley-047

[thxdave]

<snip>
In film cameras the whole mechanics inside it start to move faster as you crank the speed up but in the digital equivalent the same amounts of intervals are periodically rolling one after the other.<snip>

I'll jump in for just a tiny bit of this (because Matthew is doing a great job without me!), but your understanding of the focal plane shutter in a traditional film camera is slightly off. The "speed" or linear movement of the shutter doesn't increase as you raise the shutter speeds, but what actually happens is that the width of the shutter's opening or slit gradually gets smaller. That's why you can't exceed a certain shutter speed when you are using electronic flash. You're limited to the fastest shutter speed setting where the entire film aperture is open to the incoming light. When we transitioned from horizontally moving shutters to vertically moving ones, the sync speed jumped from 1/60th to 1/125 or higher almost immediately. I believe that this is the reason that we are seeing the "banding" issue when an EX1 is shooting in a room full of flashing strobes......but Matthew will do a better job of explaining this than me! :)

[matthew77]

I'll jump in for just a tiny bit of this (because Matthew is doing a great job without me!), but your understanding of the focal plane shutter in a traditional film camera is slightly off. The "speed" or linear movement of the shutter doesn't increase as you raise the shutter speeds, but what actually happens is that the width of the shutter's opening or slit gradually gets smaller. That's why you can't exceed a certain shutter speed when you are using electronic flash. You're limited to the fastest shutter speed setting where the entire film aperture is open to the incoming light. When we transitioned from horizontally moving shutters to vertically moving ones, the sync speed jumped from 1/60th to 1/125 or higher almost immediately. I believe that this is the reason that we are seeing the "banding" issue when an EX1 is shooting in a room full of flashing strobes......but Matthew will do a better job of explaining this than me! :)

Dave,

Exactly correct on all counts. If you've ever tried taking a flash photo at a higher speed than the "flash sync" speed on a still camera, you'll get exactly the same partial exposure as on the EX1. And mechanical focal plane shutters always move the same speed, as you said. Just the open/close timing changes.

Gogu,

There is no actual shutter in the EX1 (or the HVX or most every video camera). The sensor is always collecting light infomation. The "shutter speed" is derived from two electronic commands - the first zeroes all accumulated light (the reset) and the second reads the amount of accumulated light since it was reset (the read). Light continues to accumulate after the read, but it will all be thrown away at the next reset. At 1/100 second, only the first 1/100 second worth of light is being read (even though ultimately 1/60 second worth accumulates on the chip before the next reset), so that means motion will be frozen just like 1/100 second exposure should.

I don't understand your #2. The EX1 only records one frame rate at a time. If it records 24p, then the sensor is resetting 24 times per second, period. If it's 60p, then it's 60 times per second. Nothing is being added together.

Again, at 24p, the camera is resetting 24 times per second. It is also reading 24 times per second. The time offset between these two cycles is the exposure time or "shutter speed." The read cycle can be initiated before the reset is finished and vice versa on a rolling shutter camera, so both are often happening simultaneously on different parts of the sensor, depending on the shutter and frame rate settings.

[Gogu]

Hello again.

I haven't been around the computer for some time and I didn't read your last response.

What I was hoping to be able to explaini s this: I was trying to add up the number of 1/60 read/reset cycles at different framerates to see after how many of these cycles the second results.
For example at 1/100 @ 24fps we have 1/60 (reset) +1/100 (delay) + 1/60 (read) + 1/60 (reset) and so on....If my math is correct multiplying all this by 24 is not going to give you a full second. If we add it by 60 (as in 60fps) I think it will probably be around there...

So what does the sensor do after the 24 reset/read cycles until the second finishes ? I hope I'm making some sense now.


And after reading all your answers again i'm not sure if I understand this correctly.
If I have 1/100 shutter set, after the reset does the camera immediately start reading 1/100 light accumulation and then discards the light info from the rest of the read at the next reset OR does it wait 1/100 after the reset and then in only takes into account 1/100 light information of the total 1/60 read ?
What about a 1/48 shutter which is more than 1/60 ?

Hope you're still around and that your patience is not wearing thin because of all of my annoying questions. Thanks!!

[matthew77]

Gogu,

Please reread my last post. It's hard to explain it any better without visual aids. But I'll try.

Forget about the concept of 24 frames in one second. "One second" is irrelevant. Each frame is exactly the same, so let's look at one frame only.

Let's assume 24p at 1/60 exposure.

The total duration of the whole frame is .042 seconds. The exposure (which occurs within that time period) is .017 seconds. The read and reset times are .017 each, as is the case at all frame rates and shutter times.

at time = 0, the reset cycle begins at the top of frame.

at time = .017, the read cycle begins at the top of frame (and the reset cycle is finished, just reaching bottom of frame)

at time = .033, the read cycle reaches bottom of frame

for the next .09 seconds, nothing happens on the sensor even though the sensor is being exposed to light, all of this will be thrown out when...

the process repeats.



For the second example, 60p at 1/100 second:

The duration of the whole frame is .017 seconds. The exposure (which occurs within that time period) is .01 seconds. The read and reset times are .017 each, as is the case at all frame rates and shutter times.

at time = 0, the reset cycle begins at the top of frame. The read cycle from the previous frame is still underway.

at time = .01, the read cycle begins at the top of frame. The read cycle from the previous frame is just completed

at time = .017, the reset cycle reaches bottom of frame. At the same time the reset
cycle is beginning again at the top of frame as...

the process repeats.

There is no pause when shooting at 60fps. Read and reset immediately repeat after each has ended. Their start times are separated by the exposure time (.01 sec) and therefore they are occuring on different parts of the sensor at different times. Hope this clarifies it (though perhaps it's just more confusing).

Unfortunately the examples you chose are a bit confusing because both include 1/60 second as parameters, which are the same as the reset/read times.


I'll try one more which might make it clearer - 1fps with a 1/2 second exposure:

The total duration of the whole frame is 1.000 seconds. The exposure (which occurs within that time period) is .500 seconds. The read and reset times are .017 each, as is the case at all frame rates and shutter times.

at time = 0, the reset cycle begins at the top of frame.

at time = .017 the reset cycle reaches the bottom of frame

at time = .500, the read cycle begins at the top of frame

at time = .517, the read cycle reaches bottom of frame

for the next .483 seconds, nothing happens on the sensor even though the sensor is being exposed to light, all of this will be thrown out when...

the process repeats.

[Gogu]

Thanks so much! You re a great teacher :beer: . I understand everything now except how did you calculate each frame's length and why is the sensor "idle" for a certain amount of time in cases 1 and 3. Is it because it waits for the second to add up ?

For 24fps @ 1/48 it would be like:

at time = 0, the reset cycle begins at the top of frame.

at time = .017 the reset cycle reaches the bottom of frame

at time = .020, the read cycle begins at the top of frame

at time = .037, the read cycle reaches bottom of frame

I don't know how to do the sensor "idle" time and the whole frame time.

And in the first case shouldn't there be an adittional .017 delay between reset and read ? But maybe 1/60 shutter is equivalent with the read time so there isn't any delay in this case right ?

many thanks again!

[matthew77]

I think you got it.

The duration of each frame is a simple computation -

24p = 24fps = 1/24 second per frame. 1 divided by 24 = .0416666666 seconds for each frame

60p = 1 divided by 60 = .016666666

The sensor can be considered "idle" because that is the time that the shutter can be considered completely "closed" even though there really is no shutter and the sensor is still collecting light (which will be thrown out)

For you example (for which your math is correct) , the sensor does nothing for about the last .005 seconds, until the next frame begins.

Again, Gogu, dont think about seconds - think of each frame as an independent event.

24fps is only a fancy way of saying that each frame is photographed at .042 (1/24) second intervals.

[matthew77]

Let me just add an inconsequential thing:

24p really runs at 23.976 fps and 60p at 59.94fps, so the numbers above (which are provided for simplicity's sake) are actually slightly different.

[Gogu]

Yeey...thanks for having the patience of explaining this to me!
Now i can go brag about my exotic knowledge of digi camera framerate!

:Drogar-Smoke(DBG):

[adamr316]

Gogu,

This digi camera knowledge will help you understand the female mind much better and get you dates. Try it next time at a club or bar!

[thxdave]

Gogu,

This digi camera knowledge will help you understand the female mind much better and get you dates. Try it next time at a club or bar!

Oh man.....I am SOOO reading this thread again if THAT'S the case!! :thumbup:

[Gogu]

I will I will...keep you posted on results! :beer:

[matthew77]

It's good to know your way around a woman, and it doesn't hurt to know your way around the EX1 as well.

Good luck.