Ad
Ad
Ad
Pages: [1] 2 »   Bottom of Page
Print
Author Topic: shutter speed  (Read 5884 times)
fredjeang
Guest
« on: February 13, 2011, 05:23:13 PM »
ReplyReply

Hi,

This is a tech question regarding DSLRs video and shutter speed that I never understood.

It seems that pre-setting the (stills) shutter speed in manual mode to a multiple of 50 like 1/50th 1/100 etc...is important.
First: I still do not understand the technical link or the impact has the traditional shutter speed, in video. I've been trying many times but I can't get a technical picture in my mind.
2: Why those number? (avoiding 1/25, 1/125 etc...)

Have you ever heard about that "rule" ?

A note from Canon: Increasing the shutter speed too far beyond 1/125th sec. will result in an unusual staccato or 'stroboscopic' effect to your movies, due to the decreased motion blur from frame to frame. This can be used to effectively capture fast-moving objects in greater detail (such as race cars or other sporting events). It can also be used creatively to convey increased energy/movement, a sense of tension, or to create a surreal quality to your footage.

What I don't get is the why.
« Last Edit: February 13, 2011, 05:24:46 PM by fredjeang » Logged
langier
Sr. Member
****
Offline Offline

Posts: 649



WWW
« Reply #1 on: February 14, 2011, 08:45:01 AM »
ReplyReply

Simple reason for 1/50th second for video and partially dates back to motion picture cameras and even before...

1/50th gives you a little blur to motion with so that when you view the movie, it appears more life-like and fluid. It's not perfect but it works. Faster shutter speeds stop the blur, thus the motion is frozen and successive images take on a stop-motion, stroboscopic jerkiness look. A still image looks fine and sharp, but when combined with 24-60 fps, it's another story...
Logged

Larry Angier
ASMP, NAPP, ACT, and many more!

Webmaster, RANGE magazine
Editor emeritus, NorCal Quarterly

web--http://www.angier-fox.photoshelter.com
facebook--larry.angier
twitter--#larryangier
google+LarryAngier
Pete_G
Full Member
***
Offline Offline

Posts: 234


WWW
« Reply #2 on: February 14, 2011, 10:37:17 AM »
ReplyReply

Take a look at Spielberg's "Saving Private Ryan", especially the last 20 minutes or so, that was shot with a high shutter speed, and produces the effect Canon mentions. The technique is used (or misused) quite a lot these days. For normal action stick to 1/50th sec. The same effect is visible when directors decide to shoot action scenes in slow motion, then in editing realise this was a mistake and speed up the action by dropping frames. Slow motion uses higher shutter speeds, so the effect of shooting slow and dropping frames to create normal action produces the same effect. You can see this in most modern action films.
Logged

___________________
http://www.petergoddard.org
fredjeang
Guest
« Reply #3 on: February 14, 2011, 12:02:54 PM »
ReplyReply

Thanks guys.
Hey Pete, I saw that you shoot at Hossegor, it was my beach during my childwood...lots of souvenirs there.
Sea is tough, I remember that in summer the helicopters where rescuing all the time the drunk german and british tourists who wanted to play with the ocean completly unaware of the danger. I saw many deaths there. I learned to "read" the sea here, when you can take risks and when not, it saved my life actually.
http://www.youtube.com/watch?v=IkCLcrjrPlM&playnext=1&list=PLA7266663624D5091
I also used to go in the precise Carrefour you photograhed in Bayonne. World is small.
Cheers
« Last Edit: February 14, 2011, 05:05:43 PM by fredjeang » Logged
Pete_G
Full Member
***
Offline Offline

Posts: 234


WWW
« Reply #4 on: February 14, 2011, 05:27:00 PM »
ReplyReply

Hossegor is where we stay after driving from London for 15 hours and deciding which route to take
into Spain; we stay at a no-frills hotel by the lake, it's great, we've been doing it for years.
 
The beach and the sea are something else though.

The sea is threatening, I think it's because the beach is steep next to the water,
this makes the ocean appear to be on top of you. Scary. I took some photos but they didn't describe
this surreal view very well, pity.

You're right though, the sea is tough, I can't imagine swimming there. It's easy to understand how
drunken northern europeans who can't read the sea get, literally, "sucked in".

Looking forward to some of that southern Spanish light this year...it's been a long dark winter up here...
Logged

___________________
http://www.petergoddard.org
fredjeang
Guest
« Reply #5 on: February 14, 2011, 06:03:57 PM »
ReplyReply

Yes, you know all the coast of the Lande is totally artificial, forest included. They started to do that enormous work under the reign of Louis 14th if my memory's correct.
There are very dangerous tides called "ba´nes". They litterally suck you. People who don't know try to come back to the shore and died. But in fact, if you let yourself go with the tide you are not at risk because the ba´ne just leave you at one point and the sea will bring you back on the shore by itself on another point. But the waves are high and it can be a scary experience.
I remember the lake yes, and the town of Capbreton with the little port. I'm back 15 years ago.

Here the winter has been long too, rather cold.
Logged
bjammin
Newbie
*
Offline Offline

Posts: 22


« Reply #6 on: February 15, 2011, 04:42:05 PM »
ReplyReply

   I'm convinced that the subject of "shutter speed" when shooting video is a bit more complex than selecting multiples of 50. For example, the original United States NTSC television system employed 1/30 of a second as it's refresh rate, which was logical since US electricity is 60 cycles per second.  Try and record video of any electric light with a pronounced flicker using any shutter speed higher than 120 of a second that is not divisible by 60 and you will see that the light appears to flash brighter and dimmer.  I've experienced this effect when shooting in Europe where 50 cycles per second is the electrical standard. If you live in Europe you will find that using multiples of 50 will keep your movies shot under artificial light evenly lit unless you visit a country with 60 cycle electricity such as all of the Americas and Japan.

   Of course there is another issue.  If you select a higher shutter speed than a 50th or 60th of a second you will also observe another phenomena: an artificial looking rendering of movement. This is because no matter what shutter speed you select, a video camera will only record 24, 25, 30 or 60 frames per second for reproduction as standard video footage. All motion blur will be replaced with an artificial looking series of crisp still images.  Try shooting some video of moving water at shutter speeds above a 500th of a second and you will see what I mean; it doesn't look right.
   Now you could purchase a special "high-speed" camera and it will be able to record frame rates in the 100s per second, which when played back at one of the standard shutter speeds I'm mentioned above will reproduce as slow motion. But if your goal is to capture believable footage I recommend sticking to multiples of 50 if the electricity in your country is 50 cycles and you are shooting under artificial lighting, and stick to a shutter speed that is a multiple (or close to it) of 60 elsewhere.

bjammin
Logged
fredjeang
Guest
« Reply #7 on: February 16, 2011, 09:01:20 AM »
ReplyReply

Thanks Bjammin,

It did also appear to me that the topic was not as easy as it looks and that's why I started this thread.
On a tech point of view, regarding what you just pointed about the multiples of 50 because of the electricity standart and beware of turning into 60 in the US, I could figure that issue well from an understanding point.
But what I still can't get is the dslr itself technology.

In other words, I'm applaying rules because I've been told so and because I understand just a part of the all picture, and I don't like that. I need to fully understand why I'm doing this. When I ask arround me I never got a complete picture, nobody really explains me in details. In fact, it seems that this topic is not clear for most of the guys I know, included the very experienced videographers that went dslr recently.

What I find difficult to see in my mind is the role of applying a shutter speed for stills (let's say 1/50) while the mirror is up in live-view and the speed has been settled to 25 fps. I can't get the technology involved in the fact that pre-selecting the traditional shutter speed involved in stills affects the filming result. (then shutter speed would be completly electronic) There is something I can't get there, because what happen in-camera if you are filming not in manual mode? What does the camera choose for you?

Tried to find some info over the internet and so far nothing clear and simple to understand at the same time.  

(not sure if I'm explaining myself very well, sorry)
« Last Edit: February 16, 2011, 09:39:16 AM by fredjeang » Logged
Chris Sanderson
Administrator
Sr. Member
*****
Offline Offline

Posts: 1911



« Reply #8 on: February 16, 2011, 10:39:07 AM »
ReplyReply

In Auto mode or Aperture Priority, DSLRs use shutter speed as one of the settings to control exposure - along with ISO and aperture. Therefore DSLRs will break the 'shutter speed rule' when shooting video more often than they keep it.

Rule Number One: Always shoot video in full Manual mode.

Rule Number Two: Unless you want a 'special effect', use 1/50 (48) or 1/60 as a video shutter speed. This will look 'right' and gives reasonably smooth motion and the 'correct' amount of motion blur that audiences are used to.

Adjust Aperture to where you want it - generally two or three stops down from wide open if possible.

Select ISO to control exposure to where noise is acceptable.

And finally, use a variable ND to control the amount of light entering the lens if over-exposure threatens.

Then throw out the rules and make it look good  Grin


« Last Edit: February 16, 2011, 11:00:49 AM by Chris Sanderson » Logged

Christopher Sanderson
The Luminous-Landscape
bjammin
Newbie
*
Offline Offline

Posts: 22


« Reply #9 on: February 16, 2011, 12:06:55 PM »
ReplyReply

Chris is correct, the exposure will be partially determined by the "shutter speed," in combination with the f stop.  What is happening inside the camera is that the images are still captured 24, 25, or 30 times a second, depending upon the television standard you are using, but the amount of time committed to capturing the exposure is determined by the shutter speed selected. If the user selects a 500th of a second, then each frame of video has been exposed to light for that 500th of a second.

All video cameras spend part of their time capturing the video images and part of their time copying the images they have just produced. In truth, a camera shooting 30 frames a second will only be able to capture images in a 60th of a second, with gaps in time while the images are written to media. This is not bothersome to users because the video display is showing each frame for the entire time necessary for the images to appear seamless. They really aren't.

bjammin
Logged
fredjeang
Guest
« Reply #10 on: February 16, 2011, 02:02:12 PM »
ReplyReply

Thanks to all of you guys. Now I think I finally get the correct picture in mind.

Lu-La is a fantastic source.

Cheers
« Last Edit: March 24, 2011, 10:43:58 PM by fredjeang » Logged
mephisoft
Newbie
*
Offline Offline

Posts: 2


« Reply #11 on: February 18, 2011, 06:50:28 PM »
ReplyReply

I got my new GH2 last Friday, and will go for skiing on next Monday. I plan to use my new toy to make some action videos of my kids. This brings me to the shutter speed question.

I think I understand the 180 degree shutter rule, and that ~1/50s is the "right" sutter speed for the 24p mode. However, I am still struggling with the choice between 24p and 50i modes for shooting sports. I understand that the 24p mode has better quality due to the higher bitrate, but wouldn't 50i with 1/100s be better for fast moving objects? This question I can't answer just by theoretical reasoning, Google hasn't helped either, and I am afraid I will have no time to test before leaving for the mountains.

It would be so nice to know the answer before returning from the slopes Smiley

What is your take?
Logged
Chris Sanderson
Administrator
Sr. Member
*****
Offline Offline

Posts: 1911



« Reply #12 on: February 18, 2011, 09:18:35 PM »
ReplyReply

If you plan to use/view the footage at its originally shot frame rate, there is effectively little difference - a quibble.

If however, you plan to use slow motion effects added in post-production, the Progressive format will give better quality since its frames are complete and not interlaced. In this case, you may want to experiment with shorter shutter speeds up to reduce motion blur.
Logged

Christopher Sanderson
The Luminous-Landscape
mephisoft
Newbie
*
Offline Offline

Posts: 2


« Reply #13 on: February 19, 2011, 10:52:46 AM »
ReplyReply

It makes sense, thanks a lot!

So far I didn't think of slowing down motion, but sound as a good idea, I will definitely try and play with it.

But that brings another question: when am I better of: if I use the "clean" progressive mode footage as input for slowmo, or the 80% slow motion mode that I assume is interlaced? I have no idea of what amount of slowing-down will work best for ski-action, but I am sure that the 80% alone is not sufficient, so post processing would be needed for that as well.
Logged
slrlounge
Newbie
*
Offline Offline

Posts: 5


WWW
« Reply #14 on: April 28, 2011, 03:35:18 PM »
ReplyReply

I use the rule of 180 as a general rule of thumb. I usually just set accordingly to the certain look I want. I love shots with shallow depth of field so I like shooting below f2.8, I use ND filters to block light so I can stop down to a low aperture and therefore I get to keep my camera at 1/50th of a second.
Logged

hjulenissen
Sr. Member
****
Offline Offline

Posts: 1683


« Reply #15 on: April 28, 2011, 03:52:13 PM »
ReplyReply

I think I understand the 180 degree shutter rule, and that ~1/50s is the "right" sutter speed for the 24p mode. However, I am still struggling with the choice between 24p and 50i modes for shooting sports. I understand that the 24p mode has better quality due to the higher bitrate, but wouldn't 50i with 1/100s be better for fast moving objects? This question I can't answer just by theoretical reasoning, Google hasn't helped either, and I am afraid I will have no time to test before leaving for the mountains.
The selection of 24p/25p/30p/50i/60i/... involves device-specific compromises that may or may not work better for a given scene and a given artistic intent. In other words, simplified, general theoretical approaches may not give satisfying answers.

"50i" could consist of physical sensor lines #1,3,5 taken at time 0, then sensor lines # 2,4,6,... taken 1/50 second later. That would be what most of us considers interlacing. But then a practical camera might not have 1080 physical sensor lines. It might have 3500, and it might not have the bandwidth to read all of them 50 times/second. In that case, it might read every 4th line 25 times per second, and then use filtering/interpolation to "shift" the content up and down (and in time!) to produce 50i. Another camera might average line#1 and line#2 to produce the first line, then line#2 and line#3 to produce the first line 1/50 seconds later. It might not be optimal, but practical products seldom are....

The choice of shutter is conventionally chosen to be a 50% duty cycle (meaning that the sensor is actively integrating light half of the time). I dont think there is a theoretical justification, it is just a pragmatic, historical thing where movement is partially blurred, pňartially sharp, and where the sensor gets to integrate over some time (so as to improve noise).

-h
Logged
Ray
Sr. Member
****
Offline Offline

Posts: 8907


« Reply #16 on: May 02, 2011, 01:42:09 AM »
ReplyReply

The selection of 24p/25p/30p/50i/60i/... involves device-specific compromises that may or may not work better for a given scene and a given artistic intent. In other words, simplified, general theoretical approaches may not give satisfying answers.


This sounds very true to me, although I have little experience shooting video.

My Panasonic plasma TV has a 'sub-sampling' refresh rate of 600Hz which, it is claimed, reduces the blur in fast action. The figure of 600 presumably has been chosen because it is the lowest figure that is divisible by all the current video frame rates, 24, 25, 30, 50 & 60fps.

As I vaguely understand, but I'm not at all certain, when the video source or broadcast is 50i, for example, the TV processors may first combine each pair of interlaced frames recorded at a rate of 50 fps to make 25 progressive frames per second.

On the other hand, during fast action, combining two frames 1/50th second apart would produce a very blurry progressive image, so maybe the TV processors simply interpolate to fill in the missing lines to produce 50 fps progressive.

Each of those 25 progressive frames per second will then be displayed 600/25=24 times, or 12x in the case of 50 fps progressive, to produce a totally seamles and totally jerk-free result.

On such a display with a 600Hz refresh rate, I have difficulty understanding how using a fast shutter speed to shoot the video would produce any sense of jerkyness.

Unfortunately, I can't test this for myself at present because I am currently travelling. My video camera is the Nikon D7000, and so far I've shot all video at 24fps in aperture priority mode, setting an ISO that seemed roughly appropriate for the lighting conditions, and allowing the camera to choose the shutter speed.

I think it might be time to start experimenting with different manual shutter speeds. So far the only obvious trouble I've had relates to sound. I've been using the Sennheiser MKE 400 external microphone which fits into the D7000 hot-shoe. On one occasion, recording a cultural show in a restaurant in Pokhara, with camera fixed to gorilla pod on table, I was certain I had pressed the record button firmly, then continued eating my meal and watching the show instead of peering at the LiveView screen.

A while later, at the end of the routine, putting on my glasses to stop the video and review the result, I discovered I hadn't recorded anything. When the next song and dance routine began, I paid special attention to make sure the red record sign was active after pressing the record button, but I couldn't get the camera to record.

My memory card wasn't full. Had this Made-in-Thailand camera become faulty? I had already experienced an inability of the camera to autofocus in cold weather, not excessive cold, just normally cold around the freezing point. It was frustrating having to manually focus everything in the early mornings along the trail to the Annapurna Base Camp, until the camera warmed up in the sun.

Anyway, after searching the menu for microphone settings, suspecting the sheer loudness of the sound coming through the restaurant's loudspeakers might have something to do with the problem, I solved the problem by setting the mic to 'low sensitivity'.
Logged
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1683


« Reply #17 on: May 02, 2011, 03:08:42 AM »
ReplyReply

My Panasonic plasma TV has a 'sub-sampling' refresh rate of 600Hz which, it is claimed, reduces the blur in fast action. The figure of 600 presumably has been chosen because it is the lowest figure that is divisible by all the current video frame rates, 24, 25, 30, 50 & 60fps.
And also because it is a "large" number, sounding impressive :-)
Quote
As I vaguely understand, but I'm not at all certain, when the video source or broadcast is 50i, for example, the TV processors may first combine each pair of interlaced frames recorded at a rate of 50 fps to make 25 progressive frames per second.

On the other hand, during fast action, combining two frames 1/50th second apart would produce a very blurry progressive image, so maybe the TV processors simply interpolate to fill in the missing lines to produce 50 fps progressive.
There are many ways to deinterlace 50i/60i to 50p/60p. Typically you want to give the impression that 1080@60i has 1080 unique lines every field (which it does not), and that it has 60 full frames per second (which it does not). Some content-adaptive time/space interpolation is the usual formula, covering the fact that 50% of the information is missing.
Quote
Each of those 25 progressive frames per second will then be displayed 600/25=24 times, or 12x in the case of 50 fps progressive, to produce a totally seamles and totally jerk-free result.

On such a display with a 600Hz refresh rate, I have difficulty understanding how using a fast shutter speed to shoot the video would produce any sense of jerkyness.
You need to separate between capture resolution and presentation resolution. If you display the same frame 1000x a second, it will still be the same frame. If you try to "guess" 600 frames a second from 24 input frames, you still are limited by the information that is presented at the input.

Many film enthusiasts hate the motion interpolation modes and want to view the 1080@24p as "native" as possible. If you want the look of a movie theatre, you would probably want to flash each 24p frame in 2-3 short bursts, before moving on to the next.

-h
Logged
Ray
Sr. Member
****
Offline Offline

Posts: 8907


« Reply #18 on: May 04, 2011, 09:57:53 AM »
ReplyReply

And also because it is a "large" number, sounding impressive :-)There are many ways to deinterlace 50i/60i to 50p/60p. Typically you want to give the impression that 1080@60i has 1080 unique lines every field (which it does not), and that it has 60 full frames per second (which it does not). Some content-adaptive time/space interpolation is the usual formula, covering the fact that 50% of the information is missing.You need to separate between capture resolution and presentation resolution. If you display the same frame 1000x a second, it will still be the same frame. If you try to "guess" 600 frames a second from 24 input frames, you still are limited by the information that is presented at the input.



If we start off with the easy situation of a succession of indentical interlaced frames at either 60i or 50i, as in footage of a static scene, then combining consecutive pairs of interlaced frames will result in more image information per second  than 24p.

The problems arise where movement has taken place between one interlaced frame and the next. It seems to me that the ideal solution is dependent on computer power, sophistication of algorithms and input from a skilled operator.

It should theoretically be possible to separate static backround from moving parts in each frame so that at least all static parts of each frame, after deinterlacing, will retain at least the image resolution of 24p.

The question thus arises, if a slow shutter speed producing a degree of blur in the moving parts of the video is preferable because it produces a more natural effect, could a degree of blur resulting from consecutive interlaced frames shot at a high shutter speed, that have been combined or woven and subject to edge-detection processing and removal of artifacts, be just as good as footage shot at 24p, 25p or 30p, using state-of-the-art software?

Quote
Many film enthusiasts hate the motion interpolation modes and want to view the 1080@24p as "native" as possible. If you want the look of a movie theatre, you would probably want to flash each 24p frame in 2-3 short bursts, before moving on to the next.

I must admit I have difficulty in appreciating why this is so. It's almost like someone claiming that he hates progressive scan because he's so used to watching the flickering image of an interlaced broadcast and display.

As far as natural movement is concerned surely 60p is better than 24p. Is there some physiological reason related perhaps to saccadic eye movement, why 24p may be preferred to faster frame rates?
Logged
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1683


« Reply #19 on: May 04, 2011, 12:57:49 PM »
ReplyReply

If we start off with the easy situation of a succession of indentical interlaced frames at either 60i or 50i, as in footage of a static scene, then combining consecutive pairs of interlaced frames will result in more image information per second  than 24p.
If the scene is really static, then there is no more information than can be captured in a still-image. Therefore 24p contains as much information as 60p or "1p". 2x 60i fields should ideally contain as much information, but in practice the interlacing procedure will typically introduce some spatio-temporal lowpass-filtering (meaning that 60i will often not be quite as sharp as 30p no matter how good your deinterlacer).
Quote
The problems arise where movement has taken place between one interlaced frame and the next. It seems to me that the ideal solution is dependent on computer power, sophistication of algorithms and input from a skilled operator.

It should theoretically be possible to separate static backround from moving parts in each frame so that at least all static parts of each frame, after deinterlacing, will retain at least the image resolution of 24p.
Your practical concerns are valid, no doubt. But they all rely on simplified models of how a physical scene behaves. Those models are often valid, but not all of the time. When a dancer is spinning fast around against a detailed background consisting of an audience, and the available light cause lots of camera noise, the information about the background behind the dancer, or the dancers left face may simply not be there. The condensed data available in the video stream says how things looked at time1 and time2, but not what happened in-between, or the velocity and acceleration at any time. When one throws away 50% of the information, it is really thrown away. One can throw insane amounts of cpu cycles and Phd students at the problem, but that will only give solutions that are very good for a large percentage of scenes, not one that is perfect for all scenes.
Quote
The question thus arises, if a slow shutter speed producing a degree of blur in the moving parts of the video is preferable because it produces a more natural effect, could a degree of blur resulting from consecutive interlaced frames shot at a high shutter speed, that have been combined or woven and subject to edge-detection processing and removal of artifacts, be just as good as footage shot at 24p, 25p or 30p, using state-of-the-art software?
I dont think that practical examples shows this to be the case today, and I think that a proper theoretical analysis would show that no algorithm could guarantee it.

I still think that it is fair to expect some unknown, ideal software to be able to do what you describe 99% of the time such that 99% of us can see no difference.
Quote
I must admit I have difficulty in appreciating why this is so. It's almost like someone claiming that he hates progressive scan because he's so used to watching the flickering image of an interlaced broadcast and display.

As far as natural movement is concerned surely 60p is better than 24p. Is there some physiological reason related perhaps to saccadic eye movement, why 24p may be preferred to faster frame rates?
I think that it is purely cultural. Some are used to good experiences in the movie theatre (using stuttery 24fps) and poor tv-series (using smoother 50i/60i).

I prefer standards that are as transparent as possible. If some director wish for 24p looks, it can pretty accurately be simulated inside a 50p/60p stream (48/72p would be even better).

-h
« Last Edit: May 04, 2011, 01:00:34 PM by hjulenissen » Logged
Pages: [1] 2 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad