quick clarification

[Jon Fairhurst]

Originally Posted by Mark Hahn

...I had my f-stop fixed at f4 in a dimly lit room. I observed that it stayed at ISO 3200 and changed the shutter speed as I dialed the exposure. The video sequence shot at +2 EV showed a shutter speed of 1/6 second, which is really 1/30, right? Then I dialed the exposure down to +0 EV and it showed 1/30 second, which is also 1/30, right?

But my two video sequences showed two different light levels that matched the exposure bias each time, both during recording and in playback. How is this possible if both were actually f4, ISO 3200, and 1/30? How and where was it lying to me? Could it be that it actually used a higher ISO?

I haven't done much at 1/30 and 3200 ISO, but I think you've got it exactly right.

The camera can't expose longer than 1/30 seconds with a 30 fps frame rate. If changing the exposure actually makes things brighter, there are two possibilities:

1) Though it says 1/6 and 3200 ISO, it's really 1/30 and 12800 ISO or something, or
2) It's faking the longer shutter speed by adding some amount of the previous frame to the current frame. That could lead to some nasty motion blur artifacts with seriously long tails. (In signal processing terms, this would be an IIR, or "infinite impulse response" filter. Part of the image from the very first frame could still be there in the final frame!)

My guess is that the ISO changes. I assume that you didn't see crazy long shutter times and motion trails.

[Mark Hahn]

Originally Posted by Jon Fairhurst

I haven't done much at 1/30 and 3200 ISO, but I think you've got it exactly right.

The camera can't expose longer than 1/30 seconds with a 30 fps frame rate. If changing the exposure actually makes things brighter, there are two possibilities:

1) Though it says 1/6 and 3200 ISO, it's really 1/30 and 12800 ISO or something, or
2) It's faking the longer shutter speed by adding some amount of the previous frame to the current frame. That could lead to some nasty motion blur artifacts with seriously long tails. (In signal processing terms, this would be an IIR, or "infinite impulse response" filter. Part of the image from the very first frame could still be there in the final frame!)

My guess is that the ISO changes. I assume that you didn't see crazy long shutter times and motion trails.

We need an equivalent of your turntable study to measure ISO. Unfortunately noise is hard to measure. Of course at 12,800 and 25,600 it may not be too hard. What would be the best background, gray, black?

So if it is ISO, then maybe the entire scale of brightness is:

1) Max ISO (?) to ISO 3200 with 1/30
2) ISO 3200 to ISO 100 with 1/40
3) 1/40 to Max speed (?) with ISO 100

Could it be this simple? So the key to shooting would be stay in the second range all the time. It would be easy to get repeatable F-stop and speed. The ISO doesn't matter as much for repeatability.

You said 1/40 is really 1/48, right? If this is true, then the motion blur would be identical to that from 24 fps with 180 degree shutter. Then when the video is converted to 24 fps you'd have the perfect blur. Could Canon have been thinking this when they used 1/48?

By the way, I saw a video, which I don't have the link to right now, where a guy used a variable ND filter and he would just turn the brightness up to get into the range where the speed was right. Then he would lock the exposure and turn the brightness back down smoothly watching the ISO. He got full control pretty easily. I would have to go back to see if he claimed to be able to control different speeds reliably. The singh-ray var ND covers 2 to 8 stops, which gives great control. Unfortunately it is $300.

[Jon Fairhurst]

Originally Posted by Mark Hahn

...maybe the entire scale of brightness is:

1) Max ISO (?) to ISO 3200 with 1/30
2) ISO 3200 to ISO 100 with 1/40
3) 1/40 to Max speed (?) with ISO 100

I think this could be right. Keep in mind that some are saying that 1/30 looks just like 1/40. I don't know. I didn't test it. But I wouldn't be surprised.

The ISO doesn't matter as much for repeatability.

True. For a given scene with a given lens, keep the aperture and shutter speeds identical. Adjust ISO, lighting and/or ND filters to optimize your exposure. There's an S-curve in the camera, so it's important to keep the exposure in the sweet spot, unless you want to spend time tweaking luma curves to get things to match.

You said 1/40 is really 1/48, right?

Something like that. I've been saying 1/45, since it's someplace between 1/40 and 1/50. You can pause the video and make your best estimate.

Regardless, the motion blur will be similar to film; however, with the faster frame rate, it's a bit too smooth.

I'd actually prefer 1/80 for 30 fps, to give a more shuttered look. It might be 30 fps, but if the shutter is fast enough, it should still be a bit dreamy. If you do the math, you'll find that 24 fps at 1/48 and 30 fps at 1/80 have the same amount of time that the shutter is closed.

1/24 - 1/48 = 1/48 = 0.208333
1/30 - 1/80 = 0.208333

But it's a moot point right now. We're shooting everything at 1/40 (displayed), so we can control ISO.

By the way, I saw a video, which I don't have the link to right now, where a guy used a variable ND filter and he would just turn the brightness up to get into the range where the speed was right. Then he would lock the exposure and turn the brightness back down smoothly watching the ISO. He got full control pretty easily.

If you find the link, post it. I'd love to be able to control the ISO at a consistent 1/80.

[Mark Hahn]

Originally Posted by Jon Fairhurst

I think this could be right. Keep in mind that some are saying that 1/30 looks just like 1/40. I don't know. I didn't test it. But I wouldn't be surprised.


True. For a given scene with a given lens, keep the aperture and shutter speeds identical. Adjust ISO, lighting and/or ND filters to optimize your exposure. There's an S-curve in the camera, so it's important to keep the exposure in the sweet spot, unless you want to spend time tweaking luma curves to get things to match.

Something like that. I've been saying 1/45, since it's someplace between 1/40 and 1/50. You can pause the video and make your best estimate.

Regardless, the motion blur will be similar to film; however, with the faster frame rate, it's a bit too smooth.

I'd actually prefer 1/80 for 30 fps, to give a more shuttered look. It might be 30 fps, but if the shutter is fast enough, it should still be a bit dreamy. If you do the math, you'll find that 24 fps at 1/48 and 30 fps at 1/80 have the same amount of time that the shutter is closed.

1/24 - 1/48 = 1/48 = 0.208333
1/30 - 1/80 = 0.208333

But it's a moot point right now. We're shooting everything at 1/40 (displayed), so we can control ISO.

If you find the link, post it. I'd love to be able to control the ISO at a consistent 1/80.

First, here is the link. It is not a video but it explains the process.

Andrew Yip Nature Photography Using a variable ND filter with the 5D Mark II’s video mode

Secondly, I shot some video at 1/8 and there were no trails. So that theory is out.

Thirdly, I managed to do the ISO tests. I varied the exposure and shot a dark blue leather couch. I saved frames as 200x150 jpegs at maximum quality and used photoshop to adjust brightness (not gain) so all had histograms in the middle. I accidentally found a way to measure noise. When saving the jpeg, the file size indicated the noise.

1/30 at 3200 (3200) 12,800 bytes
1/15 at 3200 (6400) 13,263 bytes (ISO definitely changed)
1/8 at 3200 (12800) 21,318 bytes
1/4 at 3200 (25600) 22,119 bytes
1/2 at 3200 (51200) 22,647 bytes

Here are the five 200x150 patches. Judge for yourself.

http://elleh.com/3200.jpg
http://elleh.com/6400.jpg
http://elleh.com/12800.jpg
http://elleh.com/25600.jpg
http://elleh.com/51200.jpg

I think between your turntable experiments and these noise experiments we have proven that the speed goes as high as it needs in bright light and the ISO goes as high as it needs in dark light.

When doing my experiments I proved to myself that I can shoot at any ISO at 1/40 repeatedly. Now I need to see if I can start at another speed and hold that speed. If our previous thinking is correct that won't be possible and it will always jump to 1/40 when changing ISO.

The method used in the first link above can get various speeds but he does it by not changing the exposure bias after the initial setting, but by changing the actual light using the var ND. Luckily I think Canon did it right for me and I will be happy with 1/40 and not have to pay $300 for var ND. I'm just going to pay $15 for gel NDs.

A guy on DPReview said to just hold gels in front and switch them. I don't think that will work because the light will jump too much and throw off the speed setting.

[Jon Fairhurst]

Nice test!

3200 ISO (actual) looks amazingly good. Above that, not so much. This shows that we should try to avoid shooting slower than 1/30 (displayed) by adding light, when possible.

A cheaper method of creating a variable ND filter is to get two polarizers, and rotate one. When they line up, the light goes through. When they don't, the light is blocked. (Kind of like the way an LCD operates.) Of course, good polarizers that don't lose light aren't cheap, but it's nice to have a good polarizer for deep blue sky shots, and to remove unwanted reflections anyway.

Again, nice work on the super-high ISO tests.

[Ozan Biron]

Originally Posted by Mark Hahn

First, here is the link. It is not a video but it explains the process.

Andrew Yip Nature Photography Using a variable ND filter with the 5D Mark II’s video mode

Secondly, I shot some video at 1/8 and there were no trails. So that theory is out.

Thirdly, I managed to do the ISO tests. I varied the exposure and shot a dark blue leather couch. I saved frames as 200x150 jpegs at maximum quality and used photoshop to adjust brightness (not gain) so all had histograms in the middle. I accidentally found a way to measure noise. When saving the jpeg, the file size indicated the noise.

1/30 at 3200 (3200) 12,800 bytes
1/15 at 3200 (6400) 13,263 bytes (ISO definitely changed)
1/8 at 3200 (12800) 21,318 bytes
1/4 at 3200 (25600) 22,119 bytes
1/2 at 3200 (51200) 22,647 bytes

Here are the five 200x150 patches. Judge for yourself.

http://elleh.com/3200.jpg
http://elleh.com/6400.jpg
http://elleh.com/12800.jpg
http://elleh.com/25600.jpg
http://elleh.com/51200.jpg

I think between your turntable experiments and these noise experiments we have proven that the speed goes as high as it needs in bright light and the ISO goes as high as it needs in dark light.

When doing my experiments I proved to myself that I can shoot at any ISO at 1/40 repeatedly. Now I need to see if I can start at another speed and hold that speed. If our previous thinking is correct that won't be possible and it will always jump to 1/40 when changing ISO.

The method used in the first link above can get various speeds but he does it by not changing the exposure bias after the initial setting, but by changing the actual light using the var ND. Luckily I think Canon did it right for me and I will be happy with 1/40 and not have to pay $300 for var ND. I'm just going to pay $15 for gel NDs.

A guy on DPReview said to just hold gels in front and switch them. I don't think that will work because the light will jump too much and throw off the speed setting.

Im curious how 1/40 shutter @ 30p will look at when i get it converted to 24p at the transfer house.

I remember watching the first markII video footage from vincent... Some night shots u can notice grain in the sky and surrounds. And other shots looked grain free. Is it possible that the grain free shots were at 3200 iso?


Completely off topic question... how long does the stock battery last on the markII?

[Toenis Liivamaegi]

I have found that "auto lightning optimizer" helps on 50hz flicker a bit.
I've been shooting in 1/40th 1/50/th and 1/60th and nothing flickers.
But then on the other day every light flicked and I turned the lightning optimizer back to "strong".

This 5D is sort of independent thinking machine anyways, it does do something it will not tell to the users.

Oh and enabling "Highlight priority" mode will give you much more noise.

Also if you use manual lens it is easy to get the shutter to 40 or 50 or 60 and lock it while holding down the shutter button half way. It will lock the shutter to 50/60th and you can dial down the ISO while the shutter speed is/stays 50 or 60 and not something down to 30th.

But when using EF compatible lens there is no chance to get the shutter up from 1/20th when using 19mm wide lenses for example and it will not come down from 1/80th or 1/100 when using 85mm lens - this is pure logic of photography shutter speeds vs lens focal length but why on earth did they implement it to a movie mode? As you might assume the EF 50mm lens is the only one that works alright out of the box.
Our beloved capitalism is ripping us right there. The slower the apparent development the higher the profits. As there is absolutely no competition to 5D MKII why invest even an hour to make it better ;)... World economy is in so good shape that no one wants to sell anything, right? Even better, Canon announced additional +10% to +20% to existing prices from Feb Mar. In the other hand the shorter you are and the narrower is your vision the more reduced your global outlook will be. (I just watched the horrible photos of Nagasaki ground zero again to hold myself back a little.)
If you want to give an example of how the interests of the shareholders are not pro-development and innovation you can quote me and every other 5D filmmaker here.

T

[Mark Hahn]

Originally Posted by Toenis Liivamaegi

But when using EF compatible lens there is no chance to get the shutter up from 1/20th when using 19mm wide lenses for example and it will not come down from 1/80th or 1/100 when using 85mm lens - this is pure logic of photography shutter speeds vs lens focal length but why on earth did they implement it to a movie mode?
T

It makes no sense. For EF lenses (which is all I have) I am assuming that locking the lens by turning or taping the contacts will be necessary to have any real control. So far I have not tried the lens unlocked.

[Mark Hahn]

Originally Posted by Ozan Biron

Im curious how 1/40 shutter @ 30p will look at when i get it converted to 24p at the transfer house.

I think it has a good chance of looking good, since the exposure time is correct for all intents and purposes.

[Ozan Biron]

New test with iso, shutter ad manual lens.... does anyone speak spanish?

Betty_Canon 5D Mark II_Control de ISO y Profundidad de Campo on Vimeo


does anyone speak spanish or can translate?

-

[Ozan Biron]

Jon, i finally did a bit of testing with the LED dimmer light.

Jst like 1/40 shutter any ISO from 100 to 3200... I was able to get 1/50 shutter and any ISO from 100-3200.

I set exposure wheel down to 2 clicks to the left from the middle. I then placed the LED in front of the lens and set it to about 15% brightness. The camera then reads as 1/50 and ISO 640. After locking it there im able to adjust the exposure wheel up and down from 100 to 3200 iso with a constant shutter of 1/50.

For my frame rate conversion from 30p to 24p would 1/50 shutter make any different compared to 1/40? In your tests 1/40 and 1/50 almost looked the same.

[Mathieu Kassovitz]

Originally Posted by Ozan Biron

New test with iso, shutter ad manual lens.... does anyone speak spanish?

Betty_Canon 5D Mark II_Control de ISO y Profundidad de Campo on Vimeo


does anyone speak spanish or can translate?

-

If you'll scroll down the page, you'll get the translation to english.

[Jon Fairhurst]

Originally Posted by Ozan Biron

I was able to get 1/50 shutter and any ISO from 100-3200.

Unfortunately, 1/40 and 1/50 (displayed) are identical, from my tests (roughly 1/45.) I wish there were a way to lock 1/60 or 1/80 and then dial in any ISO. I'd prefer the faster shutter speed for controlled shots.

[Ozan Biron]

Originally Posted by Jon Fairhurst

Unfortunately, 1/40 and 1/50 (displayed) are identical, from my tests (roughly 1/45.) I wish there were a way to lock 1/60 or 1/80 and then dial in any ISO. I'd prefer the faster shutter speed for controlled shots.

I havent yet captured any footage yet to test. Dont have any CF cards yet. I might also do a side by side comparison between 1/40 and 1/50 going threw all the ISOs.

[Jon Oskar]

Originally Posted by Ozan Biron

Jon, i finally did a bit of testing with the LED dimmer light.

Jst like 1/40 shutter any ISO from 100 to 3200... I was able to get 1/50 shutter and any ISO from 100-3200.

I set exposure wheel down to 2 clicks to the left from the middle. I then placed the LED in front of the lens and set it to about 15% brightness. The camera then reads as 1/50 and ISO 640. After locking it there im able to adjust the exposure wheel up and down from 100 to 3200 iso with a constant shutter of 1/50.

For my frame rate conversion from 30p to 24p would 1/50 shutter make any different compared to 1/40? In your tests 1/40 and 1/50 almost looked the same.

I tried with my lite panel and can confirm that in both 1/40 and 1/50 i can record at all iso settings between 100 - 3200. 1/40 seems easier to do repeatedly. I tried with a 50mm and 18mm nikon MF lens. Although i can set 1/60 at any iso, once i hit record the camera goes down to 1/40 or 1/50 at some random ISO. (i am doing this in AV/still+movie/exposure simulation)