High ISO = more highlight detail?

The place for questions about shooting with Blackmagic Cameras.
  • Author
  • Message
Offline

Benjamin de Menil

  • Posts: 330
  • Joined: Wed Feb 04, 2015 9:04 pm

High ISO = more highlight detail?

PostThu Aug 25, 2016 1:48 pm

Is there a case for shooting at high iso in bright sunlight?

I came across a post from a Digital Bolex tech that states that "Shooting at a higher ISO means you are feeding the sensor less light. Which in turn means you have more highlight range before clipping, but this also means less shadow range before hitting the noise floor."

and similarly "When you expose for a lower ISO you feed the sensor more light. Therefore you get more detail in the shadows, but you end up losing highlight range. More light means your sensor is more likely to clip in the highlights"

It seems to me counterintuitive that one would get more detail in the highlights at a higher iSO - if the camera is set to the same exposure level. Anyone care to weigh in on this?

ps - the post was related to a firmware revision that "Switching the ISO now shifts the dynamic range of the D16 around the middle gray point. In raw at 800 ISO, Bolex Log can retain up to 6 stops of dynamic range above middle gray and 6+ stops below. Reducing ISO lowers your highlight range, but increases your shadow range and reduces the appearance of noise"
--- is this shifting on the mid point also true for blackmagic cameras?
Offline
User avatar

Valentin Remy

  • Posts: 282
  • Joined: Thu Mar 24, 2016 3:52 pm
  • Location: Belgium

Re: High ISO = more highlight detail?

PostThu Aug 25, 2016 1:58 pm

I read somewhere that indeed, with the 4.6k, you get more dynamic range in highlights in higher ISO, and more in shadows with lower ISO.

Can't remember where I saw that :/ It was a graphic illustration.
http://www.instagram.com/valentinremy.be/
Offline
User avatar

rick.lang

  • Posts: 17156
  • Joined: Wed Aug 22, 2012 5:41 pm
  • Location: Victoria BC Canada

High ISO = more highlight detail?

PostThu Aug 25, 2016 4:00 pm

It's true. You gain a stop in highlights and lose a stop in shadows. I think CaptainHook mentioned ISO 1600 shifts middle grey a stop and that's the result you see above and below middle grey.

I've posted on this previously. If your frame or the most important parts you want are areas that are brightly lit, you can improve your detail in the brightest areas by shooting with ISO 1600 on the 4.6K sensor. May be true of all BMD sensors, but I think the previous conversation about this was with regard to 4.6K.


Sent from my iPad using Tapatalk
Rick Lang
Offline
User avatar

PaulDelVecchio

  • Posts: 799
  • Joined: Wed Aug 22, 2012 5:33 am
  • Location: NY

Re: High ISO = more highlight detail?

PostThu Aug 25, 2016 4:15 pm

rick.lang wrote:It's true. You gain a stop in highlights and lose a stop in shadows. I think CaptainHook mentioned ISO 1600 shifts middle grey a stop and that's the result you see above and below middle grey.

I've posted on this previously. If your frame or the most important parts you want are areas that are brightly lit, you can improve your detail in the brightest areas by shooting with ISO 1600 on the 4.6K sensor. May be true of all BMD sensors, but I think the previous conversation about this was with regard to 4.6K.


Sent from my iPad using Tapatalk


Rick, how is ISO 1600 on your camera when exposed properly? Is it still noisy, like the 1600 on the BMCC or if properly exposed, it looks fine?

To me, I don't mind the "grain," it's the sensor pattern noise that bothers me.
Paul Del Vecchio - Director/Producer
http://www.pauldv.net
http://www.youtube.com/user/pdelvecchio814
http://www.facebook.com/pauldv
http://instagram.com/pdelv
Twitter: @pauldv
Offline
User avatar

rick.lang

  • Posts: 17156
  • Joined: Wed Aug 22, 2012 5:41 pm
  • Location: Victoria BC Canada

High ISO = more highlight detail?

PostThu Aug 25, 2016 8:11 pm

The rigorous tests I have were pre-4.0beta. I can shoot something that will illustrate the difference of each ISO for fun. Not a motion picture show as I'll vary the shutter angle to compensate for the changes in ISO.

I assume you want this in raw, but I may as well do all codecs and flavours as a 2K 16:9 window since I have the Fujinon B4 mount on the camera at this time.

Edit: in the interests of time, I shot only uncompressed raw and ProRes 444 XQ, both at ISO 200, 400, 800, 1600. The only real difference seems to be going from ISO 800 to 1600 so I scrapped the 200 and 400 clips from the little video I'm assembling. The short examines noise in a well lit scene and shooting with the lens cap on to examine noise in ProRes and raw... Wait for it.


Sent from my iPad using Tapatalk
Last edited by rick.lang on Fri Aug 26, 2016 3:22 am, edited 1 time in total.
Rick Lang
Offline

Benjamin de Menil

  • Posts: 330
  • Joined: Wed Feb 04, 2015 9:04 pm

Re: High ISO = more highlight detail?

PostThu Aug 25, 2016 9:14 pm

This is interesting news as it would imply sometimes doing the opposite of the basic tenant - brighter scene = lower iso. I would love to know if this is the case for BMPC and BMMCC.
Offline

Denny Smith

  • Posts: 13131
  • Joined: Thu Aug 01, 2013 4:19 pm
  • Location: USA, Northern Calif.

Re: High ISO = more highlight detail?

PostThu Aug 25, 2016 9:35 pm

Yes, this includes the BMCC, Pocket camera and Micro cameras, as video cameras in General. Increasing ISO on BM cameras is like increasing gain on video cameras, reducing ISO is decreasing the gain, making the sensor less sensitive or decreasing the light hitting it. Adjusting the ISO/gain does shift the mid gray level. Which is why the highlight detail and shadow detail also shift. You are moving your 12-stops of exposure range up or down the scale, kind of like Ansel Adam's Zone exposure system.
Cheers
Denny Smith
SHA Productions
Offline
User avatar

rick.lang

  • Posts: 17156
  • Joined: Wed Aug 22, 2012 5:41 pm
  • Location: Victoria BC Canada

Re: High ISO = more highlight detail?

PostFri Aug 26, 2016 10:07 pm

Take a look at a little video posted in the magenta thread that illustrates changes going from IOS 800 to 1600 in a normal bright scene and with the lens cap on.


Sent from my iPad using Tapatalk
Rick Lang
Offline
User avatar

PaulDelVecchio

  • Posts: 799
  • Joined: Wed Aug 22, 2012 5:33 am
  • Location: NY

Re: High ISO = more highlight detail?

PostSat Aug 27, 2016 3:00 am

Here's a very unscientific test. These were taken within moments of each other.

The first is ISO 800, Leica R 35mm @ f2:

Image

This is ISO 1600 , Leica R 35mm @ f2.8:

Image

What I did was set the blacks to 0 on both and in the RAW tab, I set the Highlights to -50 for both and SATURATION to +30 for both. There's a 1/4 Hollywood Blackmagic in play on both as well.

1600 clearly has more detail. Again, unscientific, as I JUST got the camera in today so doing quick shooting tests to check highlight detail, noise at different ISOs in different conditions, etc.

1600 is definitely noisier, but it's clean still! I usually end up adding grain anyway so the grain at 1600 is FAR less than what I add using Film Convert so it gets covered up. It's really not an issue either way. Also, yes, there's sensor noise at 1600 when shooting in low light, but you can EASILY crush it out and the image still looks great. I'm talking LOW light that you shouldn't be shooting in if you want a good picture. I'll post more samples later.

One other thing.... I PREFER (at the moment) shooting at ISO 1600. The image looks smoother in the highlights and overall. The rolloff is really nice at 1600. That's going to suck for outdoor shooting. I'm gonna need some stronger ND...
Last edited by PaulDelVecchio on Tue Aug 30, 2016 4:18 am, edited 2 times in total.
Paul Del Vecchio - Director/Producer
http://www.pauldv.net
http://www.youtube.com/user/pdelvecchio814
http://www.facebook.com/pauldv
http://instagram.com/pdelv
Twitter: @pauldv
Offline
User avatar

rick.lang

  • Posts: 17156
  • Joined: Wed Aug 22, 2012 5:41 pm
  • Location: Victoria BC Canada

High ISO = more highlight detail?

PostSat Aug 27, 2016 3:18 am

Good for you, Paul, putting the URSA Mini 4.6K to work the day it arrives. Agree, the highlights are better. He difference in your shot illustrates that more clearly than my shot where the top of the fence was the highlight. But in your shot and my little video, I can also see the colour is milkier at 1600 and gets creamier as you move down from 1600 to 200.


Sent from my iPad using Tapatalk
Rick Lang
Offline
User avatar

PaulDelVecchio

  • Posts: 799
  • Joined: Wed Aug 22, 2012 5:33 am
  • Location: NY

Re: High ISO = more highlight detail?

PostSat Aug 27, 2016 4:11 am

rick.lang wrote:Good for you, Paul, putting the URSA Mini 4.6K to work the day it arrives. Agree, the highlights are better. He difference in your shot illustrates that more clearly than my shot where the top of the fence was the highlight. But in your shot and my little video, I can also see the colour is milkier at 1600 and gets creamier as you move down from 1600 to 200.


Sent from my iPad using Tapatalk


I kinda like that in a way. The rolloff into the shadows seems nicer too. It's not as "contrasty." Although, if I throw on an UltraContrast 2 on there or a LowCon 1, I might think the contrast at 1600 is not enough/too milky.
Paul Del Vecchio - Director/Producer
http://www.pauldv.net
http://www.youtube.com/user/pdelvecchio814
http://www.facebook.com/pauldv
http://instagram.com/pdelv
Twitter: @pauldv
Offline

Denny Smith

  • Posts: 13131
  • Joined: Thu Aug 01, 2013 4:19 pm
  • Location: USA, Northern Calif.

Re: High ISO = more highlight detail?

PostSat Aug 27, 2016 4:34 am

At 1600 Paul, the highlights do have a nice look. Good test, thanks.
Cheers
Last edited by Denny Smith on Sun Aug 28, 2016 12:51 am, edited 1 time in total.
Denny Smith
SHA Productions
Offline
User avatar

rick.lang

  • Posts: 17156
  • Joined: Wed Aug 22, 2012 5:41 pm
  • Location: Victoria BC Canada

Re: High ISO = more highlight detail?

PostSat Aug 27, 2016 10:20 pm

I have another video ready to post in this thread later today that compares raw and ProRes 444 at all ISO and a range for exposures from ETTR to slightly underexposed. Should be informative.
Rick Lang
Offline
User avatar

timbutt2

  • Posts: 2827
  • Joined: Tue Nov 12, 2013 10:32 am
  • Location: St. Petersburg, Florida, United States of Amercia

Re: High ISO = more highlight detail?

PostSat Aug 27, 2016 11:25 pm

As much highlight detail is recoverable in 1600 ISO, I would still avoid it as much as possible due to noise. It's simply a matter of fact that the higher the ISO the more noise you introduce. Even when shooting in bright sunlight.

A lot of tests I did prove that 1600 recovered the most highlight details. However, 800 is still extremely good. Above all else however it's going to require the motivation of what you're shooting to dictate how you shoot the scene. If you need all that highlight detail then it makes sense to shoot with the intention of recovering it. Sometimes you have to let it blow out. In which case you're looking for the nicest highlight rolloff imaginable.
"I'm well trained in the art of turning **** to gold." - Tim Buttner (timbutt2)

Cameras: URSA Mini Pro G2 & Pocket 6K Pro
Past: UM4.6K, P6K, BMCC 2.5K
Computers: iMac 5K (Mid 2020) & MacBook Pro Retina 15.4in (Mid 2018)
Offline
User avatar

rick.lang

  • Posts: 17156
  • Joined: Wed Aug 22, 2012 5:41 pm
  • Location: Victoria BC Canada

Re: High ISO = more highlight detail?

PostSun Aug 28, 2016 4:47 am

I finished this video that illustrates what Tim is saying, but it was too big for me to upload to Vimeo. I cut it back as much as possible but will need to wait until next week to upload since I'm not yet a paying member of Vimeo. I think I need Vimeo Plus.


Sent from my iPad using Tapatalk
Rick Lang
Offline
User avatar

Jamie LeJeune

  • Posts: 2012
  • Joined: Tue Feb 26, 2013 4:33 am
  • Location: San Francisco

Re: High ISO = more highlight detail?

PostSun Aug 28, 2016 5:49 am

rick.lang wrote:I finished this video that illustrates what Tim is saying, but it was too big for me to upload to Vimeo. I cut it back as much as possible but will need to wait until next week to upload since I'm not yet a paying member of Vimeo. I think I need Vimeo Plus.


Sent from my iPad using Tapatalk


I use Google Drive as a free alternative whenever I'm over my quota on Vimeo. Might be an alternative if you're eager to share your latest test. Or perhaps you enjoy keeping us all in suspense : )
www.cinedocs.com
http://www.imdb.com/name/nm4601572/
Offline
User avatar

rick.lang

  • Posts: 17156
  • Joined: Wed Aug 22, 2012 5:41 pm
  • Location: Victoria BC Canada

Re: High ISO = more highlight detail?

PostSun Aug 28, 2016 5:52 am

I use DropBox for stills, but it's better to use Vimeo for its ease of viewing for everyone, I think. Sorry, Jamie. Don't lose any sleep over this!


Sent from my iPad using Tapatalk
Rick Lang
Offline
User avatar

Troy Turner

  • Posts: 95
  • Joined: Sat Nov 02, 2013 1:03 pm

Re: High ISO = more highlight detail?

PostSun Aug 28, 2016 9:07 am

PaulDelVecchio wrote:Here's a very unscientific test. These were taken within moments of each other.

The first is ISO 800, Leica R 35mm @ f2:

Image

This is ISO 1600 , Leica R 35mm @ f2:

Image

What I did was set the blacks to 0 on both and in the RAW tab, I set the Highlights to -50 for both and SATURATION to +30 for both. There's a 1/4 Hollywood Blackmagic in play on both as well.

1600 clearly has more detail. Again, unscientific, as I JUST got the camera in today so doing quick shooting tests to check highlight detail, noise at different ISOs in different conditions, etc.

1600 is definitely noisier, but it's clean still! I usually end up adding grain anyway so the grain at 1600 is FAR less than what I add using Film Convert so it gets covered up. It's really not an issue either way. Also, yes, there's sensor noise at 1600 when shooting in low light, but you can EASILY crush it out and the image still looks great. I'm talking LOW light that you shouldn't be shooting in if you want a good picture. I'll post more samples later.

One other thing.... I PREFER (at the moment) shooting at ISO 1600. The image looks smoother in the highlights and overall. The rolloff is really nice at 1600. That's going to suck for outdoor shooting. I'm gonna need some stronger ND...


That is really interesting, thanks for posting.
Offline
User avatar

Robert Niessner

  • Posts: 4924
  • Joined: Thu Feb 21, 2013 9:51 am
  • Location: Graz, Austria

Re: High ISO = more highlight detail?

PostSun Aug 28, 2016 11:23 am

PaulDelVecchio wrote:Here's a very unscientific test. These were taken within moments of each other.

The first is ISO 800, Leica R 35mm @ f2:

Image

This is ISO 1600 , Leica R 35mm @ f2:

Image

What I did was set the blacks to 0 on both and in the RAW tab, I set the Highlights to -50 for both and SATURATION to +30 for both. There's a 1/4 Hollywood Blackmagic in play on both as well.

1600 clearly has more detail. Again, unscientific, as I JUST got the camera in today so doing quick shooting tests to check highlight detail, noise at different ISOs in different conditions, etc.

1600 is definitely noisier, but it's clean still! I usually end up adding grain anyway so the grain at 1600 is FAR less than what I add using Film Convert so it gets covered up. It's really not an issue either way. Also, yes, there's sensor noise at 1600 when shooting in low light, but you can EASILY crush it out and the image still looks great. I'm talking LOW light that you shouldn't be shooting in if you want a good picture. I'll post more samples later.


If you have shot those examples in RAW, then what you have written does not make sense. In RAW the ISO settings are just metadata. If the highlights are digitally clipped you won't be able to recover neither with ISO1600 nor ISO800. If the highlights are NOT digitally clipped then you can restore them in either ISO setting. So if those two shots have the same camera settings with ISO the only difference, then the RAW file is identically.

When recording ProRes or DNxHD this is of course different.
Saying "Thx for help!" is not a crime.
--------------------------------
Robert Niessner
LAUFBILDkommission
Graz / Austria
--------------------------------
Blackmagic Camera Blog (German):
http://laufbildkommission.wordpress.com

Read the blog in English via Google Translate:
http://tinyurl.com/pjf6a3m
Offline
User avatar

rick.lang

  • Posts: 17156
  • Joined: Wed Aug 22, 2012 5:41 pm
  • Location: Victoria BC Canada

Re: High ISO = more highlight detail?

PostSun Aug 28, 2016 3:28 pm

Robert, the examples are raw with two different settings for ISO. Since Paul didn't say how he made the image appear to have the same exposure, he only has two choices, either he added ND to the ISO 1600 shot or he reduced the shutter angle when setting the ISO 1600 shot. We know the light in the scene did not vary and the aperture did not vary and his treatment in post was the same. So the light falling on the sensor is 50% less for the ISO 1600 shot. BMD have always applied a different curve in camera to the dual 11bit signals they receive from the sensor when they apply the 12bit log to the camera output. The intent of that is to give a smoother falloff of highlights when shooting ISO 1600 which Paul was illustrating.

That's my understanding of the physics and optics. Temperature and tint are true metadata that don't change the sensor data values in the raw 12bit DNG file, but changes in ISO are not metadata and 1600 applies a different curve that can make your data different in camera and on the media. In post you can chose to change the raw exposure if you wish but that doesn't make it metadata only.

Hope I haven't misinterpreted anything you've stated and of course I hope my understanding is accurate!


Sent from my iPad using Tapatalk
Rick Lang
Offline

Simon McLean

  • Posts: 141
  • Joined: Sat Jan 05, 2013 9:39 pm
  • Location: San Francisco

Re: High ISO = more highlight detail?

PostSun Aug 28, 2016 5:06 pm

So, am I understanding this right? Even with the BMCC, shooting in raw @ 1600, we will get 1 more stop above neutral grey for the highlights? For the BMCC that would mean aprox 6 stops over and 6 stops under. Wow! And is the only downside is a little bit more noise in a well lit image? cheers.
Offline
User avatar

timbutt2

  • Posts: 2827
  • Joined: Tue Nov 12, 2013 10:32 am
  • Location: St. Petersburg, Florida, United States of Amercia

Re: High ISO = more highlight detail?

PostSun Aug 28, 2016 6:51 pm

IMG_2372.jpg
ARRI Alexa ISO Dynamic Range
IMG_2372.jpg (83.9 KiB) Viewed 24478 times

Here, use the above image that ARRI put out in regards to their Alexa cameras as an example of what is happening with the Blackmagic Cameras since they do the exact same thing with their ISO settings. As you'll see from the image the ISO shifts exposure for highlights and shadows around middle grey. So you're retaining the full dynamic range of the sensor, but a higher ISO will yield more stops in the highlights above middle grey than stops of light in the shadows below middle grey.

I hope this picture is helpful. I wish someone from Blackmagic could put out their own photo that is accurate to their sensors. With the 2.5K, 4K, and 4.6K the dynamic range shift is going to be different because each has a different dynamic range. Also, the 4.6K may be close to the ARRI Alexa sensor in theory, but the dynamic range shift could be different when it comes to ISO. So don't take this picture as an accurate portrayal of what is happening with your BMD cameras. It's an approximation of what is happening.
"I'm well trained in the art of turning **** to gold." - Tim Buttner (timbutt2)

Cameras: URSA Mini Pro G2 & Pocket 6K Pro
Past: UM4.6K, P6K, BMCC 2.5K
Computers: iMac 5K (Mid 2020) & MacBook Pro Retina 15.4in (Mid 2018)
Offline

Simon McLean

  • Posts: 141
  • Joined: Sat Jan 05, 2013 9:39 pm
  • Location: San Francisco

Re: High ISO = more highlight detail?

PostSun Aug 28, 2016 7:29 pm

Thanks for posting that Tim. A very helpful chart, and explanation. The only knock I've ever had against the 2.5K image has been it's slightly steep highlight roll-off. Judging by Paul's images from his UM4.6K, one more stop makes a notable and lovely difference. I'm exited to do some testing myself.
Offline
User avatar

PaulDelVecchio

  • Posts: 799
  • Joined: Wed Aug 22, 2012 5:33 am
  • Location: NY

Re: High ISO = more highlight detail?

PostSun Aug 28, 2016 8:03 pm

timbutt2 wrote:
IMG_2372.jpg

Here, use the above image that ARRI put out in regards to their Alexa cameras as an example of what is happening with the Blackmagic Cameras since they do the exact same thing with their ISO settings. As you'll see from the image the ISO shifts exposure for highlights and shadows around middle grey. So you're retaining the full dynamic range of the sensor, but a higher ISO will yield more stops in the highlights above middle grey than stops of light in the shadows below middle grey.

I hope this picture is helpful. I wish someone from Blackmagic could put out their own photo that is accurate to their sensors. With the 2.5K, 4K, and 4.6K the dynamic range shift is going to be different because each has a different dynamic range. Also, the 4.6K may be close to the ARRI Alexa sensor in theory, but the dynamic range shift could be different when it comes to ISO. So don't take this picture as an accurate portrayal of what is happening with your BMD cameras. It's an approximation of what is happening.


This.

Some will have more success than others. I've only spent less than 1/2 a day with the camera since I'm away this weekend, so I'll have to test 1600 more, but from my quick tests, 1600 seems ok if exposed properly. I've gotten some pattern noise in darker areas in some shots, and normal "grain" in others without pattern noise. I'll have to test more but I probably would feel ok shooting 1600 when properly exposed. The pattern noise was minimal and pretty much easy to hide.
Paul Del Vecchio - Director/Producer
http://www.pauldv.net
http://www.youtube.com/user/pdelvecchio814
http://www.facebook.com/pauldv
http://instagram.com/pdelv
Twitter: @pauldv
Offline
User avatar

timbutt2

  • Posts: 2827
  • Joined: Tue Nov 12, 2013 10:32 am
  • Location: St. Petersburg, Florida, United States of Amercia

Re: High ISO = more highlight detail?

PostSun Aug 28, 2016 8:21 pm

PaulDelVecchio wrote:
timbutt2 wrote:
IMG_2372.jpg

Here, use the above image that ARRI put out in regards to their Alexa cameras as an example of what is happening with the Blackmagic Cameras since they do the exact same thing with their ISO settings. As you'll see from the image the ISO shifts exposure for highlights and shadows around middle grey. So you're retaining the full dynamic range of the sensor, but a higher ISO will yield more stops in the highlights above middle grey than stops of light in the shadows below middle grey.

I hope this picture is helpful. I wish someone from Blackmagic could put out their own photo that is accurate to their sensors. With the 2.5K, 4K, and 4.6K the dynamic range shift is going to be different because each has a different dynamic range. Also, the 4.6K may be close to the ARRI Alexa sensor in theory, but the dynamic range shift could be different when it comes to ISO. So don't take this picture as an accurate portrayal of what is happening with your BMD cameras. It's an approximation of what is happening.


This.

Some will have more success than others. I've only spent less than 1/2 a day with the camera since I'm away this weekend, so I'll have to test 1600 more, but from my quick tests, 1600 seems ok if exposed properly. I've gotten some pattern noise in darker areas in some shots, and normal "grain" in others without pattern noise. I'll have to test more but I probably would feel ok shooting 1600 when properly exposed. The pattern noise was minimal and pretty much easy to hide.

Paul,

Yes, a lot of tests I've done internally have proven that I would feel comfortable shooting 1600 ISO outdoors in bright sunlight with strong ND 1.5 or higher in front of my lens in order to capture the most highlight detail. Most of the time I barely see any noise, although it is there. As for pattern noise, not really unless the image is way too underexposed where it's plain and simple bad shooting conditions because I need to light the image.

If I'm shooting at a lower ISO (like 200) it's because I know I'm going to expose for those highlights to not blow out and I'm planning on capturing far more shadow details. One of the tests I did found that I could bring my Shadows in the RAW tab up to 100 and not introduce noise because at that lower ISO I reduce noise. Thus, my tests proved to me that shadow details are more recoverable at a lower ISO than at a higher ISO.
"I'm well trained in the art of turning **** to gold." - Tim Buttner (timbutt2)

Cameras: URSA Mini Pro G2 & Pocket 6K Pro
Past: UM4.6K, P6K, BMCC 2.5K
Computers: iMac 5K (Mid 2020) & MacBook Pro Retina 15.4in (Mid 2018)
Offline
User avatar

PaulDelVecchio

  • Posts: 799
  • Joined: Wed Aug 22, 2012 5:33 am
  • Location: NY

Re: High ISO = more highlight detail?

PostSun Aug 28, 2016 8:56 pm

timbutt2 wrote:Paul,

Yes, a lot of tests I've done internally have proven that I would feel comfortable shooting 1600 ISO outdoors in bright sunlight with strong ND 1.5 or higher in front of my lens in order to capture the most highlight detail. Most of the time I barely see any noise, although it is there. As for pattern noise, not really unless the image is way too underexposed where it's plain and simple bad shooting conditions because I need to light the image.

If I'm shooting at a lower ISO (like 200) it's because I know I'm going to expose for those highlights to not blow out and I'm planning on capturing far more shadow details. One of the tests I did found that I could bring my Shadows in the RAW tab up to 100 and not introduce noise because at that lower ISO I reduce noise. Thus, my tests proved to me that shadow details are more recoverable at a lower ISO than at a higher ISO.



Yes exactly. I did a test at 400 ISO in dark conditions. 400-1600 ISO actually. While, obviously, I had to apply a more aggressive curves adjustment to get detail out of the dark areas, they actually had less noise at 400 than at 1600, so I'm seeing the same thing. At lower ISOs, more bits are allocated to the bottom IRE values and at higher ISOs, more bits area allocated to the top ISOs from my understanding, so that explains why we're seeing these results.

I'm going to do more tests this week but I'm loving what I'm seeing out of this camera.
Paul Del Vecchio - Director/Producer
http://www.pauldv.net
http://www.youtube.com/user/pdelvecchio814
http://www.facebook.com/pauldv
http://instagram.com/pdelv
Twitter: @pauldv
Offline
User avatar

rick.lang

  • Posts: 17156
  • Joined: Wed Aug 22, 2012 5:41 pm
  • Location: Victoria BC Canada

High ISO = more highlight detail?

PostSun Aug 28, 2016 10:23 pm

Sounds like we're all doing the same kinds of tests. Hope I can upload my new tests to Vimeo on Monday.

That ARRI ISO chart may not apply to BMD's sensors in general or the URSA Mini 4.6 K sensor. We have been told the BMD sensors at ISO 1600 up an extra stop of highlights compared to ISO 800 at the cost of one stop less in shadows, as ARRI illustrates. But we've not been told it's the same balancing act at ISO 200 and 400 that ARRI illustrates. We've been told in the past, ISO 200 and 400 cost dynamic range but give more saturated colours.

Need to hear from CaptainHook on this is we can.


Sent from my iPad using Tapatalk
Rick Lang
Offline

Denny Smith

  • Posts: 13131
  • Joined: Thu Aug 01, 2013 4:19 pm
  • Location: USA, Northern Calif.

Re: High ISO = more highlight detail?

PostSun Aug 28, 2016 10:37 pm

Rick, that is why I like using the lower ISO values, especially when shooting outdoors, to get a better color saturation -- unless I am looking for a "flatter" less saturate look, of course.
Cheers
Denny Smith
SHA Productions
Offline
User avatar

Valentin Remy

  • Posts: 282
  • Joined: Thu Mar 24, 2016 3:52 pm
  • Location: Belgium

Re: High ISO = more highlight detail?

PostSun Aug 28, 2016 11:48 pm

PaulDelVecchio wrote:I did a test at 400 ISO in dark conditions. 400-1600 ISO actually. While, obviously, I had to apply a more aggressive curves adjustment to get detail out of the dark areas, they actually had less noise at 400 than at 1600, so I'm seeing the same thing.


Very interesting ! I have to shoot quite a lot in low light, I will do some tests before the next job :)

Thanks for pointing that out.
http://www.instagram.com/valentinremy.be/
Offline

John Brawley

  • Posts: 4264
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: Los Angeles California

Re: High ISO = more highlight detail?

PostSun Aug 28, 2016 11:58 pm

Some of the best "night" work I've seen was on "Zero Dark Thirty".

Most of that brilliant night work is shot at ISO 200. (Alexa)

JB
John Brawley ACS
Cinematographer
Currently - Los Angeles
Offline

William Eguienta

  • Posts: 138
  • Joined: Tue Sep 30, 2014 5:53 pm
  • Location: Montpellier, France

Re: High ISO = more highlight detail?

PostMon Aug 29, 2016 2:16 am

Interesting, i will try it today.
Oh wait, i can't because I don't have my ursa 4.6k, BMD still doesn't send camera to UK


www.kaiz3r.net
Offline
User avatar

PaulDelVecchio

  • Posts: 799
  • Joined: Wed Aug 22, 2012 5:33 am
  • Location: NY

Re: High ISO = more highlight detail?

PostMon Aug 29, 2016 5:41 am

William Eguienta wrote:Interesting, i will try it today.
Oh wait, i can't because I don't have my ursa 4.6k, BMD still doesn't send camera to UK


http://www.kaiz3r.net



I'll post some stuff then.
Paul Del Vecchio - Director/Producer
http://www.pauldv.net
http://www.youtube.com/user/pdelvecchio814
http://www.facebook.com/pauldv
http://instagram.com/pdelv
Twitter: @pauldv
Offline

Simon McLean

  • Posts: 141
  • Joined: Sat Jan 05, 2013 9:39 pm
  • Location: San Francisco

Re: High ISO = more highlight detail?

PostMon Aug 29, 2016 4:28 pm

Is shooting at ISO 1600 with a .3 ND filter not effectively the same as shooting @ ISO 800 and then lifting gamma in post? Just trying to understand this better.
Offline
User avatar

rick.lang

  • Posts: 17156
  • Joined: Wed Aug 22, 2012 5:41 pm
  • Location: Victoria BC Canada

Re: High ISO = more highlight detail?

PostMon Aug 29, 2016 6:48 pm

The camera doesn't know if you have ND 0.3 or ND 3.0. But it does know you selected ISO 1600 and will apply a different curve to the image than the ISO 800 shot.



Sent from my iPad using Tapatalk
Rick Lang
Offline

Simon McLean

  • Posts: 141
  • Joined: Sat Jan 05, 2013 9:39 pm
  • Location: San Francisco

Re: High ISO = more highlight detail?

PostMon Aug 29, 2016 7:03 pm

Hi Rick, Yes I get that, but wouldn't the end result look the same on the corrected image?
Offline
User avatar

rick.lang

  • Posts: 17156
  • Joined: Wed Aug 22, 2012 5:41 pm
  • Location: Victoria BC Canada

High ISO = more highlight detail?

PostMon Aug 29, 2016 8:29 pm

Not the same because the curve for 1600 is different than the curve for 800 with that extra stop in the highlights. If the curve is linear, you are correct, but it's curved differently so it makes a subtle difference that Paul illustrated so well.


Sent from my iPad using Tapatalk
Rick Lang
Offline
User avatar

timbutt2

  • Posts: 2827
  • Joined: Tue Nov 12, 2013 10:32 am
  • Location: St. Petersburg, Florida, United States of Amercia

Re: High ISO = more highlight detail?

PostTue Aug 30, 2016 1:51 am

Simon McLean wrote:Hi Rick, Yes I get that, but wouldn't the end result look the same on the corrected image?

Simon the only result that would remain the same is the depth of field due to the IRIS setting on the lens staying the same. ISO 800 T4 is the same as ISO 1600 T4 w/ ND 0.3 for example. And, as Rick said the curve is different so the highlights and shadows are going to have subtle differences.
"I'm well trained in the art of turning **** to gold." - Tim Buttner (timbutt2)

Cameras: URSA Mini Pro G2 & Pocket 6K Pro
Past: UM4.6K, P6K, BMCC 2.5K
Computers: iMac 5K (Mid 2020) & MacBook Pro Retina 15.4in (Mid 2018)
Offline

Simon McLean

  • Posts: 141
  • Joined: Sat Jan 05, 2013 9:39 pm
  • Location: San Francisco

Re: High ISO = more highlight detail?

PostTue Aug 30, 2016 3:09 am

Ok got it. Thanks guys.
Offline

Eddie Barton

  • Posts: 31
  • Joined: Fri Aug 24, 2012 3:48 am
  • Location: Fairfield, CA, USA

Re: High ISO = more highlight detail?

PostTue Aug 30, 2016 8:43 am

Hello, I'm the one that was quoted in the OP. I can make some EI shift charts specific to the BM cams as well if you'd all actually like to see that. It'd be based off of the current BMD Film (4K/4.6K) LUTs included with Resolve, and the manufacturer dynamic range ratings.

Also, unless they changed it from the last time I looked, there isn't a shoulder in the ISO 1600 log transform. So if you were to leave the highlights slider alone in the camera raw tab, you'd have a full stop of highlights being clipped by the curve since all of the BMD Film curves are tailored to their respective camera's native ISO. I believe the log transform in the 1600 ISO DNGs is just more for shadow detail retention, seeing as once you get down that far, you're underexposing the sensor around 4 stops, so those extra steps in the shadows are crucial when going from 16b to 12b.

And when looking at the 4.6K in raw, ISO 1600 will see around 6.7 stops in the highlights if I remember the numbers correctly. So you will see essentially what an Alexa can see at EI 500. However, the 4.6K curve is static and only allows up to 5.7 stops of range in the highlights, which means if Paul were to leave the highlights slider at 0 instead of -50, you'd likely see it clip exactly like the ISO 800 did.

Normally I would post the charts now, but it's 1:45 in the morning and I'm typing this from my phone. But I will get to it at some point tomorrow.

Sent from my VS986 using Tapatalk
Last edited by Eddie Barton on Wed Aug 31, 2016 12:49 am, edited 1 time in total.
Offline
User avatar

Troy Turner

  • Posts: 95
  • Joined: Sat Nov 02, 2013 1:03 pm

Re: High ISO = more highlight detail?

PostTue Aug 30, 2016 11:27 am

Eddie Barton wrote:Hello, I'm the one that was quoted in the OP. I can make some EI shift charts specific to the BM cams as well if you'd all actually like to see that. It'd be based off of the current BMD Film (4K/4.6K) LUTs included with Resolve, and the manufacturer dynamic range ratings.

Also, unless they changed it from the last time I looked, there isn't a shoulder in the ISO 1600 log transform. So if you were to leave the highlights slider alone in the camera raw tab, you'd have a full stop of highlights being clipped by the curve since all of the BMD Film curves are tailored to their respective camera's native ISO. I believe the log transform in the 1600 ISO DNGs is just more for shadow detail retention, seeing as once you get down that far, you're underexposing the sensor around 4 stops, so those extra steps in the shadows are crucial when going from 16b to 12b.

And when looking at the 4.6K in raw, ISO 1600 will see around 6.7 stops in the highlights if I remember the numbers correctly. So you will see essentially what an Alexa can see at EI 500. However, the 4.6K curve is static and only allows up to 5.7 stops of range in the highlights, which means if Paul were to leave the highlights slider at 0 instead of -50, you'd likely see it clip exactly like the ISO did.

Normally I would post the charts now, but it's 1:45 in the morning and I'm typing this from my phone. But I will get to it at some point tomorrow.

Sent from my VS986 using Tapatalk



Thanks, would love to see this
Offline

Simon McLean

  • Posts: 141
  • Joined: Sat Jan 05, 2013 9:39 pm
  • Location: San Francisco

Re: High ISO = more highlight detail?

PostTue Aug 30, 2016 3:54 pm

Hi Eddie, I've not upgraded to the UM4.6K yet, and am still using the BM2.5K. Love this camera. I'd love to see your charts for the 2.5K if you were going to post those. Thanks.
Offline

Eddie Barton

  • Posts: 31
  • Joined: Fri Aug 24, 2012 3:48 am
  • Location: Fairfield, CA, USA

Re: High ISO = more highlight detail?

PostTue Aug 30, 2016 7:01 pm

Okay, I'll post the characteristic curves and EI charts for all three sensors/curves. Unfortunately, I can't do gamut gamut mappings since I dont have matrices or primaries for the gamuts used in the new color spaces BM has developed. Just a few more hours as I'm still on campus.
Offline
User avatar

Douglas Bischoff

  • Posts: 45
  • Joined: Tue Nov 19, 2013 8:27 pm
  • Location: Richmond, VA

Re: High ISO = more highlight detail?

PostTue Aug 30, 2016 7:03 pm

Much appreciated!
Douglas Bischoff
Ceridwen Productions
Offline

Simon McLean

  • Posts: 141
  • Joined: Sat Jan 05, 2013 9:39 pm
  • Location: San Francisco

Re: High ISO = more highlight detail?

PostWed Aug 31, 2016 12:30 am

Yes, very much appreciated Eddie.
Offline

Eddie Barton

  • Posts: 31
  • Joined: Fri Aug 24, 2012 3:48 am
  • Location: Fairfield, CA, USA

Re: High ISO = more highlight detail?

PostWed Aug 31, 2016 12:43 am

Okay, first I'll start off with the linear, unbounded dynamic range of the cameras according to specifications given by BMD. All values are calculated from the various BMD Film LUTs that they ship with Resolve and ratings they advertise.

Original BMCC and Pocket
Image

Production 4K and URSA 4K variants
Image

The 4.6K sensor
Image

Couldn't remember if BM cameras went down to EI 100, so I left it there to be thorough.

You can see that if you have a 4.6K camera without FPN issues, you have a beast of a camera. You could push to 3200 without breaking a sweat and you'd still have a fairly clean image. There is a caveat to this range, however. The BMD Film curves were all designed to only retain all highlight information from a linear signal at the base EI. This means that when you go above the base EI, you lose dynamic range unless you use the "Highlights" slider. I don't know how it works for ProRes in camera, but I would venture to say there isn't a shoulder or highlights protection parameter present. This means that going above the camera's base ISO camera loses a full stop dynamic range. This is of course if BMD hasn't already thought of this and applied shoulders or a modified curve in camera. This isn't what I've seen with ISO 1600 DNGs from a Pocket in Resolve though, they lost that detail in the highlights.

So, here is the characteristic chart that shows the three BMD Film curves.

Image

Here is the data for all of the curves on a 10b scale:

BMD Film:
1% Black: 36
18% Gray: 392

BMD Film 4K:
1% Black: 36
18% Gray: 392

BMD Film 4.6K:
1% Black: 76
18% Gray: 420

And the linear input vs BMD Film signals

Image

I don't know how well the BM forum plays with linked images, but you can always right click and go to the higher-res images directly if need be.
Offline
User avatar

timbutt2

  • Posts: 2827
  • Joined: Tue Nov 12, 2013 10:32 am
  • Location: St. Petersburg, Florida, United States of Amercia

Re: High ISO = more highlight detail?

PostWed Aug 31, 2016 5:40 am

Thank you for sharing Eddie! Very fascinating to look at these.
"I'm well trained in the art of turning **** to gold." - Tim Buttner (timbutt2)

Cameras: URSA Mini Pro G2 & Pocket 6K Pro
Past: UM4.6K, P6K, BMCC 2.5K
Computers: iMac 5K (Mid 2020) & MacBook Pro Retina 15.4in (Mid 2018)
Offline

Benjamin de Menil

  • Posts: 330
  • Joined: Wed Feb 04, 2015 9:04 pm

Re: High ISO = more highlight detail?

PostWed Aug 31, 2016 12:45 pm

Thanks everyone for your contributions. The info in this thread will change how I approach exposure.

The highlight headroom at iso 100 is frighteningly low. I'm reticent to ever again shoot a sunny scene at below 400 iso. Or at least if I do I will be very careful about clipping.

Does film also behave like this?
Offline

Benjamin de Menil

  • Posts: 330
  • Joined: Wed Feb 04, 2015 9:04 pm

Re: High ISO = more highlight detail?

PostWed Aug 31, 2016 12:47 pm

Ps can someone confirm if blackmagic cameras clip a stop of highlights at higher than native iso?
Offline

Soeren Mueller

  • Posts: 604
  • Joined: Tue Sep 04, 2012 2:21 pm
  • Location: Düsseldorf, Germany

Re: High ISO = more highlight detail?

PostWed Aug 31, 2016 12:53 pm

I'm completely with Robert on this: How are they supposed to clip highlights when ISO is only metadata?

You only expose differently because of what you see on the monitor / the cameras "scopes", the sensor still "reads" exactly the same data. Yes, there is a slight (!) difference in the internal curve depending on which ISO you choose but it's not that significant to suddenly clip highlights.

The BM cameras don't apply analog gain or change the sensitivity of the sensor in any way when the ISO is changed in camera. (speaking of RAW)
Offline

Eddie Barton

  • Posts: 31
  • Joined: Fri Aug 24, 2012 3:48 am
  • Location: Fairfield, CA, USA

Re: High ISO = more highlight detail?

PostWed Aug 31, 2016 1:58 pm

Soeren Mueller wrote:I'm completely with Robert on this: How are they supposed to clip highlights when ISO is only metadata?

You only expose differently because of what you see on the monitor / the cameras "scopes", the sensor still "reads" exactly the same data. Yes, there is a slight (!) difference in the internal curve depending on which ISO you choose but it's not that significant to suddenly clip highlights.

The BM cameras don't apply analog gain or change the sensitivity of the sensor in any way when the ISO is changed in camera. (speaking of RAW)


Because, when you change the EI, you change how the image looks in terms of lightness. So, if you properly light for ISO 400 and the image has the correct lightness values but you notice your highlights are clipping, you either throw on ND or close down your aperture. This reduces the amount of light coming in, which is what preserves the highlights. This reduced signal is boosted digitally when you raise the EI to achieve the correct middle gray lightness. The only difference is that your highlights will still be intact in the raw file.

So, yeah, if you only changed the EI but didn't correct the exposure with NDs or aperture, then you'd have the exact same raw data at every EI. The point here is that we're exposing for an EI. So you change the light coming into the camera based on how much room you want in the highlights and shadows. It's like with film. The manufacturer rates a film stock 250 ISO or something. But the DP thinks it's too noisy and wants cleaner shadows, so they overexpose and pull the levels down after the fact. Except now, you don't have to perform the push or pull yourself, it's done for you.

Think of this as a sort of "calibrated ETTR/ETTL". When you expose to the right, you overexpose the image until clipping happens and pull down in post. The same effect is achieved when you lower the EI. The only difference is that the image will look like it's been correctly exposed on screen and you won't have to pull the exposure in post.
Next

Return to Cinematography

Who is online

Users browsing this forum: No registered users and 49 guests