Jump to content

Taming Blown Highlights


erniethemilk

Recommended Posts

Advertisement (gone after registration)

4 minutes ago, SrMi said:

Without clipping on the sensor, you could avoid clipping in the DNG by using an ISO low enough.

I'd like you to demonstrate that please. I don't believe any camera maker (except in cheap P&S) would nobble their own product by clipping the raw file above and beyond what is clipped in the sensor.

  • Like 1
Link to post
Share on other sites

28 minutes ago, SrMi said:

There is clipping on the sensor, and there is clipping in the DNG file

I'm not really understanding this part. There can no clipping on the sensor but the RAW app's handling of the RAW data gives the impression of clipping, specifically LR/ACR's 'hidden' native tone curve, whatever wacky bullshit adobe are doing with LUTs this process generation, whatever luminosity deltas adobe put into their DCP HSD tables for any particular camera, how ever the RAW app choses to work with the DNG maker notes - stuff like that. But that's more of a RAW app thing.

I think I'm miss-understanding you and you mean something different?

20 minutes ago, SrMi said:

When you see clipping in the DNG at higher ISOs, it does not mean that the sensels were saturated, but maybe the ISO "amplification" clipped the data.

 

I'm not really understanding this part either, AFAIK the DNG is created after the exposure parameters have been set and executed.. sure there's some interesting tags in DNG files, like "baseline exposure = -0.5" on the M9 which literally means that every M9 DNG you ever opened in an Adobe product has been darkened before you touched a slider (and also a tag in the M9 that amounts too anti-aliasing filter = YES

Like my amplifier analogy some posts back... you can amplify the music player until the limit of the speakers is reached and it sounds bad, but you didn't break the record - you just need to turn the volume down a bit.

Likewise if you don't clip at capture you didn't break anything by over/under amplification in post

Link to post
Share on other sites

Ok! So I just went to the bathroom!!

 

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

The first shot is 100% camera recommended, F2.4 1/60 and ISO400.

Shot 2 is F2.4 1/60 and ISO200

Shot 3 is F2.4 1/60 and ISO800

 

shot 1 and shot 2, with shot 2 receiving a 1 stop global brightness boost. (+1 ev)

 

 

shot 1 and shot 3, with shot 3 receiving a 1 stop global darkening. (-1ev)

 

Obviously this is quick and dirty (erm not that my bathroom isn't a clean place) but as we can see, the exposure was ok(ish - but under IMO) at 400 ISO, but under/over exposing it with ISO didn't make any big problems* to darken or brighten the image to match the exposure in the other two

 

I'm not sure why the reflected light on the sink isn't the same in each shot, I moved a bit between shots maybe!

But amusingly that reflection looks most obviously blown in the unedited iso200 shot!

 

*These days the problem is becoming that you could theoretically add 10 stops of push to a base ISO image from a modern camera, but the exposure slider only goes up to 5!

-------------

None of this is helping Dave (op) not clip stuff at ISO 200 !!!!

 

 

Link to post
Share on other sites

34 minutes ago, adan said:

I'd like you to demonstrate that please. I don't believe any camera maker (except in cheap P&S) would nobble their own product by clipping the raw file above and beyond what is clipped in the sensor.

Make two shots with the same exposure (same aperture and shutter speed), one at base ISO, so there is no clipping, and a second at ISO 12800. The ISO 12800 shot would likely cause a lot of clipping. Since the amount of light hitting the sensor is the same in both images, we know that the sensor's pixels have not been clipped in both shots. However, the DNG in ISO 12800 shot contains clipped data.

I include an illustration from here:

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

The red signal is determined by the exposure (shutter speed and aperture). Even though the red signal may not be clipped, the amplification can cause the blue signal to be clipped. 

Link to post
Share on other sites

27 minutes ago, Adam Bonn said:

I'm not really understanding this part. There can no clipping on the sensor but the RAW app's handling of the RAW data gives the impression of clipping, specifically LR/ACR's 'hidden' native tone curve, whatever wacky bullshit adobe are doing with LUTs this process generation, whatever luminosity deltas adobe put into their DCP HSD tables for any particular camera, how ever the RAW app choses to work with the DNG maker notes - stuff like that. But that's more of a RAW app thing.

I think I'm miss-understanding you and you mean something different?

 

I'm not really understanding this part either, AFAIK the DNG is created after the exposure parameters have been set and executed.. sure there's some interesting tags in DNG files, like "baseline exposure = -0.5" on the M9 which literally means that every M9 DNG you ever opened in an Adobe product has been darkened before you touched a slider (and also a tag in the M9 that amounts too anti-aliasing filter = YES

Like my amplifier analogy some posts back... you can amplify the music player until the limit of the speakers is reached and it sounds bad, but you didn't break the record - you just need to turn the volume down a bit.

Likewise if you don't clip at capture you didn't break anything by over/under amplification in post

The graph I included in response to Adan's question may clarify my point.

The amount of light hitting the sensor is determined by the aperture and the shutter speed (not ISO). The signal read from the sensor is proportional to the luminance from the scene. Assume that the selected exposure does not clip/saturate any of the sensor's pixels. The unclipped signal from the sensor is amplified and stored in the DNG. Too much amplification (too high ISO) will clip the data stored in the DNG.

In the amplifier analogy, you may turn your amplifier too high, which could lead to the distortions in the loudspeakers even though the source does not have distortions.

  • Like 1
Link to post
Share on other sites

1 hour ago, adan said:

I'd like you to demonstrate that please. I don't believe any camera maker (except in cheap P&S) would nobble their own product by clipping the raw file above and beyond what is clipped in the sensor.

Here are two images shot with the same exposure, one with ISO 64 and the other with ISO 3200. Both shots have no clipping in the sensor. However, the ISO 3200 DNG contains a lot of clipping.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

 

 

 

Link to post
Share on other sites

Advertisement (gone after registration)

4 minutes ago, SrMi said:

Here are two images shot with the same exposure, one with ISO 64 and the other with ISO 3200. Both shots have no clipping in the sensor. However, the ISO 3200 DNG contains a lot of clipping.

if you made that test with ISO 64 and whatever A/SS then repeated it at ISO 3200 and the same A/SS, would you expect to be able recover all the data to make the 3200 look like the 64 one?

(I think yes... )

Link to post
Share on other sites

13 minutes ago, Adam Bonn said:

(I think yes... )

Ok I just tried with with my M10 and no I'm wrong... I can't recover F4 1/500 3200 to the same image as F4 1/500 @ ISO200 (not even close to be honest)

edit

however RAW digger tells me that I've clipped a lot of data in the 3200 shot, which I take to mean as clipped sensor data

Edited by Adam Bonn
Link to post
Share on other sites

24 minutes ago, Adam Bonn said:

if you made that test with ISO 64 and whatever A/SS then repeated it at ISO 3200 and the same A/SS, would you expect to be able recover all the data to make the 3200 look like the 64 one?

(I think yes... )

No, the data has been lost at ISO 3200. I am talking about "taming the highlights" during the exposure, not after the exposure.

Edited by SrMi
  • Like 1
Link to post
Share on other sites

7 minutes ago, Adam Bonn said:

Ok I just tried with with my M10 and no I'm wrong... I can't recover F4 1/500 3200 to the same image as F4 1/500 @ ISO200 (not even close to be honest)

edit

however RAW digger tells me that I've clipped a lot of data in the 3200 shot, which I take to mean as clipped sensor data

If shooting above base ISO, clipped data in the DNG does not mean that that the data was clipped in the sensor.

 

Link to post
Share on other sites

13 minutes ago, SrMi said:

If shooting above base ISO, clipped data in the DNG does not mean that that the data was clipped in the sensor.

 

you mean that the digital post ADC push (to native ISO) is what is clipping the data?

but the data is lost because the sensor can only create a RAW from the given exposure parameters?

edit:

Sorry Srdjan, I just saw your pictorial explanation in #44, I must have missed it by clicking on the SrMi has quoted you notification... I think you answered this question already and it's yes!

Edited by Adam Bonn
Link to post
Share on other sites

23 minutes ago, Adam Bonn said:

you mean that the digital post ADC push (to native ISO) is what is clipping the data?

but the data is lost because the sensor can only create a RAW from the given exposure parameters?

edit:

Sorry Srdjan, I just saw your pictorial explanation in #44, I must have missed it by clicking on the SrMi has quoted you notification... I think you answered this question already and it's yes!

Yes, a picture is worth a thousand words 😁.

My point is that sometimes you can avoid clipping during image exposure by reducing the amplification in the camera (i.e., use lower ISO).

If the DNG is clipped, you can only pray that highlight recovery in post can reconstruct the missing data. IMO, that works up to 1EV, but it depends on the image.

 

  • Thanks 1
Link to post
Share on other sites

  • 1 year later...

Hey gang.. Leica fan but Fuji user here....  

Bring this up to Leica as fuji has a unique way to deal with blown highlights, see the image below.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!



It requires you to shoot at ISO 640 for 4 stops of high light recover, base iso is 160.  What the camera does is expose the mids and shadows at iso 640, but the highlights at iso 160, giving you incredible highlight recovery. This all happens internally in the camera, in the raw file, you just shoot an image, exposing for iso 640. 

In addition to the extra stops of highlight detail, you can control the falloff of the highlights:

This offers a very filmic falloff for the highlights.  With the extra sensor information in the form of 4 stops of highlight detail, you avoid these blobs of clipped highlights. 

This means you can avoid underexposing your Leica images in camera, effectively taking the image at the WRONG exposure, showing dark frames on your LCD... not good if you are doing client work. you rely on the COMPUTER inside your camera to expose you highlights with enough headroom to allow for a natural falloff indeed of a digital clip. 

I never understood the reason for 16 stops of dynamic range if one stop over the normal exposure (i.e. highlights, cloud details, back lit hair detail) gets obliterated, meanwhile you can expose up to peer into dust balls in your shadow where nobody looks. 

This manner of dealing with highlights result in shooting at a normal exposure, utilizing the 16 stops of DR in a more balanced way, allowing images to look more filmic, and allowing more artistic expression like hi key and backlight photography without a digital penalty. 

I thought I'd share this as a Leica fan.  Thanks for reading. 

  • Thanks 2
Link to post
Share on other sites

10 hours ago, archiea said:

Hey gang.. Leica fan but Fuji user here....  

Bring this up to Leica as fuji has a unique way to deal with blown highlights, see the image below.



It requires you to shoot at ISO 640 for 4 stops of high light recover, base iso is 160.  What the camera does is expose the mids and shadows at iso 640, but the highlights at iso 160, giving you incredible highlight recovery. This all happens internally in the camera, in the raw file, you just shoot an image, exposing for iso 640. 

In addition to the extra stops of highlight detail, you can control the falloff of the highlights:

This offers a very filmic falloff for the highlights.  With the extra sensor information in the form of 4 stops of highlight detail, you avoid these blobs of clipped highlights. 

This means you can avoid underexposing your Leica images in camera, effectively taking the image at the WRONG exposure, showing dark frames on your LCD... not good if you are doing client work. you rely on the COMPUTER inside your camera to expose you highlights with enough headroom to allow for a natural falloff indeed of a digital clip. 

I never understood the reason for 16 stops of dynamic range if one stop over the normal exposure (i.e. highlights, cloud details, back lit hair detail) gets obliterated, meanwhile you can expose up to peer into dust balls in your shadow where nobody looks. 

This manner of dealing with highlights result in shooting at a normal exposure, utilizing the 16 stops of DR in a more balanced way, allowing images to look more filmic, and allowing more artistic expression like hi key and backlight photography without a digital penalty. 

I thought I'd share this as a Leica fan.  Thanks for reading. 

That’s really cool!  It takes highlight weighted metering to another level!

I guess the best you can do on the m10 is bracket if you want the cleanest image in high dynamic range scenes. I tend to expose for the highlights that are key to my pic. Someone’s face for example. Or if close to base iso, massively underexpose and deal with a little shadow noise.
 

 

Link to post
Share on other sites

2 hours ago, costa43 said:

That’s really cool!  It takes highlight weighted metering to another level!

I guess the best you can do on the m10 is bracket if you want the cleanest image in high dynamic range scenes. I tend to expose for the highlights that are key to my pic. Someone’s face for example. Or if close to base iso, massively underexpose and deal with a little shadow noise.
 

 

Again, this is not post processing. This happens internally in the camera based on how you set it up. You get raw & jpgs looking like this straight outta camera.

These two videos explain the tech behind this mode.

https://youtu.be/RjjCa73XxsY?si=3cDnXCeQ2lWGSirO

 

https://youtu.be/HsuIApe6gvI?si=lV8A8NSNNswWeLg0

 

  • Thanks 1
Link to post
Share on other sites

With the incredible shadow reserves of the M10, I just expose for the highlights (always manually) and lift the rest in post. Here I've increased the foreground by 3 stops.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

 

  • Like 2
Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...