Reading mode: Light Dark

Dual gain exposure strategy when pushing files

Go to last post
N

No account anymore

Guest
Her is how to get the cleanest results from cameras with dual gain design, like the second and third generation Sony A7 series and A9 cameras:

Dpreview are testing ISO invariance from pushing ISO 100 files. That's ok, this is their take. But if you want to get less noisy results and less color shift, don't start at base ISO 100. Instead, start at second base ISO 640, where dual gain kicks in, unless you really, really need the extra dynamic range that ISO 100 offer.

For such scenarios, two exposures merged to HDR would be a better strategy, unless there are moving subjects that matter in the field of view.

Here are some samples. Exposure for all files are 1/2000 sec at f:8 and same white balance. Color are adjusted a tad, since the ISO 100 file was a bit more magenta when opened in Lightroom than the other files. Most important here is noise structure, though:
  • First exposure set to ISO 100, pushed 6 stop
  • Second exposure set to ISO 640, pushed 3.3 stop
  • Third exposure set to ISO 800, pushed 3 stop
  • Fourth exposure set to ISO 6400, no pushing
100% crop - note the difference in noise when pushed from ISO 100 instead of ISO 640. Exposure is exactly the same for all exposures for this ISO-invariance/dual gain test: 1/2000 sec at f:8.
100% crop - note the difference in noise when pushed from ISO 100 instead of ISO 640. Exposure is exactly the same for all exposures for this ISO-invariance/dual gain test: 1/2000 sec at f:8.



9acf8e1a0ea549a3b60beaa1abfd94f6.jpg

Similar with Nikon Z6, from Dpreview article on ISO-invariance for the Nikon.
 
Last edited:
Magnar W said:
Her is how to get the cleanest results from cameras with dual gain design, like the second and third generation Sony A7 series and A9 cameras:

Dpreview are testing ISO invariance from pushing ISO 100 files. That's ok, this is their take. But if you want to get less noisy results and less color shift, don't start at base ISO 100. Instead, start at second base ISO 640, where dual gain kicks in, unless you really, really need the extra dynamic range that ISO 100 offer.

For such scenarios, two exposures merged to HDR would be a better strategy, unless there are moving subjects that matter in the field of view.

Here are some samples. Exposure for all files are 1/2000 sec at f:8 and same white balance. Color are adjusted a tad, since the ISO 100 file was a bit more magenta when opened in Lightroom than the other files. Most important here is noise structure, though:
  • First exposure set to ISO 100, pushed 6 stop
  • Second exposure set to ISO 640, pushed 3.3 stop
  • Third exposure set to ISO 800, pushed 3 stop
  • Fourth exposure set to ISO 6400, no pushing
100% crop - note the difference in noise when pushed from ISO 100 instead of ISO 640. Exposure is exactly the same for all exposures for this ISO-invariance/dual gain test: 1/2000 sec at f:8.
100% crop - note the difference in noise when pushed from ISO 100 instead of ISO 640. Exposure is exactly the same for all exposures for this ISO-invariance/dual gain test: 1/2000 sec at f:8.
Click to expand...
The 100 iso does show more noise but looking at the plants in the window it looks like it has retained more detail, I’m guessing due to dynamic range. It might just be my observation but the 6400 images look a little cleaner than the pushed 640. The lights in the window though look more blown than in the pushed version.
 
Paul Barnard said:
The 100 iso does show more noise but looking at the plants in the window it looks like it has retained more detail, I’m guessing due to dynamic range.
Click to expand...
Dynamic range does not affect fine detail. When using exposure strategy like this, dynamic range will be well within 12 bit anyway.

Differences in fine detail for certain areas are more likely caused by variations in the structure of the noise pattern, since noise is distributed randomly.
Paul Barnard said:
It might just be my observation but the 6400 images look a little cleaner than the pushed 640. The lights in the window though look more blown than in the pushed version.
Click to expand...
There might go on some noise reduction in-camera before the raw file is written when ISO value is raised.
 
Magnar W said:
Paul Barnard said:
The 100 iso does show more noise but looking at the plants in the window it looks like it has retained more detail, I’m guessing due to dynamic range.
Click to expand...
Dynamic range does not affect fine detail. When using exposure strategy like this, dynamic range will be well within 12 bit anyway.

Differences in fine detail for certain areas are more likely caused by variations in the structure of the noise pattern, since noise is distributed randomly.
Member said:
It might just be my observation but the 6400 images look a little cleaner than the pushed 640. The lights in the window though look more blown than in the pushed version.
Click to expand...
There might go on some noise reduction in-camera before the raw file is written when ISO value is raised.
Click to expand...
The plant detail is very close to the bottom of the dynamic range. I was wondering if there is a touch more clipping at the bottom end at 640 vs 100 that meant the detail was more recoverable in the 100 shot vs the 640 shot.
 
Of course under low light that otherwise you'd need more than ISO 640 if exposed normally that less push from 2nd base ISO 640 has less penalty than more push from 1st base ISO 100 that DPR article already explained.

Personally for low light such as for evening I shoot mostly on tripod at base ISO 100, but 2nd base ISO 640 is a great choice in still subject if I have to shoot hand-held to reserve a bit more highlight in landscape type photo. But for most what I shoot in low light in wildlife, event etc I'd still shoot at normal high ISO with not much noticeable loss in highlight clipping in those applications, and at least I can check result at the normal brightness rather a very dark photo in EVF/LCD. Hope Sony will adjust the companion JPEG file from the RAW file to automatically brighten up for viewing purpose, and I don't mind to lift brightness in software for RAW files from base ISO 640.

--
https://www.flickr.com/photos/55485085@N04/albums
 
Last edited:
Paul Barnard said:
Magnar W said:
Paul Barnard said:
The 100 iso does show more noise but looking at the plants in the window it looks like it has retained more detail, I’m guessing due to dynamic range.
Click to expand...
Dynamic range does not affect fine detail. When using exposure strategy like this, dynamic range will be well within 12 bit anyway.

Differences in fine detail for certain areas are more likely caused by variations in the structure of the noise pattern, since noise is distributed randomly.
Paul Barnard said:
It might just be my observation but the 6400 images look a little cleaner than the pushed 640. The lights in the window though look more blown than in the pushed version.
Click to expand...
There might go on some noise reduction in-camera before the raw file is written when ISO value is raised.
Click to expand...
The plant detail is very close to the bottom of the dynamic range. I was wondering if there is a touch more clipping at the bottom end at 640 vs 100 that meant the detail was more recoverable in the 100 shot vs the 640 shot.
Click to expand...
There are more detail in the shadow areas in the test files than shown here.

Also, when dual gain kicks in, the sensor is getting more shadow info than at lower ISO settings. This is what dual gain is about.
 
Magnar W said:
Paul Barnard said:
Magnar W said:
Paul Barnard said:
The 100 iso does show more noise but looking at the plants in the window it looks like it has retained more detail, I’m guessing due to dynamic range.
Click to expand...
Dynamic range does not affect fine detail. When using exposure strategy like this, dynamic range will be well within 12 bit anyway.

Differences in fine detail for certain areas are more likely caused by variations in the structure of the noise pattern, since noise is distributed randomly.
Member said:
It might just be my observation but the 6400 images look a little cleaner than the pushed 640. The lights in the window though look more blown than in the pushed version.
Click to expand...
There might go on some noise reduction in-camera before the raw file is written when ISO value is raised.
Click to expand...
The plant detail is very close to the bottom of the dynamic range. I was wondering if there is a touch more clipping at the bottom end at 640 vs 100 that meant the detail was more recoverable in the 100 shot vs the 640 shot.
Click to expand...
There are more detail in the shadow areas in the test files than shown here.

Also, when dual gain kicks in, the sensor is getting more shadow info than at lower ISO settings. This is what dual gain is about.
Click to expand...
That makes sense. Thanks
 
PWPhotography said:
Of course under low light that otherwise you'd need more than ISO 640 if exposed normally that less push from 2nd base ISO 640 has less penalty than more push from 1st base ISO 100 that DPR article already explained.
Click to expand...
Depends on how bright the highlights are, and how much highlight detail you want to keep, weighted against how deep into shadow details you will go.
PWPhotography said:
But for most what I shoot in low light in wildlife, event etc I'd still shoot at normal high ISO with not much noticeable loss in highlight clipping in those applications, and at least I can check result at the normal brightness rather a very dark photo in EVF/LCD.
Click to expand...
For normal and low contrast scenes, low ISO exposures and pushing is a bad choice. Rather go for the ISO you need. Here is an example taken at ISO 64 000. The head lights of the scooter are blown, treated as shiny highlights. This is from a reportage on how teenagers move around.

Exposure 1/500 sec at ISO 64 000. A7rIII. An overcast day with dull light, no need at all for using low ISO and pushing the data. The scooter headlights are treated as shiny highlights anyway.
Exposure 1/500 sec at ISO 64 000. A7rIII. An overcast day with dull light, no need at all for using low ISO and pushing the data. The scooter headlights are treated as shiny highlights anyway.
 
Last edited:
IDK. JPEGs out of my A7R2 are so good I rarely have to make big adjustments. For me the incremental IQ gains just aren't worth the time spent. I can't remember the last time I saw noise in pulled shadows or found blown highlights in recovery. There's a little gain around ISO640 but I feel like you'd have to run it through Bill Claff level analysis to really see it.
 
sportyaccordy said:
IDK. JPEGs out of my A7R2 are so good I rarely have to make big adjustments. For me the incremental IQ gains just aren't worth the time spent. I can't remember the last time I saw noise in pulled shadows or found blown highlights in recovery. There's a little gain around ISO640 but I feel like you'd have to run it through Bill Claff level analysis to really see it.
Click to expand...
Good for you. Others might use different exposure strategies and rely on raw files. This test is primarily for them. ;-)
 
Last edited:
sportyaccordy said:
IDK. JPEGs out of my A7R2 are so good I rarely have to make big adjustments. For me the incremental IQ gains just aren't worth the time spent. I can't remember the last time I saw noise in pulled shadows or found blown highlights in recovery. There's a little gain around ISO640 but I feel like you'd have to run it through Bill Claff level analysis to really see it.
Click to expand...
Interesting thread.

Am buying the german photo magazine Color Photo (some really bright minds there), and they claim that out of camera JPGs are now so good that shooting RAW is only needed when underexposing severely (using very high ISOs that is)...

That is one way to think photography, at least when getting the exposure and white balance correct the first time around. Sony JPGs are also quite robust when it comes to post processing.

But still - pushing files RAW still seems to have the edge when done properly.
 
[QUOTE] PWPhotography said:
But for most what I shoot in low light in wildlife, event etc I'd still shoot at normal high ISO with not much noticeable loss in highlight clipping in those applications, and at least I can check result at the normal brightness rather a very dark photo in EVF/LCD.
Click to expand...
This is one thing I've always struggled with using the low light dual gain shooting strategy where you set your f stop and shutter speed and then dial back ISO to 640 to later push in post. I've read Jim's blog and I understand it, but if the differences between ISO 640 + post and just shooting at ISO 6400+ are so minute then why choose to forego live view and the ability to review in the LCD/viewfinder? Jim's a genius with this stuff, so I trust his recommendation; I just feel like I'm missing something.
[/QUOTE]
 
shortkut77 said:
PWPhotography said:
But for most what I shoot in low light in wildlife, event etc I'd still shoot at normal high ISO with not much noticeable loss in highlight clipping in those applications, and at least I can check result at the normal brightness rather a very dark photo in EVF/LCD.
Click to expand...
This is one thing I've always struggled with using the low light dual gain shooting strategy where you set your f stop and shutter speed and then dial back ISO to 640 to later push in post. I've read Jim's blog and I understand it, but if the differences between ISO 640 + post and just shooting at ISO 6400+ are so minute then why choose to forego live view and the ability to review in the LCD/viewfinder? Jim's a genius with this stuff, so I trust his recommendation; I just feel like I'm missing something.
Click to expand...
This was my gripe. Plus the "push in post" method introduces so much opportunity to mess it all up. The normal method is just much more reliable.
 
shortkut77 said:
I've read Jim's blog and I understand it, but if the differences between ISO 640 + post and just shooting at ISO 6400+ are so minute then why choose to forego live view and the ability to review in the LCD/viewfinder?
Click to expand...
The ONLY reason to use "pushing at post" exposure strategy is when the scene has important bright highlights that will be blown at higher ISO settings.

If not, use a "correct" ISO setting for your capture. That is most cases, unless you work pretty specialized.
 
Last edited:
shortkut77 said:
PWPhotography said:
But for most what I shoot in low light in wildlife, event etc I'd still shoot at normal high ISO with not much noticeable loss in highlight clipping in those applications, and at least I can check result at the normal brightness rather a very dark photo in EVF/LCD.
Click to expand...
This is one thing I've always struggled with using the low light dual gain shooting strategy where you set your f stop and shutter speed and then dial back ISO to 640 to later push in post. I've read Jim's blog and I understand it, but if the differences between ISO 640 + post and just shooting at ISO 6400+ are so minute then why choose to forego live view and the ability to review in the LCD/viewfinder? Jim's a genius with this stuff, so I trust his recommendation; I just feel like I'm missing something.
Click to expand...
In reality virtually nobody uses this method in sport/wildlife/action/event type photos as basically no much visible benefits while have very inconvenient usages (unless a FW update to brighten up associated JPEG files for viewing) on very underexposed photos or even could ruin photos when the lights on stage suddenly lighting up for example. I will still use auto-ISO mostly in those applications. It's a valid choice in low light hand-held landscape type photos, but I shoot mostly on tripod in evening at base ISO 100.

--
https://www.flickr.com/photos/55485085@N04/albums
 
Last edited:
Okay, great, so I'm not missing something. Makes sense when to use and when not to. Seems more often than not it's unnecessary and the cons outweigh the pros. Thanks all.
 
If dual gain implementation is a good thing, why is limited to two steps? Why not 10, 20, or 30 steps to cover a complete range?
 
Last edited:
sybersitizen said:
If dual gain implementation is a good thing, why is limited to two steps? Why not 10, 20, or 30 steps to cover a complete range?
Click to expand...
Because every gain stage needs some sort of cicuit to alter the capacity of the read-out-circuit and it`s already crowded in this area of the sensor.
 
Berkrathes said:
sybersitizen said:
If dual gain implementation is a good thing, why is limited to two steps? Why not 10, 20, or 30 steps to cover a complete range?
Click to expand...
Because every gain stage needs some sort of cicuit to alter the capacity of the read-out-circuit and it`s already crowded in this area of the sensor.
Click to expand...
IC components keep shrinking and getting more and more crowded. If dual gain matters in an important way, someone will do what's necessary to expand it beyond two steps.
 
sybersitizen said:
Berkrathes said:
sybersitizen said:
If dual gain implementation is a good thing, why is limited to two steps? Why not 10, 20, or 30 steps to cover a complete range?
Click to expand...
Because every gain stage needs some sort of cicuit to alter the capacity of the read-out-circuit and it`s already crowded in this area of the sensor.
Click to expand...
IC components keep shrinking and getting more and more crowded. If dual gain matters in an important way, someone will do what's necessary to expand it beyond two steps.
Click to expand...
Maybe sometime. But Keep in mind you would Need that circuit for every pixel so it will be e.g. 24 Million circuits ore even more. Dual gain was a Major improvement in sensor design that has yet to be matched by most manufacturers. Even Sony bought that Technology.
 
You must log in or register to reply here.

Similar threads

S
Any penalties for dual gain sensors (Z8 at ISO 500 vs 320 for example)
Replies
6
Views
3K
Bill Ferris
Bill Ferris
D
Can somebody here please shed some light on the dual-gain/recommended ISO on the A7RV?
2 3
Replies
54
Views
12K
daran
daran
H
Panasonic did something special with the Z6 III sensor
4 5 6
Replies
118
Views
49K
Leonard Shepherd
L
L
Sony a7 OG vs a7c
Replies
12
Views
7K
Lee Croft
L
D
a1 II vs. a9 III Field Comparison
Replies
16
Views
41K
Foxjet
F

Keyboard shortcuts

f
Forum
n
Next
w
Next unread
m
My threads
Mobile site
About
Editorial content
Cameras & Lenses
Community
All content, design, and layout are Copyright © 1998–2025 Digital Photography Review All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
When you use DPReview links to buy products, the site may earn a commission.
©GPS Media - Guides, Products, Services.
Back
Top