Sony a9 III global shutter high ISO / dynamic range tests

[Andrew Reid]

Interested to see more on this subject. Let's see how far Sony has come from the early global shutter sensor Blackmagic Production Cam 4K which couldn't do ISO 800!

Here's my findings so far:

https://www.eoshd.com/news/sony-a9-iii-high-iso-raw-test-a-noisier-6k-sensor-than-predecessor/

[BlueBomberTurbo]

With DXO, it's really not much of an issue.  What I am seeing, if all things are 100% equal in this comparison, is extra green shadow noise.  Can easily be cured with a 1 second EXIF edit, but still annoying (A9 had some, too). 

[Andrew Reid]

With such a fast sensor they have a lot more temporal resolution, so higher ISO performance if they use multi-frame NR. In the future, we won't be looking at single frame, despite The Power of One Frame ironically being the marketing slogan for the a9 III 🙂

The power of 120fps more like.

[horshack]

The link to PhotographyBlog only shows A9 III raw samples for that given shoot. Where did you find the equivalent A9 II raw taken at the same location / time?

[Andrew Reid]

15 minutes ago, horshack said:

The link to PhotographyBlog only shows A9 III raw samples for that given shoot. Where did you find the equivalent A9 II raw taken at the same location / time?

It doesn't need to be from the same time, just the same exposure relative to middle grey and same ISO

[Andrew Reid]

By the way the A9 II native ISO is 100-51,200

Which means the 250-25,600 of the A9 III clips a stop and a half of dynamic range early in both highlights and shadows.

Around 2-3 stops lost... So we're talking maybe 12 stops at best.

As for noise, it does look to be about a stop and a half worse.

[horshack]

2 minutes ago, Andrew Reid said:

It doesn't need to be from the same time, just the same exposure relative to middle grey and same ISO

Middle grey can only be established relative to the precise raw saturation level and without the necessary staged 9M3 raws available to establish saturation its precise middle grey is unknown. And saturation is more likely to be different for the 9M3 vs previous models since its an entirely different sensor/pixel architecture.

[Andrew Reid]

I looked at the images and one is clearly noisier than the other.

Works for me!

[horshack]

A poster on dpreview compared Sony's Starvis global-shutter sensor to their rolling-shutter equivalent in more typical sensors and found it had 2.4x lower FWC, which matches up with the base ISO change to 250 for the A9 III. The global-shutter sensor Starvis also had 2x the read noise 

The Starvis is their industrial line of sensors, for example security cameras. 

Link:  https://www.dpreview.com/forums/post/67351204

[Andrew Reid]

Sony are saying there's no image quality impact from global shutter, which is of course not true. The base iso 250 and lowest ISO of 2000 in S-LOG says otherwise, and it is a bit disappointing from Sony's marketing people to suggest otherwise.

I'd rather they just be honest!

Still a great camera though.

The Starvis / Pregius S - I am not too familiar with. If it is also a stacked sensor similar to the A9 III then it's a valid comparison though.

[Eric Calabros]

ISO 12800 images looks like shot by a 50mp sensor, not a 24mp one. However these are still raws. Noise impact in video will be worse due to heat.

[horshack]

On 11/9/2023 at 8:57 AM, Andrew Reid said:

Sony are saying there's no image quality impact from global shutter, which is of course not true. The base iso 250 and lowest ISO of 2000 in S-LOG says otherwise, and it is a bit disappointing from Sony's marketing people to suggest otherwise.

I'd rather they just be honest!

Still a great camera though.

The Starvis / Pregius S - I am not too familiar with. If it is also a stacked sensor similar to the A9 III then it's a valid comparison though.

Agreed. Sony is being unnecessarily cagey about this. In the following B&H video, Sony's Michael Bubolo claims Sony never discloses the dynamic range of their cameras, in response to a question about the A9 III's exact dynamic range - starts at 39:25:

Which is absolutely false because Sony discloses DR measurements in their actual product press releases, for example the claimed 15EV in their A7r V press release:

https://www.sony.com/content/sony/en/en_us/SCA/company-news/press-releases/sony-electronics/2022/sony-electronics-new-alpha-7r-v-camera-delivers-a-new-highresolution-imaging-experience-with-aibased-autofocus.html

[Andrew Reid]

Indeed, being untruthful is never a good look is it. Standard par for the course in marketing land though.

[Caleb Genheimer]

On 11/9/2023 at 7:57 AM, Andrew Reid said:

Sony are saying there's no image quality impact from global shutter, which is of course not true. The base iso 250 and lowest ISO of 2000 in S-LOG says otherwise, and it is a bit disappointing from Sony's marketing people to suggest otherwise.

I'd rather they just be honest!

Still a great camera though.

The Starvis / Pregius S - I am not too familiar with. If it is also a stacked sensor similar to the A9 III then it's a valid comparison though.

I’m just glad we have people now who know how to consistently test and report on how these CMOS sensors actually perform. It’s the core of every camera, yet all the other stuff they build around it gets all the attention. We’ve been bumming around with 12-bit ADC video modes and adequate LOG profiles for half a decade now, VERY little actual innovation in image pipeline with regards to color or dynamic range. 
 

I’m hopeful that with consistent reporting on each CMOS, the manufacturers will be pressured to actually innovate. A leap-frog all the way to 16-bit video ADC seems appropriate now that 12-bit has overstayed its welcome, or now that 8K sensors are showing up, perhaps a dual-ADC bayer pattern or ND Filter Array would be appropriate. 
 

This global shutter sensor is an important step, and I’m glad to see it. It indicates that they’re paying attention to demand for sensor improvements, and although it DOES still have trade-offs, they’ve set a new benchmark for hybrid cameras in the category of shutter performance. Critically, I think they learn a lot from *actually* bringing a camera to production, and hopefully the next time out, they’re sticking to global shutter, and innovating further to minimize the trade-offs. Additionally, Sony builds sensors for Panasonic, and it seems like they have some form of ~2-3yr tech swap agreement. If one comes up with something, the other seems to eventually add it into their sensors a few years later. I think Panasonic remains the leader in hybrid bodies for video users, and if they take a swing at full frame global shutter, it will be implemented really well.

[kye]

7 hours ago, Caleb Genheimer said:

I’m just glad we have people now who know how to consistently test and report on how these CMOS sensors actually perform. It’s the core of every camera, yet all the other stuff they build around it gets all the attention. We’ve been bumming around with 12-bit ADC video modes and adequate LOG profiles for half a decade now, VERY little actual innovation in image pipeline with regards to color or dynamic range. 
 

I’m hopeful that with consistent reporting on each CMOS, the manufacturers will be pressured to actually innovate. A leap-frog all the way to 16-bit video ADC seems appropriate now that 12-bit has overstayed its welcome, or now that 8K sensors are showing up, perhaps a dual-ADC bayer pattern or ND Filter Array would be appropriate. 
 

This global shutter sensor is an important step, and I’m glad to see it. It indicates that they’re paying attention to demand for sensor improvements, and although it DOES still have trade-offs, they’ve set a new benchmark for hybrid cameras in the category of shutter performance. Critically, I think they learn a lot from *actually* bringing a camera to production, and hopefully the next time out, they’re sticking to global shutter, and innovating further to minimize the trade-offs. Additionally, Sony builds sensors for Panasonic, and it seems like they have some form of ~2-3yr tech swap agreement. If one comes up with something, the other seems to eventually add it into their sensors a few years later. I think Panasonic remains the leader in hybrid bodies for video users, and if they take a swing at full frame global shutter, it will be implemented really well.

Most sensors do a full readout in the highest bit-depth at 24/25/30p, but at higher frame rates typically reduce the bit-depth of the read-out.

Assuming this was to save a bit on data rates and processing, then that means they have been making progress - they just spent it all on resolution instead of bit-depth.  Yet another hidden cost of this preposterous resolution pissing contest that the entire industry is doing, with consumers cheering all the way down.