SL2-S Noise

[hdmesa]

Advertisement (gone after registration)

3 hours ago, SrMi said:

Same noise when normalizing the output to the same dimensions, not at the pixel level. Typically, noise is not compared at pixel levels.

SL2-S added a completely different sensor technology/provider, you cannot compare it to SL2's sensor (FSI vs. BSI). Also, most BSI sensors seem to have dual-conversion gain, which reduces noise at higher ISOs. It is unrelated to BSI.

https://www.adorama.com/alc/faq-whats-a-back-side-illuminated-sensor/

I don't understand what you're saying really. You propose that BSI sensor technology is separable from the sensor?

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[LocalHero1953]

12 hours ago, SrMi said:

Thom Hogan has written about it here. The fill factor on FF sensors was already so high that the improvements with BSI are barely noticeable.

Also, when Sony (a7r > a7rII) and Nikon (D810 > D850) switched from FSI to BSI, there were no noticeable improvements in noise.

That article is from 2015. I looked for more recent commentary and found others that said the opposite. However neither Hogan nor the other writers give actual data on Fill Factor for FSI and BSI sensors - they appear to be just expressing opinions or summarising others. Do you have any links for Fill Factor data for sensors?

[SrMi]

6 hours ago, hdmesa said:

https://www.adorama.com/alc/faq-whats-a-back-side-illuminated-sensor/

I don't understand what you're saying really. You propose that BSI sensor technology is separable from the sensor?

What I am saying is that dual conversion gain technology is unrelated to BSI. When a manufacturer switches from FSI without DCG to BSI with DCG (similar performance otherwise, not like SL2 vs SL2-S), there will be differences at higher ISOs after DCG kicks in. DCG can be implemented in FSI sensors as well.

[SrMi]

6 hours ago, LocalHero1953 said:

That article is from 2015. I looked for more recent commentary and found others that said the opposite. However neither Hogan nor the other writers give actual data on Fill Factor for FSI and BSI sensors - they appear to be just expressing opinions or summarising others. Do you have any links for Fill Factor data for sensors?

To the best of my knowledge, there has not been much change since 2015 in that regard.

I will share the fill factor if I can find more information.

In the meantime, I take a pragmatic approach and observe whether there was any improvement in max DR when Sony and Nikon switched from FSI to BSI sensors. Per P2P there were none, nor have I seen anyone reporting improvements.  The latest such switch occurred when Sony upgraded a6600 (FSI) to a6700 (BSI). The switch to BSI did not improve noise.

[hdmesa]

4 hours ago, SrMi said:

What I am saying is that dual conversion gain technology is unrelated to BSI. When a manufacturer switches from FSI without DCG to BSI with DCG (similar performance otherwise, not like SL2 vs SL2-S), there will be differences at higher ISOs after DCG kicks in. DCG can be implemented in FSI sensors as well.

Dual gain with FSI can be illustrated with Canon's R5, which required Canon to bake in noise reduction to the RAW files from ISO 100-640 in order to keep up with comparable BSI sensors. BSI and the additional light collected by sensors of this technology work hand in hand with other technologies used to design the sensor to create the end results. Not sure why you're splitting hairs over statements about BSI sensors being better at high ISO noise than FSI since as a general rule, since all current BSI sensors are dual gain and few FSI sensors are.

We generally refer to modern sensors as either FSI, BSI, stacked, or global. That doesn't mean there are not other shared technologies that improve performance. For example, stacked sensors are also BSI, but when speaking about performance, I should be able to refer to them as stacked without someone calling out the fact that it's mostly the dual gain or BSI that does this or that. Of course we know that the stacked part of the sensor tech is related only to speed.

Bottom line is the FSI sensor in the SL2 is neither dual gain nor BSI and the results align with that. The BSI dual gain (and whatever else) sensor in the SL2-S has much nicer noise character at high ISO in very low light. Calling out whether that's due to BSI, dual gain, or a combination thereof – I don't understand the relevance of that nuance in the context of this thread.

Edited January 20, 2024 by hdmesa

[SrMi]

3 minutes ago, hdmesa said:

Dual gain with FSI can be illustrated with Canon's R5, which required Canon to bake in noise reduction to the RAW files from ISO 100-640 in order to keep up with comparable BSI sensors with dual gain. BSI and the additional light collected by sensors of this technology work hand in hand with other technologies to create the end results. Not sure why you're splitting hairs over statements about BSI sensors being better at high ISO noise than FSI since as a general rule, since all current BSI sensors are dual gain and few FSI sensors are.

You are comparing the FSI sensor from manufacturer A with the BSI sensor from manufacturer B. You will not see any relevant noise difference if you compare FSI and BSI sensors from the same manufacturers (Sony and Nikon). In my book, that is proof enough that BSI does not contribute to noise reduction in a meaningful way. The reason is probably the advance in microlens design that allowed nearly 100% fill factor with FSI as well.

Sony a6600 (FSI) and a6700 (BSI) have dual-conversion gains. There is no relevant difference in noise between the two cameras.

 

[hdmesa]

Advertisement (gone after registration)

Just now, SrMi said:

You are comparing the FSI sensor from manufacturer A with the BSI sensor from manufacturer B. You will not see any relevant noise difference if you compare FSI and BSI sensors from the same manufacturers (Sony and Nikon). In my book, that is proof enough that BSI does not contribute to noise reduction in a meaningful way. The reason is probably the advance in microlens design that allowed nearly 100% fill factor with FSI as well.

Sony a6600 (FSI) and a6700 (BSI) have dual-conversion gains. There is no relevant difference in noise between the two cameras.

 

Why do you think Sony moved to BSI sensor technology if there was no relevant difference between them?

Regardless, I get what you're saying, I just don't understand why it's relevant in the context of this thread. Should I just refer to the SL2 and SL2-S sensors without using the term FSI or BSI? 

[SrMi]

3 hours ago, hdmesa said:

Why do you think Sony moved to BSI sensor technology if there was no relevant difference between them?

Regardless, I get what you're saying, I just don't understand why it's relevant in the context of this thread. Should I just refer to the SL2 and SL2-S sensors without using the term FSI or BSI? 

BSI sensors have the advantage of better corners (less vignetting and cast) and faster readout. 

Also, here is a relevant discussion on DPR from 2017.

See #116 why it started. I was irked by the false statement that BSI alone helps with noise in FF sensors.

[hdmesa]

1 hour ago, SrMi said:

BSI sensors have the advantage of better corners (less vignetting and cast) and faster readout. 

Also, here is a relevant discussion on DPR from 2017.

See #116 why it started. I was irked by the false statement that BSI alone helps with noise in FF sensors.

Seems more like to me that Sony is using the more efficient light collection of the BSI sensor to support other improvements while retaining current noise performance. Trying to think of an analogy: engines that leverage new efficiency of operation to increase performance over fuel consumption. In the a6700, Sony may have set a goal to prioritized readout speed since BSI let them do that while maintaining the noise performance that had with FSI. With their 60mp shared sensors, the BSI tech was leveraged to increase resolution with noise being kept lower than it would have had the sensor been 60mp FSI. 

[hdmesa]

 

@SrMi, does this align with what you're saying?

https://www.dpreview.com/forums/thread/4550474

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!

[SrMi]

6 minutes ago, hdmesa said:

 

@SrMi, does this align with what you're saying?

https://www.dpreview.com/forums/thread/4550474

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!

Yes, faster readouts but not better sensor efficiency.

Also better corners. I wonder if it made Leica’s life easier when they added BSI sensor to M11.