Read noise measurements

[Dem Bell]

Hi folks. I wonder why the input-referred read noise reported on photons2photos for two Fuji cameras X-S10 and X-T4 that are supposed to share the same sensor is so much different:

https://www.photonstophotos.net/Charts/RN_e.htm#FujiFilm X-S10_14,FujiFilm X-T4_14

There is about 1.5x factor difference.

Comparing to other 24Mp-26Mp Fuji cameras of the same elk, it looks like the X-T4 is the odd one out with quite high read noise. Can this be down to a systematic error (e.g. it was measured in hot weather, sample variation) or it is actually telling us something about the X-T4?

How would you interpret this difference?

 

[robert1955]

Hi folks. I wonder why the input-referred read noise reported on photons2photos for two Fuji cameras X-S10 and X-T4 that are supposed to share the same sensor is so much different:

https://www.photonstophotos.net/Charts/RN_e.htm#FujiFilm X-S10_14,FujiFilm X-T4_14

There is about 1.5x factor difference.

Comparing to other 24Mp-26Mp Fuji cameras of the same elk, it looks like the X-T4 is the odd one out with quite high read noise. Can this be down to a systematic error (e.g. it was measured in hot weather, sample variation) or it is actually telling us something about the X-T4?

How would you interpret this difference?

Interesting question. I added the X-Pro3 and X-E4 which also have the [or 'a'?] 26MP sensor:

https://www.photonstophotos.net/Charts/RN_e.htm#FujiFilm X-S10_14,FujiFilm X-T4_14

There the X-E4 stands out in the other direction.

But if we look at the Read noise in DN chart on Photos to Photos these differences all but disappear:

https://www.photonstophotos.net/Charts/RN_ADU.htm

 

[Dem Bell]

Wow, the XE4 does swing in the other direction, same sensor minus the IBIS I believe:

https://www.photonstophotos.net/Cha... X-Pro3_14,FujiFilm X-S10_14,FujiFilm X-T4_14

which makes me think that these data are prone to some systematic error.

 

[ThrillaMozilla]

Hi folks. I wonder why the input-referred read noise reported on photons2photos for two Fuji cameras X-S10 and X-T4 that are supposed to share the same sensor is so much different:

https://www.photonstophotos.net/Charts/RN_e.htm#FujiFilm X-S10_14,FujiFilm X-T4_14

There is about 1.5x factor difference.

Comparing to other 24Mp-26Mp Fuji cameras of the same elk, it looks like the X-T4 is the odd one out with quite high read noise. Can this be down to a systematic error (e.g. it was measured in hot weather, sample variation) or it is actually telling us something about the X-T4?

How would you interpret this difference?

Good question. Clerical error?

The photon transfer curves show the S10 as having 6% higher error (probably an insignificant difference).

T4: 1.658 DN x 2.039 e-/DN = 3.38 electrons

S10: 1.791 DN x 2.007 e-/DN = 3.59 electrons

In addition, there are some odd disclaimers:

"In particular read noise values may be lower due to black level clipping.
...Noise values (the colored dots) at high signal levels may be underestimated due to the way data is gathered."

 

[bclaff]

Conversion from Digital Numbers (DN) to electrons requires an accurate value for gain.

(Equivalent to unity gain, FWC, etc.)

The technique I use is subject to a lot of variation depending mostly on the quality of the images provided. This seems particularly true for Fujifilm for some reason.

 

[ThrillaMozilla]

Conversion from Digital Numbers (DN) to electrons requires an accurate value for gain.

(Equivalent to unity gain, FWC, etc.)

The technique I use is subject to a lot of variation depending mostly on the quality of the images provided. This seems particularly true for Fujifilm for some reason.

But why don't the numbers from here agree with numbers from here? See my previous post. The numbers don't seem to agree at all, and not just from these cameras.