Reading mode: Light Dark

X-Pro2 Sensor Measurements

Go to last post
B

bclaff

Forum Pro
Messages
14,415
Solutions
24
Reaction score
13,409
Location
Metro-West Boston, MA, US
I now have sensor measurements for the Fulifilm X-Pro2 posted at my site.

a2d9d5edcfa14d3381d032bc1b494238.jpg.png

Note the drop at ISO 800, this is dual conversion gain technology that is becoming more and more widespread.

This shows up in the Photographic Dynamic Range (PDR) as a boost at ISO 800:

3fa4c7402da24ea89e792f22550ba104.jpg.png

--
Bill ( Your trusted source for independent sensor data at http://www.photonstophotos.net )
 
Thanks a lot.

Could you enlighten us in idiot speak for dummies what would entice Fuji to order a sensor with the dual gain "kick" around 800 ISO while Sony chose to position it around 400 ISO ? Or is it simply related to their base ISO difference ?
 
Last edited:
MayaTlab0 said:
Thanks a lot.

Could you enlighten us in idiot speak for dummies what would entice Fuji to order a sensor with the dual gain "kick" around 800 ISO while Sony chose to position it around 400 ISO ? Or is it simply related to their base ISO difference ?
Click to expand...
For example, the new Sonly ILCE-6300 (A6300) looks like this:

93d773d5ffff4ae589debb39041b4214.jpg.png

It's true that conversion gain changes at ISO 400 rather than ISO 800.
But also note the base ISO is ISO 100 rather than ISO 200.
So yes, then the question is; why such a high base ISO for the X-Pro2.

I don't know that answer except that all the Fujifilm X-Trans sensors seem to have higher base ISO than most other brands; typicall ISO 200 rather than ISO 100.
--
Bill ( Your trusted source for independent sensor data at http://www.photonstophotos.net )
 
nice work Bill!

What did you calculate for FWC?

edit: interesting to see the A6300 having lower read noise
 
Last edited:
bclaff said:
MayaTlab0 said:
Thanks a lot.

Could you enlighten us in idiot speak for dummies what would entice Fuji to order a sensor with the dual gain "kick" around 800 ISO while Sony chose to position it around 400 ISO ? Or is it simply related to their base ISO difference ?
Click to expand...
For example, the new Sonly ILCE-6300 (A6300) looks like this:

93d773d5ffff4ae589debb39041b4214.jpg.png

It's true that conversion gain changes at ISO 400 rather than ISO 800.
But also note the base ISO is ISO 100 rather than ISO 200.
So yes, then the question is; why such a high base ISO for the X-Pro2.

I don't know that answer except that all the Fujifilm X-Trans sensors seem to have higher base ISO than most other brands; typicall ISO 200 rather than ISO 100.
--
Bill ( Your trusted source for independent sensor data at http://www.photonstophotos.net )
Click to expand...
So if I understand properly there isn't a difference in how the underlying sensor operates (appart from, obviously, the CFA), just that the shift in base ISO, whatever the reason for it may be, causes the shift where the conversion gain changes ?
 
Thank you so much for your service to the community. Very much appreciated.

I have a specific question, and it pertains to ISO-invariance since that's a hot topic these days...

Does your analysis allow to figure out how ISO-invariant a sensor is and whether skipping in-camera analog amplification and instead doing external digital amplification would basically give similar results or not?

What baffles me a bit is the observation that the read noise of the X-E2 is lower at lower ISOs than that of the X-Pro2, but the read noise of the X-Pro2 is lower at ISOs of 800 and above. The X-E2 is quite ISO-invariant in practice, but the jury is still out on the X-Pro2. My hunch is that it's possible that one would benefit from using in-camera gain at ISOs of 800 and higher with the X-Pro2, rather than doing purely digital amplification.

Any thoughts?
 
nixda said:
Thank you so much for your service to the community. Very much appreciated.
Click to expand...
You're welcome.
nixda said:
I have a specific question, and it pertains to ISO-invariance since that's a hot topic these days...

Does your analysis allow to figure out how ISO-invariant a sensor is and whether skipping in-camera analog amplification and instead doing external digital amplification would basically give similar results or not?
Click to expand...
In my analysis I try to separate digital from analog gain. Typically this is only a factor at the "extended" ISO setting.

Here's one way to look at ISO in-variance, Input-referred Read Noise

e58645567d2f44afacd0b9e4d48a180f.jpg.png

My rule of thumb is that the sensor becomes "ISO invariant" where input-referred read noise drops by less than 1/3 stop per stop of ISO setting.

Due to the dual conversion gain architecture I'd say that ISO 200 is ISO invariant up to ISO 640 and that ISO 800 is invariant up to ISO 12800 (where digital takes over).

You can also look at my Shadow Improvement chart :

68d6146e85ec462c8ed4c16eaed4bb15.jpg.png

Again, notice little improvement from ISO 200 to ISO 640, and essentially none from ISO 800 to ISO 12800 (higher ISO settings are false readings due to digital scaling).

Hope this helps.

--
Bill ( Your trusted source for independent sensor data at http://www.photonstophotos.net )
 
bclaff said:
nixda said:
Thank you so much for your service to the community. Very much appreciated.
Click to expand...
You're welcome.
bclaff said:
I have a specific question, and it pertains to ISO-invariance since that's a hot topic these days...

Does your analysis allow to figure out how ISO-invariant a sensor is and whether skipping in-camera analog amplification and instead doing external digital amplification would basically give similar results or not?
Click to expand...
In my analysis I try to separate digital from analog gain. Typically this is only a factor at the "extended" ISO setting.

Here's one way to look at ISO in-variance, Input-referred Read Noise

e58645567d2f44afacd0b9e4d48a180f.jpg.png

My rule of thumb is that the sensor becomes "ISO invariant" where input-referred read noise drops by less than 1/3 stop per stop of ISO setting.

Due to the dual conversion gain architecture I'd say that ISO 200 is ISO invariant up to ISO 640 and that ISO 800 is invariant up to ISO 12800 (where digital takes over).

You can also look at my Shadow Improvement chart :

68d6146e85ec462c8ed4c16eaed4bb15.jpg.png

Again, notice little improvement from ISO 200 to ISO 640, and essentially none from ISO 800 to ISO 12800 (higher ISO settings are false readings due to digital scaling).

Hope this helps.

--
Bill ( Your trusted source for independent sensor data at http://www.photonstophotos.net )
Click to expand...
So, if I understand correctly, you're saying that if one needed less than 2 stops of amplification of data initially captured at base ISO, one might as well just stay at base ISO and do the amplification digitally. And, if one needed more than two stops of amplification to get the desired image brightness, then one might as well stay at ISO800 and do any subsequent amplification digitally.
 
nixda said:
bclaff said:
nixda said:
Thank you so much for your service to the community. Very much appreciated.
Click to expand...
You're welcome.
nixda said:
I have a specific question, and it pertains to ISO-invariance since that's a hot topic these days...

Does your analysis allow to figure out how ISO-invariant a sensor is and whether skipping in-camera analog amplification and instead doing external digital amplification would basically give similar results or not?
Click to expand...
In my analysis I try to separate digital from analog gain. Typically this is only a factor at the "extended" ISO setting.

Here's one way to look at ISO in-variance, Input-referred Read Noise

e58645567d2f44afacd0b9e4d48a180f.jpg.png

My rule of thumb is that the sensor becomes "ISO invariant" where input-referred read noise drops by less than 1/3 stop per stop of ISO setting.

Due to the dual conversion gain architecture I'd say that ISO 200 is ISO invariant up to ISO 640 and that ISO 800 is invariant up to ISO 12800 (where digital takes over).

You can also look at my Shadow Improvement chart :

68d6146e85ec462c8ed4c16eaed4bb15.jpg.png

Again, notice little improvement from ISO 200 to ISO 640, and essentially none from ISO 800 to ISO 12800 (higher ISO settings are false readings due to digital scaling).

Hope this helps.
Click to expand...
So, if I understand correctly, you're saying that if one needed less than 2 stops of amplification of data initially captured at base ISO, one might as well just stay at base ISO and do the amplification digitally. And, if one needed more than two stops of amplification to get the desired image brightness, then one might as well stay at ISO800 and do any subsequent amplification digitally.
Click to expand...
Yes (provided you are post processing).

--
Bill ( Your trusted source for independent sensor data at http://www.photonstophotos.net )
 
Hi Bill,

I have measured the read noise and the gain of my XE1 and I was pretty much thrilled when someone pointed me to your results on the XPRO2. At first I saw that my RN numbers started close to yours. However after thinking about it I realize that I am not sure what the DN number is. I have looked at your website but didn't find the definition. In the case of my XE1 I measured a standard deviation of 0.63 at Iso200 with a gain (EBU/e) of 1/8. What would be the corresponding value of DN in that case? Thanks.
 
Last edited:
rlx said:
Hi Bill,

I have measured the read noise and the gain of my XE1 and I was pretty much thrilled when someone pointed me to your results on the XPRO2. At first I saw that my RN numbers started close to yours. However after thinking about it I realize that I am not sure what the DN number is. I have looked at your website but didn't find the definition. In the case of my XE1 I measured a standard deviation of 0.63 at Iso200 with a gain (EBU/e) of 1/8. What would be the corresponding value of DN in that case? Thanks.
Click to expand...
If you're willing to gather the data I'd be happy to do a fully analysis of the X-E1 with you.
Just send email or Private Message.

As for your data, a standard deviation of 0.63 is probably too low. I am suspicious.
If I have time I could also look at your data and computation.
 
Very interesting.

I knew the change n conversion gain happens around/before ISO800. When I saw the A6300 measurements, i expected the same curve, but shifted on the ISO scale.

Another interesting question would be, where those cameras are ISO wise (saturation based). Since dxo doesn't do Fuji (X-Trans), do you have any idea if anyone does?

Last but not least: what up with your X-E2 values? They are not really in line with all the other Fujis?

thanks for taking your time and doing all this!
 
uMad said:
Another interesting question would be, where those cameras are ISO wise (saturation based). Since dxo doesn't do Fuji (X-Trans), do you have any idea if anyone does?
Click to expand...
I have my own technique but it's pretty tedious.
I would need to collaborate closely with someone to pull it off.
uMad said:
Last but not least: what up with your X-E2 values? They are not really in line with all the other Fujis?
Click to expand...
Not sure what you mean. Perhaps we should discuss by email or private message.
uMad said:
thanks for taking your time and doing all this!
Click to expand...
You are welcome.
 
Thanks Bill for getting this done so quickly. Your results appear pretty much as expected which is a good thing.

Bob
 
many thanks for that study, from your results, i see that sony A6300 has more DR and lower read noise than xpro2, am i right?
 
at base iso, where i shoot %99 of the time, may we see visible advantage with A6300 sensor, where is the clear gap between xpro2? , or is it very subtle difference? i'm just trying to figure out how much difference in your graphs could be visible in photos.thanks.
 
ijustloveshooting said:
at base iso, where i shoot %99 of the time, may we see visible advantage with A6300 sensor, where is the clear gap between xpro2? , or is it very subtle difference? i'm just trying to figure out how much difference in your graphs could be visible in photos.thanks.
Click to expand...
Your question is simple, but the answer not so much so !

Unless the dynamic range of the scene being photographed exceeds that of one camera but not the other, then you are not likely to see any difference.
If the dynamic range is within the capability of both cameras then the results may be indistinguishable.
So the value of the "better" camera are for those cases where one camera can get a good image and another cannot.
This depends quite a bit on your personal needs as a photographer.

In this case I image there are factors other than dynamic range that would go into any decision about whether one camera or the other is better for your purposes.

Sorry if this is not more helpful !
 
You must log in or register to reply here.

Keyboard shortcuts

f
Forum
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