Fujifilm X-Pro3 Sensor Measurements at PhotonsToPhotos

Started 3 months ago | Discussions
bclaff Veteran Member • Posts: 9,432
Fujifilm X-Pro3 Sensor Measurements at PhotonsToPhotos
11

These results are likely to cause confusion.

The X-Pro3 appears to have improved PDR but it is an illusion.

It's well known that Fuji implements the ISO setting in an unusual way.
In this case they have given ISO 320 to ISO 12800 an addition 1 stop boost.

Fuji uses a proprietary Exif tag called "Raw Exposure Bias" to indicate how much they have shifted the raw data from "standard".

Adjusted in Excel for Raw Exposure Bias we see that the X-Pro3 and X-T3 are essentially identical. (They shift to dual conversion gain at a slightly different point).

-- hide signature --

Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )

Fujifilm X-Pro3
If you believe there are incorrect tags, please send us this post using our feedback form.
alekt449 Regular Member • Posts: 156
Re: Fujifilm X-Pro3 Sensor Measurements at PhotonsToPhotos
5

Someone at Fuji said "We are sick of pixel peepers lets have fun with them "

Sugyn Regular Member • Posts: 127
Re: Fujifilm X-Pro3 Sensor Measurements at PhotonsToPhotos
5

Doesn't this undermine the useability of the testing. How do we know when a camera company is undermining the results. Shouldn't you be normalizing the data? Any other camera companies doing this?

OP bclaff Veteran Member • Posts: 9,432
Re: Fujifilm X-Pro3 Sensor Measurements at PhotonsToPhotos
1

Sugyn wrote:

Doesn't this undermine the useability of the testing. How do we know when a camera company is undermining the results.

Agree.

Shouldn't you be normalizing the data?

There has been discussion over the years about making the x-axis at PhotonsToPhotos more like that at DxOMark where they use their "Measured ISO" value.

The problem is that I have no consistent way of doing that.

For example I have tested many cameras not tested by DxOMark and in any case they are no longer testing cameras at a reasonable pace.

I looked into Adobe Baseline Exposure but found the quality of their data well below that of DxOMark partly because they seem to have shifted standards at some point but retained old legacy data.

And then there are companies like Fujifim who have a proprietary (but known) value in their Exif information.

So I have opted to continue using the ISO Setting as this is what the photographer has (some) control over.

Any other camera companies doing this?

Yes, Olympus for example, and you can get a good idea by looking at the sort-able table (I would sort on stops descending) beneath the chart DxOMark Derived Measured ISO at PhotonsToPhotos. You can find similar information at DxOMark but it's not so easily accessible (IMO). Keep in mind that DxOMark only tested a limited number of Fujifilm cameras.

-- hide signature --

Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )

OP bclaff Veteran Member • Posts: 9,432
Re: Fujifilm X-Pro3 Sensor Measurements at PhotonsToPhotos
2

alekt449 wrote:

Someone at Fuji said "We are sick of pixel peepers lets have fun with them "

It's unclear what they were thinking ... this could even be an error in the firmware.

-- hide signature --

Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )

rwbaron Forum Pro • Posts: 14,332
Re: Fujifilm X-Pro3 Sensor Measurements at PhotonsToPhotos
2

bclaff wrote:

Sugyn wrote:

Doesn't this undermine the useability of the testing. How do we know when a camera company is undermining the results.

Agree.

Shouldn't you be normalizing the data?

There has been discussion over the years about making the x-axis at PhotonsToPhotos more like that at DxOMark where they use their "Measured ISO" value.

The problem is that I have no consistent way of doing that.

For example I have tested many cameras not tested by DxOMark and in any case they are no longer testing cameras at a reasonable pace.

I looked into Adobe Baseline Exposure but found the quality of their data well below that of DxOMark partly because they seem to have shifted standards at some point but retained old legacy data.

And then there are companies like Fujifim who have a proprietary (but known) value in their Exif information.

So I have opted to continue using the ISO Setting as this is what the photographer has (some) control over.

Any other camera companies doing this?

Yes, Olympus for example, and you can get a good idea by looking at the sort-able table (I would sort on stops descending) beneath the chart DxOMark Derived Measured ISO at PhotonsToPhotos. You can find similar information at DxOMark but it's not so easily accessible (IMO). Keep in mind that DxOMark only tested a limited number of Fujifilm cameras.

So this anomaly exists only for the X-Pro3 in your charts at this time?

-- hide signature --
 rwbaron's gear list:rwbaron's gear list
Fujifilm X-T2 Fujifilm X-T20 Fujifilm XF 18-55mm F2.8-4 R LM OIS Fujifilm XF 55-200mm F3.5-4.8 R LM OIS Fujifilm XF 10-24mm F4 R OIS +8 more
OP bclaff Veteran Member • Posts: 9,432
Re: Fujifilm X-Pro3 Sensor Measurements at PhotonsToPhotos

rwbaron wrote:

...

So this anomaly exists only for the X-Pro3 in your charts at this time?

Yes, if by anomaly we mean that the curve is offset in two stages rather than the usual one.

In general all the Fuji curves lie to the right of where they "ought" to be.

In this case the values from ISO 320 to ISO 12800 are even further right than those at other ISO settings.

-- hide signature --

Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )

panther fan Regular Member • Posts: 380
Re: Fujifilm X-Pro3 Sensor Measurements at PhotonsToPhotos

bclaff wrote:

rwbaron wrote:

...

So this anomaly exists only for the X-Pro3 in your charts at this time?

Yes, if by anomaly we mean that the curve is offset in two stages rather than the usual one.

In general all the Fuji curves lie to the right of where they "ought" to be.

In this case the values from ISO 320 to ISO 12800 are even further right than those at other ISO settings.

I disagree you can clearly see see Fuji offset in the GFX50S above ISO 1600

panther fan Regular Member • Posts: 380
Re: Fujifilm X-Pro3 Sensor Measurements at PhotonsToPhotos
5

bclaff wrote:

Sugyn wrote:

Doesn't this undermine the useability of the testing. How do we know when a camera company is undermining the results.

Agree.

Shouldn't you be normalizing the data?

There has been discussion over the years about making the x-axis at PhotonsToPhotos more like that at DxOMark where they use their "Measured ISO" value.

The problem is that I have no consistent way of doing that.

For example I have tested many cameras not tested by DxOMark and in any case they are no longer testing cameras at a reasonable pace.

I looked into Adobe Baseline Exposure but found the quality of their data well below that of DxOMark partly because they seem to have shifted standards at some point but retained old legacy data.

And then there are companies like Fujifim who have a proprietary (but known) value in their Exif information.

So I have opted to continue using the ISO Setting as this is what the photographer has (some) control over.

Any other camera companies doing this?

Yes, Olympus for example, and you can get a good idea by looking at the sort-able table (I would sort on stops descending) beneath the chart DxOMark Derived Measured ISO at PhotonsToPhotos. You can find similar information at DxOMark but it's not so easily accessible (IMO). Keep in mind that DxOMark only tested a limited number of Fujifilm cameras.

I am sorry that the next paragraph might sound kinda harsh. And I am very thankfull for the data provided for free on the internet.

But that kinda makes the data almost useless for comparing cameras. People want to compare DR at a given light level, not at a given arbitrary label. Is there no way to fix that in the future?

OP bclaff Veteran Member • Posts: 9,432
Re: Fujifilm X-Pro3 Sensor Measurements at PhotonsToPhotos
1

panther fan wrote:

bclaff wrote:

Sugyn wrote:

Doesn't this undermine the useability of the testing. How do we know when a camera company is undermining the results.

Agree.

Shouldn't you be normalizing the data?

There has been discussion over the years about making the x-axis at PhotonsToPhotos more like that at DxOMark where they use their "Measured ISO" value.

The problem is that I have no consistent way of doing that.

For example I have tested many cameras not tested by DxOMark and in any case they are no longer testing cameras at a reasonable pace.

I looked into Adobe Baseline Exposure but found the quality of their data well below that of DxOMark partly because they seem to have shifted standards at some point but retained old legacy data.

And then there are companies like Fujifim who have a proprietary (but known) value in their Exif information.

So I have opted to continue using the ISO Setting as this is what the photographer has (some) control over.

Any other camera companies doing this?

Yes, Olympus for example, and you can get a good idea by looking at the sort-able table (I would sort on stops descending) beneath the chart DxOMark Derived Measured ISO at PhotonsToPhotos. You can find similar information at DxOMark but it's not so easily accessible (IMO). Keep in mind that DxOMark only tested a limited number of Fujifilm cameras.

I am sorry that the next paragraph might sound kinda harsh. And I am very thankfull for the data provided for free on the internet.

But that kinda makes the data almost useless for comparing cameras. People want to compare DR at a given light level, not at a given arbitrary label. Is there no way to fix that in the future?

Well, I think "useless" is "kinda harsh"; most comparisons are fine it's just some manufacturers like Fujifim (and Olympus) what have "non-standard" ways of doing ISO setting in their raw data.

-- hide signature --

Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )

OP bclaff Veteran Member • Posts: 9,432
Re: Fujifilm X-Pro3 Sensor Measurements at PhotonsToPhotos

panther fan wrote:

bclaff wrote:

rwbaron wrote:

...

So this anomaly exists only for the X-Pro3 in your charts at this time?

Yes, if by anomaly we mean that the curve is offset in two stages rather than the usual one.

In general all the Fuji curves lie to the right of where they "ought" to be.

In this case the values from ISO 320 to ISO 12800 are even further right than those at other ISO settings.

I disagree you can clearly see see Fuji offset in the GFX50S above ISO 1600

Yeah, the analog ISO range for the 50S is from ISO 100 to ISO 1280; ISO 1600 is an "extended" ISO. ISO 1600 and up are implemented using Raw Exposure Bias.

-- hide signature --

Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )

deednets Forum Pro • Posts: 10,449
Re: Fujifilm X-Pro3 Sensor Measurements at PhotonsToPhotos

bclaff wrote:

panther fan wrote:

bclaff wrote:

Sugyn wrote:

Doesn't this undermine the useability of the testing. How do we know when a camera company is undermining the results.

Agree.

Shouldn't you be normalizing the data?

There has been discussion over the years about making the x-axis at PhotonsToPhotos more like that at DxOMark where they use their "Measured ISO" value.

The problem is that I have no consistent way of doing that.

For example I have tested many cameras not tested by DxOMark and in any case they are no longer testing cameras at a reasonable pace.

I looked into Adobe Baseline Exposure but found the quality of their data well below that of DxOMark partly because they seem to have shifted standards at some point but retained old legacy data.

And then there are companies like Fujifim who have a proprietary (but known) value in their Exif information.

So I have opted to continue using the ISO Setting as this is what the photographer has (some) control over.

Any other camera companies doing this?

Yes, Olympus for example, and you can get a good idea by looking at the sort-able table (I would sort on stops descending) beneath the chart DxOMark Derived Measured ISO at PhotonsToPhotos. You can find similar information at DxOMark but it's not so easily accessible (IMO). Keep in mind that DxOMark only tested a limited number of Fujifilm cameras.

I am sorry that the next paragraph might sound kinda harsh. And I am very thankfull for the data provided for free on the internet.

But that kinda makes the data almost useless for comparing cameras. People want to compare DR at a given light level, not at a given arbitrary label. Is there no way to fix that in the future?

Well, I think "useless" is "kinda harsh"; most comparisons are fine it's just some manufacturers like Fujifim (and Olympus) what have "non-standard" ways of doing ISO setting in their raw data.

So for "most" companies your charts are useful, except Fuji and Olympus?

Since Panther didn't ask about other companies and this is about the X-pro3, are you saying that for the X-pro3 your chart is in fact not particularly relevant? E. G. you can't compare cameras using your charts when it comes to the X-pro3?

Another Fuji cheats, but this time even more than ever before? Christoph 21 still around?

Any benefits in all this?

Deed

 deednets's gear list:deednets's gear list
Sony RX1R II Fujifilm X-T3 Fujifilm X-Pro3 Fujifilm 50mm F2 R WR Fujifilm XF 14mm F2.8 R +7 more
OP bclaff Veteran Member • Posts: 9,432
Re: Fujifilm X-Pro3 Sensor Measurements at PhotonsToPhotos
1

deednets wrote:

bclaff wrote:

panther fan wrote:

bclaff wrote:

Sugyn wrote:

Doesn't this undermine the useability of the testing. How do we know when a camera company is undermining the results.

Agree.

Shouldn't you be normalizing the data?

There has been discussion over the years about making the x-axis at PhotonsToPhotos more like that at DxOMark where they use their "Measured ISO" value.

The problem is that I have no consistent way of doing that.

For example I have tested many cameras not tested by DxOMark and in any case they are no longer testing cameras at a reasonable pace.

I looked into Adobe Baseline Exposure but found the quality of their data well below that of DxOMark partly because they seem to have shifted standards at some point but retained old legacy data.

And then there are companies like Fujifim who have a proprietary (but known) value in their Exif information.

So I have opted to continue using the ISO Setting as this is what the photographer has (some) control over.

Any other camera companies doing this?

Yes, Olympus for example, and you can get a good idea by looking at the sort-able table (I would sort on stops descending) beneath the chart DxOMark Derived Measured ISO at PhotonsToPhotos. You can find similar information at DxOMark but it's not so easily accessible (IMO). Keep in mind that DxOMark only tested a limited number of Fujifilm cameras.

I am sorry that the next paragraph might sound kinda harsh. And I am very thankfull for the data provided for free on the internet.

But that kinda makes the data almost useless for comparing cameras. People want to compare DR at a given light level, not at a given arbitrary label. Is there no way to fix that in the future?

Well, I think "useless" is "kinda harsh"; most comparisons are fine it's just some manufacturers like Fujifim (and Olympus) what have "non-standard" ways of doing ISO setting in their raw data.

So for "most" companies your charts are useful, except Fuji and Olympus?

Generally you can make comparisons within a manufacturer, which is one common use of the charts.

For some, like Fuji and Olympus, the reader should be aware that ISO settings may not be comparable.

Since Panther didn't ask about other companies and this is about the X-pro3, are you saying that for the X-pro3 your chart is in fact not particularly relevant? E. G. you can't compare cameras using your charts when it comes to the X-pro3?

I'd say the X-Pro3 chart is not really relevant; I'd substitute something like the X-T3 instead.

Another Fuji cheats, but this time even more than ever before? ...

It's the first time I've seen Fuji do this in the middle of their ISO range.

Any benefits in all this?

I'm having trouble imagining any benefit.

-- hide signature --

Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )

Truman Prevatt
Truman Prevatt Forum Pro • Posts: 10,332
Re: Fujifilm X-Pro3 Sensor Measurements at PhotonsToPhotos
9

deednets wrote:

bclaff wrote:

panther fan wrote:

bclaff wrote:

Sugyn wrote:

Doesn't this undermine the useability of the testing. How do we know when a camera company is undermining the results.

Agree.

Shouldn't you be normalizing the data?

There has been discussion over the years about making the x-axis at PhotonsToPhotos more like that at DxOMark where they use their "Measured ISO" value.

The problem is that I have no consistent way of doing that.

For example I have tested many cameras not tested by DxOMark and in any case they are no longer testing cameras at a reasonable pace.

I looked into Adobe Baseline Exposure but found the quality of their data well below that of DxOMark partly because they seem to have shifted standards at some point but retained old legacy data.

And then there are companies like Fujifim who have a proprietary (but known) value in their Exif information.

So I have opted to continue using the ISO Setting as this is what the photographer has (some) control over.

Any other camera companies doing this?

Yes, Olympus for example, and you can get a good idea by looking at the sort-able table (I would sort on stops descending) beneath the chart DxOMark Derived Measured ISO at PhotonsToPhotos. You can find similar information at DxOMark but it's not so easily accessible (IMO). Keep in mind that DxOMark only tested a limited number of Fujifilm cameras.

I am sorry that the next paragraph might sound kinda harsh. And I am very thankfull for the data provided for free on the internet.

But that kinda makes the data almost useless for comparing cameras. People want to compare DR at a given light level, not at a given arbitrary label. Is there no way to fix that in the future?

Well, I think "useless" is "kinda harsh"; most comparisons are fine it's just some manufacturers like Fujifim (and Olympus) what have "non-standard" ways of doing ISO setting in their raw data.

So for "most" companies your charts are useful, except Fuji and Olympus?

Since Panther didn't ask about other companies and this is about the X-pro3, are you saying that for the X-pro3 your chart is in fact not particularly relevant? E. G. you can't compare cameras using your charts when it comes to the X-pro3?

Another Fuji cheats, but this time even more than ever before? Christoph 21 still around?

Any benefits in all this?

Deed

I doubt if it has anything to do with cheating.  The issue is ISO is not all that relevant with today's CMOS sensors with on sensor amplification.  ISO is a relic of film and even then there was ambiguity.  For example many people when they calibrated film for use in the Zone System four TriX to be ISO 200, not 400.  ISO is suppose to be related to sensitivity.  But even in B&W film - the actual ISO was dependent on the developer and development protocol.

Dynamic range is a measure of the amount of light difference between the minimal detectable signal to saturation. The minimal detectable signal can often be subject to some sort of debate depending on what is defined as detectable.  But saturation is pretty clear.  Maybe a better criteria would be the sensor noise floor to saturation.  On sensor amplification have the impact of raising the noise floor which decreases the dynamic range and amplification is applied by raising the ISO values.

What even makes the discussion more confusing is the claim that many of todays sensors are "ISO-less", which means you are as well off applying no amplification to the image on sensor and then raising the entire image in post processing.  In that case you are not raising the noise floor reducing the resulting dynamic range you are working with the full sensor dynamic range at "base ISO."  If this is true - all we really need to compare cameras is the sensor dynamic range at base ISO.

Of course ISO is mostly for JPEG's and the standard defines it for JPEG's not Raw and the appropriate on sensor amplification is required for an acceptable JPEG image.

It is amazing how much bandwidth has been wasted over time over the ISO debates.  It is important if you shoot JPEG and not so much if you shoot Raw and process.  A sensor has a base sensitivity and dynamic range.  As on sensor application is applied through raising ISO, then the noise floor goes up and the dynamic range goes down.  If one simply set the ISO at base and then shoots and raises the values after the fact from the raw files - with today's sensors as DPR points out often - you get about the same resulting dynamic range in the resulting images or ISO-less sensors.

-- hide signature --

"If you learn only methods, you’ll be tied to your methods, but if you learn principles you can devise your own methods." Ralph Waldo Emerson
___
Truman
www.pbase.com/tprevatt

 Truman Prevatt's gear list:Truman Prevatt's gear list
Fujifilm X-H1 Fujifilm XF 35mm F1.4 R Fujifilm XF 56mm F1.2 R XF 90mm Fujifilm 50mm F2 R WR +8 more
robert1955 Veteran Member • Posts: 4,742
Re: Fujifilm X-Pro3 Sensor Measurements at PhotonsToPhotos
1

Truman Prevatt wrote:

deednets wrote:

bclaff wrote:

panther fan wrote:

bclaff wrote:

Sugyn wrote:

Doesn't this undermine the useability of the testing. How do we know when a camera company is undermining the results.

Agree.

Shouldn't you be normalizing the data?

There has been discussion over the years about making the x-axis at PhotonsToPhotos more like that at DxOMark where they use their "Measured ISO" value.

The problem is that I have no consistent way of doing that.

For example I have tested many cameras not tested by DxOMark and in any case they are no longer testing cameras at a reasonable pace.

I looked into Adobe Baseline Exposure but found the quality of their data well below that of DxOMark partly because they seem to have shifted standards at some point but retained old legacy data.

And then there are companies like Fujifim who have a proprietary (but known) value in their Exif information.

So I have opted to continue using the ISO Setting as this is what the photographer has (some) control over.

Any other camera companies doing this?

Yes, Olympus for example, and you can get a good idea by looking at the sort-able table (I would sort on stops descending) beneath the chart DxOMark Derived Measured ISO at PhotonsToPhotos. You can find similar information at DxOMark but it's not so easily accessible (IMO). Keep in mind that DxOMark only tested a limited number of Fujifilm cameras.

I am sorry that the next paragraph might sound kinda harsh. And I am very thankfull for the data provided for free on the internet.

But that kinda makes the data almost useless for comparing cameras. People want to compare DR at a given light level, not at a given arbitrary label. Is there no way to fix that in the future?

Well, I think "useless" is "kinda harsh"; most comparisons are fine it's just some manufacturers like Fujifim (and Olympus) what have "non-standard" ways of doing ISO setting in their raw data.

So for "most" companies your charts are useful, except Fuji and Olympus?

Since Panther didn't ask about other companies and this is about the X-pro3, are you saying that for the X-pro3 your chart is in fact not particularly relevant? E. G. you can't compare cameras using your charts when it comes to the X-pro3?

Another Fuji cheats, but this time even more than ever before? Christoph 21 still around?

Any benefits in all this?

Deed

I doubt if it has anything to do with cheating. The issue is ISO is not all that relevant with today's CMOS sensors with on sensor amplification. ISO is a relic of film and even then there was ambiguity. For example many people when they calibrated film for use in the Zone System four TriX to be ISO 200, not 400. ISO is suppose to be related to sensitivity. But even in B&W film - the actual ISO was dependent on the developer and development protocol.

Dynamic range is a measure of the amount of light difference between the minimal detectable signal to saturation. The minimal detectable signal can often be subject to some sort of debate depending on what is defined as detectable. But saturation is pretty clear. Maybe a better criteria would be the sensor noise floor to saturation. On sensor amplification have the impact of raising the noise floor which decreases the dynamic range and amplification is applied by raising the ISO values.

What even makes the discussion more confusing is the claim that many of todays sensors are "ISO-less", which means you are as well off applying no amplification to the image on sensor and then raising the entire image in post processing. In that case you are not raising the noise floor reducing the resulting dynamic range you are working with the full sensor dynamic range at "base ISO." If this is true - all we really need to compare cameras is the sensor dynamic range at base ISO.

Of course ISO is mostly for JPEG's and the standard defines it for JPEG's not Raw and the appropriate on sensor amplification is required for an acceptable JPEG image.

It is amazing how much bandwidth has been wasted over time over the ISO debates. It is important if you shoot JPEG and not so much if you shoot Raw and process. A sensor has a base sensitivity and dynamic range. As on sensor application is applied through raising ISO, then the noise floor goes up and the dynamic range goes down. If one simply set the ISO at base and then shoots and raises the values after the fact from the raw files - with today's sensors as DPR points out often - you get about the same resulting dynamic range in the resulting images or ISO-less sensors.

(Emphasis mine) and it is pretty useless [and not intended for]  for numerical comparisons.

 robert1955's gear list:robert1955's gear list
Fujifilm X-E3 Fujifilm XF 16mm F1.4 R WR Fujifilm 50mm F2 R WR Samyang 12mm F2.0 NCS CS XF 90mm
Iuvenis Senior Member • Posts: 1,275
Re: Fujifilm X-Pro3 Sensor Measurements at PhotonsToPhotos

Any benefits in all this?

I'm having trouble imagining any benefit.

Isn't the advantage that if you set auto ISO, you are less likely to lose highlight data?

As I understand it, if you set ISO manually, you can keep your ISO at 640 on an X-T3 even as the light levels dim, then push in post processing. That essentially gives the same amount of shadow noise as setting the ISO dial to higher settings, but reduces the chance you clip any highlights.

However, there are good reasons to use auto ISO. First, it means the camera automatically switches to base ISO if light levels are high enough. Second, it means the EVF can be used to give a visual indication of the brightness of the scene using your chosen settings.

I use the auto ISO, and when shooting in low light, the camera assumes I want to brighten the shadows more than I actually intend. For example, a photograph of someone's face lit by candlelight - I want them to be surrounded by darkness, the important part of the photograph is their face. In that case, I may even end up clipping highlights on the face, so I often dial down the exposure compensation dial to compensate.

With this new scheme, does it not mean that you can set auto ISO and still achieve some (ie. one stop) of the benefits of manually setting ISO at the point of the dual gain switch? So in the scenario I've given, I end up with 1 stop of exposure compensation by default and, more on the dial if I would like?

 Iuvenis's gear list:Iuvenis's gear list
Fujifilm X-T2 Fujifilm XF 16mm F1.4 R WR Fujifilm XF 56mm F1.2 R XF 90mm Fujifilm 16-55mm F2.8R LM WR +3 more
random78 Regular Member • Posts: 288
Re: Fujifilm X-Pro3 Sensor Measurements at PhotonsToPhotos

bclaff wrote:

deednets wrote:

bclaff wrote:

panther fan wrote:

bclaff wrote:

Sugyn wrote:

Doesn't this undermine the useability of the testing. How do we know when a camera company is undermining the results.

Agree.

Shouldn't you be normalizing the data?

There has been discussion over the years about making the x-axis at PhotonsToPhotos more like that at DxOMark where they use their "Measured ISO" value.

The problem is that I have no consistent way of doing that.

For example I have tested many cameras not tested by DxOMark and in any case they are no longer testing cameras at a reasonable pace.

I looked into Adobe Baseline Exposure but found the quality of their data well below that of DxOMark partly because they seem to have shifted standards at some point but retained old legacy data.

And then there are companies like Fujifim who have a proprietary (but known) value in their Exif information.

So I have opted to continue using the ISO Setting as this is what the photographer has (some) control over.

Any other camera companies doing this?

Yes, Olympus for example, and you can get a good idea by looking at the sort-able table (I would sort on stops descending) beneath the chart DxOMark Derived Measured ISO at PhotonsToPhotos. You can find similar information at DxOMark but it's not so easily accessible (IMO). Keep in mind that DxOMark only tested a limited number of Fujifilm cameras.

I am sorry that the next paragraph might sound kinda harsh. And I am very thankfull for the data provided for free on the internet.

But that kinda makes the data almost useless for comparing cameras. People want to compare DR at a given light level, not at a given arbitrary label. Is there no way to fix that in the future?

Well, I think "useless" is "kinda harsh"; most comparisons are fine it's just some manufacturers like Fujifim (and Olympus) what have "non-standard" ways of doing ISO setting in their raw data.

So for "most" companies your charts are useful, except Fuji and Olympus?

Generally you can make comparisons within a manufacturer, which is one common use of the charts.

For some, like Fuji and Olympus, the reader should be aware that ISO settings may not be comparable.

Since Panther didn't ask about other companies and this is about the X-pro3, are you saying that for the X-pro3 your chart is in fact not particularly relevant? E. G. you can't compare cameras using your charts when it comes to the X-pro3?

I'd say the X-Pro3 chart is not really relevant; I'd substitute something like the X-T3 instead.

Another Fuji cheats, but this time even more than ever before? ...

It's the first time I've seen Fuji do this in the middle of their ISO range.

Any benefits in all this?

I'm having trouble imagining any benefit.

This definitely makes it tricky to use these charts specially for Olympus and Fuji. When I had first looked at PDR results for EM1 Mark II, I was shocked to see it performing better than latest APS-C sensors. However looking at DXOMark charts I realized that this is just because of the ISO labeling.

For Fuji DXOMark doesn't even publish the results so there isn't any other resource apart from your charts to look at Fuji DR. Also your photographic DR charts are more useful and relevant compared to DXOMark numbers anyway. However given this ISO inconsistency issue I don't know how much faith I can put in the PDR results when I compare say Fuji X-T3 to say Sony A6400. So if there was any way to normalize the results in a way similar to how DXOMark does with measured ISO, that would indeed have been very welcome.

 random78's gear list:random78's gear list
Canon EOS M Nikon D7000 Sony a6000
Truman Prevatt
Truman Prevatt Forum Pro • Posts: 10,332
Re: Fujifilm X-Pro3 Sensor Measurements at PhotonsToPhotos

robert1955 wrote:

Truman Prevatt wrote:

deednets wrote:

bclaff wrote:

panther fan wrote:

bclaff wrote:

Sugyn wrote:

Doesn't this undermine the useability of the testing. How do we know when a camera company is undermining the results.

Agree.

Shouldn't you be normalizing the data?

There has been discussion over the years about making the x-axis at PhotonsToPhotos more like that at DxOMark where they use their "Measured ISO" value.

The problem is that I have no consistent way of doing that.

For example I have tested many cameras not tested by DxOMark and in any case they are no longer testing cameras at a reasonable pace.

I looked into Adobe Baseline Exposure but found the quality of their data well below that of DxOMark partly because they seem to have shifted standards at some point but retained old legacy data.

And then there are companies like Fujifim who have a proprietary (but known) value in their Exif information.

So I have opted to continue using the ISO Setting as this is what the photographer has (some) control over.

Any other camera companies doing this?

Yes, Olympus for example, and you can get a good idea by looking at the sort-able table (I would sort on stops descending) beneath the chart DxOMark Derived Measured ISO at PhotonsToPhotos. You can find similar information at DxOMark but it's not so easily accessible (IMO). Keep in mind that DxOMark only tested a limited number of Fujifilm cameras.

I am sorry that the next paragraph might sound kinda harsh. And I am very thankfull for the data provided for free on the internet.

But that kinda makes the data almost useless for comparing cameras. People want to compare DR at a given light level, not at a given arbitrary label. Is there no way to fix that in the future?

Well, I think "useless" is "kinda harsh"; most comparisons are fine it's just some manufacturers like Fujifim (and Olympus) what have "non-standard" ways of doing ISO setting in their raw data.

So for "most" companies your charts are useful, except Fuji and Olympus?

Since Panther didn't ask about other companies and this is about the X-pro3, are you saying that for the X-pro3 your chart is in fact not particularly relevant? E. G. you can't compare cameras using your charts when it comes to the X-pro3?

Another Fuji cheats, but this time even more than ever before? Christoph 21 still around?

Any benefits in all this?

Deed

I doubt if it has anything to do with cheating. The issue is ISO is not all that relevant with today's CMOS sensors with on sensor amplification. ISO is a relic of film and even then there was ambiguity. For example many people when they calibrated film for use in the Zone System four TriX to be ISO 200, not 400. ISO is suppose to be related to sensitivity. But even in B&W film - the actual ISO was dependent on the developer and development protocol.

Dynamic range is a measure of the amount of light difference between the minimal detectable signal to saturation. The minimal detectable signal can often be subject to some sort of debate depending on what is defined as detectable. But saturation is pretty clear. Maybe a better criteria would be the sensor noise floor to saturation. On sensor amplification have the impact of raising the noise floor which decreases the dynamic range and amplification is applied by raising the ISO values.

What even makes the discussion more confusing is the claim that many of todays sensors are "ISO-less", which means you are as well off applying no amplification to the image on sensor and then raising the entire image in post processing. In that case you are not raising the noise floor reducing the resulting dynamic range you are working with the full sensor dynamic range at "base ISO." If this is true - all we really need to compare cameras is the sensor dynamic range at base ISO.

Of course ISO is mostly for JPEG's and the standard defines it for JPEG's not Raw and the appropriate on sensor amplification is required for an acceptable JPEG image.

It is amazing how much bandwidth has been wasted over time over the ISO debates. It is important if you shoot JPEG and not so much if you shoot Raw and process. A sensor has a base sensitivity and dynamic range. As on sensor application is applied through raising ISO, then the noise floor goes up and the dynamic range goes down. If one simply set the ISO at base and then shoots and raises the values after the fact from the raw files - with today's sensors as DPR points out often - you get about the same resulting dynamic range in the resulting images or ISO-less sensors.

(Emphasis mine) and it is pretty useless [and not intended for] for numerical comparisons.

Yup!

-- hide signature --

"If you learn only methods, you’ll be tied to your methods, but if you learn principles you can devise your own methods." Ralph Waldo Emerson
___
Truman
www.pbase.com/tprevatt

 Truman Prevatt's gear list:Truman Prevatt's gear list
Fujifilm X-H1 Fujifilm XF 35mm F1.4 R Fujifilm XF 56mm F1.2 R XF 90mm Fujifilm 50mm F2 R WR +8 more
rwbaron Forum Pro • Posts: 14,332
Re: Fujifilm X-Pro3 Sensor Measurements at PhotonsToPhotos
2

random78 wrote:

bclaff wrote:

deednets wrote:

bclaff wrote:

panther fan wrote:

bclaff wrote:

Sugyn wrote:

Doesn't this undermine the useability of the testing. How do we know when a camera company is undermining the results.

Agree.

Shouldn't you be normalizing the data?

There has been discussion over the years about making the x-axis at PhotonsToPhotos more like that at DxOMark where they use their "Measured ISO" value.

The problem is that I have no consistent way of doing that.

For example I have tested many cameras not tested by DxOMark and in any case they are no longer testing cameras at a reasonable pace.

I looked into Adobe Baseline Exposure but found the quality of their data well below that of DxOMark partly because they seem to have shifted standards at some point but retained old legacy data.

And then there are companies like Fujifim who have a proprietary (but known) value in their Exif information.

So I have opted to continue using the ISO Setting as this is what the photographer has (some) control over.

Any other camera companies doing this?

Yes, Olympus for example, and you can get a good idea by looking at the sort-able table (I would sort on stops descending) beneath the chart DxOMark Derived Measured ISO at PhotonsToPhotos. You can find similar information at DxOMark but it's not so easily accessible (IMO). Keep in mind that DxOMark only tested a limited number of Fujifilm cameras.

I am sorry that the next paragraph might sound kinda harsh. And I am very thankfull for the data provided for free on the internet.

But that kinda makes the data almost useless for comparing cameras. People want to compare DR at a given light level, not at a given arbitrary label. Is there no way to fix that in the future?

Well, I think "useless" is "kinda harsh"; most comparisons are fine it's just some manufacturers like Fujifim (and Olympus) what have "non-standard" ways of doing ISO setting in their raw data.

So for "most" companies your charts are useful, except Fuji and Olympus?

Generally you can make comparisons within a manufacturer, which is one common use of the charts.

For some, like Fuji and Olympus, the reader should be aware that ISO settings may not be comparable.

Since Panther didn't ask about other companies and this is about the X-pro3, are you saying that for the X-pro3 your chart is in fact not particularly relevant? E. G. you can't compare cameras using your charts when it comes to the X-pro3?

I'd say the X-Pro3 chart is not really relevant; I'd substitute something like the X-T3 instead.

Another Fuji cheats, but this time even more than ever before? ...

It's the first time I've seen Fuji do this in the middle of their ISO range.

Any benefits in all this?

I'm having trouble imagining any benefit.

This definitely makes it tricky to use these charts specially for Olympus and Fuji. When I had first looked at PDR results for EM1 Mark II, I was shocked to see it performing better than latest APS-C sensors. However looking at DXOMark charts I realized that this is just because of the ISO labeling.

For Fuji DXOMark doesn't even publish the results so there isn't any other resource apart from your charts to look at Fuji DR. Also your photographic DR charts are more useful and relevant compared to DXOMark numbers anyway. However given this ISO inconsistency issue I don't know how much faith I can put in the PDR results when I compare say Fuji X-T3 to say Sony A6400. So if there was any way to normalize the results in a way similar to how DXOMark does with measured ISO, that would indeed have been very welcome.

I agree with what you've posted but for all intents and purposes aren't we splitting hairs? Fuji uses Sony sensors but with their 6x6 mosaic overlay so the characteristics should be quite similar if not identical. A tenth or even a quarter stop is not something to agonize over.

Bob

-- hide signature --
 rwbaron's gear list:rwbaron's gear list
Fujifilm X-T2 Fujifilm X-T20 Fujifilm XF 18-55mm F2.8-4 R LM OIS Fujifilm XF 55-200mm F3.5-4.8 R LM OIS Fujifilm XF 10-24mm F4 R OIS +8 more
random78 Regular Member • Posts: 288
Re: Fujifilm X-Pro3 Sensor Measurements at PhotonsToPhotos
1

rwbaron wrote:

random78 wrote:

bclaff wrote:

deednets wrote:

bclaff wrote:

panther fan wrote:

bclaff wrote:

Sugyn wrote:

Doesn't this undermine the useability of the testing. How do we know when a camera company is undermining the results.

Agree.

Shouldn't you be normalizing the data?

There has been discussion over the years about making the x-axis at PhotonsToPhotos more like that at DxOMark where they use their "Measured ISO" value.

The problem is that I have no consistent way of doing that.

For example I have tested many cameras not tested by DxOMark and in any case they are no longer testing cameras at a reasonable pace.

I looked into Adobe Baseline Exposure but found the quality of their data well below that of DxOMark partly because they seem to have shifted standards at some point but retained old legacy data.

And then there are companies like Fujifim who have a proprietary (but known) value in their Exif information.

So I have opted to continue using the ISO Setting as this is what the photographer has (some) control over.

Any other camera companies doing this?

Yes, Olympus for example, and you can get a good idea by looking at the sort-able table (I would sort on stops descending) beneath the chart DxOMark Derived Measured ISO at PhotonsToPhotos. You can find similar information at DxOMark but it's not so easily accessible (IMO). Keep in mind that DxOMark only tested a limited number of Fujifilm cameras.

I am sorry that the next paragraph might sound kinda harsh. And I am very thankfull for the data provided for free on the internet.

But that kinda makes the data almost useless for comparing cameras. People want to compare DR at a given light level, not at a given arbitrary label. Is there no way to fix that in the future?

Well, I think "useless" is "kinda harsh"; most comparisons are fine it's just some manufacturers like Fujifim (and Olympus) what have "non-standard" ways of doing ISO setting in their raw data.

So for "most" companies your charts are useful, except Fuji and Olympus?

Generally you can make comparisons within a manufacturer, which is one common use of the charts.

For some, like Fuji and Olympus, the reader should be aware that ISO settings may not be comparable.

Since Panther didn't ask about other companies and this is about the X-pro3, are you saying that for the X-pro3 your chart is in fact not particularly relevant? E. G. you can't compare cameras using your charts when it comes to the X-pro3?

I'd say the X-Pro3 chart is not really relevant; I'd substitute something like the X-T3 instead.

Another Fuji cheats, but this time even more than ever before? ...

It's the first time I've seen Fuji do this in the middle of their ISO range.

Any benefits in all this?

I'm having trouble imagining any benefit.

This definitely makes it tricky to use these charts specially for Olympus and Fuji. When I had first looked at PDR results for EM1 Mark II, I was shocked to see it performing better than latest APS-C sensors. However looking at DXOMark charts I realized that this is just because of the ISO labeling.

For Fuji DXOMark doesn't even publish the results so there isn't any other resource apart from your charts to look at Fuji DR. Also your photographic DR charts are more useful and relevant compared to DXOMark numbers anyway. However given this ISO inconsistency issue I don't know how much faith I can put in the PDR results when I compare say Fuji X-T3 to say Sony A6400. So if there was any way to normalize the results in a way similar to how DXOMark does with measured ISO, that would indeed have been very welcome.

I agree with what you've posted but for all intents and purposes aren't we splitting hairs? Fuji uses Sony sensors but with their 6x6 mosaic overlay so the characteristics should be quite similar if not identical. A tenth or even a quarter stop is not something to agonize over.

Bob

Well yes and no - X-T3 uses a 26MP sensor so it is different from the sensor used in Sony cameras  -- it may still be a Sony sensor but it is not the same one as used in A6xxx series. Plus the other thing is that every now and then there are more extreme cases such as Fuji XPro3 and E-M1 Mark II where the difference is much higher. Another example is Panasonic G9 which has PDR going neck to neck with A6400 but we have no clue how much of that is real and how much of that is down to ISO labeling difference.

 random78's gear list:random78's gear list
Canon EOS M Nikon D7000 Sony a6000
Keyboard shortcuts:
FForum MMy threads