Fujifilm X-T100 Photographic Dynamic Range (PDR) at PhotonsToPhotos

bclaff wrote:

A little behind some other recent models but pretty good performance:

Note that this is not a dual conversion gain sensor.

I'm always looking to add more measurements here's what I have/don't have.

A red box indicates that I have no measurements.

I'm also missing Fixed Pattern Noise (FPN) for the following: X-E2, X-Pro1, X-T1

If you can help collect the necessary RAF files get in touch by email or Private Message.

Bill, thank you for doing this, very impressive results for the x-t100, what I notice when using dpr's exposure lattitude and iso invariance tests is just how good the x-t100 stacks up compared with the competition in real life though. For example you could surmise from your charts that the em1 ii has a better sensor, reality is much more different though! Using dpr if we first look at lattitude we can see that x-t100 is very impressive compared to m43, I tested my own camera against the G9 and it was comparable in dr (as per your test) but in lattitude the em1 ii and g9 are clear evidence of pushing shadows compared to the larger sensor. What is so impressive here is how it stacks up against the x-t2 and of course the new FF 46mp z7.

As a second example going to 1/320, well it looks even more impressive. In the real world again the shadow pull with the x-t100 is every bit as good as demonstrated here.

Moving on to iso invariance its not difficult to see the full 1 stop, probably more improvement over m43 and again, probably on par with the z7 and better than the x-t2

Its difficult to understand Fuji's concept behind the x-t100, I am not complaining its a super star, but it measures well, but in practice it appears difficult to see anything meaningfully better.

tradesmith45 wrote:

... For the X-T100, the Shadow Improvement chart shows little shadow improvement

This nearly flat line indicates that the sensor is ISO In-variant:

w/ higher ISO & the Read Noise charts show much higher noise compared to say an X-T20 or X-T2.

As I stated in the original post, and as is evident by the Read Noise in DNs :

The X-T100 does not have dual conversion gain so it does not get the read noise improvement at ISO 800.

Would it be correct to conclude the X-T100 will have much worse S/N for astro imaging than the other 2?

I'm not sure it "much" more. PDR certain falls further behind at ISO 800 and above.

So does a Shadow Improvement of less than 1 EV mean the shadows are suppressed at higher ISO (to reduce noise)? The Canon 6Dii shows much higher than 1 EV shadow improvement w/ higher ISO.

No. Shadow Improvement is primarily indicative of how much read noise is upstream/downstream from the amplifier. The X-T100 is "ISO In-variant" whereas the 6D Mark II is not.

Regarding your request for RAW images for measuring fixed pattern noise, I have an X-T10 similar but not the same as an X-T1. Is that helpful to you?

No. I'm looking for exactly those models.

Is fixed pattern noise included in the read noise charts?

Yes. Read noise is as observed so Dark Signal Non-Uniformity (DSNU), a form of Fixed Pattern Noise (FPN), is included.

The FPN test is summarized on the Sensor Heatmaps page (nothing to do with heat, this form of visualization is called a heatmap.)
For example, here's the X-20 :

Best viewed "Original Size"

The false color is simply to help you see if there are any patterns.
The illuminated test does show some horizontal bands; not bad though.

My simple comparison of FPN - a 2 & 4 min. dark frame boosted 4EV during RAW conversion- shows a substantial difference between the XT10 & my XT2 w/ the later showing fewer pixels w/ FPN but they are much brighter.

In my test the X-T10 looks more vertical while the X-T2 more horizontal.

Best viewed "Original Size"

If we look to dpr's tests again in terms of lattitude we can see that the x-t100 is very clean, I like that I have the full dr at iso200, it certainly helps with lower light and getting shutter speed up and is a great compromise. What is interesting is just how well it compares to other aps-c milc, x-t2/a6500 and m43

Here we see 1/160 for iso100 and 5ev for the a6500 and I maintained the em1 ii to show how far behind m43 is, but the shutter speed to another stop on the x-t100 and x-t2 but the x-t100 is very impressive

If I was to need higher shutter and dr with base iso 100 then pushing the Sony is not possible, m43 is gone in comparison and x-t2, well the results are clear.

bclaff wrote:

tradesmith45 wrote:

... For the X-T100, the Shadow Improvement chart shows little shadow improvement

This nearly flat line indicates that the sensor is ISO In-variant:

w/ higher ISO & the Read Noise charts show much higher noise compared to say an X-T20 or X-T2.

As I stated in the original post, and as is evident by the Read Noise in DNs :

The X-T100 does not have dual conversion gain so it does not get the read noise improvement at ISO 800.

Would it be correct to conclude the X-T100 will have much worse S/N for astro imaging than the other 2?

I'm not sure it "much" more. PDR certain falls further behind at ISO 800 and above.

So does a Shadow Improvement of less than 1 EV mean the shadows are suppressed at higher ISO (to reduce noise)? The Canon 6Dii shows much higher than 1 EV shadow improvement w/ higher ISO.

No. Shadow Improvement is primarily indicative of how much read noise is upstream/downstream from the amplifier. The X-T100 is "ISO In-variant" whereas the 6D Mark II is not.

Regarding your request for RAW images for measuring fixed pattern noise, I have an X-T10 similar but not the same as an X-T1. Is that helpful to you?

No. I'm looking for exactly those models.

Is fixed pattern noise included in the read noise charts?

Yes. Read noise is as observed so Dark Signal Non-Uniformity (DSNU), a form of Fixed Pattern Noise (FPN), is included.

The FPN test is summarized on the Sensor Heatmaps page (nothing to do with heat, this form of visualization is called a heatmap.)
For example, here's the X-20 :

Best viewed "Original Size"

The false color is simply to help you see if there are any patterns.
The illuminated test does show some horizontal bands; not bad though.

My simple comparison of FPN - a 2 & 4 min. dark frame boosted 4EV during RAW conversion- shows a substantial difference between the XT10 & my XT2 w/ the later showing fewer pixels w/ FPN but they are much brighter.

In my test the X-T10 looks more vertical while the X-T2 more horizontal.

Best viewed "Original Size"

The Fat Fish wrote:

The X-T100 is a gem. I used one with my X-E3 for a day and preferred the images coming off the X-T100. I'd love that sensor in an X-T3 spec'd body.

I hope they are working on it too

tradesmith45 wrote:

bclaff wrote:

tradesmith45 wrote:

... For the X-T100, the Shadow Improvement chart shows little shadow improvement

This nearly flat line indicates that the sensor is ISO In-variant:

w/ higher ISO & the Read Noise charts show much higher noise compared to say an X-T20 or X-T2.

As I stated in the original post, and as is evident by the Read Noise in DNs :

The X-T100 does not have dual conversion gain so it does not get the read noise improvement at ISO 800.

Would it be correct to conclude the X-T100 will have much worse S/N for astro imaging than the other 2?

I'm not sure it "much" more. PDR certain falls further behind at ISO 800 and above.

So does a Shadow Improvement of less than 1 EV mean the shadows are suppressed at higher ISO (to reduce noise)? The Canon 6Dii shows much higher than 1 EV shadow improvement w/ higher ISO.

No. Shadow Improvement is primarily indicative of how much read noise is upstream/downstream from the amplifier. The X-T100 is "ISO In-variant" whereas the 6D Mark II is not.

Regarding your request for RAW images for measuring fixed pattern noise, I have an X-T10 similar but not the same as an X-T1. Is that helpful to you?

No. I'm looking for exactly those models.

Is fixed pattern noise included in the read noise charts?

Yes. Read noise is as observed so Dark Signal Non-Uniformity (DSNU), a form of Fixed Pattern Noise (FPN), is included.

The FPN test is summarized on the Sensor Heatmaps page (nothing to do with heat, this form of visualization is called a heatmap.)
For example, here's the X-20 :

Best viewed "Original Size"

The false color is simply to help you see if there are any patterns.
The illuminated test does show some horizontal bands; not bad though.

My simple comparison of FPN - a 2 & 4 min. dark frame boosted 4EV during RAW conversion- shows a substantial difference between the XT10 & my XT2 w/ the later showing fewer pixels w/ FPN but they are much brighter.

In my test the X-T10 looks more vertical while the X-T2 more horizontal.

Best viewed "Original Size"

Thanks Bill for the reply.

Let me restate what I think you're saying & what the data says regarding use for astro. I'll use XT100 vs XT20.

Single stage amplification & invariance means the XT100 probably should be used differently from the XT20. To match read noise levels (not always important for astro when shot noise dominates), the XT100 must be set to ISO 318 & boosted in post to match the XT20 @ ISO 800. At ISO 318, the XT100 will have a bit more DR but less?? shadow detail?

This is an area where translating from detailed data to practical visibility can be difficult. Having to struggle more to reveal shadow detail & retain star color for my XT2 compared to a rented XPro2, seems to me differences around 0.5 EV are meaningful for astro. But this also brings up issues of copy variation & measurement error. For example your charts of Shadow Improvement for an XPro2, XT2 & XT20 are slightly different - are those differences meaningfully accurate?

Regarding FPN, its not clear to me that the short exposure test you use for read noise reveals the types of problems encountered with astro & much longer exposures. For example a comparison of FPN from the D5300 & D7500 over at Stargazers Lounge reached the opposite conclusion (D5300 better) from what your Sensor Heat map data shows. Your comments here would be helpful. I use your info often & simply want to understand the meaning.

Here's my dumb (uncalibrated for black point) comparison of my XT10 & a rented XPro2. I can maybe see the horizontal vs vertical difference. In practice photo terms, the long -2-4 min.- exposures I use for starlit landscapes from the XP2 are fewer but more obvious compared to the XT10. I more often have to clone these out from my XT2.

Thanks again.

Would you like a dark frame from my x-t100, if so how long what iso?

tradesmith45 wrote:

...

Let me restate what I think you're saying & what the data says regarding use for astro. I'll use XT100 vs XT20.

We may be wandering off topic since this is not the astrophotography forum

Single stage amplification & invariance means the XT100 probably should be used differently from the XT20. To match read noise levels (not always important for astro when shot noise dominates), the XT100 must be set to ISO 318 & boosted in post to match the XT20 @ ISO 800. At ISO 318, the XT100 will have a bit more DR but less?? shadow detail?

You may want to look a Input-referred Read Noise rather than Read Noise in DNs.

If you click on cameras in the legend you'll get a table with some relevant information.
At ISO 318 the X-T100 has a capacity of 15,248e- with an Engineering Dynamic Range (EDR) of 12.2
At ISO 800 the X-T20 capacity is only 6,065e- with a slightly higher EDR of 12.3

So by giving up capacity you get a little deeper into the shadows.

This is an area where translating from detailed data to practical visibility can be difficult. Having to struggle more to reveal shadow detail & retain star color for my XT2 compared to a rented XPro2, seems to me differences around 0.5 EV are meaningful for astro. But this also brings up issues of copy variation & measurement error. For example your charts of Shadow Improvement for an XPro2, XT2 & XT20 are slightly different - are those differences meaningfully accurate?

The auto-scaling of the charts makes them look more different than they are.
I'd say there's no meaningful differnence there.

Regarding FPN, its not clear to me that the short exposure test you use for read noise reveals the types of problems encountered with astro & much longer exposures. For example a comparison of FPN from the D5300 & D7500 over at Stargazers Lounge reached the opposite conclusion (D5300 better) from what your Sensor Heat map data shows. Your comments here would be helpful. I use your info often & simply want to understand the meaning.

My FPN testing is specifically for things like DSNU and PRNU.
You're probably more interested in dark current and amp glow; I don't measure those.

Here's my dumb (uncalibrated for black point) comparison of my XT10 & a rented XPro2. I can maybe see the horizontal vs vertical difference. In practice photo terms, the long -2-4 min.- exposures I use for starlit landscapes from the XP2 are fewer but more obvious compared to the XT10. I more often have to clone these out from my XT2.

Thanks again.