X20 and underexposing

tomhongkong

Veteran Member
Messages
5,714
Solutions
2
Reaction score
4,535
Location
HK
It has been stated that there are circumstances when the X20 underexposes, and in the ensuing attempts it makes to correct for this it loses detail, creates watercolour etc. This seems especially true of shots of landscapes in low/poor light.

There are some samples in this forum which certainly look like that.

However there are several glowing reviews which do not mention it, and many users are quite happy. If it occurs it may be user induced, or for a particular auto setting.

In order to try to get to the bottom of it, it would be useful if X20 users who have experienced this problem could say how the X20 was set when it occured, and preferably include a sample with exif data.

If there is some setting which we should avoid, it would clarify the matter.

I have been hanging off in buying an X20 until the IQ is confirmed, and this seems to be the question mark against it at present

Many thanks

tom
 
tomhongkong wrote:

It has been stated that there are circumstances when the X20 underexposes, and in the ensuing attempts it makes to correct for this it loses detail, creates watercolour etc. This seems especially true of shots of landscapes in low/poor light.

There are some samples in this forum which certainly look like that.

However there are several glowing reviews which do not mention it, and many users are quite happy. If it occurs it may be user induced, or for a particular auto setting.

In order to try to get to the bottom of it, it would be useful if X20 users who have experienced this problem could say how the X20 was set when it occured, and preferably include a sample with exif data.

If there is some setting which we should avoid, it would clarify the matter.

I have been hanging off in buying an X20 until the IQ is confirmed, and this seems to be the question mark against it at present

Many thanks

tom
This is not an underexposure by metering, it is the way how Fuji processes images from X-trans sensors. Fuji is exposing the sensor as if lower ISO is selected (like for ISO 1000, when 1600 is selected on the camera), and then shadows and midtones are gradually lifted during processing (both JPEG and to the lesser extent, RAW). This trick leaves more highlight protection area in the exposure, and makes the sensor seem less noisy. It works fine with APS-C sensors, with their much bigger natural dynamic range and low noise, but in using the same technique with the smaller 2/3 sensor, Fuji obviously miscalculated it can pull it off. The noise and smearing in low contrast scenes is the result of always pulling shadows and midtones up in the images, and then exposing noise issues of the much smaller sensor, and then throwing loads of selective NR to control the excessive noise. One thing for X20 owners to try would be to neutralize the default shadow/midtone lifting Fuji uses (and deliberate underexposure), by selecting Shadow Tone Hard setting (to pull down artificially lifted shadows), select +2/3 Ev or 1 Ev always dialed in, and use Highlight Tone Soft to eliminate potential highlight clipping problems because Fuji is probably using hard curve in highlights to increase apparent contrast of images, which is always reduced by aggressive shadow lifting. Maybe it will not work, if tone mapping used by Fuji is variable depending on the scene, or so complex it can't be easily compensated during shooting. Then RAW is the only solution, with careful monitoring to expose to the right in histogram.
 
It is not exposing as if a lower ISO was selected, that would mean longer exposure time. If anything it's the other way around. In those situations where the X20 seems to underexpose it uses too short an exposure time and then makes up for it via digital gain/tone-mapping. Still several images I have seen also look too dark, others do not.

Trevor's comparison examples done with Average metering were exposed 1/3 EV to 1/2 EV longer by the X20 compared to his X10 for roughly the same scene (even minimal differences have a big impact with Average metering, though). So if the X20 uses less analog gain for the same ISO setting compared to the X10 (which uses higher gain than what the user selects according to DXO) then in Trevor's "Average" example the metering made up for it.

All a bit quirky...
 
Timur Born wrote:

It is not exposing as if a lower ISO was selected, that would mean longer exposure time. If anything it's the other way around. In those situations where the X20 seems to underexpose it uses too short an exposure time and then makes up for it via digital gain/tone-mapping. Still several images I have seen also look too dark, others do not.

Trevor's comparison examples done with Average metering were exposed 1/3 EV to 1/2 EV longer by the X20 compared to his X10 for roughly the same scene (even minimal differences have a big impact with Average metering, though). So if the X20 uses less analog gain for the same ISO setting compared to the X10 (which uses higher gain than what the user selects according to DXO) then in Trevor's "Average" example the metering made up for it.

All a bit quirky...
I meant to say Fuji is processing the sensor as less sensitive then stated on the dial, meaning when ISO 1600 is selected, Fuji is adjusting sensor gain as if ISO 1000 is selected, creating deliberate underexposure as part of their processing approach with X-trans. I am not sure is it because the X-trans CFA array is less transparent then Beyer, so sensitivity is lost, or deliberate choice in less gaining of the sensor. In the other forum there are posts about this gain/sensitivity difference (or loss) with X-trans.
 
That's why I corrected you. :-P

That being said, I don't think this is the whole picture (pun intended). Several of the images I had access to do not suffer from too low a gain, but from too short exposure time, which in turn leads to more noise. So there is underexposure happening, at least in some situations.
 
Wasted headroom is happening in any mode but perhaps SR+ is the worst.

Just another thing for a firmware tweak - I hope!
 
Whichever way it is stated, the result seems to be that sometimes the shot is underexposed, and that the shadow/mid tones are lifted to compensate, leaving more headroom for the highlights but poor IQ in the shadow/mid tone areas.

However, it obviously does not occur all the time as we have seen some good shots from the X20, and one of my objectives was to try to identify when it did occur (and therefore whether there was a way round it....average metering is one suggestion)

I did read the other thread suggested by Timur, and it further confirms that the problem exists, without identifying what causes it.

tom
 
Last Thursday, I got my X20...... and was extremely disappointed because nearly all my X20 landscape test-photos were severly underexposed.

Today, I learned from an excellent technical support person of Fujifilm, that I should have shot my landscape photos in "Spot Metering" mode ....not in the X20 "Multi Metering" mode (which is a kind of Matrix Metering mode).

To a non-expert like me, that seemed to be quite an odd recommendation (with my Nikon D5000, I shoot nearly all my Landscape Photos in Matrix Metering Mode ...and most of the time they look nice - that has also been true for all digital cameras that I used since 2003).

But the recommendation of shooting landscape Photos with the X20 "Spot Metering" Mode really helped.

Even if I have now a work-around (that I do not like much): I really wish that an otherwise nice camera like the X20 provides a Mutli Metering Mode that functions well for Landscape Photos (and also, of course, for other types of Photos).

Do you think, that there is a good probability, that we will get for the X20 a good "Multi Metering" mode via a Firmware Fix? Or is it usual that Fujifilm Cameras do-not/can-not provide a "Multi Metering"/"Matrix Metering" generating well-metered Landscape Photos?

Also, I find it quite odd, that all of the Automated Modes ("Auto" Mode, "Advanced SR Auto" Mode, and the "Landscape Scene" of "SP Scene Position") of the X20 can use only a Metering Mode (the "Multi Metering" Mode) ... that often results in poorly metered Landscape Photos.

Crossing my fingers.
 
Last edited:
Robert Eckerlin wrote:

Last Thursday, I got my X20...... and was extremely disappointed because nearly all my X20 landscape test-photos were severly underexposed.

Today, I learned from an excellent technical support person of Fujifilm, that I should have shot my landscape photos in "Spot Metering" mode ....not in the X20 "Multi Metering" mode (which is a kind of Matrix Metering mode).

To a non-expert like me, that seemed to be quite an odd recommendation (with my Nikon D5000, I shoot nearly all my Landscape Photos in Matrix Metering Mode ...and most of the time they look nice - that has also been true for all digital cameras that I used since 2003).

But the recommendation of shooting landscape Photos with the X20 "Spot Metering" Mode really helped.

Even if I have now a work-around (that I do not like much): I really wish that an otherwise nice camera like the X20 provides a Mutli Metering Mode that functions well for Landscape Photos (and also, of course, for other types of Photos).

Do you think, that there is a good probability, that we will get for the X20 a good "Multi Metering" mode via a Firmware Fix? Or is it usual that Fujifilm Cameras do-not/can-not provide a "Multi Metering"/"Matrix Metering" generating well-metered Landscape Photos?

Also, I find it quite odd, that all of the Automated Modes ("Auto" Mode, "Advanced SR Auto" Mode, and the "Landscape Scene" of "SP Scene Position") of the X20 can use only a Metering Mode (the "Multi Metering" Mode) ... that often results in poorly metered Landscape Photos.

Crossing my fingers.
This is a design issue of the camera, X-Trans sensors are underexposed, and then boosted during processing by the camera. This is just how Fuji implements X-trans, it works with bigger sensors, not so much with the smaller one in X20. Therefore, firmware fix of the issue would be a pretty big change, like a complete re-working of the design philosophy of X-trans, not very likely to happen. As I said before, one thing to try would be to neutralize the default shadow/midtone lifting Fuji uses (and deliberate underexposure), by selecting Shadow Tone Hard setting (to pull down artificially lifted shadows), select +2/3 Ev or 1 Ev always dialed in, and use Highlight Tone Soft in settings to eliminate potential highlight clipping problems, because Fuji is probably using hard curve in highlights to increase apparent contrast of images, which is always reduced by aggressive shadow lifting. It might work, but maybe not, depending on how much Fuji's doctoring of the exposure curve is messed up.
 
DS21

Thank you very much for your answer. I am not at all an expert, and I am therefore perhaps asking a dumb question.
If the X20 and its sensor

- are capable to provide a good metering in "Spot Metering" mode,

- why would they not be capable to also provide a good metering in "Matrix Mode" (under the assumption that at Fujifilm they understand how to implement a good Matrix Mode metering)?
 
Robert Eckerlin wrote:

DS21

Thank you very much for your answer. I am not at all an expert, and I am therefore perhaps asking a dumb question.
If the X20 and its sensor

- are capable to provide a good metering in "Spot Metering" mode,

- why would they not be capable to also provide a good metering in "Matrix Mode" (under the assumption that at Fujifilm they understand how to implement a good Matrix Mode metering)?
Spot metering evaluates only a small portion of the scene, the small area around the focus point. It should not work welll in landscape shots, but if it does work with X20 without series overexposure or underexposure (depending on do you focus on dark or light part of the scene), it just proves X20 metering is not working as it should. For example, if you focused on the darker part of the scene, spot metering would expose it higher and make it seem like it works in fixing underexposure. Fuji knows (or used to know) how to make good matrix metering, X20 is just messed up design, because they tired to use the X-trans sensor like on their bigger sensor APS-C cameras, but with smaller sensor they ended up doing a lot of fudging and tricks to compensate for small sensor smaller dynamic range (by deliberately underexposing), and noise (stronger NR and watercolor effects).
 
Something I noticed with the x10 is that if you have external flash set to On and you don`t use external flash, it will under expose in AP mode, even when it should give a decent exposer.
 
No, it's just short for aperture priority, which is what A on the dial is even shorter for.
 
Robert Eckerlin wrote:

Last Thursday, I got my X20...... and was extremely disappointed because nearly all my X20 landscape test-photos were severly underexposed.
It is not just the X20 - my X10 does exactly that.

What I have noticed is with *any* sky in the frame it exposes that sky perfectly...leaving the 'land' dark. Looking at the resultant histogram the highlights (sky) are just reaching the righthand edge so it is making every effort not to blow the sky out which of course leaves the 'land' dark.

I've compared with my LX3 which does expose the 'land' correctly but overexposes the sky to compensate - not something I had really noticed much until I got the X10 and saw the quality of the skies it was reproducing!

I often sacrifice the sky a bit and dial in anything from +0.3ev to +1ev or point the camera down (to avoid the sky), half press the shutter to take the exposure then reframe for the shot.
 
I've used spot metering for the last 30 years and also with my new x20 where it works perfectly. You don't use spot metering at random and hope things turn out right you have to carefully choose what you want to meter. It demands a bit more thought but it works 99% of the time.
 
yellodog wrote:

I've used spot metering for the last 30 years and also with my new x20 where it works perfectly. You don't use spot metering at random and hope things turn out right you have to carefully choose what you want to meter. It demands a bit more thought but it works 99% of the time.
I find the spot metering quite accurate. It feels like it performs better than my old D7000 and a colleague's D3100. In practice, there really is no automagic answer to the perfect scene metering. I mean, even WB is still iffy in some situations with my DSLRs and let's not forget to mention the Canon Rebels!

The real secret is to know your scene and understand how the camera will interpret it. Point-and-shoots are usually used by people who have no idea (nor need) of what happens in the background. Once a shot is taken, you will actually see a LOT of bad metering, and harsh flash exposure, no matter what P&S model but to the "common eye" it's "I don't see what's wrong".

Liz.
 
yellodog wrote:

I've used spot metering for the last 30 years and also with my new x20 where it works perfectly. You don't use spot metering at random and hope things turn out right you have to carefully choose what you want to meter. It demands a bit more thought but it works 99% of the time.
yellodog,

The X20 has been the only digital camera that I ever used, that was not able to expose correctly the majority of my landscape photos in Matrix Metering Mode.

And by the way: none of the auto shooting modes of the X20 support "spot Metering": neither the "Auto" shooting mode, nor the "Advanced SR Automode", nor the "SC Scene Position Mode" (for Scene Modes such as "Landscape"). What for are all these auto Shooting Modes of the X20, if they can not even expose correctly most Landscape photos?
 

Keyboard shortcuts

Back
Top