X-pro 2 sooc waxy skin?

[adzman808]

I've owned all 3 generations of X-Trans sensors.

 

There was a work a round for the 'waxy skin' skin on X-Trans II

 

Remember?

 

Shoot ISO 1600 raw with -2 stops EV, then use the in camera raw convertor to push the file 2 stops to get a SOOC jpeg

 

But why or how did this work? (ie why was this suggested as a solution, NOT did you think it fixed it)

 

Well X-Trans II (& I) is only "ISOless" to ISO 1600 then digital push afterwards, it was the camera's internal push of the post ADC data that made the image

 

X-Trans III is (apparently) 100% "ISOless" (with dual gain tech according to many, namely two base ISOs 200 and 800)

 

So different things are going on 'under the hood' with X-Trans III

 

I'm NOT saying that the XP/XT2 has no high ISO smoothing issues, but it's worth noting the sensor in these cameras is very different to the previous generations.

[petergabriel]

Didn't know about isoless to 1600 on x trans 1 & 2, but that explains why my x-pro 1 looks best from iso1600 and below.

[petergabriel]

@flesix

Great shots. Illustrates the waxy skin on x trans 2 expertly. It is beyond me why some still can't see the issue.

[kimcarsons]

I've owned all 3 generations of X-Trans sensors.

 

There was a work a round for the 'waxy skin' skin on X-Trans II

 

Remember?

 

Shoot ISO 1600 raw with -2 stops EV, then use the in camera raw convertor to push the file 2 stops to get a SOOC jpeg

 

But why or how did this work? (ie why was this suggested as a solution, NOT did you think it fixed it)

 

Well X-Trans II (& I) is only "ISOless" to ISO 1600 then digital push afterwards, it was the camera's internal push of the post ADC data that made the image

 

X-Trans III is (apparently) 100% "ISOless" (with dual gain tech according to many, namely two base ISOs 200 and 800)

 

So different things are going on 'under the hood' with X-Trans III

 

I'm NOT saying that the XP/XT2 has no high ISO smoothing issues, but it's worth noting the sensor in these cameras is very different to the previous generations.

 

I think you've confused yourself regarding what ISOless means. A dual gain system is explicitly NOT "ISOless", that is to say, you will get different results shooting at ISO 200 and boosting to 1600 in post than you would shooting at 800 and boosting to 1600 in post. ISOless means, very specifically, that shooting at the base ISO and boosting in post produces EXACTLY the same result as shooing at the desired ISO in camera.

Edited December 11, 2016 by kimcarsons

[adzman808]

I think you've confused yourself regarding what ISOless means. A dual gain system is explicitly NOT "ISOless", that is to say, you will get different results shooting at ISO 200 and boosting to 1600 in post than you would shooting at 800 and boosting to 1600 in post. ISOless means, very specifically, that shooting at the base ISO and boosting in post produces EXACTLY the same result as shooing at the desired ISO in camera.

Thanks

 

Yes I know what isoless means and its a term I don't personally like... (hence why I used "" around the word) PERSONALLY I feel that the term should be pre or post ADC push

 

I haven't got around to testing the claims of the dual gain tech on the XP2..

 

DPR claim the XP2 (and other cameras sharing the sensor) is ISOless, and Thom Hogan says it's dual gain tech.

 

The semantics come into play when we talk about having two base ISOs (aka your point about 800 pushed 1 stop vs 200 pushed 3), so if we say (or more accurately as you said) that a sensor with 2 base ISOs cannot be ISOless what you actually mean is that pushing files will not be consistent if you cross the 2nd ISO threshold

 

However leaving semantics aside, what we find is that """ISOless""" actually means zero pre ADC brightening, and in this respect the sensor found in the latest Fujis (allegedly) qualifies

 

See why I dislike the term?

[kimcarsons]

Thanks

 

Yes I know what isoless means and its a term I don't personally like... (hence why I used "" around the word) PERSONALLY I feel that the term should be pre or post ADC push

 

I haven't got around to testing the claims of the dual gain tech on the XP2..

 

DPR claim the XP2 (and other cameras sharing the sensor) is ISOless, and Thom Hogan says it's dual gain tech.

 

The semantics come into play when we talk about having two base ISOs (aka your point about 800 pushed 1 stop vs 200 pushed 3), so if we say (or more accurately as you said) that a sensor with 2 base ISOs cannot be ISOless what you actually mean is that pushing files will not be consistent if you cross the 2nd ISO threshold

 

However leaving semantics aside, what we find is that """ISOless""" actually means zero pre ADC brightening, and in this respect the sensor found in the latest Fujis (allegedly) qualifies

 

See why I dislike the term?

 

Fair enough. Now to dig a little deeper: I don't think the gain structure has anything whatsoever to do with the waxy skin tones issue. The workaround you mention contains a hint to the actual cause of the problem, though. The reason such a trick works is that Fuji's in-camera processing that determines the strength of the noise reduction effect applied is mapped to the ISO as shot, not taking the "push/pull processing" gain into account. None of Fuji's cameras allow the user to control the strength of noise reduction for each ISO, but I guarantee you there is such a mapping in place in the firmware (some cameras from other manufacturers do allow the user to edit this curve, which is welcome). The result is that at ISO 800, you may like the look of NR 0, but at ISO 6400 you find that NR -2 looks better (bearing in mind that NR 0 at 6400 is really applying a stronger effect than NR 0 does at ISO 800 due to the non-editable strength curve), but if you want to enforce this preference, you'll have to change the setting yourself, every time you change ISO. I'm not aware of any camera that allows the user to configure luminance and chrominance NR separately, but it is obvious that (to some extent due to the requirements of X-Trans demosaicking) Fuji's chrominance NR is too aggressive. If all of this sounds like a huge headache, well, yes, I agree, and that's one of the many reasons why most serious photographers shoot RAW, so they can worry about all this later when processing the images on a desktop computer.

Edited December 11, 2016 by kimcarsons

[adzman808]

Fair enough. Now to dig a little deeper: I don't think the gain structure has anything whatsoever to do with the waxy skin tones issue. The workaround you mention contains a hint to the actual cause of the problem, though. The reason such a trick works is that Fuji's in-camera processing that determines the strength of the noise reduction effect applied is mapped to the ISO as shot, not taking the "push/pull processing" gain into account. None of Fuji's cameras allow the user to control the strength of noise reduction for each ISO, but I guarantee you there is such a mapping in place in the firmware (some cameras from other manufacturers do allow the user to edit this curve, which is welcome). The result is that at ISO 800, you may like the look of NR 0, but at ISO 6400 you find that NR -2 looks better (bearing in mind that NR 0 at 6400 is really applying a stronger effect than NR 0 does at ISO 800 due to the non-editable strength curve), but if you want to enforce this preference, you'll have to change the setting yourself, every time you change ISO. I'm not aware of any camera that allows the user to configure luminance and chrominance NR separately, but it is obvious that (to some extent due to the requirements of X-Trans demosaicking) Fuji's chrominance NR is too aggressive. If all of this sounds like a huge headache, well, yes, I agree, and that's one of the many reasons why most serious photographers shoot RAW, so they can worry about all this later when processing the images on a desktop computer.

The (well known and well publicised) workaround that I mentioned (but didn't invent!!) is exactly what you say, a way of getting the internal FW to use a different NR algorithm

 

Always remember though... RAW offers a LOT more options, but raw SW is still reading the meta data (think "recipe book") that the camera inbeds into the raw file, so various raw sw interprets high iso RAFs differently. For example on my xt1 with C1 and a 6400 shot raf... c1 would default to adding grain as part of its default NR setting, but it didn't do this with base ISO shots

 

So raw does give one th freedom to work how one wishes, but with the caveat that one needs to understand ones raw convertor and how its reading fuji meta data

[kimcarsons]

The (well known and well publicised) workaround that I mentioned (but didn't invent!!) is exactly what you say, a way of getting the internal FW to use a different NR algorithm

 

Always remember though... RAW offers a LOT more options, but raw SW is still reading the meta data (think "recipe book") that the camera inbeds into the raw file, so various raw sw interprets high iso RAFs differently. For example on my xt1 with C1 and a 6400 shot raf... c1 would default to adding grain as part of its default NR setting, but it didn't do this with base ISO shots

 

So raw does give one th freedom to work how one wishes, but with the caveat that one needs to understand ones raw convertor and how its reading fuji meta data

 

Adding fake grain? Wow, that's taking it a bit too far if you ask me. The point is that you have a choice of software with which to process the RAWs---and aren't just stuck with whatever limitations Fuji wants to enforce. That choice includes Free Software options which don't try to do all the thinking for you and won't add NR, grain, distortion correction or anything else unless you explicitly request it.