OOC JPGs are lens (distortion) corrected!
That's no surprise.
Early sample wide-angle shots looked far too 'straight' - I thought they were quite obviously 'corrected' in camera by software/firmware.
No surprise, no. But people kept disputing that in-camera and/or RAW converter lens correction would happen and now I know for sure. Additionally I have a way to see true lens distortion at various angles to make better educated decisions about what angle to use where high detail matters (digital correction means loss of detail).
There is so much 'digital processing' that occurs anyway - I really don't think that lens distortion correction is any worse than all the other routine de-mosaic interpolation, diagonal> orthogonal re-sampling, NR, sharpening, etc, that are always carried out anyway.
'Raw' should always 'ideally' record/save all sensor data where ever possible - ideally allowing the user to change choices/options later.
I agree, and this is why I see the "10 mb RAW" file anomaly as a bug rather than a feature. But given how poorly all RAW converters work with full 18 mb DR 200/400 RAW images it currently is a bug that turned into a feature.
I see it to be quite the opposite - the fact that the raw data halves in size, is most likely only done for some good and sound reason. Not a bug, nor a feature as such, but a rational/logical result of a process like internal 'pixel binning'.
As you probably remember - I've previously suggested that it's likely that the camera may switch to a combination of software/ISO DR with 'pixel-pair binning' at higher ISO, instead of 'dual/split-exposure DR'.
I have not seen any indication yet that ISO DR uses different methods at different ISOs.
...except where the RAF files change by almost a half in size - that's certainly an 'indication' of something.
There are indications for half the sensor being underexposed and indications for the full sensor being underexposed, but the outcome is pretty much consistent. Even high ISO 18 mb RAWs still need to be binned by the RAW converter as it seems (did a quick and dirty comparison of 10 mb vs 18 mv RAW at ISO 1600 + DR 400), so again no apparent difference as far as ISO is concerned.
When looking at the raw data - 'ISO DR' data will always 'appear' like underexposure, indistinguishable from the under-exposed
half of the 'dual exposure-time DR' method.
Also, it is possible that the camera could employ 'dual-ISO DR' - which would produce identical data to that of 'dual exposure-time DR'.
One sure way to differentiate the two possible methods, as you know, is to detect the dual shutter speed timings the way you've done previously with a moving highlight subject/target.
Both ISO based DR and EXPosure based are a combination of hardware underexposure + curves. It doesn't really make sense to call one hardware and the other software.
Yes, indeed I mostly agree there - although it is sometimes necessary to use terms that others here are more familiar with.
BTW, ISO DR 400 pulls down highlights by about -2.5 EV, not just 2 full stops. So it seems that after hardware ISO/amplification underexposure happened the highlights are further pulled down by curves (there is headroom left).
I'm not convinced - I wonder if you may be mistaken there. It's complicated/confusing to try to evaluate 'EV' within 'curved' data.
It doesn't make sense to shift the 'normal exposure's highlights down further than the extra EV/DR gained by the lesser-exposed part.
It also seems like ISO based DR indeed uses only half the sensor for ISO/amplification tricks, which then has to be properly combined with the overexposed half just like EXPosure time based DR.
I don't think that is likely at all - I don't see your reasoning.
The reasoning is that RAW converters have a hard time combining both halves of the sensor only with 18 mb RAW files , regardless of the camera setting! When ISO DR would expose both halves the same then I fail to see the reason why only 18 mb files under-saturate (plus increased noise) in RAW software, but not in pre-combined RAW/JPG.
O.K. - I'm seeing the way you're thinking a bit more clearly now.
I agree it is possible that the 'ISO DR' could use two different ISO/amplification settings - although the actual 'exposure' is still the same (same shutter speed and aperture), so there is no
"overexposed half" as such.
Some further and more detailed tests would have to be done to take a closer look at this. If you've got spare time on your hands, feel free to to some of your own. I specifically need to take a look at DR 100 18 mb files, too.
I'm currently writing some code that enables me to directly access and analyse X10 RAF files' sensor data.
Unfortunately I don't own an X10 (or any other Fuji camera) - but I have downloaded a few RAF files to work on.
Maybe you'd like to volunteer to supply me with some more specific 'DR' sample files - if, as and when, I'm ready/able to analyse them...?
When the camera doesn't use it's 'dual/split-exposure' method, then all pixels are exposed the same, so they should all be the same signal/exposure and all equally useful and valuable.
Yes, they should. But since the outcome is
very different this is an indication that ISO DR might also be using split exposure for the sensor halves.
Split gain/ISO, yes, quite possibly - but that's not split 'exposure' - the 'exposure' (EV) is the same.
If the 'gain/ISO' is split then I should be able to see and confirm this using the code I'm working on.
