The Advantages of DR400 vs DR200 vs DR100 - High Contrast Scene

Started 10 months ago | Discussions thread
Luego
Contributing MemberPosts: 941
Like?
Timur's summary of Fujifilm X10's functions and issues...
In reply to Trevor G, 10 months ago

Trevor, perhaps you want to refresh your memory. There was a time when you were just a student while Timur "the master of X10" did all the research.

At least provide the link to Timur's summary before you state misleading information about the X10.

http://www.dpreview.com/articles/3785306838/summary-of-fujifilm-x10s-functions-and-issues

L usually provides higher detail than M at all ISOs, but at ISO higher than 400 M size provides about 1 step worth of noise-reduction better luminance noise and overall less chroma blotching and white-noise shift.

- EXR DR (where half the sensor is exposed shorter and later) is only used at ISO lower than DR (ISO 100 + DR 200, ISO 100-320 + DR 400). Maximum exposure time is limited to 1/4 second in M-anual mode then. In A-perture mode this leads to a bug with EXR DR when exposure time is chosen longer than 1/4 second by the camera, switch to ISO 400+ or turn off DR if you need that much exposure. EXR DR does not replace clipping whites of the longer exposed half with information of the shorter exposed half, but only attenuates the clipping white to gray and then blends both halves (seen as a grey halo with a thin outline in between around specular highlights).

- HR DR (High Resolution) is used with ISO equal or higher than DR (i.e. ISO 400 + DR 400) and does not use the EXR capabilities of the sensor for DR. It can be used for both L and M images, but when combined with M then the X10 will use EXR SN pixel-binning in combination with DR. HR DR works by underexposing at lower ISO (amplification) and then pulling up shadows and mids via curves (at higher bit-rate RAW level before creating the JPG). It's the same as doing it manually via post-processing, but saves the extra work.

- Regardless of DR setting maximum exposure time in all other modes than M and A is limited to 1/4 second, this can lead to underexposure if you reach the limits of aperture and ISO. The camera will show the red underexposure letters, but still refuse to use longer exposure time.

- DR 400 provides higher dynamic range at all ISOs, but at the costs of less saturation in the deepest shadows and dark outlines around specular highlights (not just orbs). DR 200 can be a good compromise, but you have to choose the right setting for the right situation. EXR DR may increase the total dynamic range of the camera at ISO 100-200. HR DR does not increase the total dynamic-range of the camera, but it provides higher dynamic-range per ISO setting (dr normally decreases with higher ISO, so HR DR provides higher dr per ISO by internally using lower ISO before applying curves).

- EXR DR cannot be combined with flash, only HR DR is available when the built in flash is used (and likely the same with external flash).

- AUTO DR leads to overexposure (on my unit) even when all settings are the same as when you had set them manually. You cannot compensate this via -EV, because the overexposure only happens when AUTO DR chooses DR 200/400, not when it chooses DR 100. But the DR value chosen by AUTO DR is only visible _after_ you took the shot.

- AUTO ISO can also lead to a very slight change of exposure (but minimal), can be both more or less exposure. In combination with AUTO DR this can lead to even more overexposure, though.

- AUTO ISO (1600 and 3200) do not use their highest possible setting when the flash is used, even when this leads to underexposure. So AUTO ISO (1600) will max out at ISO 800, and AUTO ISO (3200) will max out at 1600.

- "+MOTION" IS only works in combination with AUTO ISO. It will increase ISO when it detect motion or camera-shake, but will allow lower ISO compared to not using "+MOTION" IS when the frame is steady (aka the X10 chooses higher ISO than necessary when "+MOTION" IS is turned off).

- AF-C may focus at a point slightly off center even when the cross-hair on aim and playback suggest the focus to be on the perfect center spot. This means that you may see a different focal plane on the image than what you aimed at even when the X10 confirmed to be in focus via green cross-hair. AF-S would rather report a failed AF in such a situation.

- "Spot" metering always meters on the center and not on the AF spot. The manual describes this properly, but it's still confusingly named.

- "Face Detection" automatically switches metering to "Multi", but in reality it meters and exposed on the detected face! Unfortunately the "superintelligent" flash does not measure on the face, which can lead to overflashed faces in some situation (especially if the face moved out of the center).

- "Tracking" AF does not switch metering and the metering does not follow the tracked object (aka not like Face Detection).

- "Tracking" AF works _better_ when power-saving is turned _off_. This is because power-saving not only dims the screen, but also reduces the refresh/frame-rate and seemingly this not only affects the output, but also the processing of on-screen information. According to the manual power-saving is turned off automatically for "Face Detection", but for "Tracking" it needs to be done manually.

- Aiming at a frame with bright highlights inside the metering area (depending on current metering setting) leads to the aperture closing down during aiming to provide a better preview on screen. Once you (half)press the shutter the aperture performs an "open-close-open" sequence that takes some time and is not related to auto-focus. The wider your chosen aperture to be for the shot the longer this sequence takes. Using AUTO ISO also has some very slight impact on the lengths of the sequence.

- Shooting "RAW + JPG" in M size at ISO at least double as high as DR leads to RAW file size being cut in half the size (does not happen with "RAW" only), from 18 mb to 10 mb. More important, this leads to SN pixel-binning happening inside the camera before RAW creation! The difference between pixel-binning inside the camera vs. inside the raw software is that the automatically applied curves to shadows and mids differ and thus the ranges/behavior to changes in the raw software is different (same value have vastly different outcome).

- RAW Converter EX/Silkypix seems to mostly throw away the information of the longer exposed sensor half of EXR DR raws (not just clipped highlights, but everything)! What you will get is the same as if you had exposed the whole image later and shorter. If any motion was present during exposure then some of the motion related regions may show wrong/missing colors and kind of dithering artifacts. This seems to be different with different images. What it does effectively is to put priority on the shorter (and thus worse) exposed sensor half and then may or may not add in some information from the longer exposed half.

-*-

Some clarification on dynamic range and how dr "expansion" works on the X10 via the use of DR 200/400 modes (EXR DR vs. HR DR):

According to DxO lab the X10 offers a maximum dr of 11.29 EV at ISO 100 and drops down to 8.15 EV at ISO 1600. This means that the higher the ISO the lower the dynamic-range.

*EXR DR = half the sensor is exposed shorter and later*

EXR DR is only usable for M sized images at ISO 100 + DR 200 or ISO 100-320 + DR 400, aka ISO value lower than DR value! All other combinations of ISO + DR use the other DR method called "High Resolution" (HR) DR. Additionally the maximum exposure time is limited to 1/4 second, likely to avoid motion artifacts that can appear because of the time-difference between the sensor halves (1/4 also seems to be the about limit of the image-stabilization's effectiveness).

Theoretically using EXR DR 400 at ISO 100 should extend the dynamic-range of the X10 to around 13.3 EV. In practice I suspect that it is a bit lower than that, because at least in-camera clipped highlights are rather ungracefully blended instead of sophistical replaced with the other sensor half. Before the blending happens the underexposed half seems to have its mid-tones lifted to match the longer exposed one. And because half the sensor is underexposed and collect less photons you also lose some information in the deepest shadows, usually in form of some lost saturation, at least in JPGs.

Once you use ISO 400 (9.8 EV dynamic range according to DxO) the theoretical dynamic range in combination with EXR DR would be around 11.8 EV. But usually you are using shorter exposure times with higher ISO or shot a lower light scene, so half the sensor would be even more underexposed and thus the practical gain would be nowhere higher than the original 11.3 EV that ISO 100 give at DR 100. No more benefit + possible drawbacks = makes no more sense to use EXR DR at ISO 400 and higher.

Raw converter software is responsible for properly blending the two sensor halves and applying curves. This is what most software still fails at, including RAW Converter EX, Silkypix and Lightroom. Capture One currently seems to do it best, but lacks support for 12 mp DR 100 files in return. No software offers sophisticated control over the process.

*High Resolution (HR) DR = full sensor exposing at lower amplification/ISO*

HR DR has several benefits over EXR DR and doesn't need an EXR sensor at all (which is the reason why you find it in the X100, too). Internally it exposes the shot at 1 (DR 200) or 2 (DR 400) stops lower ISO to protect highlights and then applies curves to the higher bit-rate raw data before creating the final JPG. The total highlight protection of HR DR JPGs is about 2.5 EV, which likely is done by pulling highlights further down by 0.5 EV from the raw data that offers that much headroom.

Since the camera uses lower ISO internally it gains dynamic-range, as can be seen by the DxO measurements (lower ISO = higher dynamic-range). Again I doubt that 2 stops of dynamic-range are gained in practice, especially since each stop of ISO does not correspond to a full stop of dynamic-range.

The benefits of HR DR over EXR DR are:

- Can be used with L sized images and can be combined with EXR SN pixel-binning with M sized images.

- Exposes the whole sensor at the same time and thus avoids motion artifacts of EXR DR (which happen due to differently exposed halves with EXR DR).

- Exposes the whole sensor for the full time and thus allows more photons to reach the AD converters. Interestingly HR DR still leads to loss of saturation in the lowest shadows just like EXR DR, so analog amplification seems to play a role, too.

- Allows full maximum exposure time to be used (8 seconds at ISO 400) vs. 1/4 second maximum of EXR DR.

- Is supported by all major raw converter software, even more so when pre-binned 10 mb raw files are used.

The reason why HR DR is only usable at ISO equal or higher to DR is simple. In order to use 1 or 2 stops lower ISO/amplification you need to start at higher ISO. The sensor simply is not capable of using lower ISO than 100, so you need ISO 200 for DR 200 (=ISO 100) and ISO 400 for DR 400 (=ISO 100).

And again, there is no "hardware" vs. "software". Both DR modes use hardware level exposure tricks (EXR uses shorter exposure time, HR uses lower ISO/amplification) and both modes seem to apply tone-curves afterwards. Only EXR DR is a "true" EXR sensor mode, but then HR DR allows to combine DR with EXR SN pixel-binning, too.

There are different types of RAW files on the X10:

- 12 mp size L with or without HR DR (RAW or RAW+JPG). It's the job of the RAW converter to demosaic the image to 4000x3000 pixels and apply tone-curves for HR DR (the camera just underexposes internally). Some converters throw away half the sensor information and only create an M size image (Capture One).

- 12 mp size M with EXR DR (RAW+JPG). It's the job of the RAW converter to demosaic the image to 2816 x 2112, apply curves to the shorter exposed half and blend both halves. Silkypix/EX throws away most (sometimes all) of the longer exposed half information and mostly (sometimes only) creates an image out of the longer exposed half (not just highlights stuff). These files handle differently in each converter, some better (Capture One), some in between (Lightroom), some worse (Silkypix).

- 12 mp size M with HR DR (RAW+JPG). It's the job of the RAW converter to demosaic the image to 2816 x 2112, apply curves for HR DR and pixel-bin both halves together for better SN.

- 12 mp size M without HR DR (RAW+JPG). It's the job of the RAW converter to demosaic the image to 2816 x 2112 and pixel-bin both halves together for better SN.

- 6 mp size M with HR DR (RAW+JPG, ISO 2x DR). The camera seems to do the pixel-binning and likely even curves! Some RAW converters cannot load these files at all (Capture One), those that can should all be able to handle them well. But as a consequence to the camera's pre-production these files react differently to RAW converters' controls (exposure, highlights etc) compared to 12 mp files. This means that the same RAW converter setting will have a different outcome.

Reply   Reply with quote   Complain
Post (hide subjects)Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow