RAW Histogram "calibration"

Submitted by niccoc1603 on Sun, 08/26/2018 - 01:08
Forums: 
General discussion

Hi there,
 
I am a new user of FRV.
I am trying to get along with the RAW histogram and get it properly calibrated for my camera SONY ALPHA A7III
from the manual I read:
the underexposure indication limit (meaning the dynamic range of the camera) can be set even more flexibly, depending on the ISO settings. For this, one needs to turn on the Preferences – Image Display - Exposure - ISO-based underexposure limit setting and set three parameters for the camera: underexposure level (dynamic range) for ISO 200, ISO 1600, and “for high ISO.”
Calculating the Underexposure Limit (dynamic range) will happen in the following manner

  • For ISO values of 1600 and lower, down to ISO 6, the linear change is such that the line passes through the value for ISO 200. In other words, the line passes through two user-defined points: ISO 1600 and ISO 200.
  • For ISO settings 1601 and above, the dynamic range decreases by one stop for each stop of ISO setting increase, but the limit is always greater or equal to the value that is entered for high ISO.

A7III is ISO-invariant and DR can be seen here
So basically ISO invariant from 100 to 500 and from 640 onward.
How can I properly calibrate RAW histogram in FRV?
Right now, all default settings I have a scale from -10/UE-4.6/+3. Does it already take into account the camera properties?
Thanks

lexa

Sun, 08/26/2018 - 02:02

Dear Sir:

Lower (shadow) end of usable dynamic range is matter of

 - personal taste (noise tolerance)

 - typical image editing/retouching patterns (do you usually 'open shadows' /made it brighter/?)

 - output media size (the larger the print, the less acceptable noise)

 - output media type (e.g. matte paper will hide some shadow color noise due to ink spread)

So, it is very hard to suggest some specific numbers for specific camera. Default values (9/7/3) are conservative enough for today full-frame cameras, and rather optimistic for micro-4/3 and large sensor compacts.

Please note:

  1. FRV defines dynamic range from 'overexposure limit', so it is full DR, +3EV (from midtone to saturation) is already included in calculation. So, 4.6stops @ISO1600 looks too conservative to me. I have not used A7III, but for A7RIII I prefer something about 6-7EV@ISO1600, depending on print size.
  2. Published DR values may be normalized for some standard output size (e.g. 8Mpix in DXOMark default plot). This allows inter-camera comparison 'in same conditions', but overestimates DR range for the case when full resolution is used (due to large pring and/or cropping)

--

Alex Tutubalin/FastRawViewer team

niccoc1603

Sun, 08/26/2018 - 02:09

Many thanks for your quick reply.
 
According to A7III ISO/DR chart I set the values as follows:
10.5EV @ISO200, 8.5EV @ISO1600, 3.5EV @high ISO
is it too agrressive?

lexa

Sun, 08/26/2018 - 06:12

Dear Sir:

as mentioned in my initial reply, useable DR depends on may aspects, both personal (noise tolerance) and output type/size (larger prints are less tolerant to noise because on small prints values are averaged, so noise is less).

So, any specific value suggestion are incorrect because specific subjective (personal) and objective restrictions are not taken into account.

Meanwhile, 10.5EV @ISO200 looks ok (for me!) for A7III (assuming it is more or less the same as my A7R-III). I mostly do not use high ISO values on any camera, so could not suggest anything for higher ISOs (I use 6-7EV for ISO 1600 point, but this value is rarely used in my real culling, because used ISO range is mostly limited to 100,200, 640).

 

--

Alex Tutubalin/FastRawViewer team

niccoc1603

Sun, 08/26/2018 - 07:00

thank you
If I understand correctly, the UE value is the exposure below which shadow recovery does not "look good" anymore, right?
 

lexa

Sun, 08/26/2018 - 07:52

Generally - yes. Devil is in details as usual: if you decrease print size (to, for example, 4x6 in, or for web-sized FullHD (2Mpix) or even less), then several pixels will be averaged to make single output pixel, so noise will decrease due to pixel averaging.

That's why it is impossible to select single 'UE level' suitable for all cases.

--

Alex Tutubalin/FastRawViewer team

niccoc1603

Sun, 08/26/2018 - 08:44

in an ISO-invariant regime, should the UE value be constant?

lexa

Sun, 08/26/2018 - 11:57

In ISO-invariant mode  DR drops by 1 stop on 1-stop ISO change

--

Alex Tutubalin/FastRawViewer team

Howard (not verified)

Tue, 11/20/2018 - 20:51

This thread raised two questions in my mind.
1. What is the signficance of the 117 mark when viewing JPG's? Isn't the middle either 118.9 or 127.5?
2. You said:
"FRV defines dynamic range from 'overexposure limit', so it is full DR, +3EV (from midtone to saturation) is already included in calculation.". 
Is it assuming 3.0 stops over some unspecified mid-point or is it based on the maximum value that RawDigger would see or what? In other words, how does it decide what "EVzero" is for the histogram?  I've noticed that for some shots, I can increase the exposure to more than +3.0 before hitting an overexposure indication, but usually not.  not much. Also, I can't find "overexposure limit" on the preferences, just some things that sound like they might be the same but not that exact label.
 
Thanks.

LibRaw

Tue, 11/20/2018 - 22:07

Dear Sir:

The most common practice with cameras, in our experience, is to render the midpoint (the raw data number resulting from exposing according to a simple metering, like spot or centre-average) as 117 (assuming they use sRGB and Adobe with simple gamma 2.2, here is the supporting calculation: 0.18^(1/2.2)*255 = 116.9).

FastRawViewer determines dynamic range down from clipping point, clipping point being determined either by the explicit number in the file metadata ("camera provided highlight limit" in FastRawViewer Preferences -> "Image display" -> Exposure tab, this may be a bit conservative), or by recognizing the histogram shapes characterisical to clipping - please have a look at https://www.rawdigger.com/howtouse/rawdigger-histograms-overexposure-shapes

The EV0 point is based on our knowledge of calibration of in-camera meters, for each individual camera model. The standard suggests close to 3 EV headroom (the "distance" between the midpoint and clipping). Thus the generic calibration for raw is about 12.7%, not 18% as many beleive. Camera designers often deviate from this generic recomendation ensuring even more headroom to protect the highlights (thus the myth of cameramakers "overstating ISO speed"). The less "noisy" is the system, the more headroom in highlights could be afforded; except mostly for "extended" ISO: for extended low ISO the calibration is offset to leave less room in highlights compared to regular ISO speed settings, for extended high ISO the opposite may take place; there are other exceptions, too. Adobe are referring to this calibration game while explaing the need for Baseline Exposure. Please have a look at https://www.rawdigger.com/howtouse/deriving-hidden-ble-compensation and pp. 93, 88, 139 of FastRawViewer Manual.

You can see the value of Baseline Exposure in the "BLE" field in EXIF panel of FastRawViewer.

For your convenience, FastRawViewer Manual comes with the installation (main Menu - Help - "PDF Manual") and is fully searchable. You can also download it separately in higher resolution from https://www.fastrawviewer.com/download - direct link is http://updates.fastrawviewer.com/data/FastRawViewer-Manual-ENG.pdf

Add new comment