Trevor G: Now I understand...

Started Jan 12, 2013 | Discussions thread
Mark H
Mark H Veteran Member • Posts: 3,612
X10 RAW/JPEG headroom measured 0.2 EV...

Trevor G wrote:

Kim Letkeman wrote:
Your test just shows that there is one more stop of latitude in the RAW file, ´╗┐which is true of almost every sensor by definition´╗┐ ...

The test shows nothing of the sort. It's not designed to show the clipping point, but to show (and successsfully at that) the considerable difference between RAW and JPEG with highlight recovery on a high contrast scene.

I use another, full contrast scene to evaluate clipping point. I've shown numerous examples from that but I have never established the native clipping point of the sensor, at DR100 in M size and ISIO100.

The X10 sensor, in RAW, DR100, ISO100 and M size clips before +0.7EV. It's perfectly fine at +0.3EV

In other words the native RAW highlight headroom for the X10 is around 0.5EV, which is typical of small sensor cameras. In this case we are treating the X10 as if it didn't have an EXR sensor.

NOTE from MOD: This post was edited to avoid further "disputes" in this thread. Please stick to the topic.

I find neither your, nor KL's statements (quote: "...about 1 more stop..(EDIT BY MOD)") are correct .

As I said before (EDIT BY MOD), camera designers simply do not leave upward of 1EV of sensors' highlight headroom DR unused/discarded at lowest their ISO. To do that would be just idiotic, given that all cameras strive to maximise DR/minimise noise.

I have now actually measured the X10's sensor's ISO-100/DR100% RAW highlight headroom versus it's JPEG output - and the result, is only about +0.2 EV.

The method...

Find a 'RAW + JPEG' image pair with a white area at just below '255' i.e. values of 253~254.

Find the corresponding 12 bit RAW values in the 'Green' channel of the RAW image file.

The corresponding RAW value was an average of '3264'.

Knowing the X10's JPEG tone curve (from DPReview's DR test plots) the JPEG '253.5' = '3264' RAW extrapolates to JPEG '255' = '3330' RAW.

The maximum X10 RAW signal value is 3839 (4095-256 offset) - [Note: this max/saturation value actually appears to reduce, gradually throughout the image area, most likely due to the pixels' signals drooping/decaying slightly whilst waiting their turn to be A/D converted.]

The RAW full highlight headroom is therefore, at the very most, about Log2(3839/3330) = +0.2 EV

That is about as accurate a measurement method as can be - and far better than relying on exposure compensation adjustments and camera and/or RAW developing software.

I see no reason for any other camera to be significantly different (at base ISO) - again, to repeat myself - camera designers simply wouldn't waste significant chunks of sensor highlight DR at base ISO for no good reason - and even the relatively small value of 0.2EV seems something of an inexplicable waste not to use it within the JPEG.

N.B. Again, as I've said before (e.g. here ) - where programs such as ACR/Lightroom appear to recover significantly greater headroom than this, it is often achieved by 'smart-guesstimating' what the clipped RAW channel(s) (usually the 'green' channel) might have been by extrapolation and/or manual colour correction.

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