Wide Dynamic Range and colour management

[stoneage]

I have been reading this recently:-
http://blender.stackexchange.com/questi ... 5862#55862

It seems to be a very useful addition to colour management, giving more control over the scene referred values.

Also, I get the virtual greycard I always wanted


There is more discussion about it and other colour management here:-
http://chat.stackexchange.com/rooms/348 ... abbit-hole

[surena]

Really interesting!

[David Bluecame]

Hello,

I agree, this is very interesting. Thank you for bringing this up!

However, every time I read something about middle gray, I shiver. It is not at all clear to me whether middle gray is a liner value of 0.180, 0.214, 0.216 or any other value.

For example there are several different definitions at Wikipedia: https://en.wikipedia.org/wiki/Middle_gray

Plus the value 0.180 in the post you referred to... (it is 0.018 in their post but I suppose that's a typo?)

[stoneage]

Yes. I believe the intention was to make a wider range of values available to people making 'photo-real' renders, so they settled on the 0.18 convention to mimic the photographer's 18% grey.

Yeah, it must be a typo. I pointed at it in The Rabbit Hole, so hopefully someone who can edit will fix it.

[Samo]

As far as I know, digital cameras do not map a wide dinamic range linearly into low dinamic range but compress "out of range data" mainly into the low and upper end of the histogram. The sensor in my DRSL camera has got about 6 EV stops of range, +3 EV and -3 EV from the middle tone. The camera software will tonemap linearly about -/+ 2 EV stops from the middle point but the upper and lower stop of that 6 EV range will be compressed mainly into the upper and lower part of the histogram making an histogram spike. Anything beyond that +3 EV / -3 EV range will become finally over and under exposure respectively. This is an important issue since, as we have seen lately, brightness is closely related with noise, in the sense that we don't need to sample bright pixels for a montecarlo noise we can not see anyway, so rendering should be done always with final tone mapping curves IMO to take advantage of a differential adaptive algorithm. If not, we can find unexpected noise if we tonemap in an editing app. I will try to post several examples later as this is an issue I would like to explain in my grey 18 tutorials.

[stoneage]

My cheap DSLR has a dynamic range of nearly 13 EV. That's closer to six or seven stops either side of 'middle grey'. As far as I know it does use a transform curve and isn't linear.

The wide dynamic range LUT seems to give artists more information about the rendered values that reach the display.

:/

[Samo]

My cheap DSLR has a dynamic range of nearly 13 EV. That's closer to six or seven stops either side of 'middle grey'. As far as I know it does use a transform curve and isn't linear.

The wide dynamic range LUT seems to give artists more information about the rendered values that reach the display.

:/

Hi

Are you sure? I have a canon 1100D and it clips range either 3 EV above or below mid tone at ISO100. It probably has got more dynamy range but the usable one without post edition is just that. Six or seven stops are faily common for any entry digital camera and higher range is only available in full frame high end cameras. Range also depends on ISO values.

[stoneage]

Are you sure?

I can't find the Nikon link I looked at earlier, but I found these:-

http://www.imaging-resource.com/PRODS/n ... part-i.htm

http://www.trustedreviews.com/nikon-d33 ... ict-page-3

http://www.dxomark.com/Reviews/Nikon-D3 ... erformance

If I save .jpg instead of RAW the range will be less.