Well I randomly picked the first and third articles you quote to read and amazingly both of them appear to support the generally understood position that RAWS exposed to the right, short of clipping, produces optimum results rather than the opposite you are suggesting.
a) Dpreview article: The OPENING WORDS in the first article you quote actually say the opposite. 'Sensors are linear devices. If you double the amount of light, the sensor output will double, as long as the pixels are not full[1]. Once a pixel reaches full capacity, it will give a constant or "clipped"' b) camera_characterisation.pdf: Have you actually picked an article that has pretty much nothing to do with the point but full of maths in the hope no-one would want to read it? :-) I did have a quick look and found almost nothing of relevance, but what I did find were side comments supporting the general perception that linearity is poorest at low signal values i.e. you want to overexpose to reduce these. There are also some graphs showing this much more clearly but not reproducible here. A quote from this article is as follows: 'The Sony DXC-930 camera which we used for our experiments is quite linear for most of its range, provided it is used with gamma disabled. However, in all three channels it has a substantial 12 response to no light (camera black) as well as a slight non-linearity for small pixel values' I'll accept that very long exposures have other things going on but in general voltage stored in photo sites is directly proportional to the number of photons they trap, as explained in that first article. Then that voltage is read by an ADC, as you say an ADC is not perfect but will typically have an accuracy expressed as a % of full voltage range. That means the ADC will be more accurate when it measures the higher voltages and less accurate on the lower values. Just about everything argues that for optimum results (if you are prepared to go to the trouble) the way is to shoot RAWs exposing as much as possible but without clipping/over-saturating the sensor wells and I haven't (yet) seen any creditable study to the contrary. I haven't looked at the other two articles you quote as they appear to be just blog posts. -----Original Message----- From: Aahz Maruch [mailto:[email protected]] Sent: 12 July 2013 14:27 To: [email protected] Subject: Re: [Darktable-users] Bracket shot workflow? On Fri, Jul 12, 2013, Rob Z. Smith wrote: > From: Aahz Maruch [mailto:[email protected]] >> >> It's a little trickier than that because (per _Light: Science & >> Magic_), you do lose some detail when an exposure is shifted too far >> to the right (overexposure) even when there are no blown highlights > > This does sound just wrong. Could you post the reference so I can see > where that opinion comes from? That reference is a book by Fil Hunter, Paul Fuqua, and Steven Biver (borrowed it from my local library, yours may have it, too). Quacking around for "camera sensor linearity", there are similar comments online: http://www.dpreview.com/glossary/camera-system/sensor-linearity http://harvestimaging.com/blog/?p=1154 http://kobus.ca/research/publications/02/camera_characterization/camera_characterization.pdf http://www.qsimaging.com/blog/?p=81 It makes sense if you think about it: the ADC is not going to be perfect, and it will be less perfect as you approach the ends of the dynamic range. -- Hugs and backrubs -- I break Rule 6 http://rule6.info/ <*> <*> <*> Help a hearing-impaired person: http://rule6.info/hearing.html ------------------------------------------------------------------------------ See everything from the browser to the database with AppDynamics Get end-to-end visibility with application monitoring from AppDynamics Isolate bottlenecks and diagnose root cause in seconds. Start your free trial of AppDynamics Pro today! http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk _______________________________________________ Darktable-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/darktable-users The content of this email is private and confidential, and unless otherwise stated only the intended recipient may use the content of this email for its intended purpose. If you are not the intended recipient, you may not retain, copy, forward or disclose the information herein, and we ask you to notify the sender or contact our Customer Services department on 0844 633 1000 or at [email protected] The copyright and all other intellectual property rights subsisting in or to the contents of this email belong to NHBC or are used with the permission of the owner and all such rights are reserved. Recipients are asked to note that opinions, conclusions and other information in the contents of this email that do not relate to the official business of NHBC are neither given nor endorsed by NHBC. This email has been scanned for viruses, but NHBC does not accept any liability in respect of loss or damage caused by any virus which is not detected by its virus detection systems. Data Protecti on Act 1998. NHBC is the Data Controller for the purposes of the Act. Your personal details will be stored and processed in accordance with the Act for the purposes of dealing with your enquiry or claim and for research and statistical purposes. If you make a claim under a Buildmark policy you agree to your data being passed to others involved with your claim such as the original builder, or a consultant or remedial works contractor that we may employ in connection with your claim(s) and matter ancillary to your claim(s). Other than disclosure provided for in this statement, we will not pass any data about you to any other party without your permission unless we are required to do so by law. NHBC, the National House-Building Council, is a company registered by guarantee in England, registration number 320784, and it is authorised and regulated by the Financial Services Authority. ------------------------------------------------------------------------------ See everything from the browser to the database with AppDynamics Get end-to-end visibility with application monitoring from AppDynamics Isolate bottlenecks and diagnose root cause in seconds. Start your free trial of AppDynamics Pro today! http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk _______________________________________________ Darktable-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/darktable-users
