On Wed, May 22, 2013 at 6:23 PM, Ciprian Dorin Craciun <[email protected]> wrote: > On Thu, May 23, 2013 at 12:50 AM, Bruce Walker <[email protected]> wrote: >>>>> The "underexposure" is exactly the problem: in most cases although >>>>> the JPEG (or the embedded JPEG in the RAW that we see the histogram >>>>> for) is overexposed, the actual RAW data is under exposed, to the >>>>> point that almost 25% of the histogram contains nothing. >>>> >>>> Since this strange effect only occurs after you tweak the camera >>>> settings to achieve this elusive UniWB thing, I'd respectfully suggest >>>> reseting your JPEG settings back to normal. >>> >>> On the contrary, this effect I've noted is **before** making any >>> "special" settings, i.e. straight "normal" settings. >> >> Ciprian, can you describe a scene or circumstances in which you have >> observed this very odd behavior? Maybe an example image? I'm >> non-plussed because in all of my shooting I've _never_ experienced >> that. And I can safely say that I've shot in just about every known >> lighting condition. [Known to me. :-)] > > At the link below you'll find the following: > * the `.dng` which is the raw image; (this is the only option for > K-30 for RAW;) > * the `.jpeg` from the camera; (I was shooting RAW+;) > * the `.thumb.jpg` which is the extracted JPEG from the `.dng` via > `dcraw`; > * the `.ppm` obtained from `dcraw` without WB; > * the `.txt` obtained from `rawshack`; > > http://data.volution.ro/ciprian/e3a8d2a8f1098b9053f28369c7a42a36/ > > The picture was taken in a bright morning, at a distance of about > 30cm, without touching any of the image parameters except of shooting > in RAW, and manual mode. (This is the "best" out of other shots.)
Great. I pulled these two images into Lightroom and grabbed the histogram from each. For anyone following along, here they are: https://www.dropbox.com/sh/fnoxm59ofnjlyyd/vBFifxkWts#/ First is the RAW (DNG), second is the JPEG. Yes, the JPEG indicates some highlight clipping. If the blinkies were on I would have dialed in 1/3-stop reduced exposure. Then I'd have compensated in Lightroom with a touch of increased exposure. Maybe. > Although I must make a small correction to my initial statement: I > was under the wrong impression that the RAW was **completely** > underexposed, which seems it was not the case. However in half of the > "higher" levels (from 1024 to 4095) there lies less 10% of the pixels, > meanwhile the midlevels (256 to 1024) contain almost 50% of the > pixels. (I've taken the red channel as the scene is dominated by red.) There's a huge difference between bright/dark images and over/under-exposure. And I've heard of pixel-peeping, but histogram-counting? > However my main statement that the JPEG is blown out, meanwhile > the RAW is somewhat underexposed still holds, as by looking at the > JPEG histogram you have the impression of an overblown red channel, > meanwhile the raw histogram says otherwise. This RAW image is pretty much perfectly exposed. Emphasis on _perfectly_. No clipping of highlights; trace clipping of shadows; data spread nicely across the entire histogram. It's a really great exposure -- and a very good looking plant too, by the way. An _excellent_ exposure. Now, I stand by my original rebuttal: toss your UniWB crud and go shooting. What you need is less theory and more practice. :-) -- -bmw -- PDML Pentax-Discuss Mail List [email protected] http://pdml.net/mailman/listinfo/pdml_pdml.net to UNSUBSCRIBE from the PDML, please visit the link directly above and follow the directions.
