I've tried developing the same RAW in Rawtherapee this morning, but cannot generate a histogram that looks like the histogram from ACR6.2 (shown above). I also have not yet been able to recover the whispy cloud and dust bunny in the "blown" areas of the sky that can be recovered in ACR6.2/Elements.

Maybe ACR6.2 is doing some very clever processing, modelling data in the (clipped) blue and green channels from (unclipped) data in the red channel.

I need to revisit this ETTR stuff again and, perhaps using Rawdigger, see if I can discover exactly what the in camera histograms and digital preview blinkies are telling me.

Andrew

No, none of that. If they have asked for "Back to basics" then give them exactly that. What's the nearest you can get to a disposable film camera .....

* Kit lens set at 28mm or 28mm prime
* All settings at factory defaults
* Must be taken at ISO 100, Sunny 16
* In camera JPEGs - no RAW
* No external processing other than copying in-camera JPEG onto memory stick

Yeah, basic .......... I like that

Andrew

Yes, it loads ONLY RAW files - if I'm going to try to speed read, maybe I should take more time over it.

I downloaded Rawdigger and have had a brief play with it. This has left me thoroughly confused. Let me explain ......

I loaded an image into Rawdigger that I took as RAW some months back (i.e. not a test photo taken for the purpose). The photo has a large area of sky that is very bright - the sun is just off frame to the left. With RGB rendering selected, Rawdigger shows a large area of sky as being blown. The Rawdigger histogram shows the green and blue channels as being seriously clipped.

If I load the RAW into Photoshop Elements 8 (which uses ACR6.2), the histogram appears to mirror the conclusion from Rawdigger, that there's some serious clipping going on here - but if I drag the exposure slider one stop to the left, the histogram looks much better with no clipping at all. This is a screen dump of the histogram from ACR6.2.




It doesn't look clipped to me!! Note the two black triangle thingies at the top corners - none of the RGB channels are (significantly) clipped. How odd - this suggests to me that there wasn't any real clipping at all, just a lot of bright (but not clipped) pixels that made the (pre-exposure correction) histogram appear clipped. If there was real clipping, then the right hand edges of the (exposure corrected) histogram curves would be like a cliff-edge - and they are not!

If I open the image in the Elements editor and look at the histogram for each RGB channel, they all look quite normal with no obvious clipping. There are only a few tens of pixels at RGB levels above 250. So still no real clipping. So, what is going on with Rawdigger? It seems to be speaking with forked tongue!!

I need to read the Rawdigger manual properly and maybe checkout forums discussing Rawdigger. At the moment, my tentative conclusion is that Rawdigger isn't really showing what it purports to (though I concede that my understanding might be defective). Is it possible that the highlight clipping it shows on the RGB rendered image (and on its histograms) is what the JPEG preview image is like?

Any thoughts?

Andrew

OK, thanks, maybe I should give it a look.

Andrew

I think it depends which old lens - some are not very good.

Although I have occasionally seen some CA in photos taken by the kit lens, my copy really is very good in that respect. According to reviews I've seen, Pentax's other normal length zooms (16-45mm, 17-70mm and 16-50mm) all have MUCH higher levels of CA even than the Mk1 18-55mm lens.

I suppose it is to be expected that there will be more sample variability in lower priced lenses (such as the kit) - so not surprising that different folks have different experiences of the lenses they have.

Andrew

I'm not familiar with Rawdigger, but from a quick scan of the webpage, it seems to say you have to do the RAW conversion and then save as a TIFF before you can use it. This makes is a pain to use, especially as you can see the RAW histogram in ACR and which channels (if any) are clipped.

My question was directed at DrO ... I was wondering how he knew when actual clipping occured on the K5.

Andrew

Wow, I don't have twelve - just this mottley few .....

* Sigma 24mm f/2.8 Superwide II (used quite a lot)
* Kenlock 28mm f/2.8 (not used)
* Chinon 28mm f/2.8 (not often used)
* Pentax SMC M 50mm f/1.7 (used *fairly* regularly)

I have tested all of these (using test charts and also with real subjects) against the ALII kit lens . Only the Pentax M50 is clearly and consistently sharper everywhere, at f/5.6. The Sigma 24mm, which is famed for being sharp, is not obviously sharper in the centre of the frame but is definitely softer than the kit lens in the corners. The Chinon is similar to the kit lens at f5.6, but is soft wide open, so is not really worth its place in my camera bag. The Kenlock is ........... not very good.

Perhaps I have an exceptionally good copy of the kit lens.

Andrew

How did you check whether actual clipping occured?

Just yesterday (after reading an artical about ETTR), I took a series of test photos as RAW at one third stop increments, starting from just below the exposure at which there was visible clipping in the histogram and/or blinking in the Digital Preview. I then loaded each RAW in turn into Photoshop Elements/ACR6.2 and dragged the exposure slider to the left, until the rightmost edge of the histogram curve parted company with the axis. If the rightmost end of histogram curve looks like a cliff, then you have highlight clipping. If instead, the curve gradually falls to zero and there is no cliff, then you don't have clipping.

For the K30, it was only once I had loaded a RAW taken at 1.33 stops above the level at which clipping first appeared in the K30's histogram or as blinking in the Digital Preview, that I started to see genuine highlight clipping. This was the article that I happened across yesterday that spurred me into doing the testing. Seemingly you can also push the Nikon D800E about 1.3 stops beyond the level at which blinkies start to appear in a Digital Preview before you get real highlight clipping. Interesting that the K5 is different .....


Andrew

I've always done that, but recently have reconfigured the RAW/Fx button on my K30 for Digital Preview.

One thing I've discovered recently is that Digital Preview and test exposure histograms are very conservative and you don't actually get clipped highlights unless you open up at least 1.33 stops AFTER they start to show the first signs of clipping (on the K30).

Andrew




Andrew

I agree with John's advice. Also, make sure that you compare the results of your new (to you) old primes with those from a 18-55mm kit lens. Although the primes may go wider, there are very few oldies that are significantly better than the kit lens at f/5.6 and onwards (Pentax SMC M 50mm f/1.7 is an obvious example).

Andrew