[2351HD]

Royce for that very reason that I expose for the JPEG preview and I set all JPEG image settings in the camera to low, low contrast, low saturation etc.

This allows me to see the RGB histogram as close to the raw data as possible, since live view only gives a luminance histogram.

[rob8888]

Hi Royce. Thanks for the analysis. The blue island frame is around 1.5 stops away from any clipping. It's been output in LR at +.36, contrast +5 with highlight and shadow recovery moderate, and a gentle grad on the top. This was one of the first frames that drew it to my attention to the problem as I was uploading jpegs to my website. This was the best I could get it after hours of jpeg fiddling, checking profiles and workflow to find what was going wrong is on this link:

http://www.naturespic.com/NewZealand/image.asp?id=51509

Worst I've seen in years of trannie scanning and various digital cameras. Just can't seem to do anything with it?

[itshimitis]

Royce for that very reason that I expose for the JPEG preview and I set all JPEG image settings in the camera to low, low contrast, low saturation etc.

This allows me to see the RGB histogram as close to the raw data as possible, since live view only gives a luminance histogram.

The risk with this strategy is that it isn't 100% correct - it's close, but not 100%, butt he only way of getting separate R, G, B readings so is a good guide.

[Kolor-Pikker]

If anybody wants to have a look at a file for me and see what you can see I'd be grateful. Here's one .DNG raw to download here (just happens to be right up against the highlight limit, but this doesn't always need to be so):

Rob8888, I took a quick look at your sample DNG file. The magenta area in the sky is that way because both green channels in the raw data are completely clipped in that region -- zero data. Highlight recovery is attempting to pull back some detail, but there's only red and blue channel info to use; the result is magenta. So totally clipped green is the immediate cause.

Why the green channels are clipped is another question. Is it as simple as an overexposed frame? You say underexposing doesn't help -- underexposing by how much? How does the RGB histogram look at the time of capture? Admittedly the camera histo is only showing data from the JPEG thumbnail, but I'm curious how you see the G channel vs. R & B in-camera at the time of capture.

Hi Royce. Thanks for the analysis. The blue island frame is around 1.5 stops away from any clipping. It's been output in LR at +.36, contrast +5 with highlight and shadow recovery moderate, and a gentle grad on the top. This was one of the first frames that drew it to my attention to the problem as I was uploading jpegs to my website. This was the best I could get it after hours of jpeg fiddling, checking profiles and workflow to find what was going wrong is on this link:

Worst I've seen in years of trannie scanning and various digital cameras. Just can't seem to do anything with it?

Ok so I checked out the supplied image, and also compared it to all the images with skies I ever shot with the Z, and here are some findings:



I usually leave quite a bit of headroom for highlights, the second histo is an image that has the sun in the frame, but the large spike is still located where it actually falls in most of my images, on the third line of the histogram.

And here is a snapshot I took in NY that has the brightest, bluest sky I ever shot with this camera so far:



The key thing here is that the sky is not the brightest part of the image, you can see that the blue spike in the histogram is closer to the third line than the edge, and as a result nothing I do with the sky results in banding.

I also did some experiments with the LED lamp, and while there are no noteworthy images to show, there certainly are issues that occur with certain wavelengths of blue light when using highlight recovery, but not magenta, although it causes the histogram to blow out sooner. I could do some more extensive testing to narrow it down, but I'm thinking this might be too controlled of a test, the lamp outputs only the wavelength you set it to, but sunlight contains all wavelengths of light.

When I say that this camera doesn't like overexposure I mean it. If I'd shot the photo of the beach that you supplied I wouldn't even bother trying to recover the sky, rather you should have used different settings - ISO100 instead of 200 and bump the aperture to at least f/9 to f/11 - as a result the brightest part of the image should have ended at around the same point as on my images, and any residual spikes on the histo are for specular highlights (chrome for example) and the sun.

---

Last edited by Kolor-Pikker; 02-27-2015 at 06:27 AM.

[Royce Howland]

Rob8888, willing to share the DNG for that blue island photo? I'll take a look at it on the weekend as well. If it was 1.5 stops down from clipping point, then I assume that indeed no channels are clipped. Then it's a separate issue from the others where the green channel is substantially clipped.

One case would be how to best process for preservation of the areas of continuous tone in blue. The other case how to reconstruct the sky when a critical part of the image data is actually missing from the capture. For the latter, a couple of things I think in terms of would be copying a portion of the R or B channel into the G channel; or doing some work in Lab mode to reverse the magenta and green shifts rather than trying to fight the colours in RGB mode.

[Kolor-Pikker]

I don't often have to contend with problem colors, but sometimes you can just take the easy way out and paint in the color yourself... what's stopping you from loading the image up in photoshop, selecting a paintbrush in color mode and painting in from a picked color? or select the sky and drag a gradient in color mode to overwrite the colors. As long as the tonal range is smooth and unaffected by this banding, which you can check by switching to black and white, then you can simply recolor the sky as you please.

[itshimitis]

I think different people have different ethical views on how much editing to do to an image. I'm not sure whether I would do what you suggest. Two reasons I'm not sure whether I'd want to or secondly whether I'd be any good at doing it in a smooth way.

[Kolor-Pikker]

I think different people have different ethical views on how much editing to do to an image. I'm not sure whether I would do what you suggest. Two reasons I'm not sure whether I'd want to or secondly whether I'd be any good at doing it in a smooth way.

I like to think of photography as the starting point for making an image, what happens after I take the shot is more important, although a technically good shot as a starting point is an essential part of it. As nice as it would be sometimes to just tweak some settings in your Raw converter and be done with it, if I have to pass an image through the meat grinder and re-light scene elements, replace bits, and re-color others to get the ideal end result, then so be it.

I'm an artist among other things, so I don't shy away from sometimes putting images under the knife, since photography comes with a lot of problems you wouldn't have to contend with if you just painted the scene in the first place. So long as it looks believable, it doesn't matter how you attained the result, the exception being reportage, which is in my opinion the only aspect of photography where ethics really matter.

Not knowing whether I want to do something and whether I could do it cleanly are thoughts that got ripped out of my head and tossed in the trash after working with people who breath down your neck to get the job done, and you have no choice but to do it and in a good way.

[itshimitis]

I like to think of photography as the starting point for making an image, what happens after I take the shot is more important, although a technically good shot as a starting point is an essential part of it. As nice as it would be sometimes to just tweak some settings in your Raw converter and be done with it, if I have to pass an image through the meat grinder and re-light scene elements, replace bits, and re-color others to get the ideal end result, then so be it.

I'm an artist among other things, so I don't shy away from sometimes putting images under the knife, since photography comes with a lot of problems you wouldn't have to contend with if you just painted the scene in the first place. So long as it looks believable, it doesn't matter how you attained the result, the exception being reportage, which is in my opinion the only aspect of photography where ethics really matter.

Not knowing whether I want to do something and whether I could do it cleanly are thoughts that got ripped out of my head and tossed in the trash after working with people who breath down your neck to get the job done, and you have no choice but to do it and in a good way.

I wasn't judging - I think we all have our own boundaries, some of them ethical, some of them technical ability. I have some weaknesses and one of them is technical ability in Photoshop. I still have a lot to learn in many areas of photography, not least in the technical side of editing.

[ndevlin]

Thanks for making that file available Rob. I can see exactly what you mean, with a strange magenta-ish arc developing in the sky when the exposure is lowered in the blue channel. Everything about this behavior looks like a clipped channel, despite the histogram displaying as within gamut for the relevant parts of the image. I have experienced nothing of the sort in almost 9,000 frames, including pretty heavy post-processing of sky-tones, so I wonder whether this isn't an issue somehow specific to your camera.

I think the only way you can isolate the problem is to shoot a scene side-by-side with two cameras, and/or do a decreasing exposure ramp on images of this sort as a test to see if a reduced exposure solves the problem.

Strange and perplexing indeed.

Good luck and let us know what comes of this. And sorry I can't offer more help.

- N.

[texandrews]

In photography there are indeed very different attitudes to what is acceptable, even outside of jounalism, so itshimitis is correct about that, as I'm sure you know

I'm an artist among other things, so I don't shy away from sometimes putting images under the knife, since photography comes with a lot of problems you wouldn't have to contend with if you just painted the scene in the first place.

And as a fellow artist, trained as a painter but having worked in and taught many media, I completely agree with you. After the 20th century, there is just nothing sacrosanct any longer about any medium. Thus looking at it from my angle, I find some of the more restrictive attitudes about what is ok and not ok in photography---cropping or not, processing beyond a certain point or not, & etc---just personal issues at best and precious at worst. I sort of understand how having to work on a photograph for "hours" might be off-putting (or impossible for some commercial work), but then again I sort of don't: when it takes days or weeks to finish a drawing, painting, sculpture, installation, or print run, "hours" seems like a fabulous deal! In fact, considering what it used to take me in a wet darkroom to get 5 identical 16x20 inch archival prints, dried and spotted and ready for framing, "hours" still seems cheap.

So long as it looks believable, it doesn't matter how you attained the result,...

I'm not even certain about the believable bit!

... the exception being reportage, which is in my opinion the only aspect of photography where ethics really matter.

I agree, if we lump in fashion and certain medical and technical photography, even though some of this latter is enhanced necessarily. I think there are some well documented things going on in fashion photography/post-processing that are very questionable ethically, even as outrageous as some fashion photography gets. In art? Well, Cubism is now over 100 years old....it is to students today as Impressionism was to me when I was a student. Next year DADA will be 100 years old, so there's some perspective....

[Kolor-Pikker]

And as a fellow artist, trained as a painter but having worked in and taught many media, I completely agree with you. After the 20th century, there is just nothing sacrosanct any longer about any medium. Thus looking at it from my angle, I find some of the more restrictive attitudes about what is ok and not ok in photography---cropping or not, processing beyond a certain point or not, & etc---just personal issues at best and precious at worst. I sort of understand how having to work on a photograph for "hours" might be off-putting (or impossible for some commercial work), but then again I sort of don't: when it takes days or weeks to finish a drawing, painting, sculpture, installation, or print run, "hours" seems like a fabulous deal!

I don't want to work on stuff for hours either, having gone through various phases I'm pretty certain I don't have the patience for any of it either, but getting a good result doesn't have to take long if you know the tools well.

[lm4187]

rob8888,

I doubt that your camera has a problem. As I see it, there are several things going on with your images that are not unique to the 645z.

1. Dynamic range issues:
a. The 645z has a dynamic range of 14, give or take (haven't seen any exact measurements, yet).
b. The NZ scenery can have a dynamic range exceeding 14 (as can many bright sunny scenes around the world). Hence the overexposure.
c. When you convert to JPG, you're cramming 14-bits per channel from the camera into 8-bits per channel in the JPG image.

2. Spatial resolution issues.
Converting to JPG also loses spatial resolution, not only due to smaller image size, but also to compression artifacts in the JPG algorithms. JPG conversion throws out details that don't impact on visual perception. It can also induce banding because intermediate color values in a gradual, smooth area can be deleted--giving discontinuities in the color gradation. Discontinuity means edges. Our eyes (and some nice image processing by the brain) are exceptionally good at detecting edges (smooth gradations, not so much). Hence the potholes in your color gamut.

Banding can also be induced by monitor display settings or software display settings.

For the dynamic resolution issues--have you tried shooting HDR series? For washed-out skies--polarizing filter to help tame the dynamic range between sky and foreground? Also, better to underexpose than overexpose--as others have noted above, clipped values can't be recovered and really confuse highlight recovery algorithms. The 645z is reported to have very good shadow recovery.

For the spatial resolution issues--some software seems to work better than others in RAW conversions.

I hope this explanation doesn't sound too preachy. I wish I were going through a similar learning curve with a 645z; but I'm still learning my K-3.

So, good news: probably not the 645z.

Bad news: trying to cram an analog world into square, digital pixels is not for the faint-hearted.

[One last thought: have you tried saving RAW+ and comparing the camera's JPG to your software-processed image?]

Cheers,
Lou

[rob8888]

I'll upload the blue island DNG when I get to a computer.

Can I politely ask we focus on the issue instead of processing ethics? I have tried rescuing images by fixing that area but it's impossible with gradient skies - it'll do your head in.

Lastly, I have one of the largest collections of NZ images in the world. I've grown up dealing with the huge dynamic range our clear atmosphere & harsh sun angles creates. I've shot everything from KM25/Velvia & all kinds of scanners and a long history of digital cameras. I have NEVER had so much difficulty as I have with the 645z in this area. I'm really scratching my head. It shouldn't be this hard it this point in digital development.

Otherwise I love the 645z (some of the glassware and CA not so much though).

[Royce Howland]

I'm definitely willing to post a couple of trial rescues, Rob, ethics aside. I tried a quick & dirty fix on your first shared DNG, using the method of copying part of the R or B channel into the G channel in Photoshop. It definitely fixes the green/magenta rainbow issue but introduces a different effect. The new effect is one that could be more legitimate, i.e. I've seen skies looking like that here. But I don't know how the sky "should" look in your scene; so I'm not sure how the fix translates to either what you saw, or how you wanted to develop the final look.

The original cause is definitely over-exposed G channels in the raw data; this isn't a guess, it's from looking at the raw image data using the RawDigger app. The DNG file G channels are definitely blown, no matter what the Adobe histogram says. That's definitely creating the magenta/green effect. Why the over-exposure is happening, I can't say. It could be a flaw with your camera, it could be something else. I've shot a bunch with my 645Z with a lot of blue skies over here, and never saw anything like this so far in my frames. You're right, it shouldn't be that difficult...