Luminous Landscape Forum
Equipment & Techniques => Digital Cameras & Shooting Techniques => Topic started by: MalcolmL on February 15, 2011, 08:09:26 pm
-
I have a long standing difficulity with the large smooth areas of skies - both cloudy and blue when it comes to tonal manipulation to attain a better tonal zoning in digital photographs.
This is a problem I have encountered both with HDR suites and also with the shadow and highlight sliders of standard photoeditors such as Adobe Photoshop Elements or Silkypix.
In attempting to bring down the highlights so I have some detail in the sky or even darken a sky,what starts out as a full spectrum of tonal gradation of gets squeezed down to perhaps a gamut of 6-12 tones with awfully pixelated demarcations bewteen each zone.
This often gives a semi- solarised type effect. I was advised to stop shooting io JPEG to avoid compression issues that might cause this, but even in RAWs converted to TIFFs I cannot resolve this issue.
This is now the ONLY issue I have personally yet to solve now in terms of image quality in digital files.
Should I just be exposing for the skies and lightening the shadows of is there somethong more fundamental I am missing ??
-
Are your TIFFs 16 bit images?
-
I have batch converted in XN view which I believe saves in 8 bit - I will need to check that when I get back to my home PC. Does 16 bit get round this issue ??
Maybe I should be doing my dynamic range adjustment in RAW and then saving as 16 bit TIFFS ??
Maybe exposing for the skies rather than averaging EV for the whole scene is the way because I doubt that the issue of limited gamut range arises in the shadow areas ????
Thanks
-
Maybe I should be doing my dynamic range adjustment in RAW and then saving as 16 bit TIFFS ??
Ya think? Seriously, you need to get to the point where the raw files (not JPEGs) are as close as you can get them BEFORE processing the TIFF files and yes, 16 bit is a lot better for post processing than 8 bit.
-
Thanks - sounds like good advice - will try that way.
malcolm
-
Hi,
That can also depend on your monitor. Or you highlights being clipped. If you posted one of the raw files for download it would help.
One you convert to 8 bits you loose a lot of information which cannot be regained using 16 bit processing. One of the good reasons to shoot raw.
Best regards
Erik
I have a long standing difficulity with the large smooth areas of skies - both cloudy and blue when it comes to tonal manipulation to attain a better tonal zoning in digital photographs.
This is a problem I have encountered both with HDR suites and also with the shadow and highlight sliders of standard photoeditors such as Adobe Photoshop Elements or Silkypix.
In attempting to bring down the highlights so I have some detail in the sky or even darken a sky,what starts out as a full spectrum of tonal gradation of gets squeezed down to perhaps a gamut of 6-12 tones with awfully pixelated demarcations bewteen each zone.
This often gives a semi- solarised type effect. I was advised to stop shooting io JPEG to avoid compression issues that might cause this, but even in RAWs converted to TIFFs I cannot resolve this issue.
This is now the ONLY issue I have personally yet to solve now in terms of image quality in digital files.
Should I just be exposing for the skies and lightening the shadows of is there somethong more fundamental I am missing ??
-
My TIFF files are all 24 bit. I checked
I have cracked this one. I edited the dynamic range of a RAW file in a RAW editor and converted to a TIFF. I then took the original RAW and converted to a 24 bit TIFF and gave it the same overall treatment. Attached are crops from the sky (I tweaked the contrast in the crops to magnify any aberrations). I have also included thumbs of the whole files.
Thanks for your help
-
Working straight through in 16-bit is necessary if you're to preserve tonal gradations. Absence of red-channel information is already a challenge. But there is another factor in HDR to consider.
HDR involves supersampling a scene. This process depends in most cases on the scene being static across the range of samples. While blue sky gives the appearance of being a static field of blue, it is anything but. You can mitigate this a bit by making a separate HDR for the sky, lower the strength, and adding some smoothing in Photomatix. You would want to maintain a 16-bit workflow throughout.
-
My TIFF files are all 24 bit. I checked.
You do realize that 8-bit and 24-bit are the same thing? So you are losing bit depth in the TIFF conversion. If you saved from raw to 16-bit (48-bit counting 3x color) TIFF, there should be no problem.
-
No I did not realise this at all.
I need to do some homework !
However by editing in RAW (see above) it seems that I have solved most of the issues. However this still leaves the issue of the final convertion to a TIFF which really needs to be in 16 bit. I can print from RAW however using the Sony RAW editor. But other editors have features that Sony RAW does not
Malcolm
-
The biggest problem I found when switching from film/slides to digital is how it treats skies. There are a great number of software aproaches to combat this, but I haven't really been satisfied with any of them.
This site seems to be mostly about things one can do with computer and printing technology, with very little attention to actually taking pictures. But my approach is to get it right, or as close as possible, at time of capture.
This includes lens selection (some lenses treat skies better), use of filters (CPL and GND) and exposure manipulation. I'll expand upon these, if anyone is interested, but I doubt it.
From the computer play perspective, the best thing I've found is combining two widely exposed images with a simple cut-and-paste merger in Photoshop. This only works if you have a distant and flat horizon, with no trees or other items running vertically though both parts.
An example is posted here. The sky, by the way, was gray and almost flat. Exposer difference between rocks and sky were almost three stops.
-
Also be sure to use the "linear" or "neutral" control to get the most linear capture you can for HDR purposes. Using tone curves with images input HDR is a mess, and the data just doesn't shake out right.
-
Thank you !! - I can but agree. This is where I have too have thoughts - digital is great - image quality is FAR better than film (per unit area of sensor / film in terms of sharpness alone). But we all know there is more to it than sharpness. Tonal range has always been the issue despite the higher intrinsic dynamic range of good sensors (that is about 11-12 EV). It is pointless capturing that 11-12 EV unless it can be carried through post processing to print and this is where the current nemesis lies in digital. It is a thorny issue that has been conveniently ignored by the industry (mostly).
HDR was an attempt to address this - but as has just been said ""Using tone curves with images input HDR is a mess, and the data just doesn't shake out right"". I would never input tone modulated images into HDR. I am not sure if HDR fully addresses the issues and I need to do some in depth testing to see if HDR wins over simple tonal adjsutment in RAW exposed for the skies (see below)
So we are left with ""my approach is to get it right, or as close as possible, at time of capture"". Yes and yes !
BUT you say ""This only works if you have a distant and flat horizon, with no trees or other items running vertically though both parts."" - well OK but that rather restricts what you can do does it not ??
Not really ideal at all is it (??)
My contention is this - in those shots where there is a lot of sky and we want to get that data into the final print we expose for the sky. The shadows areas will be dark but there seems to be little VISIBLE loss of image quality when shadows are lightened in the sliders. All this done in RAW of course. I am totally prepared to accept that there IS data loss when the shadows are lightened but its is not that visible. What counts is what the eye can see on thhe final print. Horribly banded skies DO show on the print - I have rejected many a print on that basis.
Then there is converting the image for other manipulations in standard photoeditors (I do not use full Photoshop although maybe I need to bite that bullet).
Saving as a 16 bit TIFF from RAW for a FF camera gives file sizes of about 135 megabytes - I tried that for the first time today. I have just bought a 2 terrbyte external drive to store these huge bloated files. Maybe we will all be buying 10 terabyte drives to store all this quality, or maybe we should all go back to large fomat film and pay 300 dollars a single drum scan (I personally never had the luxury of large format film).
One thing I feel pretty sure about is that the issue of tonal quality in digital photography has been to a large extent sidelined as its rather troublesome and uncomfortable.
-
Hi,
You can alway try the graduated filter in Lightroom or Photoshop ACR. You could also checks this article:
http://echophoto.dnsalias.net/ekr/index.php/photoarticles/46-fixing-sky-with-luminosity-mask
Photoshop CS5 has an option for HDR tonemapping 16 bit images, I normally use local adjustment with default values.
Here are some of my HDR images: http://echophoto.smugmug.com/Special-methods/HDR/HDR/ but these are made from multiple exposures.
Best regards
Erik
Thank you !! - I can but agree. This is where I have too have thoughts - digital is great - image quality is FAR better than film (per unit area of sensor / film in terms of sharpness alone). But we all know there is more to it than sharpness. Tonal range has always been the issue despite the higher intrinsic dynamic range of good sensors (that is about 11-12 EV). It is pointless capturing that 11-12 EV unless it can be carried through post processing to print and this is where the current nemesis lies in digital. It is a thorny issue that has been conveniently ignored by the industry (mostly).
HDR was an attempt to address this - but as has just been said ""Using tone curves with images input HDR is a mess, and the data just doesn't shake out right"". I would never input tone modulated images into HDR. I am not sure if HDR fully addresses the issues and I need to do some in depth testing to see if HDR wins over simple tonal adjsutment in RAW exposed for the skies (see below)
So we are left with ""my approach is to get it right, or as close as possible, at time of capture"". Yes and yes !
BUT you say ""This only works if you have a distant and flat horizon, with no trees or other items running vertically though both parts."" - well OK but that rather restricts what you can do does it not ??
Not really ideal at all is it (??)
My contention is this - in those shots where there is a lot of sky and we want to get that data into the final print we expose for the sky. The shadows areas will be dark but there seems to be little VISIBLE loss of image quality when shadows are lightened in the sliders. All this done in RAW of course. I am totally prepared to accept that there IS data loss when the shadows are lightened but its is not that visible. What counts is what the eye can see on thhe final print. Horribly banded skies DO show on the print - I have rejected many a print on that basis.
Then there is converting the image for other manipulations in standard photoeditors (I do not use full Photoshop although maybe I need to bite that bullet).
Saving as a 16 bit TIFF from RAW for a FF camera gives file sizes of about 135 megabytes - I tried that for the first time today. I have just bought a 2 terrbyte external drive to store these huge bloated files. Maybe we will all be buying 10 terabyte drives to store all this quality, or maybe we should all go back to large fomat film and pay 300 dollars a single drum scan (I personally never had the luxury of large format film).
One thing I feel pretty sure about is that the issue of tonal quality in digital photography has been to a large extent sidelined as its rather troublesome and uncomfortable.
-
[...] as has just been said ""Using tone curves with images input HDR is a mess, and the data just doesn't shake out right"". I would never input tone modulated images into HDR. I am not sure if HDR fully addresses the issues and I need to do some in depth testing to see if HDR wins over simple tonal adjsutment in RAW exposed for the skies (see below)
So we are left with ""my approach is to get it right, or as close as possible, at time of capture"". Yes and yes !
So when you used ViewNX, you were using the "linear" picture control, and not the "standard" as I suggested? You would want to do that. Just double checking. So you would go on to now make a set of 16-bit (x3=48) captures and take those into Photomatix.
Then you can do several things in Photomatix. There's exposure fusion and averaging, none of which require tonemapping. There is HDR with the many possible applications of tonemapping, starting from the lightest strength.
But in the end, you might get away with a good camera and a full 14-bit capture as the starting point.
-
One thing I feel pretty sure about is that the issue of tonal quality in digital photography has been to a large extent sidelined as its rather troublesome and uncomfortable.
Not at all...there are many, many threads here on LuLa about just this subject; controlling scene contrast within the dynamic range of a sensor and how to process raw files to optimize the final tone curve of the image for printing. It's definitely one of the top 5 topics...
You are indeed hamstrung by using the tool set you have...Camera Raw in Photoshop and Lightroom can easily do local tone and color corrections at the raw processing stage...there are also methods of bracketing and blending the multiple exposures to control the tonal range.
The other thing you need to understand is that if you under expose and image to maintain a "normal sky" you will be underexposing the capture and the shadows will indeed start blossoming with noise. Depending on the ISO is may or may not be exaggerated...but what you also don't seem to grasp is that there is far more data and detail in the brightest portions of raw captures...as long as you don't clip the highlights, you can easily tone them down in the raw processing. So you need to adjust how you expose digital to get the most from your sensor.
-
Hi,
To fully utilize the sensor expose to the right using the histogram. Maximum exposure without clipping. Jeff Schewe had some nice demo on the issue somewhere.
The classic article is here: http://www.luminous-landscape.com/tutorials/expose-right.shtml
The reasoning may not be absolutely scientifically stringent but the suggestions are good,
Best regards
Erik
Not at all...there are many, many threads here on LuLa about just this subject; controlling scene contrast within the dynamic range of a sensor and how to process raw files to optimize the final tone curve of the image for printing. It's definitely one of the top 5 topics...
You are indeed hamstrung by using the tool set you have...Camera Raw in Photoshop and Lightroom can easily do local tone and color corrections at the raw processing stage...there are also methods of bracketing and blending the multiple exposures to control the tonal range.
The other thing you need to understand is that if you under expose and image to maintain a "normal sky" you will be underexposing the capture and the shadows will indeed start blossoming with noise. Depending on the ISO is may or may not be exaggerated...but what you also don't seem to grasp is that there is far more data and detail in the brightest portions of raw captures...as long as you don't clip the highlights, you can easily tone them down in the raw processing. So you need to adjust how you expose digital to get the most from your sensor.
-
Jeff Schewe had some nice demo on the issue somewhere.
Here it is: Un-debunking ETTR (http://schewephoto.com/ETTR/index.html)
There's also an example of just how much image detail can be contained in the extreme highlights of an image...
-
Thank you Schewe - . There are quite powerful controls in Sony RAW which include dynamic range, histogram adjustments and shadow / highlight adjusters I can use, so perhaps I am not so hamstung as might be.
Howerver I am only just starting to grasp some of the basics of dynamic range physics and I clearly have my homework cut out for a while.
Malcolm
-
Thank you Schewe - . There are quite powerful controls in Sony RAW which include dynamic range, histogram adjustments and shadow / highlight adjusters I can use, so perhaps I am not so hamstung as might be.
Well, it doesn't have localized raw processing adjustments like ACR or LR...but you can certainly learn how to process multiple version of the raw file into different TIFF files for blending using layers...the problem there is that ideally you want to be using 16 bit and I think Element's 16 bit functionality is more limited than Photoshop's...I'm thinking you might want to look into upgrading to Photoshop if you want the optimal control over your images...
-
This site seems to be mostly about things one can do with computer and printing technology, with very little attention to actually taking pictures. But my approach is to get it right, or as close as possible, at time of capture.
This includes lens selection (some lenses treat skies better), use of filters (CPL and GND) and exposure manipulation. I'll expand upon these, if anyone is interested, but I doubt it.
Hi Ron,
Yes I really would like you to expand on how you "get it right.. at the time of capture". As I am always very interested in doing just that, which remininds me of what a seasoned old veteran in my camera club once said to me when I asked him how I could attain the correct exposure for any given shot, his very wise answer was then delivered back to me in a very North of England accent - "Tha needs ter gerrit reet in't box O' leet" - so to interpret for all those of you who do not understand the intracies of the Barnsley dialect - "You need to get it right in the box of light" - and that is what I have tried to do ever since.
So yes I really would like you tell me how you approach this please.
Photobloke
-
Interesting suggestions and educational thread. I too am a firm believer of getting things right in the field, to avoid manipulation afterwards. When shooting landscapes, skies, etc, a couple of old school graduated neutral density filters are some of the most useful accessories one can have. It takes some practice, but after a while it becomes second nature in one's "mechanical" photographic process.
The landscape is not going anywhere, so the extra minute or so of adjusting the filter and exposure are hardly a nuisance, and can save you a lot of time later on in the computer. When I shot slide film, it was a bit tricky to do all the "slide the filter up and down while pressing the depth of field preview button to locate the transition area" kind of dancing. With live view, these days, it is a lot easier.
-
Thanks
Yes getting it right at the time of the pressing of the button is crucial - the right composition, the right timing, the correct exposure. As in the days of film there is then rendring the image - in film it was processing - at the neg. stage then in the darkroom. Choice of development time, chemicals, pushing, dodging and burning. The better the negative the less manipulation required but some manipulation was usually required.
So the workflow with digital is electronic and (IMHO) is easier with greater range and flexibility but it is still important. A good digital ngative can be ruined by clumbsy PP.
With the advice gleaned from this posting I have found that major manipulation of the files in RAW (16 bit) then saving as a TIFF (8 bit) with fine tuning if required later in TIFF, holds on to almost all the data in the original file and produces lovely final prints, with minimal noise and no posterisation of highlights.
I am now leaving my camera on all its default settings - high dynamic range set to off, colour on neutral and exposing for the average brightness in each frame. ALL the gross manipulation is then done in RAW on the PC. Clearly the better the original capture the less invasive this manipulation will be, but with good software we can make a good capture really shine.
MalcolmL
-
Stay in 16-bit mode until the very final step; you'll have every advantage that way. Editing 8-bit files is only just possible, but never optimal unless you have nothing better. And even then, typically you'd convert back to 16-bit mode during editing to control for accumulated arithmetic error.
-
Also, try this. It's free and it does great things with bracketing and blending:
http://www.luminous-landscape.com/forum/index.php?topic=17775.0
I find it works better than the commercial programs because it simply does one thing. Drop the images in with little to no processing done to them and then manipulate the output.
http://www.guillermoluijk.com/article/nonoise/index_en.htm
His web site with info on how to use it.
Why you want to use this is because as the photos show, it results in very very clean and almost perfect looking HDR blending. To the point where it not only blends the two images together, but makes the image have virtually no noise or artifacts. What's left is amazingly easy to process in Lightroom or a similar program. It's HDR, it's clean, and it just works.
Post #60 is a good example in that thread of what you are looking for. The sky in the final product looks perfect.
-
I said what I did about people not wanting to hear about time-of-capture because it is what I've seen on various sites and especialy here. I posted three or four long posts about it on another thread, and would like to repost them here, but I can't find it. After going to all that trouble with no responses, I mentioned there was more to it, but if there was no intrest, I'd quit posting. Someone posted that "that's probably best, because it doesn't address the OP's original question." Well, it did actually, but not in the "what button should I click on the computer to make up for not getting it right at time of capture" type of usual question. So I figured, why bother.
Nevertheless, I have a couple PMs asking for my thoughts and a few posts here expressing interest, so I don't mind sharing what I've learned about this during almost four decades, but if anyone knows the location of those other posts, it would be a lot easier to repost them here and finish it up.
Most want to shoot on auto and process on manual, but I have come to the conclusion that should be reversed, not because I don't know much about PP, but because I do. What many don't realize is that everything you do in PP doesn't just affect the target, but other things across the image spectrum. If you change, for example, the red channel, it doesn't just change red, but the hues of all the four primary colors. I learned this years ago when I was the editor of a high-quality magazine and also had to serve as the photo editor, which included color correcting chromalons made from slides. It's really pretty complicated.
Some PP is necessary, and just as during the film era, I can alway improve a shot slightly in the darkroom (PP), but my approach is to handle any shot that needs more than the most minor adjustments by fixing it with the delete button. This includes sharpness, because oversharpening causes all kinds of problems, most notably artifacts. Often, my final criteria for keeping a RAW image for processing it to blow it up to 100 percent.
In the meantime, someone asked me, in a PM I think, to further explain the cut-and-paste method I used for the picture included in a previous post. This is something, like HDR, that has to be planned ahead of time and includes two widely exposed images, one to enhance the sky and the other to expose the land. Both, or all, images are taken on a tripod with a remote or delay for mirror-up function. It only works when you don't have a tree or other complicated image running through both exposures, but a majority of scenic shots have a distant horizon that works with this method. In the case of the posted shot, it works because the rocks have even edges.
In PS, I first open the image that is exposed for the sky, select it with the "quick-select" tool and and select copy from the edit menu. Then I close that image and open the one exposed for the land. In the edit menu I click on paste, and move it into place. If both images were processed to the same size, it fits pretty well, but if there are some thin (usually lighter) spots along the edges of the two, they can be easily and quickly repaired with either the "spot-healing brush" or the "clone" tool.
Here's another shot of Elephant Rocks using the same method. Again, this is a good illustration of this method, because the sky was almost uniform grey and fairly bright, which is the toughest challenge for digital. No filters of any kind were used.
-
Ron, I think your approach is one valid approach, and it seems to come from years of shooting slides in which the obvious fact is that you have (pretty much) the one chance to "get it right," and that is when you trip the shutter.
As much as I respect that (and I do), the very idea of "getting it right" is philosophically laden with assumptions about what is "right." What you demonstrate is one aspect of "right." But there are other senses of getting it right, including the entire range of shooting for what you know you will get at the end of the process, for whatever process you are using.
New sensor technology has changed the operant assumptions a bit. With the introduction of the D7000, we now have a practically ISO-less camera where the read noise is fixed and linear. In other words, one image taken with analog gain at ISO1600, and another image taken at ISO100 and given 4 stops of digital gain, are demonstrably equivalent.
And for others, there's ETTR, which gives you the optimal signal, and this is another important sense in which one can "get it right."
-
Ron, thanks for the summary. The technique you describe is indeed one way of doing it. I think we have today enormous flexibility, with a combination of "old" tools (GND filters) and "new" tools (combining images in post-processing). Which ones to apply will depend on the objectives, and how many f-stops are the sky and land apart. Sometimes, filters are enough to get the "good" result with one shot. Sometimes, it requires taking two or more images, and combining them later on ,as you describe. Other times, you can manage with just one RAW file, processing it twice, and combine the images.
Personally, if I can get it in the field, I will attempt to get it.
-
I said what I did about people not wanting to hear about time-of-capture because it is what I've seen on various sites and especialy here. I posted three or four long posts about it on another thread, and would like to repost them here, but I can't find it. After going to all that trouble with no responses, I mentioned there was more to it, but if there was no intrest, I'd quit posting.
http://www.luminous-landscape.com/forum/index.php?topic=50736.msg419969#msg419969
-
The first consideration for color is the equipment you use. I'm not a devoted fan of any manufacturer. I started with Contax and Zeiss lenses, shot Nikons for decades and switched to Pentax a few years ago. While most of my reasons for switching were based upon the bottom line (value for the money), one of the deciding reasons was the "Pentax look," which appeals to me.
Much of this is based upon individual tastes, and I'm not trying to imply that Pentax is better. All the top cameras are very good, and all the top lines feature in-camera adjustments you can make to boost color saturation, contrast and adjust the hue. Play around with these until you get the look you want. Even though my Pentax gives me the closest to the look I want right out of the box, I adjusted the sat. and contrast +1 to further enhance it.
Everything you do between the in-camera settings to the final adjustment of the printer effects the color, but my approach is to make the most minor adjustment along the way, building upon each, and I've found that adjusting in-camera applies a more even and holistic approach than doing it in PP. Every image is a bit different, and while I often bump contrast in PP, seldom do I change any of the color channels.
Even though I shoot RAW only, I convert to JPEG using the software that comes with the camera, so it automatically applies these setting to the JPEG, according to the Pentax idea of color. I suspect this is a good practice, regardless the make of your equipment, because the software they package with the camera is designed specifically for that model. Others may be easier to use and give you far more options, but they are designed for a broad range of makes and models.
All digital cameras are escentually computer controlled, hand-held processors in the first place, so it makes sense to me to start my control over processing within the camera and keep PP to the minimum.
Also play around with the various modes on your camera. Landscape mode, for example, is designed to enhance greens and contrast. That's the one I settled upon for landscapes, but you may like something else better in yours. Everyone should experiment with their equipment under the most controlled test possible.
Modern cameras are actually very powerful processors, with a great number of options, and all of them can be adjusted to get the look you want (or something very close) at time of capture, so get to know yours well by conducting a lot of test with various modes and adjustments
It takes quite some time to get to know a camera (or lens) to get the most from it.
-
You choice in lenses is very influencial upon color. The design of the lens and the coatings cause lenses to "render" images in a certain way. Zeiss is the most consistent in the way they render colors and contrasts than any of my experience. Within the four or five most popular DSLR manufacturers, there is a wide variance in the way their lenses produce colors and contrasts. I find this true even within the Pentax line, which probably puts more emphisis into quality glass than some others. It is ture even among the Limiteds. Among my current lenses, I have two that render images as if they have built in polorizers, especially the DA 15 Limited, which produces images that remind me of Kodachrome 64, and occasionally even Kodachrome 25.
Lenses designed specifically for landscapes (usually wider) generally are better. One of the first things I do when getting a new lens is run it through a series of test to determine the sweet spot. Generally, the wider the lens the higher the sweet spot and DOF for comparable f-stops, which means you can get more depth and detail into them.
My point is, to increase the richness of colors at the time of capture, examine the line of lenses compatable with your system. Look at a broad range of images produced by them, and you'll start to see patterns of color and contrast.
It took me a long time to realize the importance of glass, and my advice now is to spend more on a lens than on a body. The body gets you there, but it is the eye (lens) that sees. Most any Zeiss is good. Among the various camera makers, and Sigma, one needs to shop with a critical eye, but within each of them are lenses that rival, and occasionally even surpass, Zeiss.
Before buying a more powerful PP software, invest in a high-quality lens. Bodies and software packages are upgraded every year, but lenses last a lifetime.
-
The general advice with digital is to shoot to the right. I think this comes from product photographers and people who concentrate on making a few pennies from stock photography, both of which want more shadow detail than the human eye notices, or even wants to notice. I think landscape photography is the opposite. Shadows add greatly to the mood, richness and drama of the image, and shooting and/or processing to the left greatly enhances color richness. Shooting to the right, or overexposing, and then bringing it back in PP starts with dull colors. Shooting to the left, or underexposing, starts with rich colors. Just -1/3 stop makes a considerable difference with some cameras, while more is needed on other models. On the same scene with a tripod, mirror-up function, try bracketing exposure in AV, then without any color manipulation in PP compare them full screen on your computer to see the difference in color depth and richness.
While it is true that exposing to the right allows you to manipulate an image more in PP without introducing a lot of noise, while at the same time retaining more detail in the shadow areas, as long as you shoot at the lowest ISO possible, noise is not a problem. Besides, if you get it as close as possible at time of capture, you don't need to manipulate it as much in PP and risk introducing noise and artifacts. It is not just noise, but everything, that is better at lower ISOs, including color.
When you combine the richer color of lower ISOs with a slight underexposure to further enhance those colors, you begin with a rich image. Your shadows, of course, will be darker and lack detail, but I think that for landscapes, this is a good thing. Check out the way Ansel Adams and other masters utilized shadows. Doing this is even more important with color than with b/w, because color lacks contrast in comparison.
Various cameras expose differently. Those designed for higher ISO capabilities tend to overexpose a bit to begin. If your camera tends to overexpose in a plastic, product photography sort of way, you may need to underexpose a full stop to get what you want, so it is important to do some controlled tests and get to know your camera's exposure values well. At any rate, if you need to adjust exposure more than -1/3 stop in PP, you're not getting it right at time of capture.
Underexposing at time of capture also lets you capture a scene with a wide DR without blowing highlights, which can't be recovered. Bumping contrast also increases color slightly to the eye.
The most obvious way to bump color is with saturation, but I avoid this, or bump it very slightly in PP, for reasons already mentioned.
-
Thank you, Mike, for the link.
I'm not suggesting this is the only way to shoot landscapes. Take or discount whatever you want. I'm just sharing my way and welcome further suggestions, because I'm still learning too. Most of my suggestions are to experiment with your own equipment to achieve the results that suit your tastes. The saying: garbage in, garbage out, is an exaggeration, but there is some truth in it. I'm not suggesting one do no PP. It is a necessary componet to the finished product, just as are color profiles and printer settings. My main point is that if one puts as much time into figuring out the processing computer (camera) one holds in one's hands as one does with a desktop, the end results will be much better and more reliable.
The camera is where it all begins. Make it about photography first, computer games second.
What was posted from the other thread is an overview of equipment. Next I'll get into actually shooting, because beyond all the sophisticated auto functions and the power of PP software packages, the most important thing remains what one puts into the frame and when they snap the shutter. This isn't just about composition, but color enhancement and DR control as well (they are both closely related).
Right now, however, I need to do a photo shoot of kayakers that should take all afternoon. Maybe I'll add more later this evening.
-
Hi,
It's actually coming from landscape shooters who now what they are doing.
The idea is very simple, by exposing to the right you maximize the information in the image and it is decided in postprocessing how that information will be utilized. Digital sensors are essentially linearly proportional devices, so exposing to right does not effect anything except the number of bits actually used by the sensor, detail available in the darks and noise in the image. Exposing to the right is very much like using low ISO film, except that tonality is not affected.
To my best knowledge, the term ETTR was coined by a quite knowledgeable photographer name Michael Reichmann who also happens to be owner of this site.
Best regards
Erik
The general advice with digital is to shoot to the right. I think this comes from product photographers and people who concentrate on making a few pennies from stock photography, both of which want more shadow detail than the human eye notices, or even wants to notice.
-
Hi,
I don't think lenses play a major role in taming skies and dynamic range. It is very well possible that lenses have different color renditions but I would suggest that can effectively be taken care of by properly calibrating the camera/lens combination with a color checker card and using proper white balancing.
Some lenses may be better than others regarding veiling flare. Fixed focus lenses with fewer elements and non moving groups may have less flare, that is more contrast, than lenses of more complex design.
My mostly used lenses are Zeiss designs (labeled "Zeiss") and they are quite good but I don't think they are exceptional.
Best regards
Erik
You choice in lenses is very influencial upon color. The design of the lens and the coatings cause lenses to "render" images in a certain way. Zeiss is the most consistent in the way they render colors and contrasts than any of my experience. Within the four or five most popular DSLR manufacturers, there is a wide variance in the way their lenses produce colors and contrasts. I find this true even within the Pentax line, which probably puts more emphisis into quality glass than some others. It is ture even among the Limiteds. Among my current lenses, I have two that render images as if they have built in polorizers, especially the DA 15 Limited, which produces images that remind me of Kodachrome 64, and occasionally even Kodachrome 25.
Lenses designed specifically for landscapes (usually wider) generally are better. One of the first things I do when getting a new lens is run it through a series of test to determine the sweet spot. Generally, the wider the lens the higher the sweet spot and DOF for comparable f-stops, which means you can get more depth and detail into them.
My point is, to increase the richness of colors at the time of capture, examine the line of lenses compatable with your system. Look at a broad range of images produced by them, and you'll start to see patterns of color and contrast.
It took me a long time to realize the importance of glass, and my advice now is to spend more on a lens than on a body. The body gets you there, but it is the eye (lens) that sees. Most any Zeiss is good. Among the various camera makers, and Sigma, one needs to shop with a critical eye, but within each of them are lenses that rival, and occasionally even surpass, Zeiss.
Before buying a more powerful PP software, invest in a high-quality lens. Bodies and software packages are upgraded every year, but lenses last a lifetime.
-
To my best knowledge, the term ETTR was coined by a quite knowledgeable photographer name Michael Reichmann who also happens to be owner of this site.
I don't think Mike "coined" the term but he sure popularized the concept. The concept actually was told to Mike by Thomas Knoll, the co-author of Photoshop and founding engineer for Camera Raw as explained in this article (http://www.luminous-landscape.com/tutorials/expose-right.shtml).
-
Jeff,
Thanks for putting this right. I was pretty sure i was not on solid ground saying Mike "coined" the term, but the first time I saw it was in the article you refer to.
Best regards
Erik
I don't think Mike "coined" the term but he sure popularized the concept. The concept actually was told to Mike by Thomas Knoll, the co-author of Photoshop and founding engineer for Camera Raw as explained in this article (http://www.luminous-landscape.com/tutorials/expose-right.shtml).
-
Everything you do between the in-camera settings to the final adjustment of the printer effects the color, but my approach is to make the most minor adjustment along the way, building upon each, and I've found that adjusting in-camera applies a more even and holistic approach than doing it in PP. Every image is a bit different, and while I often bump contrast in PP, seldom do I change any of the color channels.
Even though I shoot RAW only, I convert to JPEG using the software that comes with the camera, so it automatically applies these setting to the JPEG, according to the Pentax idea of color. I suspect this is a good practice, regardless the make of your equipment, because the software they package with the camera is designed specifically for that model.
Many people who do RAW processing start with a "neutral" or "linear" capture with no other adjustments besides WB and levels to start. This linear file is the cleanest data to start with. A manufacturer's stock "looks" are based on tone curves applied to the data, which produce a "pleasing" look, but make the job of postprocessing harder. If you start with curved data, you are always one generation away from the clean data, and you can scarcely "uncurve" it if you try without loss. Start with linear data, and /then/ apply curves and adjustments, and your colors will come out as straight as they can.
-
Ron, thinking further, I see you as an experienced photographer who shows signs of wrestling just slightly with the role that computers are to play in your future photography. You are carrying over some of your assumptions from film photography in the process. For example, you advocate the slide-shooter's principle of underexposing just slightly to increase color saturation. Unfortunately, that assumption is flawed in digital photography. As much as there may seem to be a principle of exposure that is constant between film and digital, in the end this isn't the actual case. Of course the camera manufacturers do their level best to maintain the illusion of "just like film" by using film concepts interchangeably in the digital realm, but it's just an illusion.
You can do what you want to do, but do it a little better if you tease out some of the concepts involved. For example, if you shoot ETTR, use linear settings for capture, and apply just the few things you like, I think you will achieve better results -- better by your own standards. Those 12-14 bits really "add up" in more ways than one. And it doesn't have to take a long time, unless you gain a taste for doing more close and detailed work with a fine brush.
-
Hi Ron,
Thanks for telling us how you do this and giving us the benefit of your wealth of experience, as well as the rest of the information as and when you post it.
Photobloke.
-
I'm not some old-timer stuck in the past, suffering from dementia. Well, I am an old-timer, but I've been on computers since the early 1980 Macs, both for work and pleasure. I've worked in computer design with Photoshop for years and with PageMaker and Quark before that, plus a couple of programs designed specifically for newspapers. I've served as an editor, photo editor and published my own magazine. I don't claim to know it all, and I will admit I haven't kept up with that end for the past few years, concentrating mostly on photography, but I still have friends who do it for a living, and every one of them has told me often that "the less one does in PhotoShop the better." The common tendency is to overdo it, and the most frustrating part of their job is correcting all the click and slides others make, even professionals.
Yes, the idea of underexposing comes from Kodachrome days, but I rarely underexpose more -1/3 with digital, bringing it back during PP. I do this most of the time and print at 30X40 with no noise at all. If you think it doesn't make a difference, conduct the test I described. I will add that with the Nikons I shot for decades, one needs to underexpose more than with the Pentax I now use. I don't expect anyone to take my word for anything, because I don't either, especially across the internet. What I suggest is that you conduct your own tests and make up your on minds.
Again, this look is not for everyone. Many prefer the more plastic look of digital, which I think is great for product photography, but not landscapes. Why do some want to expose more detail in the shadows than the human eye notices, or even wants to notice, in a landscape anyway? Shadows in landscapes add drama. That's why we wait for low light angles and avoid shooting at mid-day.
-
I'm not some old-timer stuck in the past, suffering from dementia. Well, I am an old-timer, but I've been on computers since the early 1980 Macs, both for work and pleasure. I've worked in computer design with Photoshop for years and with PageMaker and Quark before that, plus a couple of programs designed specifically for newspapers. I've served as an editor, photo editor and published my own magazine. I don't claim to know it all, and I will admit I haven't kept up with that end for the past few years, concentrating mostly on photography, but I still have friends who do it for a living, and every one of them has told me often that "the less one does in PhotoShop the better." The common tendency is to overdo it, and the most frustrating part of their job is correcting all the click and slides others make, even professionals.
Yes, the idea of underexposing comes from Kodachrome days, but I rarely underexpose more -1/3 with digital, bringing it back during PP. I do this most of the time and print at 30X40 with no noise at all. If you think it doesn't make a difference, conduct the test I described. I will add that with the Nikons I shot for decades, one needs to underexpose more than with the Pentax I now use. I don't expect anyone to take my word for anything, because I don't either, especially across the internet. What I suggest is that you conduct your own tests and make up your on minds.
Again, this look is not for everyone. Many prefer the more plastic look of digital, which I think is great for product photography, but not landscapes. Why do some want to expose more detail in the shadows than the human eye notices, or even wants to notice, in a landscape anyway? Shadows in landscapes add drama. That's why we wait for low light angles and avoid shooting at mid-day.
Certainly nobody thinks you're stuck in the past, and even less, suffering from a brain disorder. :-) And certainly nobody thinks one should over-do work in post, by definition.
Think of it this way. What happens in your camera after you press the shutter release /is/ photoshop in essence. The raw capture, applying curves, white balance, sharpening, saturation--this is your camera on photoshop (cut to shot of fried egg in a hot skillet). I can't help but think that you yourself would choose and apply filters of your own with more finesse. Cool? Call it a hunch.
-
Yes, the idea of underexposing comes from Kodachrome days, but I rarely underexpose more -1/3 with digital, bringing it back during PP.
Never shot many negs huh?
In point of fact, when shooting raw, you are better off exposing to the right if the scene's contrast range fits inside the dynamic range of your sensor. Controlling the image's tone curve (and whatever color saturation you may want) is easy to accomplish in the raw processor way before you ever have the image open in Photoshop. You are indeed leaving potential image quality on the table when you intentionally under expose. Since the raw capture is linear, you have tones of levels in the highlights and far less in the shadows. By under exposing and then lightening the image, you are indeed getting more noise in your shadows which you might be compensating for by darkening the tones.
Just how much potential image detail is their in the extreme highlights with digital? A lot...see Un-debunking ETTR (http://schewephoto.com/ETTR/index.html) for both an explanation of ETTR and how much image detail you can tease out of even over exposed images.
In point of fact, digital capture is a bit more like shooting negs than chromes–lot more dynamic range with digital than chromes.
-
I'm tired of defending myself, and fell like what I'm saying challenges some pecking order here or something. I've been shooting professionally for almost four decades. Of course, I've shot negs.
This is about getting more out of skies at time of capture. I guess you expose that to the right too.
If you can't even listen to a different approach without inuendos, never mind.
-
I'm tired of defending myself, and fell like what I'm saying challenges some pecking order here or something. I've been shooting professionally for almost four decades. Of course, I've shot negs.
This is about getting more out of skies at time of capture. I guess you expose that to the right too.
If you can't even listen to a different approach without inuendos, never mind.
All I care about personally is whether you get the best information available for what you want to do.
-
I guess you expose that to the right too.
Yes...where...appropriate...
You really need to understand what digital capture does and doesn't give you...
When you are in the field, you must decide what the optimal raw capture exposure will be...
If you are habitually underexposing, then yes, you are leaving IQ on the table. That's easy to prove.
On the other hand, if IQ is important to you, then it makes sense that the digital capture is optimally exposed. If you, by nature, under expose by 1/3 of a stop because of some sort of film based experience, then yes, I would say that you are producing images at less than an optimal raw exposure.
Sorry to break your bubble...
Do you want optimal results or do your want to force fit your images to your expectations?
There is nothing ethically wrong with post processing images to meet your visual expectations. Assuming you know HOW to do the post processing...that's the real challenge...apparently your post-processing skills are lacking.
That's all I'm saying...if you habitually under-expose raw captures, then you are leaving IQ on the table. As long as you understand and accept that, fine. Otherwise, you might want to adapt your approach to raw captures.
For other people, I would caution that there is a better way...
And if time in the industry is any sort of criteria, I think I may have you beaten...been there, done that, got the TEE-shirt and the scars...
-
Jeff,
I don't take your point. It may be a question of definition but in my view we would always expose to the right, thus maximizing the number of photons detected by the sensor. In essence ETTR means that the histogram is moved to the right but not allowed to clip nonspecular highlights. In what situation would it be preferable to expose less than maximum possible?
Best regards
Erik
Yes...where...appropriate...
-
Excellent thread guys! Every post counts though not all the words in them. However, I need to say that you are here a little anal about the issue. At some point, reading some of these posts seemed like you were discussing exposure of 8X10 chromes. It is digital, isn't? Can't anybody do a "free" brief bracketing of 3 exposures and get the best part out of the 3 to pp? Am I missing something here?
Great thread anyway and thanks.
Eduardo
-
In essence ETTR means that the histogram is moved to the right but not allowed to clip nonspecular highlights. In what situation would it be preferable to expose less than maximum possible?
The key word is clip. If the contrast range of the scene is greater than the dynamic range of the sensor, by definition you can't capture the full range without clipping. If highlight detail is important to a shot, the you have to expose for the highlights and let the shadows go. In this case, ETTR isn't a good idea. I agree that whenever the scene range is within the dynamic range of the sensor, it's a good idea to ETTR as long as doing so doesn't interfere with the needs of F stop or shutter speed.
-
It's a bit of a generalization but camera manufacturers own software is rarely in the same league as the offerings from Adobe, Apple, Iridient etc. I used the Sony software when I had a Sony Alpha 900 on test and it was not very inspiring, sorry. Give the others a go, you might be pleasantly surprised.
-
Hi,
I don't think that you get the idea. The idea with ETTR is that you maximize exposure, so that you get least possible noise. So even if you bracket you would not process the image that looks good but the one which is near clipping. It's essentially very similar to exposing for the shadows with negative film. Just an example, blue sky. You can expose blue sky for natural look, but you can also increase exposure one step extra. In postprocessing you would than reduce "exposure" by one step, so you get the sky right, but you would gain one stop in dynamic range and reduce noise in the shadows.
Best regards
Erik
Excellent thread guys! Every post counts though not all the words in them. However, I need to say that you are here a little anal about the issue. At some point, reading some of these posts seemed like you were discussing exposure of 8X10 chromes. It is digital, isn't? Can't anybody do a "free" brief bracketing of 3 exposures and get the best part out of the 3 to pp? Am I missing something here?
Great thread anyway and thanks.
Eduardo
-
Excellent thread guys! Every post counts though not all the words in them. However, I need to say that you are here a little anal about the issue. At some point, reading some of these posts seemed like you were discussing exposure of 8X10 chromes. It is digital, isn't? Can't anybody do a "free" brief bracketing of 3 exposures and get the best part out of the 3 to pp? Am I missing something here?
Great thread anyway and thanks.
Eduardo
I agree. Memory is so cheap nowadays there should be little risk of running out of memory during a day's shooting. This was not the situation when I bought my first DSLR about 7 years ago. I recall I paid about A$ 800 for a 1GB compact flash card. On accasions, realising I would run out of memory before the end of the day if I continued shooting in RAW mode, I would switch to jpeg mode.
Nowadays I frequently bracket 3 exposures, not necessarily for merging to HDR but to provide a choice for slection of the best one for processing.
Skies can be a problem when there's a deceptively bright patch where the sun is trying to break through the cloud cover but not quite succeeding. In such circumstances the choices seem to be, (1) a totally blown patch of sky, (2) a correctly exposed sky but noisy shadows on the land, (3) bracket exposures for merging to HDR, (4) buy a Nikon D7000 or Pentax K5 ;D .
-
Hi,
Personally I hate bracketing. The reason is really that when I work trough a days shooting I end up with 600 pictures instead of 200. Doing the job right at capture time just takes a glance at the histogram but saves a lot of PP time.
Just my 2 cents...
Best regards
Erik
I agree. Memory is so cheap nowadays there should be little risk of running out of memory during a day's shooting. This was not the situation when I bought my first DSLR about 7 years ago. I recall I paid about A$ 800 for a 1GB compact flash card. On accasions, realising I would run out of memory before the end of the day if I continued shooting in RAW mode, I would switch to jpeg mode.
Nowadays I frequently bracket 3 exposures, not necessarily for merging to HDR but to provide a choice for slection of the best one for processing.
Skies can be a problem when there's a deceptively bright patch where the sun is trying to break through the cloud cover but not quite succeeding. In such circumstances the choices seem to be, (1) a totally blown patch of sky, (2) a correctly exposed sky but noisy shadows on the land, (3) bracket exposures for merging to HDR, (4) buy a Nikon D7000 or Pentax K5 ;D .
-
Hi,
Personally I hate bracketing. The reason is really that when I work trough a days shooting I end up with 600 pictures instead of 200. Doing the job right at capture time just takes a glance at the histogram but saves a lot of PP time.
Just my 2 cents...
Best regards
Erik
Erik,
That seems a trivial problem to me. Make the normal exposure the first one in the series of 3. Spend as much time as you like getting it right. The other two bracketed shots simply become a sort of insurance.
When sorting through the images at the end of the day, it should not be necessary to examine all 600, just the first image of every group of 3. It's only in the event of a miscalculation that you might choose to use one of the other two exposures.
Bracketing also provides the advantage of it being less likely that one 'misses the moment' as a result of spending too much time trying to get a good ETTR.
Cheers!
-
Hi!
I see your point, and you are of course right. For me, with my way of working, it is causing a mess.
Best regards
Erik
Erik,
That seems a trivial problem to me. Make the normal exposure the first one in the series of 3. Spend as much time as you like getting it right. The other two bracketed shots simply become a sort of insurance.
When sorting through the images at the end of the day, it should not be necessary to examine all 600, just the first image of every group of 3. It's only in the event of a miscalculation that you might choose to use one of the other two exposures.
Bracketing also provides the advantage of it being less likely that one 'misses the moment' as a result of spending too much time trying to get a good ETTR.
Cheers!
-
Hi,
Phase One owners are happy...
I use Adobe Lightroom and I love it. Won't say it's the best raw processing engine, I simply don't know (before defining best you need to define good). For me it's a workflow solution taking care of the image from the cradle to the grave. I don't really now about the grave part ;-)
I also tested Iridient Raw Developer from time to time, it's good but I'm pretty much sold on Lightroom. At a time I was considering writing image management software for my own needs, but when Aperture came out I realized that the software was there, but I was on PC and that Microsoft OS (and Linux). So when Lightroom came out I jumped on it. Later I switched to Apple but stayed with Lightroom.
Best regards
Erik
It's a bit of a generalization but camera manufacturers own software is rarely in the same league as the offerings from Adobe, Apple, Iridient etc. I used the Sony software when I had a Sony Alpha 900 on test and it was not very inspiring, sorry. Give the others a go, you might be pleasantly surprised.
-
I also tested Iridient Raw Developer from time to time, it's good but I'm pretty much sold on Lightroom. At a time I was considering writing image management software for my own needs, but when Aperture came out I realized that the software was there, but I was on PC and that Microsoft OS (and Linux). So when Lightroom came out I jumped on it. Later I switched to Apple but stayed with Lightroom.
Best regards
Erik
I'm a Lightroom affectionado too, but have you tried the highlight recovery tools in Aperture? Seems to be able to pull more back than LR for some reason. These is even a histogram opion that shows you how much raw data exists above 255. I still prefer LR but that might be something to do with familiarity too!
-
That's is exactly what I do. I call my workflow the "in situ-lazy brain". At least for my way of shooting it is easier for me to do "the thinking" in front of the computer when usually I have time to my heart's content, not so the same while shooting outdoors. It can be pedestrians, animals, vehicles, extreme weather, darkness, sunshine, stylists, posers, AD's, you name them! - What I mean, I hate to be waited of or wait! :'(
Yes I come home with 3X the pictures to download and review. Small nuisance!
Thanks guys for all the good posts.
Eduardo
Erik,
That seems a trivial problem to me. Make the normal exposure the first one in the series of 3. Spend as much time as you like getting it right. The other two bracketed shots simply become a sort of insurance.
When sorting through the images at the end of the day, it should not be necessary to examine all 600, just the first image of every group of 3. It's only in the event of a miscalculation that you might choose to use one of the other two exposures.
Bracketing also provides the advantage of it being less likely that one 'misses the moment' as a result of spending too much time trying to get a good ETTR.
Cheers!
-
I don't think Mike "coined" the term but he sure popularized the concept. The concept actually was told to Mike by Thomas Knoll, the co-author of Photoshop and founding engineer for Camera Raw as explained in this article (http://www.luminous-landscape.com/tutorials/expose-right.shtml).
Thomas Knoll is an iconic figure in digital photography, but the rationale given in the quoted article is incorrect. The brightest f/stop of a 12 bit digital capture contains 2048 possible levels, but the actual number of perceivable levels is drastically limited by noise. Shot noise increases with exposure, but the signal:noise is highest in the brightest f/stop. For example, here is a 200x200 flat frame of the green1 channel of the Nikon D3. The mean pixel value is 8296, but the standard deviation is 175.
(http://bjanes.smugmug.com/Photography/DXO/111G1/1214705269_yf7xX-O.png).
As explained by DXO in their tonal range evaluation, it makes no sense to have a quantization precision much smaller than noise:
(http://bjanes.smugmug.com/Photography/DXO/DXOtonalRangeDef/1214705571_yi5iQ-O.png)
The actual tonal range for the D3 is 8.72 bits or 422 levels, much smaller than 16183 levels implied by the 14 bit readout.
(http://bjanes.smugmug.com/Photography/DXO/D3TonalRange/1214705533_AKPVG-O.png)
Nonetheless, ETTR is a valuable procedure since it increases signal:noise and the effect is most dramatic in the darker tones than the highlights as explained by Emil Martinec in his excellent post (http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/noise-p3.html#ETTR) on noise, where he specifically refers to Micheal's article.
Regards,
Bill
-
Thomas Knoll is an iconic figure in digital photography, but the rationale given in the quoted article is incorrect.
Actually, I think it would be unfair to paint Thomas as the author of the "rationale" written about in the article...the article was written by Michael not Thomas. I'm pretty sure Thomas knows that the reason ETTR works is that more photons = better signal to noise.
-
The brightest f/stop of a 12 bit digital capture contains 2048 possible levels, but the actual number of perceivable levels is drastically limited by noise. Shot noise increases with exposure, but the signal:noise is highest in the brightest f/stop. For example, here is a 200x200 flat frame of the green1 channel of the Nikon D3. The mean pixel value is 8296, but the standard deviation is 175.
Regards,
Bill
The distribution of pixel photons/electrons in a real image (no flat frames) is not Poisson, in the usual sense you imply here.
Sincerely,
Joofa
-
Quote ronkruger
All digital cameras are escentually computer controlled, hand-held processors in the first place, so it makes sense to me to start my control over processing within the camera and keep PP to the minimum.
Unquote
You maybe technically correct in your statement but I believe that this is the wrong way to go about things. IMO the camera should be used for getting exposure and composition right with all the controls set to as near zero as possible. You then import the image to the computer and process. The computer is undoubtedly a faster, more flexible and more powerful processor of images than any camera. If you get your settings - over contrast and saturation - wrong in camera then you have sort them out in the computer which isn't easy so zeroing them in camera is best. Too much contrast and saturation affects exposure and they will cause over exposure. If you then underexpose in camera to prevent this then as Schewe points out you will have further problems. Ron you posted an image that I am re posting. Can you explain why is it the sky in the background is dull and not producing much light but the rocks in the foreground are light and under two of the rocks there are shadows that are from different directions. It looks like you added the sky to the image and forgot about the shadows. Forgive me if I am wrong in this assertion :) I think as Schewe and others points out you will have to up date your thinking. :)
-
The distribution of pixel photons/electrons in a real image (no flat frames) is not Poisson, in the usual sense you imply here.
Sincerely,
Joofa
A point which is quite irrelevant to the present discussion.
-
The distribution of pixel photons/electrons in a real image (no flat frames) is not Poisson, in the usual sense you imply here.
Sincerely,
Joofa
Yes, we take pictures of real scenes, not flat fields. Since the signal:noise varies with photon flux, the flux has to be held constant for a proper experiment and we have to sample enough photons to get a valid sample size so as to yield a statistically valid standard deviation. A real scene could be broken down into regions and the standard deviation obtained by the process of integration. Shot noise is the predominant source of noise for all but the darkest regions of the image, where read predominates (ignoring thermal noise and PRNU), and shot noise does follow a Poisson distribution. DXO does give a full SNR plot and also a value for 18%, which is often taken to represent average scene reflectance. I really don't get your point.
Regards,
Bill
-
>Yes, we take pictures of real scenes, not flat fields.
This much I agree on :)
>Since the signal:noise varies with photon flux, the flux has to be held constant for a proper experiment and we >have to sample enough photons to get a valid sample size so as to yield a statistically valid standard deviation. A >real scene could be broken down into regions and the standard deviation obtained by the process of integration. >Shot noise is the predominant source of noise for all but the darkest regions of the image, where read >predominates (ignoring thermal noise and PRNU), and shot noise does follow a Poisson distribution. DXO does give >a full SNR plot and also a value for 18%, which is often taken to represent average scene reflectance. I really >don't get your point.
>Regards,
>Bill
But how does the rest of your post equate to the reality of shooting an image, processing it and printing it? I doubt that most of the posters will be thinking in these terms when doing so. ::)
-
But how does the rest of your post equate to the reality of shooting an image, processing it and printing it? I doubt that most of the posters will be thinking in these terms when doing so. ::)
I think that in real world photography, the SNR will be worst in the shadows and can be improved by ETTR without clipping of channels. Hue twists introduced by non-linear exposure controls in some raw converters may be problematic in some cases.
Regards,
Bill
-
But how does the rest of your post equate to the reality of shooting an image, processing it and printing it? I doubt that most of the posters will be thinking in these terms when doing so. ::)
And in addition to what Bill already answered, noise in real image scenes is most noticeable in the areas with little detail, e.g. smooth gradients. When there is a lot of detail, it is usually harder to notice noise anyway.
Cheers,
Bart
-
Yes, we take pictures of real scenes, not flat fields.
With this understanding we can assume that experiments done using flat fields are not fully describing natural image situations.
Since the signal:noise varies with photon flux, the flux has to be held constant for a proper experiment and we have to sample enough photons to get a valid sample size so as to yield a statistically valid standard deviation.
For a flat field that makes sense. But the issue in the reality is that we have only a single photo and how do we define a measure of SNR, especially one coming from the shot noise? Classroom experiments and displaying power spectrum of flat fields, as many have done, does not answer that question, except reproducing well-known facts regarding white noise known in the signal processing for decades. At this stage we must have an awareness that if we are to tackle the reality of natural images, as in photography, then some of the basic models need to be developed further. It is possible that the situation may become too complicated, and some assumptions have to be put in. But, at the first stage there should be a realization that flat fields are not we are looking for.
A real scene could be broken down into regions and the standard deviation obtained by the process of integration.
Are you saying that you would quote me 5, 10, 20, ..., numbers, complete with area coordinates from which the respective SNRs in the seemingly uniform patches were derived, regarding a simple question if I ask what is the measure noise in this picture that I took with my favorite camera?
Shot noise is the predominant source of noise for all but the darkest regions of the image, where read predominates (ignoring thermal noise and PRNU), and shot noise does follow a Poisson distribution.
The operative word I used in my original message is the "usual" Poisson process, as done in flat fields. A natural image may still be described by a more complex Poisson process, with some very interesting properties.
DXO does give a full SNR plot and also a value for 18%, which is often taken to represent average scene reflectance. I really don't get your point.
Here is the point: What exactly is the meaning of standard deviation of shot noise being the square root of mean signal in digital imaging? As my example in the following links shows it is only applicable to group of images:
http://www.luminous-landscape.com/forum/index.php?topic=51782.msg427055#msg427055 (http://www.luminous-landscape.com/forum/index.php?topic=51782.msg427055#msg427055)
Not applicable to a single image, in general, which is usual photography - does one acquire 100 images of the same static scene just to get a better handle on shot noise? What is weakly applicable in the case of a single image is that the sqrt of a pixel electron/photon count gives an approximate handle on the standard deviation of noise, if the count is high enough. But this number is not the actual noise value on that pixel in a single picture, as sqrt is an average measure of noise if a large number of pictures had been acquired. Some interpretation of the validity of the sqrt of pixel count as being shot noise, which is almost always done without even the blink of an eye, is needed. Under what circumstances is that valid? For a single image that will basically boil down to transitioning from a temporal statistic to area statistic (i.e. spatial) in the neighborhood of a pixel. For example, if the signal variation in a pixel neighborhood is smooth enough, then what is the equivalence of pixel sqrt count to the area statistics. This issue is not as trivial as it is usually treated.
Sincerely,
Joofa