The Grid. And The Windy Pixel HDR FAQ and tutorial!

I’m using today’s post to start a FAQ section that will evenutally try to address all the great questions we’ve been getting from readers. Since we started The Windy Pixel, we have received several emails asking about HDR photography and how it is done. But first – today’s post!

Early in the morning, when the neighborhood is still sleepy enough to make very little noise, you can hear the city’s low, constant hum. I used to hear it in the ‘burbs growing up, leaving very early for school. Without any local distraction, the noises of the expressways and airports blend into a distant white noise. High above the neighboring south side buildings, the huddle of downtown skyscrapers on the horizon behind me, the University of Chicago hospital’s turrets framed slag heaps on the horizon and the vast western grid, hammering out its one note song.
Prints available.

Photos by Justin Kern – Feel free to use images with links and credit – no commercial use without permission.

Okay – on to the tutorial! As I intend to go into a lot of detail in the tutorial, I will put the FAQ: HDR up top – so those of you who don’t want to make HDR photographs yourself don’t have to plod through a lot of Justin’s ramblings.

What is HDR photography?

HDR stands for “high dynamic range.” This is, of course, a relative term and tacitly implies that “normal” photography is low dynamic range. This has to do with something called bit depth (i.e. the number of shades of gray between white and black) and the dynamic range (i.e. the range of luminance recorded between the brightest and darkest pixels) in digital photographs. Bit depth and dynamic range are not the same thing (more on that in a minute). Any form of digital information exists in discrete quanta. Most photos you see on your computer screen are JPEGs – this file format allows variable compression to maximize the amount of visual data stored per byte – i.e. they are small files. It also just so happens that a JPEG photograph can represent 256 different shades between (and including) black and white. Furthermore, the difference in brightness levels on a monitor between black and white is somewhere between 256:1 and 1000:1. All this to say, JPEG looks pretty close to what photographic prints look like to our eye under decent lighting. In other words, what we are used to seeing from the days of film.

But, when you look at the sky, does the ground become so dark that you can barely make out any detail? Of course it doesn’t, and that is because you’re eyes see millions of shades of gray between the bright disc of the sun and the backs of your eyelids at night. Dynamic range is frequently described in units of EV (or exposure value) where each EV is twice the value of the previous EV. This is easier understood in practice – if you have a camera that can record data across 12 EV, and you expose so the sun is just barely pure white, anything that is less than 4096 times darker (pretty much everything except the sky around the sun) will be pure black. And that will likely be a very boring photograph. The human eye probably has something closer to 32-bit capabilities (that’s 4,294,967,296 colors!), and if that weren’t enough, it can probably capture a dynamic range of 1,000,000:1 or approximately 20 EVs. The simple truth is truth is that film or JPEGs don’t really capture all that much information – photographs never really look like the real thing.

Some photographers use this to their advantage – Ansel Adams overcame the limitations of film by creating an elegant and scientific process of mapping light values into the negative and then onto the print … in his head. If you’re like me then you don’t have the creative genius of Ansel Adams. You probably don’t have a donkey, a view camera and a lifetime to dedicate to documenting the disappearing American west (and Yosemite and Yellowstone are overrun these days anyways!). But who wants to compete with the master anyways – you’re living in the age of pixels and HDR photography! Simply put, HDR photography is the process of taking multiple exposures of a scene at different shutter speeds, so that you have very dark exposures and very light exposures, and using a computer to put the light values your camera’s sensor recorded into a single photograph. That photograph is a representation of every light value from the darkest shadow to the brightest highlight without losing details in each. HDR photography is a crutch for me. I use it to transmit my emotional response to a landscape to you via a computer screen.

But HDR photographs are JPEGs – so how does any of that techno-babble above mean anything when I’m still looking at a computer monitor and you’re still showing me JPEGs?

Great question. The answer is that your brain is pretty damned fancy, and it can interpret more than it sees. Once you make an HDR photograph, you use a computer program to “tone map” the light values into the draconian space afforded by JPEG. This process causes reversals, or some “dark” things are lighter than some “light” things. When I see an HDR photograph, I feel that our brains are capable of reversing some of the tone mapping process and seeing the scene for what it was more clearly.

Why even bother? Seems like a lot of work.

Because if you were with me up on the roof when I was taking the photograph above then you’d probably agree with me; but the world from up there didn’t look like any of these:

It looked like this:

Okay then – how do I make HDR photographs myself?

First let me say that I’m a total amateur – I am not a professional photographer and likely never will be. There are many other great HDR tutorials out there as well as a few books published by professionals. I have borrowed shamelessly from these tutorials in my attempt to find my own HDR style … I’m still looking . (Stuck in Customs and Before the Coffee deserve special mention here)

Hang on … (drags out soap box … climbs on top) – I am a feverish defender of HDR photography. If you make HDR photographs yourself or have seen much HDR on the web, you’ve probably noticed that there are people out there who *hate* HDR. I think a lot of this is sour grapes – HDR is relatively new and draws a lot of attention because of that fact. Moreover, there is a lot of bad HDR out there (just as there is a lot of bad “normal” photography). So, especially on photo sharing sites like flickr, it is sometimes the case that lousy HDR photographs get attention while exceptional LDR photographs are ignored.  I feel there is something to be learned from every photographic technique.  I wasn’t much of a photographer when I used film, I was always bored by my photographs.  Digital is a pure medium, and in that medium almost anything is possible.  Don’t write off a technique because you think people do it poorly or use it too much.  Keep your eyes and your mind open.

This is my attempt at an HDR explanation and tutorial. I’ve tried to illustrate each step with as many screen shots as I felt necessary, but I will qualify this tutorial with the following disclaimer: I am assuming that you have a rudimentary knowledge of Adobe Photoshop – nothing fancy, don’t worry!  You will also need a camera and a tripod (or something motionless on which to rest the camera).  Cameras that take “RAW” files (or files that have not been processed into JPEGs) are ideal – but don’t fret, all DSLRs and many smaller point and shoot cameras take RAW files.  Most importantly, I ask that if you have a question about this tutorial you do not hesitate to post a comment or email us using the Contact link at the top of the blog!

Here’s what you do:

1.) Put something interesting in front of the camera. The how, what, why and where are your decisions, but I would recommend looking for dramatic light levels.  Twilight is my favorite time to photograph.  It is a highly coveted light and has been affectionately named “blue hour” for obvious reasons.  The Irish call it the gloaming and it is a magical time not to be missed for the world.  My very favorite HDRs were all taken in those precious moments before the sun rises and I’m always angry with myself when I sleep 1 minute too long and miss that split second of perfect light.  Don’t be misled by the cheap and tawdry sunrise and sunset (kidding), the real show happens just before and just after.

2.) Take some images. Bracket two or more exposures (many DSLRs and other cameras have a “bracketing” button that allows you to tell the camera to take 2-9 frames spaced by 1 or more EVs). Always shoot RAW. There are some photographers (though not many) who would tell you to take JPEG because blah blah blah. RAW is a digital negative (if there is such a thing). You could shoot JPEG for two years and then regret not having the extra processing leeway and quality of a RAW file like I did, or you could shoot RAW right now and forever. Your choice.

3.) Convert RAW files to TIFF. Use Adobe Lightroom or another RAW converter (Photoshop, Aperture, or your camera manufacturer’s software) to convert the files to 16-bit TIFFs.  Even the manufacturer of Photomatix will tell you, HDR programs are no good at handling RAW files.  Yes, they give you a passable result, but you have almost no control over the white balance etc.  A proper RAW convertor is what you need to make all your pre-HDR adjustments and to get the best quality HDR possible.  You’ll notice in the screenshot below that I have five exposures bracketed by 1 EV each – there is the metered exposure and a -2, -1, +1, +2 EV exposure opened in Lightroom:

The advantage of taking photographs as RAW files is that the light values present in the scene have not been adapted to a curve, they are linear.  If you haven’t already, set the tone curve to “Linear” in the develop module:

4.) Export into Photomatix/HDR MAX/etc. Open the TIFF files you generated in an HDR software.  There are more and more options available all the time – I use Photomatix and HDR MAX (here I’ll be showing you how to use the most popular, Photomatix).  I should also point out that Adobe Photoshop CS3 and 4 have the capability of generating tone mapped HDR photographs, and I will include a tutorial on that in the future.  The makers of Photomatix have written a plug-in for Lightroom that allows you to export the TIFFs directly to Photomatix (if you have saved the TIFFs, open Photomatix and go to “Process” –>  “Generate HDR”):

5.)  Generate HDR. Photomatix will ask you what you would like to do with these photographs, click “Generate HDR” here the program will allow you to align the images in case you moved the camera as you were exposing.  If you used a tripod or rested the camera on something stationary, you can unclick the radio button next to “Align images” as it will just take more time.  If, however, you hand-held the images, aligning the images will significantly improve your results (these align features have gotten much better recently and can save some otherwise lousy handlheld HDRs).  Once your computer’s CPU has fired away for a while you’ll see the program trying to show you a 32-bit image on your very-not-32-bit monitor.  It will look very strange.  There is a box called the “HDR Viewer” which gives you luminance data as you drag your mouse over the image preview.  What you are going to do is click the button that says “Tone Mapping”:

6.)  I can haz tonemapz right? K thx bai. You are now presented with a small image preview and a lot of bells and whistles.  This is where things get fun (I bet you thought they never would!) – here are all the sliders and adjustments that allow you to customize the way the program maps the 32-bit light values into the 8-bit JPEG space.  Play around and find the look you want:

Look at the window that says “Tone Mapping Settings” – you’ll notice I’m using the “Details Enhancer” – I prefer this method and have never used “Tone Compressor.”  There are seven sliders:  “Strength,” which adjusts how powerful the reversal effect that I mentioned above is – the higher this is the more dramatic the HDR.  “Color Saturation,” pretty obvious – the tone mapping process reduces color saturation and you’ll need to play with this to get the colors back to the way you remember them.  “Luminosity,” determines how much light comes through the whole image, it’s effects can be subtle but think of it close to the “brightness” adjustments you’re used to in any other image editor.  “Microcontrast,” helps you pop all those little details you see with your eye – if this is set to a high value you will see that tone reversals can happen on a smaller and smaller scale, increasing local contast.  “White Point” and “Black Point” will move the pixel values up and down in the JPEG space (the histogram shows you a density distribution of pixels across that space – the values all the way on the right are pure white, and those on the left are pure black).  Because you are working with an image that your computer can’t display, and a preview that is a tiny fraction of the size of the full image, the histogram serves a vital function – it allows you to predict how much contrast and luminosity are in the final image.  “Gamma” is the same as  any other image editor and will help you correct brightness – but I always leave it at 1.  You’ll notice I’ve cranked the settings to their max in this image – but that is certainly not the case with every image.  You’ll find that each slider needs to be adjusted for each image – it’s fun to play around with these after all.

Finally, you’ll see that therea re two other menus in the “Tone Mapping Settings” box: Color Settings and Smoothing Settings.  These allow you to make adjustments to color temperature and how the program handles blacks and smooths light between parts of the image.  I’m not showing these here because typically they require less adjustment.  I keep “Micro Smoothing” at somewhere between 1 and 10 which greatly helps to reduce noise and the halo artifacts you sometimes see in people’s HDR images.  Click “Process” and the computer will crank away again and show you the result – save as a JPEG or a TIFF (your choice – I use JPEG).

7.) Fixing problems in Photoshop. Open the tone mapped image in Photoshop.  Sometimes you’ll notice that there are highlights or shadows in the image that look strange.  Or perhaps you’ll notice that the sky looks very silly.  What you’re going to do next is open up JPEG images of the under or overexposed frames to correct these featues.  Although the sky looks fine in the example image, I’m showing you how I fix problems because it is the most frequent weak spot in my tone mapped images.  I’ve opened the -1 EV image along side the tone map, and copied it as a layer on top of the tonemapped image.  Next, I’ve clicked Layer –> Add Layer Mask –> Hide All.

You should now see the tone mapped image; in the layers palette (bottom right of the screen grab below) you’ll see an underexposed layer with its black mask next to it.  When you create a layer mask you’re telling the program what parts of the image you want to see.  Black = transparent, white = visible.  Make sure you’ve selected the black mask so that when you make adjustments you are changing what is black and what is gray/white.  Next, you’ll see on the tool palette on the left I’ve selected the paint brush and changed the Opacity (on the tool bar above the image) to 30%.  This is going to let me carefully and slowly adjust what is showing.  Paint over the sky so that you begin to see the underexposed image’s sky coming through (if you look at the black mask next to the underexposed layer you’ll see the top part is a gray color – that is how it looks after adjustment).  Next flatten the image (Layer –> Flatten image.

8.) You can stop here if you like the result. Everything after this step is up to your personal taste and allows you to put your own spin on the workflow.  You can use any number of creative post processing techinques to add more/less contrast or to accentuate details.  Some people, myself included, use a Photoshop plugin called “Lucis Arts” to give it a look that some call animated.  This software is rather expensive (it’s gotten more so since I bought a license) – especially for a photoshop plugin.  There is a competing product called “Topaz” with which I have no experience.  Do some cyber sluething and decide if you like Lucis or Topaz before you purchase (it is also a flashpoint for some people – with the hating of different techniques).  Just for fun I’m going to detail a techinique already built into photoshop that has a Lucis Arts quality to it and I’m going to combine it with a techinique called the Orton Effect, whose roots lay way back in the chemical darkroom.  (Lucis by the way, is a differential hysteresis algorythm – the company holds a patent, but if you are a programmer and a math wiz you can write one yourself.  If you search on flickr you’ll find someone whose done just that).

Enough talk – the tutorial … must focus.  Click Layers –> New Layer –> Duplicate Layer three times, until you have four copies of the same layer.  I’ve named my layers to indicate what I’m using each one for 1-High Pass, Soft Light; 2-Multiply, Gaussian Blur; 3-Screen; 4-Background:

9.) The Orton Effect: Select the Screen layer and change the blending mode (drop-down menu in the layers palette) to Screen.  Select the Multiply layer and change its blending mode to (you guessed it) Multiply.  While still on the Multiply layer, go to Filters –> Blur –> Gaussian blur and change the amount of blur to your liking.  With 12 megapixel images, 3-5 looks best (if you hide the topmost layer you’ll be able to see the effect):

10.) Popping contrast with a High-Pass layer.  Your HDR has a cool fuzzy and dramatic look to it now, but you’ve lost some detail.  Heres 1 of 1000s of ways to bring it back.  Select the “High Pass, Soft Light” layer and go to Image –> Adjustments –> Desaturate.  Then go to Filter –> Other –> High Pass and select a value between 3-10 (this is a fun time to play around with the settings).

This layer will look like an Etch-A-Sketch drawing until you change the blending layer to “Soft Light.”  Things are starting to look pretty good.

11.) Curves, saturation and you’re done! If needed, give the image a final Curves adjustment and/or tweak the saturation (sometimes it gets a little too high).  Flatten the image and tada!:

The difference between this image and the one I posted is up to the processing and what route you take.  You can make these things your own with a little experimentation!  Have fun and please write with any questions!