Here’s some steps I’ve used for generating some motion/video out of HDR stills. There are many different workflows and I’ll touch on some of them from an overview perspective now. The first one is the workflow I employed for an older clip shot at Eden Gardens State Park in North West Florida. For reference, this is the clip (overview to follow below it).

HDR Video – Eden Gardens 2008 from Michael James on Vimeo.

I shot it with a Nikon D3 with a Nikon 14-24mm f/2.8G lens.  It sat on top of a fluid head tripod I use for video work. I also had a shutter release cable attached to the D3. I set the camera to automatic bracketing and once I determined the mid point I switched from aperture priority to full manual so that each of the AEB sequences I was about to fire would be EXACTLY the same.  I determined I would not need to go a full 9aeb and set the D3 to fire off 7 shots with 1EV steps between.  I lowered the settings of the camera to shoot JPEG in a size that is larger than HD.

Also of critical importance is locking down white balance.  I chose Daylight here (an obvious choice), but will often dial in an exact Kelvin setting for other lighting situations (although Kelvin WB selection is generally not available in lower end DSLRs).

If I wanted to shoot RAW I would not have been able to have captured the entire tilt move in just over 60 seconds due to the buffer issues, but the lower JPEG setting meant I would never tap out the buffer and could shoot burst after burst easily.  Also, because the steps were only 1EV jumps between each of the 7aeb captures, I was comfortable only shooting JPEG.  I would NOT attempt shooting JPEG for anything larger than 1EV jumps between each shot for this type of approach.  Time was critical here to capture the entire tilt move in a short period of time.  (I shoot RAW for all my other typical HDR workflows, including HDR Timelapse)

The shutter release of the camera was set to continuous so that I could hold down the shutter release cable button with one hand to fire off a full 7 shots at 9fps while keeping my other hand on the tripod handle to make small incremental movements to tilt up after each 7aeb capture.  I do it this way to be quick enough to capture the entire tilt move seen in the video in a little over 60 seconds.

To repeat myself slightly now…, I have the camera ready to capture the brackets in continuous shutter release mode and then prepared for the first AEB capture by aiming the camera slightly below the horizon into the deep shadows.  I then fired off a burst by holding down the shutter release cable button and quickly moved the tripod a very small (unmeasured and done by eye) amount; slightly tilted up for the next capture.  Then again held the shutter release cable button to fire off another 7aeb and repeated this process until the tilt to the sky was complete.  This took all of about one minute to capture.

Back home I dump the images to a folder on the computer and then I launched photoshop to begin the merge to HDR process.  You don’t have to use photoshop as the program to merge to HDR, but back in 2008 when I put this together that was the process I used.  I had a few scripts that I had created that had photoshop merge to HDR and then use custom settings to tonemap out to 16 bit TIFFs.

I employ two alternative workflows these days.  One of which is where I merge to HDR in applications that batch merge and have them save .EXR files for each merge.  Those .EXR files are then imported into either Adobe After Effects or Eyeon Fusion and then I use open source plugins to do the tonemapping in those compositing programs as well as the ensuing optical flow.

But in 2008 those plugins did not exist yet and this process I’m still describing was the workflow I used for the Eden Gardens sequence.

What you see below in the screen shot (click to enlarge or right click and open in a new window) is one of those 7aeb sequences in the merge to HDR dialogue. I’m just showing you that to show you that the dynamic range of the scene was pretty large for an outdoor shot and the sky was blown out when you could see the leaves straight ahead and the leaves were pitch black when the sky was properly exposed.

Like I said, there are several programs out there that can automate the process of merging and tonemapping your brackets and I plan to cover how to use them in this capacity in some future posts and training, but the process I employed here was a home grown merge to HDR and tonemapping recipe I used for Photoshop CS3.  In the end I had 30+ Tiffs from that automated process to then use as the skeleton for an image sequence.  I then take those Tiffs to the next step for optical flow treatment.

Before I move to that next step, here is a larger view (click for larger image) of a single tonemapped frame, along with the 7 shots above the top of it showing you the dynamic range covered for that one image.

The next step I did then was to take the TIFFs into Apple Shake which treats the group of single images like a sequence.  I then used custom settings (see image below; click to enlarge) to expand the number of frames to be created BETWEEN each of the frames I was importing into it.  In other words, Apple Shake was now going to create the frames and guess at the pixel movement between the frames I shot in the garden.  There are other programs such as After Effects, Final Cut Pro, Fusion, etc that can accomplish the same thing with optical flow, but I used Shake in this instance.

I also recropped the image in Shake to match the ratio of HD footage and had it export the sequence as 1080p ProRes footage which I planned to send to Final Cut Pro to add audio and titles (see image below; click to enlarge).

Then I imported the 1080p footage into Final Cut Pro (see image below; click to enlarge) and added sound and some titles to export to various formats (including a web version which exists on Vimeo).

Today I edited the HDR Timelapse sequence of the Las Vegas sunset.  I had shown one frame from the sequence over the weekend and posted a larger version previously as well.  This is one single tonemapped edit from the HDR Timelapse sequence. (Video link to Vimeo below).

Today I had a real estate shoot cancel, so I got a bunch of past real estate shots edited and then took some time to piece together the Las Vegas Sunset HDR Timelapse.  Because I chose to merge to hdr and tonemap that one single image over the weekend (above) using Photomatix Pro, I thought I’d use Photomatix Pro again for the entire sequence.  What the hell.  Its not for a client so a little over saturation won’t matter here.  I made a few mental mistakes in the pipeline because I generally use other apps, but the results are O.K. for government work  :)

This was the capture/post flow:

Nikon D3  + Nikon 85mm f/1.4D shot through my hotel window on a tripod utilizing the in camera HDR Timelapse (You can combine a timelapse with bracketing in the D3).  I chose to go with a 7AEB with 1EV steps and the camera was firing off that bracket every 15 seconds.  Of the 7AEB I killed off the final over exposed image of each bracket.  I felt adding that latitude would brighten the image too much.  So each merge to HDR was 6 shots spaced 1EV apart.  The total number of tonemapped shots was 90.  Basically covering a span of  22 mins 30 seconds worth of time during the sunset. Both the Merge to HDR and Tonemapping was done in the batch feature of Photomatix Pro.  I then took the tonemapped images into Final Cut Pro for time and color treatment before exporting for YouTube & Vimeo.

Because I shot this sequence in portrait mode, I opted to make slight color correction differences to two versions and post one video with both playing together.  I should have pushed on image a lot harder to show variation.  Now that I’ve watched it online it seems very subtle.

I also simultaneously shot a HDR Timelapse a foot away from the D3 with a Canon T2i in horizontal mode as wide as the kit lens will go (18mm for that EF-S, but on a crop body).  I shot for a full hour with the T2i.  About 15 minutes prior to setting up the D3 and the same after the D3 exhausted the compact flash I had in it.  It will be interesting to see those results versus the D3.

I’m no expert when it comes to compression/encoding.  I can only say that the video looks a hell of a lot better in FCP than online!

Vimeo 720p Version LINKED HERE (opens in a new window)

I have a lot to share regarding NAB and HDR, from cameras to software. For now just a quick peak of one frame of a sunset HDR Timelapse sequence. I haven’t processed the sequence yet, I just grabbed one frame to quickly process (Nikon D3 + Nikon 85mm f/1.4D set to AEB).  I decided to do what I never do…  use Photomatix Pro to get a saturated and more surrealistic looking tonemapped image for feel.  Because that’s how I feel about Vegas.  Warm and glowing.

I got a ton of shots, but didn’t shoot as often as I had thought because of a few time consuming reasons. I had food poisoning one night/day as well as both speaking briefly in the Post Pit at NAB and also spending two days trying to see all there was to see at the NAB Show itself (which is so big, that words fail to describe).

More to come next week, but I have back to back to back shoots lined up with builders this week for properties that have to be shot right away.  Its torture not being able to process the Vegas stuff right away, but all in good time.

Another quick HDR capture I chose to tonemap via Photomatix Pro
(Those two images will likely be the last I run through Photomatix Pro for some time. I’m just not thrilled with the way it handles color and has more than subtle hue shifts that cause saturation/hue issues)
http://www.digitalcoastimage.com/vegas/test-50mm.html

More to come soon…

Dynamic range of sensors varies from camera to camera. You might be surprised by the (updated) chart below because the dynamic range of sensors is not as directly related to camera costs as you might think. You’ll notice that some mid level DSLRs are way ahead of the curve over other camera’s pro level cameras when it comes to dynamic range of the sensor.  This test was conducted by DXOmark.com and you can go there to read about how the tests were conducted.

|||—— UPDATED FOR NEW CAMERAS TESTED BY DXOmark.com —–|||

What is important and relevant to HDRI is the dynamic range of your sensor. It really matters.  Unless you taking large and tight brackets of each scene then it really, really, REALLY matters even moreso than you think.  If you are following a technique taught by someone using say a D3x and they tell you all you need is X shots, but you are using a camera at the bottom of the chart below and taking the same number… there is no way you can match that person’s results. Its mathematically impossible.

I get asked all the time how many shots do I take when I shoot, but nobody asks me with what camera.  Everyone assumes they are all about the same for HDR.  They aren’t.

Gear used for capturing brackets for HDR is even more critical if you are only taking 3 shots.  It will directly impact the quality of your HDR file. IF YOU ARE SHOOTING FOR VFX WORK THIS IS SLIGHTLY LESS CRITICAL THAN IF YOU WILL BE TONEMAPPING FOR REALISM (only slightly less though).

I’ve captured and tonemapped over 13,000 images for architecture, real estate and commercial shoots.  I’ve forfeited all my vacation time the last 4 years just to do intensive testing about what works better.  Gear, post production, you name it.

I’m pretty obsessive compulsive about trying to get a realistic looking image for real estate interiors and I’ve bought and/or rented about 40% of the cameras on this list below. DXOmark.com seems on the mark as far as my personal results are concerned.

Dynamic Range of Camera Sensors (I chose to focus on modern day digital models only)

No test is perfect. DXOmark has attempted to do this objectively, but because ISO/Noise and how cameras can handle scenes, colors and other differing factors, this should not be the only way to base your purchasing decision.  For example…

Someone could use the Canon 1000D (also known as the Canon XS) plus a piece of gear to get pro HDR captures.

If you bought that entry level Canon 1000D/XS for $400+ and a PROMOTE CONTROL for $299, (which will soon be adding HDR Timelapse to its functions in addition to HDR capture and standard Timelapse), then you could take very large and tight brackets with the Promote Control attached to your 1000D/XS that will capture well beyond what the in camera AEB capabilities of every single camera on that chart above.  Yes, including a D3x or 1Ds Mark III.  Sometimes the piece of “kit” you add to the mix can make a lower end device perform very well (dynamic range speaking here).  Of course the pixel quality of a D3x/1DsMarkIII is easily better than a Canon XS, but at the cost of $8000 it should be.

HDR Timelapse is only supported in a fully automatic way using the most expensive cameras. They allow you to “Set and Forget” so you can leave it on a tri-pod to capture away. Some cameras have timelapse, but can not combine both timelapse along with capturing brackets for HDR.  Automated HDR Timelapse is something that exists in the semi-pro to pro cameras that costs many thousands of dollars.  That is about to change.

Promote Control

I covered the Promote Control in detail the summer of 2009 ( BLOG POST HERE ) and even created some getting started videos which are embedded in that blog post.  A link to their website is also listed on that blog post if you want to view which cameras are supported.

The Promote Control allows extended HDR capture with cameras that both don’t have a HDR mode or a limited HDR mode (such as 3AEB).  The Promote Control also has a separate Timelapse feature as well, but you couldn’t use both HDR Mode and Timelapse Mode at the same time.  Soon you will be able to do just that.  The Promote Control is being worked on (the firmware specifically) to allow HDR Timelapse.

Because the Promote Control connects via USB, this means cameras that only cost hundreds will be able to do what pro bodies that cost thousands can do.  Automated HDR TImelapse capture. The developer has indicated it is now at the top of the cue for the next firmware update!!!