I’ve been meaning to do this for some time now.  I think it will be helpful for some to understand what cameras have wider dynamic ranges and I’ll try to explain why this is important for HDR capture. The chart I show at the end is showing the amount of dynamic range (stops) that were measured on the sensor. The larger the number the wider the range of light captured (hence dynamic range).

I gathered data from DXOMark.com and then put it into a chart format.  I only looked at the digital cameras of the last 5 years (from now) so some will not appear here.  DXO Mark  does not cover every single camera.  For example, the Sigma SD14 is not listed here and I use that camera for professional work.  The SD14’s Dynamic Range is (feels) huge when I am editing shots from it.  I’d love to see where it would have fell in this data set, but DXO Mark doesn’t cover the Sigma line.

The first thing that will blow your mind is how far ahead of the curve the Fujifilm S3 Pro really was (and still is) in terms of capturing plenty of light in one shot.  It is no wonder that so many wedding photographers I’ve bumped into rave about it and the S5 Pro.

What is important to derive from the chart (relevance) is if you are trying to follow a High Dynamic Range Imaging workflow taught for VFX or Tonemapping.  Obviously if your Guru has a D3x and you are copying their steps using a Canon or Olympus… mathmatically there is no way you can get the same results.

I cringe when I see someone post (or say on a podcast) that you should always take “X” shots…. or you ONLY need “X” shots at “X” EV.  Then later I find out they are shooting with a D3x.  UGH!

They don’t even realize they are accidentally misleading the droves of followers who will now go attempt to recreate images made with shots taken with their own cameras which have inferior sensors for capturing a wide dynamic range with the same number of shots taken.

If there is anything I have learned in the last 3+ years of day to day HDR capture and editing pipeline… there is NEVER, EVER, EVER a scene exactly like the last.  That said, I’ve gotten to the point where I know when to take 5 shots where 1EV jumps will be perfectly fine and other scenes require at least 11-12 shots taken with 0.7 EV jumps or possibly even 15-17 shots at 1/3rd EV steps.

Even basic landscape shots vary greatly depending on sun/shade and then considering the content trying to be captured.  I’d really have to do a workshop or video to show you what I mean and how to easily evaluate where to start your bracketing for the base exposure’s exposure compensation and approximately how many shots and how wide/tight to make the EV steps.

What this chart (and blog post) doesn’t show is how many shots each camera can handle in an AEB sequence.  I”m going to cover that in a future post because that is as important (if not far more important) than the DR of a single image captured.

Even though tonemapping uses pieces of each of the images captured, if you take fewer shots to cover the same dynamic range of a scene someone else takes or if your sensor has a lower DR than that of another camera taking the exact same number of shots, then your tonemapping will be pulling from noisier shadows of the underexposed image(s)  and pulling from more blown out highlights of the over exposed image(s), thereby causing undesirable saturation and IQ issues.  The extreme example of this is when you try to process an image from one RAW (faux HDR).

Also, if you shoot JPEGs for sequences instead of RAWs, then you will also need more shots with tighter EV jumps to match a RAW workflow.  If you plan on doing a lot of pre-processing of images prior to merging them to a HDR file, then you will also benefit from tighter stops… and there are very good reasons for this kind of workflow as well.

Also… I am so focused on extremely high DR scenes given I shoot interiors for a living, that this becomes absolutely paramount for professional HDR workflows.  You can get away with a lot on landscape shots. Landscapes for me is childs play.  I’m not saying that to be cute, they are just a total cake walk compared to shooting dark interiors with blazing bright light spilling in through windows or open doors.  Apples to Oranges.

I have never encountered a scene in over 10,000 tonemapped HDRs now that did not benefit from taking more shots with tighter EV jumps.  Low contrast landscapes where the sun is not in frame are not the kinds of shots I’m talking about here.  I’m talking high contrast landscapes to interior shots with large DR scenes.

I’m also talking about creating more realistic and more natural tones for tonemapped images.  If you are gunning for surreal then I would say the easiest way to accomplish that is to ignore anything I teach.  Because if you stretch your shots too wide (large EV jumps) … you are guaranteed to easily get oversaturated and surreal hues that will look like technicolor art renderings.  If that is your final goal, then don’t shoot with a sensor with a high dynamic range and then purposely max out the EV steps.  So for you,  virtually any Olympus or Canon body and using 2.0 EV steps is perfect.  Because 2.0 EV jumps is too big and Canon’s and Olympus’s sensors are of the narrowest range in the industry… so that combo will help you create surreal arty photos using AEB sequences.

To create more realistic renditions of a scene (tonemapping) you can of course use the same Canon, but simply tighten up the steps and increase the number taken to appropriately cover the dynamic range of a scene.

For architectural work with blazing bright light spilling into windows and deep dark shadows… I won’t go over 1EV steps when I bracket.  The place the tonemapping suffers the most from shortfalls of a capture is that window/door area itself.  The edges.  In fact I rarely go over 0.7 EV jumps for such scenes.  When I do test out 1.3 and higher EV jumps for those shots, I inevitibly get an image that has more noise being pulled from the shadows and the window/door edges suffer quality issues.  Also,  large EV jumps will cause the saturation of the upper range to experience strange hues (false to our eyes, not the sensor).

This is all a result of the tonemapping pulling from images that have larger jumps between due to using wider EV steps.  It is forcing the tonemappers to pull from larger areas of luminosity from each individual image.  So you can imagine that this is where using a camera with a larger dynamic range becomes very valuable!!!  Garbage pulled, garbage displayed.

By feeding a merge more images, the tonemapping pulled from the sweet spots of the histogram of each shot.  Hence why the more the shots taken with smaller EV jumps, the less often the tonemapping pulls from the wide ends of a single image’s luminance range (where shadow noise is worse and highlights are more blown out).

The chart below  is especially useful if you are considering a camera that only has 3 AEB.  Look at the DR of say a Nikon D5000 versus a Canon Rebel series.  Heck a D5000 has a wider range than any of Canon’s flagship bodies.

You could tripod off two cameras (like a Nikon D5000 versus a Canon 500D) from the same location, but the same 3 AEB capture using say a +/-2.0EV is going to be a dramatic quality difference.  Consider the fact that the D5000 is capturing a full 1EV more per image! That means when a tonemapper is dealing with a large DR scene and it was only fed 3 images, it really, really, really, REALLY matters what the quality and range of each of those 3 independent images can deliver.  Take a moment and think about that.

See why I am always groaning about Canon now?  It has everything to do with HDR capture issues (for professional work).  I am actual quite keen about Canon otherwise.  I own three 580EX II flashes and a ST-E2 wireless unit.  I have created my best portraits in my portfolio with Canon’s cameras!!!  I really do love the look off of Canon’s sensors, but for HDR work… they are behind Nikon and Sony for sensor DR.

Now look at a D3x versus Canon’s flagship model.  Same story.  Same number of shots on a canon versus the Nikon D3x will result in less range captured.  However…

… let us not forget that regardless of the sensor’s Dynamic Range reported below… each of these cameras perform differently in high ISO environments.  There are multiple considerations in the end.  Dynamic Range, Noise Performance and AEB capabilities of the camera.  Actually FPS is very critical for certain shooting conditions too.  The faster the FPS capture, the less of an issue of things moving/zooming in and out of frame during a capture. This chart only considers the raw data collected showing the full dynamic range captured on each camera’s sensor.

Enough chatter… chart below speaks volumes.
The larger the number, the more (stops) dynamic range of a scene captured:

———————————————————————————————-

Camera Sensor Dynamic Range

The 2009 follow-up article to this blog post is linked below

Camera Sensor Dynamic Range (continued)