The term HDRI has been a common phrase in the VFX community for almost 20 years, but the term has always been extremely vague, where "high" could mean anything.
#SUN CORONA WITH HDR FULL#
In this section I've described what is essentially a Full Dynamic Range Image (FDRI), meaning it has no clipped value pixels and therefore the entire dynamic range is captured. This more than covers the entire dynamic range of the scene and the final HDRI ends up between 20-30 EVs.
#SUN CORONA WITH HDR PLUS#
For high contrast situations, like direct sunlight, I only need the first 9+9 exposures plus the additional 9 for the sun. Note: I’ve never needed over 40 EVs of dynamic range in any of my HDRIs. Combining the dynamic range of 23 exposures adds to way over 30 stops of dynamic range per HDRI. I’ve made a simple chart plotting these two bracketing methods, based on the dynamic range on my camera’s sensor: Now, let’s say I add the extra bracket of 5 at the end, how many stops of dynamic range is shot with 23 exposures at +-1EV? My camera sensor shoots just under 15 stops of dynamic range per exposure. In outdoor scenes, I mostly stick to 9+9 brackets, since this covers completely underexposed (all black) to completely overexposed (all white). The resulting HDRI goes from complete black to complete white in 99% of the cases, having gone from 1/8000s to 30s exposure. My camera just happens to have bracket modes of 3, 5 and 9 exposures, so I can quickly shoot 2x9 brackets and add another 5 at the end, should the scene need it (dark, high contrast environments). I usually shoot 18 or 23 exposures per angle, depending on the scene. But, this means 7 exposures are insufficient to get the full dynamic range and therefore adding more exposures is necessary. Generally, +-1EV is good enough to ensure accurate coverage, in most situations. More exposures mean better data, but that doesn't mean you have to shoot +-1/3EV either. If, however, you bracket the same amount of exposures far enough apart that you do cover the entire range (like +-3EVs), you can end up lacking the important overlapping data between exposures and that can make it really difficult to get a smooth and accurate interpolation between the exposures.
Too few exposures can result in clipped shadows and highlights in the HDRI, which means you're not capturing the entire dynamic range of the scene. It's actually common to shoot no more than 7 exposures and move on (Weta Digital has recently started shooting 14 exposures instead of 7, in their so-called “Extended HDRIs”).
#SUN CORONA WITH HDR MOVIE#
On a movie set, there's only so much time a VFX supervisor or on-set data wrangler can spend shooting HDRIs before the light crew starts disassembling the set. It is my belief that many HDRI photographers shoot too few exposures, either because they don't pay attention to clipping, or perhaps the scene they’re in changes too quickly, making it difficult to spend too much time on one HDRI. Okay, let’s get to it and start with the most important one, dynamic range: I will stop and explain here and there, but please check out my links at the end if you feel like reading up on these subjects a bit more. Keep in mind that I expect my readers to have a basic understanding of photographic terms like stops, bracketing, and clipping. Instead, I’ll try and explain a few simple, but very important, aspects of HDRI creation in the hopes that we’ll continue the discourse agreeing on the key principles laid out in this post. I’ve come to understand that it’s not for me to try and lecture the reader on photography, colors and computer science. I’ve written about this subject so many times but never hit Publish. I’ve always had difficulty writing, especially when it comes to the subject of HDRIs, as it is both technical and artistic. This is a post I’ve been wanting to write for a long, long time.