Trade Offs When Shooting HDRIs for Image Based Lighting.

Now that I’m shooting these timelapse HDRI sky hemispheres with the Nikon D800E, I have some added challenges associated with the usual capture of a HDRI sequence. Because I’m shooting for image based lighting solutions, I need the following:

  1. Maximum image sharpness, so the sky can be used as an animated background environment
  2. Maximum dynamic range, for accurate, contrasty image based lighting that produces complex shadows
  3. Fast shooting speed, so
    1. there is no ghosting from cloud movement, and
    2. so that I can choose whether to create time lapses featuring slow moving clouds or faster, more dramatic ones
  4. Minimal lens flares, halos or streaks, so that lens flare effects can be added in post. They just look ugly too!
  5. Perfectly timed frame intervals (an intervalometer)

Solutions

I have two options to control the D800 to solve all the above problems. The camera can’t do it on its own, because it only shoots auto bracketed with 1 ev steps (enter shared pain of all HDR photographers here) and I need each frame sequence to be shot at an exact time interval for the timelapse sky footage to appear smooth.

The first option is the Promote Control, an intervalometer that takes HDR bracketed sequences over a set time period. Its one drawback is that it can’t adjust the aperture or ISO during the bracketing sequence, so once I choose an aperture, I’m stuck with it. Therefore I can’t shoot 1 second at f3.5 all the way down to 1/8000th at f22. It has to be from say, 2-4 seconds at f3.5 all the way to 1/8000th, again at 3.5.

The second option is to hook up a laptop with software that controls the camera via usb, but this is slow. The benefit to this is I can tell the camera to shoot whatever combination of f-stop and shutter speed I want, resulting in capture of the complete dynamic range in any situation. The huge draw back here is the speed that shooting takes place – only 1 frame a second! This really kills ghosting and image sharpness and limits us to only super fast moving time lapses.

I feel that while using the 1 second capture time between frames might have been fine 9-10 years ago with reduced sensor sizes, with the resolution of the D800, we are going to see ghosting for sure. So I’m somewhat firmly in favour of the Promote Control.

Trade offs when shooting with the Promote

The choice then becomes, what aperture do we choose if we have to go for f3.5 or f22? The toss up between the top 5 points above is this –

  • Using a wide open aperture helps with lens flares and sharpness (ok yes, f8 to 11 is optimal for sharpness, but not flares), but detracts from dynamic range by about 2-3 stops. Sometimes you will never notice this range missing, especially if the shots take place during the afternoon where the sun isn’t as bright, or when it is partially or fully obscured by clouds. Since this is a rather popular time of day for image based lighting, this is currently my best guess at the ideal shooting set up.
  • The second option is to shoot with the aperture stopped down as much as possible. This reduces sharpness by a bit, increases dynamic range and makes lens flare much worse. Oh, and the 4-8 second end exposure required will probably create more ghosting or muddied/blurry looking clouds.
  • The third option might be to go for f/8 or 11, this would give us the sharpest possible image, but some level of lens flares and ghosting due to 4 second shutter speed on the longest exposure.

The points above are actually rather hard to balance when you’re restricted from changing aperture, and I’d appreciate your opinion if you are an end user of image based lighting (please comment below or email me!).

Yet another point is that I could shoot at f22 during day shots, with the assumption that most virtual cameras aren’t going to be looking right into the sun, therefore the lens reflections and ghosting won’t be as apparent.

Summing up

As you might have gathered, I’m leaning towards shooting with the Promote at f3.5 mostly during the afternoon/sunset anyway, due to the more appealing sun angle and colour, therefore the loss of dynamic range will be minimal. I’m also talking to the team who created the Promote and hoping that perhaps in a future update we might see aperture bracketing along with shutter speed bracketing!

I’d love to know what you think, and what you’d prefer to use in terms of the above trade-offs. So please comment below, and if you’d like to continue following along, click here to sign up for the newsletter. I’ll notify you whenever the making-of is updated, or when samples and tutorials become available and so on.

Thanks for reading!

Links and Resources

Promote Control – Intervalometer, HDR Capture and Bulb Ramping.

Timelapse HDRI Sky Hemispheres – Overview of the project, based on Debevec’s Direct HDR Capture of the Sun and Sky

I’m still investigating software solutions!

3 comments

  1. I’m by no means an expert either shooting HDRs or using image based lighting – let’s say I’m a semi-ediucated enthusiast. I’ve always been under the impression that you should always have a fixed aperture, and only vary the exposure, because the focal length of your shots will change from exposure to exposure if you are changing the aperture. This will lead to some ugly haloing and other artifacts when you merge to HDR. This probably isn’t going make much difference in sky shots, seeing as how the sky is at infinity for all intents and purposes, but I mention it here because it will change the nature of your other undesired quality – the lens flare.

    As for the image based lighting, I’d like to see your examples done with a lighter material on the ground plane – the dark grey makes it difficult to evaluate shadows and secondary light bounces. Just saying.

    • Hey James, thanks for chiming in. I just know that Debevec changed aperture for his time-lapse HDR skies, so I guess I just assume that’s correct 🙂 I’m actually thinking I’m going to have to move up to f8 or f11+ regardless though, as I couldn’t capture a midday sun properly at f3.5. I may have even clamped off one of my first tests too.

      I’m in the process of doing some more tests to judge the effect of losing 1, 2 or 3 stops and doing comparison renders. I’ve also just shot a HDR “by hand” going from f3.5 to f22, so will see if there’s any halo’ing. I think though if you shoot most of it at f3.5, all the way to 1/8000th, and then begin closing the aperture, then halo’ing might be less of an issue, will soon see 🙂

      And yeah, lens flares are such a pain. Would love some kind of intelligent filter that could detect and remove them! I’ve read that Photosphere does it, but the link is down. If someone has it lying around I’d love to have a go, since obviously hand removing them in footage is impossible 🙂 Actually just found via HDRLabs, HDRExpose apparently can remove them, so I’m downloading their trial now.

      Will change that ground plane too, thanks! Email me if you’d like to get your hands on a few frames to test for yourself. Otherwise I hope to have some samples out for everyone in a week or two.

      Cheers!

  2. BTW I just updated the Blender post with another, brighter ground plane/higher exposure.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.