This page is about the “quality” of light when using an HDRI for image based lighting. This is not an ACES only topic, but as the ACES RRT can handle high dynamic range imagery better than a standard sRGB view transform, it felt right to have it as a part of Learning ACES.
On this page I will focus on:
Mistakes I made setting up a scene in Blender 🙂
Setup of a new test scene in Blender 2.8 beta to check out HDRs.
Comparison of Theta-S vs. DSLR HDR image quality when it comes to dynamic range.
“Fake” in a new sun into a low dynamic HDRI and the side effects of doing that.
A collection of HDRIs from the Theta S, render results and balanced renderings.
Preparation: finding the right exposure in Blender without an HDRI
When taking a photo all materials, “shaders” and the light are already set by nature.# A camera is used to a find the right exposure and take a picture of a scene.
In Blender I can set materials, shaders and the intensity of the light. Without measuring the light it is easy to mess up a scene’s exposure.
When I started thinking about the UNDERSTANDING GAMUT WITH ACES AND BLENDER page I created this very simple scene with the cubes and the ground having each 0.180 for the base color in either R, G, B or all of them. A white (RGB = 1.0/1.0/1.0) Blender sunlight with a strength of 15.0 is lighting the scene.
The addition of the Colorchart gives a reference about the exposure of the rendering. Suddenly the image looked over exposed. After adjusting the exposure of the rendering to match the 18% grey patch – the image feels under exposed.
My day to day job is compositing with Nuke & Flame. How many times I had to work with renderings which felt somehow “wrong”. Playing around in Blender helped me to better understand lighting in 3D.
Using HDRIs for lighting
Especially easy to use and small 360° cameras like the Ricoh Theta are very tempting to use for creating an HDR instead of carrying around a big tripod, a panohead, the DSLR, the lenses, ND filters etc. And for indoor situations these small cameras can be great. But for capturing direct sunlight theses cameras are not well equipped. The comparison of the two cameras might be unfair – but it is often what you get if you don’t specify otherwise.
Prepping a scene for the HDRs.
The Blender test scene with only pure red, green and blue cubes are not really natural colors. Thats why I added more cubes and spheres and assigned them the all the colors from the colorchart patches. Check the link to the ColorChecker – Wikipedia.
The Theta-S HDRI – Seville – Parque Miraflores
The sun in the Blender scene is turned off and the entire lighting is only coming from the HDRI.
The Canon 5D HDRI – Hamburg – Hafencity
The balanced results of both renderings stacked.
Theta S HDRI – “fake” a new sun.
To make more out of the rendering with the Theta-S HDRI, I merged the sun from the Canon 5D HDRI and replaced the “weak” sun.
Conclusion
Comparing the last two images doesn’t show a lot of difference in the quality of light. Is the Theta-S HDRI with the sun taken from the Canon HDRI accurate? For sure not, but it looks good. But if the job is to comp in a CG car into the “Hamburg Hafencity” scenery and the camera plates & HDRI were taken only minutes apart – then it might be easier to place the 3D car with a realistic light in the scene when the camera plates and HDRI have the highest quality.
On the other hand at some point you will also get the desired look with a poorer quality HDRI – it might just take a lot longer to get there.
Examples with good and bad Theta-S HDRIs.
Last but not least – here are some more renderings of bad and good HDRIs that I made with the Theta S in the last years. Somehow I look forward to the new Theta-Z1, although it still cannot compete with a DSLR in bright sunlight of course.
The following examples are HDRI, the rendered image and the balanced result with the middle grey patch at about 0.180.