I have read and experimented with HDR equirectangular photos for backround lighting and I’ve come to the conclusion that, being such huge, difficult to produce files, the benefits are not worth the problems. I can achieve the same effects with a good 8bit JPG and a sunlamp or area light and have better control over the final results.
I find that a good outdoor equirectangluar JPG does illuminate the objects in the scene, and the sun in the photo will cast subtle, soft shadows, so when I add a sun lamp (lined up with the photo sun), shadows have a proper soft blue light fill. I understand that with an HDR, using the proper math nodes, I can get sharp shadows without the sun lamp, but I do not see any advantage, given the disadvantages of HDR photos, and given the ease of using the sun lamp.
I’d be interested in learning reasons why I might be wrong. Thanks for any clues.
In my opinion, there are cases in which the use of a jpeg can give a good result, in large open spaces I guess, nevertheless the high range of hdri is really necessary for realistic lighting; think of the presence of buildings or trees in the environment image that cast shadows on the scene, a jpeg will only give diffuse light from the sky, without contrast, because the difference in value between the sky and darker parts is too low.
Without a magical “align selected sun to brightest pixel of background” button, using a sun lamp is certainly not “easy”. It is, however, the way I do it too. Even builtin sky texture can’t be magically aligned, but this is en route. Unfortunately, most HDRs are clipped, and you have to either scale the brightest part up to the value of the sun, or supplement with a separate sun lamp. The first makes for very easy previews, but I’ll rely on the sunlamp in the end since it cleans up much faster than a brightened HDR. 8bit will not be enough to capture even the brightness of the outer atmospheric halo of the sun, and obviously not the disk itself (difference will only be visible in weak reflections - and sun being so small, it’s probably not worth it even). I don’t even think 8bit will be enough if sun lights up/shines through some clouds.
With an ideal HDR pano, you shouldn’t need any math nodes.
You’re not wrong, unless you’re integrating CG on a backplate where it needs to match the lighting exactly.
As others may have said, even painting out the sun on a perfect HDRI can be a good strategy. Chances are the sun has moved anyway from the time the HDRI was taken to when the backplate was shot for film or TV. So much is faked with lighting on set that we certainly shouldn’t fear “faking” things in our fake CG world.
At the end of the day we’re all looking for the best images, not who is technically correct. Doing things “technically correct” should make your life easier and more predictable. As long as you don’t aimlessly encourage people to ignore good HDRI because “it doesn’t matter” then you’re not wrong. It does matter, and so does a pretty picture.
The issue-at-bar is that JPEG and its brethren are image files, designed to be small in size and efficiently displayed on a target device. The compression can be “lossy,” and the image is mapped into the gamut of an output-device. There is no concept of “blacker than black” or “whiter than white.” Nor is there need for it.
HDRI, as the term is used today, is really designed to capture data for subsequent processing. File-size is presumed not to be a concern, and loss is unacceptable. Now, the data can pass through many processing stages without loss, and you can normalize the data between files as needed without affecting them … until you finally reach the end of your pipeline and there produce a “deliverable” file which might well be a JPEG or a MOVie. Only then are you purposely addressing “display-device concerns” and “throwing away information.” (Which, of course, you didn’t “throw away” at all.) You can generate as many deliverables as you need, each with different parameters, all from the same unblemished source.
If you’re only doing a “one-stage production line,” and don’t mind re-rendering if the image is wrong, there might be no advantage to you to use HDRI files, because they’re not gaining you anything that you need in this workflow. The way that you describe your project, this might well be the case here for you.
Thanks to all for your replies. Sundialsvc4, some of what you say about HDR, I would say about RAW, but I understand the “whiter-than-white” idea and that it is why blender can make sunshine from such an image.
Here’s an experiment comparing an HDR (rather lo res) and a JPG copy. The main differences are the huge size of the HDR and the noise in the HDR. With the JPG, I have to set the sun in the right position, but that only takes about a minute.
Perhaps the HDRs might be important for interiors, but even there, I suspect I could imitate the effect with a jpg and some area lights properly positioned using the same method that I use to align the sun. Note the nice blue-lit shadows in both renders.
if it works for you, great! That’s really all that matters.
That being said, there are some pretty important differences that are difficult to fake. Reflections on anything that isn’t a mirror ball will be crushed pretty significantly. As you mentioned, interior lighting is more time consuming to fake. Placing a sun light manually isn’t much time, but if you are trying to pick out a look that works for your scene, that can be a lot of adjustment.
if you are only ever going to be rendering outdoor images with a single simple sun light, then party on! Again, if it works for you, that’s all that matters.
Try adding fresnel driven sharp reflections to the diffuse sphere, and maybe make it dark grey to easier see the difference - even the clouds would probably be whiter than white. Full diffuse and full glossy is not the way to test. Also, I would never use jpg in combination with filmic, as full white never reaches full white with filmic, it has to be considerably whiter to reach full. HDR’s also support twiddling with the scene exposure, whereas jpgs would just be horrible (sun area would become uniformly greyer rather than narrowing in to a smaller full white spot).
The noise in HDR comes from that a HDR is much harder to sample than a regular light. As a result, when I do use HDR I’m often only using it for secondary ray effects (reflection, refraction), using diffuse only area lights for the rest with a color that mimics what I’m getting from preview the HDR.
Shooting and storing in raw is used to allow for tweaking the exposure and brightness/curve settings in a non destructive way - it captures slightly more data than 8bit, but you can’t compare it to HDRs, clipped or not.
You can turn a hdr into a ldr using a color mix node with clamp enabled.
HDR lighting is only really “important” when you’re trying to mimic measured set lighting. Otherwise it’s just a very fast and convenient way of achieving complex lighting. And of course, when HDRs are good, a good way to test assets to work in all lighting conditions. I’d encourage not using HDRs to light finished scenes, mostly due to the noise, but I have to discourage using jpgs as a “lighting source” as you’re asking for trouble down the line (filmic, exposure, exr rendering, compositing etc).
The test images that you posted above, concerning a instance where hdr lighting has no reason to be used, and where even a simple dome light would suffice, seems to show that you did not take answers into any consideration.
No, it does not show that at all. It shows a clarification of my own question. You do not have a method, other than your imagination, to guess at what I have or have not taken into consideration. That I don’t say something doesn’t mean I am ignoring something.
Are you the official scold in this forum? Do you read minds?
Nope, I just read your question:
“I have read and experimented with HDR equirectangular photos for backround lighting and I’ve come to the conclusion that, being such huge, difficult to produce files, the benefits are not worth the problems. I can achieve the same effects with a good 8bit JPG and a sunlamp or area light and have better control over the final results.
. . . .
… but I do not see any advantage, given the disadvantages of HDR photos, and given the ease of using the sun lamp.
I’d be interested in learning reasons why I might be wrong. Thanks for any clues.”
Thanks JohnW, I can start with an 8 bit JPG and produce a 32 bit RGB openexr in photoshop, but I don’t find the term “float”. Can you give me a clue on that? Is it something I can find in Photoshop? My reading tells me that is where the “whiter than white” / “blacker than black” comes from.
This probably explains everything. No, you can’t. What you can do is start with a 32 bit .exr and produce an 8bit .jpeg. It doesn’t work the other way.
Where did you acquire the hdri used in the render posted earlier?
