IRay: Light source parameters VS Tone Mapping (Camera parameters)
Hello,
I know that there are several ways to influence the lighting in your pictures, and two of those are;
- Directly at the light source; in case of emmissive surface, playing around with the light parameters such as light temperature, IES, and Luminance;
- And the Tone mapping, playing with ISO, F/Stop and Shutter speed.
My question is if there are advantages (quality, render speed) to do one over the other, or the two simultaneously, or if they are pretty much redundant?
For example, in an indoor scene, using a emissive light bulb surface, and I want to add more light to the room. Should I raise the luminance or should I raise ISO? What are the pros and cons of one or the other?
Using DAZ3D.
Thanks
Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
Changing tone mapping does not in any way influence the actual light in the scene, any more than changing the f/stop or shutter speed on a real camera will boost the light around you. Both will alter the appearance of the recorded image, but within reasonable limits, for the purposes of render speed, it's always good to use higher values of luminance. Much of what the tone mapper does comes after the scene is rendered internally. You help Iray by giving it the light it needs to begin with.
As noted in many threads here, tone mapping has no effect on the real-world effects of ISO, f/stop, or shutter speed. There are merely conveniences for setting the overal exposure value.
Until such time as I get a video card that can handle the rigours of tone mapping, I have chosen to provide brighter lights, turn off tonemapping all together, render to canvasses and do my tonemapping in PhotoShop. Postwork=blasphemy. Well, I was going to hell anyway,
Thanks Tobor,
I understand that Tone Maping has nothing to do with light. I imagine this as in real life, tone maping emulates the "camera". It's jsut passive, standing there, and receiving the information of the world it is pointed at. That information are the calculations of the computer and mesh light parameters and surfaces parameters are what is important for those calculs. Tone maping just receives the calculation results and filters those results like a real camera would to. Let's say it's an instanteneous Photoshop job after the computer figured out the whole parameters.
I learned from your answer that Tone Maping has no influence on rendering speed. That's one good thing I learned. How about rendering quality? How about those grainy surfaces we get in dim light scenes?
I asked all this because without any further knowledge on my side, I think the 2 systems to be redundant.
For any given timeframe, render quality is based on the amount of light in the scene. In general, the more light, the more ray paths that Iray has to work with, so the faster the pixels are converged. (Convergence is when multiple samples to the same pixel are compared against one another, and over time, their values are deemed substantially similar, and therefore that pixel is "done.")
Iray's tone mapper acts both before and after the render, so some values do have an influence in the rendering, but for the most part, the bulk of what you see as the exposure is after the rays have rendered. With this in mind, overlight the scene like you're shooting an old Technicolor movie. If the scene is too bright, use tone mapping to bring it to the proper levels. This will (usually*) yield a faster render, which is what you want. The faster it is, the quicker you can achieve the quality you want.
What some people tend to suggest is that if the scene is too dark from the start, then adjust the tone mapping. That's not going to help Iray's rendering engine any. Boost the light, either by adding more lights, if needed, or increasing the lumens of those lights you already have.
* Obviously, at some point, adding more light isn't going to solve every issue. Some kinds of troublesome scenes, like those were the primary light sources are indirect, remain difficult to calculate for Iray in any case.
Erm... that doesn't follow. An Iray render is an Iray render, whether or not you adjust the Tone Mapping — doing that has no effect at all on how much work your computer's doing to run the render. It only affects what the finished render looks like, and your computer doesn't care about that.
As SpottedKitty says, this may not be giving you the performance benefit you think it does. But if you do use canvases (there are many benefits to it, the least of which is image control without having to re-render), and you turn the internal tone mapper off, do remember to play with the Nominal Luminance control. This provides a hint to properly render the scene. When the nominal luminance value is set to 0 (default), Iray will estimate the nominal luminance value from the tone mapper settings. But if you turn the tone mapper off in favor of tone mapping externally, nVidia advises to provide a nominal luminance to compensate.
Great advice Tobor! Thank you.
That "graininess' you are experiencing has little, if anything, to do with the lighting; this is mostly a function of when Iray deems the render to be 'complete'. The controls in the 'Progressive Rendering' node are used to adjuest that, especially the 'Completion' group.
In my renders, if I encounter unacceptable 'grain', I bump up the 'Max Samples' value from its default of 5000 to its maximum setting of 15,000, and then set the Max Time parameter to some utterly insane number, like, oh, 135,000 and change. (Just click and drag; more is better...!)
Of course, this is going to affect the render time but, usually, patience is a virtue....
For dark scenes this is simply covering up the basic problem, which is insufficient rays to trace. Iray is a production renderer aimed at commercial use, and it's made to solve these problems without having to always throw more time them. Fix the underlying issue, and the render is likely to take less time. Patience is a virtue, but spending time when you don't need to is not. The various rendering settings are there to compensate and improve, but they always work best when the scene is tuned to begin with.
So – and I have heard more than one story on this – it sounds like the tone mapping settings are very close in behavior to the ISO setting in a digital camera (or, for that matter, a film camera). You can turn the ISO setting to some very large number, but in a very low light setting, that results in noise, random pixels that read lighter than they should. In a film camera, high-ISO films are grainier, for a similar reason. More light responsiveness with the same kind of sensing surface has to be paid for in some way. As with cameras, then, the only way to get a clean image is to add more light.
Or am I still misunderstanding?
Pretty much...
More light (amount) generally means cleaner, quicker, more efficient renders.
As Mjc says, similarnet effect, but they come about it in very different ways.
In the Iray implementation we have in D|S, and unlike some other unbiased renderers, from the Tone Mapping settings there is no artificial film grain added with higher ISO values, no camera blur produced by slow shutters, and no depth of field effects with f/stop. All these three do is vary the exposure value in proper reciprocal fashion. (Actually, in 4.8 at least, there was a bug where EV went the wrong way when adjusting the ISO value. The bug was reported by many, but I don't know if they fixed it in 4.9.)
The technical programmer docs for Iray talk about how the tone mapper is both pre- and post-lens, meaning it probably has influence both before and after the render. From experimentation, I can see the tone mapper has some general influence in setting nominal luminance for the render, but the bulk of what it does is very much like the various adjustment layer controls in Photoshop, such as Exposure and Curves. Just like in Photoshop, if you have a crappy image to begin with, you're not going to get great results simply by tweaking adjustment layers. So with Iray, you get the best (and generally fastest) results when the "exposure" is correct to begin with. This means a scene that has enough light "photons" for Iray to do its job. Without as many rays to trace, it just follows that pixel convergence is going to be slow.
All that said, some things like indirect light are just difficult to do. Fortunately, in addition to the architectural filter, which is designed as an indirect light hint, we now have light/sky portals, which provided added improvements. There's no reason not to use all the tools at our disposal to improve render quality and speed. Simply having it render longer is not a valid long-term solution.
How would one know an appropriate value to use for nominal luminance once turning off Tone Mapping?