Iray emmissive lights
kaotkbliss
Posts: 2,914
in The Commons
I was just wondering if there was a difference in render speed with the different Luminance Units
Daz 3D is part of 
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
Yes, but you'll blowout and flatten your render's surfaces. In the Render Settings, under Tone Mappings at the top is Exposure Value or EV and personally I feel that gets renders done faster withough as much as a bright blow-out falttening effect. Try decreasing in increments of 1 from the default of 13.0 for brighter pictures and decreasing by 1 for darker pictures.
I'll ask a probably dumb question. We all know that emmisive lighting can be painfully slow to render, so how come Iray skydomes, (aren't they just an emmisive lightsource?), render so quickly.
I'm revisiting a comic I had started and in order to keep area consistency throughout the whole series (however long that lasts) I'm laying out the entire city. All the signs and street lights and such are all emissive lights and I know my renders are going to be slow (I've already done some test renders and adjusted the lights for appropriate brightness) Currently I'm using the kcd/m2 Unit but I'm wondering if changing it to something like W or lm with a higher value would actually render a little faster.
On a side note, currently the scene is using 1.6 GB in geometry and objects/buildings are only used 1 time, the rest is instanced. Next I'm going to be working on reducing texture sizes (texture atlas doesn't seem to work well with the urban future 2 building :(
For reducing texture sizes, I highly recomend https://www.daz3d.com/scene-optimizer. It gives you a lot of control over what gets reduced and by how much. I have a bad habit of blowing through the memory on my 6gb 1070, and I've successfully used Scene Optimizer several times to reduce things enough to keep the card in play. And honestly, I haven't really been able to see any visual difference on the ones I've used it on, even rendering at 6827×3840. Plenty of difference in render time, nut no noticable detail loss on the ones I've used it on. But I've also only used it on large, wide angle scenes, so the fine details in skin and such wouldn't show anyway. I've never tried it on closeup portraits however. But you can set it to only reduce what you want it to, so it's possible to only do the background if you are doing close ups and are worried about losing detail (that was my biggest concern with getting Optimizer, but so far I've been very happy with it)
I've come to terms that I'm going to be doing CPU renders with this project since I've only got a 4GB card But I still want to speed it up as much as I can and I think the biggest area is the emissive lighting. And if there's only a fraction of a difference between the unit types, multiply that by the actual number of glowies and it could really add up.
The units don't matter as long as you keep the overall lighting value the same. That is 100 mph isn't the same speed as 100 kph. 100 lm/cm2 isn't the same light level as 100 lm/m2
@fred9803 "so how come Iray skydomes, (aren't they just an emmisive lightsource"
Iray Skydome is actually a misnomer. It's been used because people were familiar with 3DL Skydomes. Iray skys/lights are a HDR file that may have an associated jpg image. There is no geometry associated with an Iray sky/HDR/Environment light. It's the geometry/verticies that make emissives slower. A primitive plane
Emmissive lights that are just a single polygon are actually pretty fast to render, which is why most of the lights in the ghost light sets are single poly.
If the lamps of the street lights are high poly then making them emmissive will result in slow render times. You are better off using a single near invisible plane as your light source which is positioned just below the street light. That is how the strip light works in the ghost lights 2 kit, and it looks pretty good and renders very fast.
Thank you both FastBike and Havos, looks like I'll be making some primitives and removing all the current emissives.
Would it be better to remove emissive lights where possible and use a point light instead.
For example a street light, replace the texture of the light with a glass shader, then place a point light inside...
The thing about emission is that it's based on polygons.
So a 1 polygon plane is going to have a very low impact, while trying to make the entire skin of a High Resolution Subd4 Genesis 8 figure glow is going to crush your computer.
Low polygon emission lights aren't going to be noticably slower than spotlights with geometry, go for it.
Just be aware that if you are converting a room, that light sources might be attached to large sections; if that section is given High Resolution of several subd levels, so will the lights that are part of it.
As someone who has always CPU rendered with DAZ Studio if you have gotten used to the speed of GPU rendering with small scene ain't nothing you can do to CPU render at that speed. Accept that.
Now what you could do is divide up your big scene so that it can render on your GPU and composit all the renders together. All you have to go is create different groups in your scene and assign different oobject to the different groups and click the 'eye icon' on the group you don't want to render to hide that group and then render. Repeat for each group and composit together.
Because they're not Emissive lights, they're Environment lights. It's very easy to set up an Environment light for a fast render; it's also (much too) easy to set up an Emissive light for a slow-as-molasses render.
Why don't lights parented to an object instance with that object?
I have a light pole parented to a street section. When I instance that street section, the light pole is instanced with it.
But if I put a point light or spotlight on the pole and instance the pole, only the pole is instanced even though you can instance lights...
To answer this specifically: no, as long as the amount of light stays the same. As Fastbike noted, you have to alter the luminance value to correspond to the units. I recommend cd/cm^2, which is candles per centimeters square. Firstly, centimeters is the defaults units of measure in D|S. Secondly, the apparent brightness of the emissive object remains the same as you enlarge or reduce it, since the unit is based on area.
The render can be faster if more and brighter light is added. This has to do with avoiding areas of indirect-only light. That kind of lighting takes the longest to render. The more ray paths into an area, the more samples there are; the more samples, the faster the render. It takes hundreds, or even thousands, of repetitive samples to each pixel to render the scene.
I'm pretty sure that changing the Luminance Units setting for Emmissive surfaces will NOT -- per se -- require any longer or shorter Render Times to achieve a given targeted "quality" in the resutling image.
First, some background information -- which does not seem to be explained very well at all in the Daz documentation -- that makes it easier to figure out the answers to questions like the one the Original Poster (OP) asked ...
You'll notice in the Emissions section of the Surfaces | Edit tab that the parameters listed there include both:
By analogy, it may help to think of these in terms of "How much milk do you want" ... say, 12 Units? But 12 of what? Pints or gallons?
With gallons, you get a lot more total milk than you do with pints.
The "trick" to choosing such settings may be recognizing that there are two different kinds of Luminance Unit listed to choose from, but they're not really labeled very obviously as being in two categories:
The Units of this type include:
The conceptual formula is: Units of Area x Amount Per Unit = Total Output.
So two emissive surfaces of the same overall size (e.g., 2x10 = 20 and 4x5 = 20) will put out different total amounts of light only if their Luminance settings are different (e.g., 2x10x100 = 2,000 v. 4x5x250 = 5,000)
With this kind of Unit, the amount of light output per unit of area for all options offered is cd - for Candela, which is just another (very old - but still very useful) way of measuring "X" amount of light based upon how strong it is compared to a "standard" candle. .
The Units of this type include:
So -- assuming you haven't pulled a brain cramp yet ... :-) -- you can see that which Luminance Unit you choose for a given emissive surface can easily change the amount of light by a factor of 10 to 100,000 !!! So, yeah, which option you pick generally will make a HUGE difference in how "bright" a scene is.
Now ... to your specific situation (to the extent that I undersand it correctly) and the question of Emission settings v. Render Time ...
Iray "likes a lot light" when it comes to Render Time -- which is to say, it takes the Render Engine more Iterations to calculate the proper (and firely-/noise-free) appearance of a surface if the light hitting that surface is "weak" rather than "strong" (largely because it has to calculate more "bounces" of light rays to "figure out" how that under-lit part of the surface should look).
But "jacking up" the light settings on an emissive surface may well cause more problems than it solves, since -- as others have pointed out -- when you jack up the light on part of an object (or part of a scene), any parts that were already at the top of their exposure range also get "jacked up" and you lose their color and detail, etc. So now you have "dial down" other lights to keep various parts of the object/scene "balanced", and now those parts may have dark, Render Time problems.
I'm not as confident about this next point, but ... Turning up the Exposure Value (EV) setting in the Environment |Tone Mapping tab to offset any "jacking up" of emissive-surface settings probably will NOT result in materially lower Render Times for a given "quality" of final image. I'm not sure whether the Render Engine "sees" the light hitting surfaces before any changes EV settings) or it "sees" the light after it's "squelched" by a higher EV setting. (Keep in mind that increasing the EV number reduces the apparent light-level/exposure in the rendered image, while decreasing EV makes things brighter.)
You should be able to "clock" the relative difference, if any, with a simple scene of one emissive surface and one primitive cone or cube, etc. In either case, "jacking up" the light output of emissive surfaces should still leave you with the:"imbalance" problem across parts of the object/scene.
I'm pretty sure that the comments by others are correct that the number of polygons in the emissive surface can have a relatively LARGE impact on whatever portion of total Render Time is related to tracing light from your emissive source(s). I haven't "clocked" that specifically in any formal way, but I've had lots of "rough feel" experience that says this is the case -- enough so that I almost always use a one-face primitive plain as the basis for my emissive lights. Even if you need to place emissive light in multiple places that can't practically be covered by a single plain, your total "render load" from quite a few single-faced emissive plains is very likely to be less than many other options.
Now let me reach back almost 40 years to my movie/tv lighting days .... with maybe a few modern-technology adaptations availalbe in a virtual world ...
How to light a large, sprawling (and perhaps especially outdoor) scene/environment -- and keep shots made in various parts of it from various angles consistently exposed?
My guess is ... You'd be best off by lighting the overall "town" with a Dome Light of some sort -- probably an HDRI version of good quality, possibly from a set of multiple versions with a coordinated "look" for different times of day/night. From a lighting perspective (as opposed to what "sky" you mgiht actually see in the renders), your objective in this strategy is to use the Dome to "lay down" a consistent level of "base illumination" across the "whole town" (or at least the exterior portions of it. To help assure consistent exposure, etc., the level of "whole environment" lighting probably should NOT be changed much (if at al) from shot to shot.
By "base illumination", I mean :"just enough" to lift "the deep shadows" -- at least in the "wider open" areas of the environment -- to the point that you just start to recognize decent-sized shapes, etc.
THEN ... find those sections in the environment that both "matter" to your shots and are still much darker than the "open areas" (because the buildings or other objects shield them from the Dome, etc.) ... and place (probably pretty big) single-poly emissive plains in those key, shielded-darker areas. In essence, you're trying to "even out the darker parts" by raising their light levels up to the overall Base illumination level.
"Hanging" such emissive plains "high up" and/or "far away laterally" may often be needed for various reasons. And because light from such sources (like real light) falls off with "the square of the distance" between light source and subject, that can often mean REALLY STRONG light settings. (For example, DOUBLING the distance between source and subject cuts the intensity of the light hitting the subject to 50% (1 EV Step); and DOUBLING the distance again cuts light at the subject to 25% (2 EV Steps).
So you might find the "stronger" Luminance Unit options of cd / square centimeter or kcd / meter a more workable standard.
Keep in mind that Surface settings related to Refraction or Cutout can be used to make emissive lights "invisible" to your cameras while still providing illumination.
In both of the previous steps, it can be very useful to use what I call "exposure poles". These can be primitive cylinders -- of "whatever" usefull diameters -- that all share the same "material", which should be 10 "stripes" or :"bands" ranging in even steps from Black (0,0,0) to White (255,255,255). If you get from Black to White and still have more pole to cover, just repeat the series of stripes/bands again. (If you set the pattern up as a "tile", you can easily apply it to most any object regardless of size or shape),
What you're really doing here is creating a 3D version of a photographers "grayscale card" -- but one that can be (virtually) "three stories high" if you need it.
Just place a "pole" anywhere you'll need a "light reading" of the environment. Scatter the all over the place (probably as instances) if you need to, especially since you'll be turning them off at render time, and they're so geometrically simple that they eat up almost no memory.
They can't give you a direct "light reading" like a true light meter can. But by doing a quick "spot render" on a pole after making a lighting adjustment, you can -- at the very least -- see readily if characters or other objects placed in that space will "disappear in the mud" or "get blown out in the highlights", and with a useable quantification of "by how much" in whichever light/dark direction. (Just count how many bands/steps of exposure either "disappear" or "blow out". That tells you roughly how much you need to dial the light up or down, respectively. Then run another spot render ... and "rinse and repeat as necessary".)
Any two parts of the environment where you see (or can't see) the same bands/steps of the pole are almost surely lit to effectively the same levels. So keeping the exposures for different shots consistent when they're made in different parts of the environment should become a lot easier.
Now throw in your light sources for creating the story-telling "ambience" of your shots -- street lamps, lights from store/office windows, etc. You can use the exposure poles here, too, both for shot-specific exposure and to stay consistent with shots made in other parts of the environement. Most of these lights may well be of the traditional Parametric type (spotlights, etc.), rather than emissives.
Finally ... and for each shot separately ... add / subtract / adjust various light sources to make that shot "sing". But if you keep an eye the visible v. "lost" bands on the exposure pole(s) you set up for each shot, you should be able to achieve workable (and for sure "recoverable) consistency in exposures for all of your shots.
And because you've got a Base Level of Illimination throughout the environment, few portions of it should create a "drag" on Render Times while Iray grinds through more Iterations to render them acceptably.
Hope all this helps.
Been meaning to write up most of this for our internal purposes for a while anyway. So it seemed like covering it somewhat rigorously while I wait for some test render to "cook" made sense.
Thank you for the useful read. All I pretty much knew is: The more light the better and /m2 mesh lights scaled with size.
I've not used any HDRI lights before (And I don't think I've downloaded any) but it sounds like I really should. I had planned on just using Iray's natural sun for lighting, then adjusting the iso as needed.
I consider the Sun & Sky option in the Render Settings | Environment tab to be "dome-ish" for these purposes -- in effect, if not exactly underlying "technology". That could work well, too, for widespread Base Illumination purposes.
Good look with your project. Hope you'll share some images from it when they're ready!
In case you (or other readers) are not that familiar with conventional photography ...
Exposure Value relates to how much light is present in a scene. ISO relates to how "sensitive" the film (or image sensor) is to light -- higher ISO meaning you need less light to get a proper exposure, and lower ISO meaning you more light.
Somewhat obviously, I guess, "proper exposure" requires balancing out these two, which is done through a combination of Shutter Speed (how long the available light is allowed to "hit" the film/sensor) and/or Lens f-stop (how wide the aperture behind the lens is, which determines how much of the available light is allowed to "get through" to the film/sensor -- sort of like the pupil of your eye).
So the conceptual "formula" is:
(Amount of Light Hitting Subject) x (ISO) x (Shutter Speed) x (f-stop) = (Perceived Exposure)
This explains why -- if you change any of the related settings for one of these four factors in the Environment | Tone Mapping tab -- one or more of the other settings will change automatically.
Keep in mind that the labeling of f-stops -- for example "f/22" -- can be a bit misleading. You might think that "22" would mean more light gets to the film/sensor than "11", but the opposite is true. This is because the actual number is 1 divided by 22 (1/22), to indicate that the aperture (like a "pupil") of the lens is "stopped down" to 1/22 of whatever its widest-open configuration can be, given the dimensions and other design factors of the camera and lens. So, as with any number expressed as a fraction, the bigger the denominator (bottom number) in the fraction is, the smaller the value of the overall fraction. It may help to think of it this way: When the sun is very bright, the pupil in your eye "squints" (it gets smaller) so that less light can get through to the photosensors in the back of your eye (the film, so to speak) -- so that the "exposure" remains balanced. In a dark setting, just the oppposite happens.
POTENTIAL KEY TIME/FRUSTRATION SAVER:
(I've found that -- if you don't more-or-less "get" the stuff covered below -- you can waste a lot of time fiddling with various settings for rendering image from 3D models.)
There is also -- in the Parameters tab for any Camera object you add to your scene -- an f-stop setting (available in the Camera section of the listed Parameters). THIS f-stop setting does NOT affect "exposure". This Parameter affects only the Depth of Field (DOF) in your rendered image. DOF is the name for how "deep" the zone of "acceptable focus" around the point of "exact" focus will be in an image -- in other words, it determines which parts of the scene will appear "sharp" and which will be blurred.
For a given straight, line-of-sight distance between the Camera and the point of exact focus (which usually is set to match the distance to the object of greatest interest in the scene) ... the "higher" the f-stop (smaller aperture), the broader the DOF; and the "lower" the f-stop (bigger aperture), the narrower the DOF.
The amount of light present has nothing directly to do with DOF, since DOF is a function of how rays of light reflected from objects at various distances away from the camera bend around differently as they pass through a lens -- regardless of whether those rays are "strong" or "weak". In conventional photography (film or digital), however, the amount of light present can indirectly drive exposure settings since -- for example -- low-light conditions can force you to "open up" the aperture (make if wider -- a lower f-stop setting) so that enough light gets through to the film/sensor to create a good exposure. It works just the opposite way with bright light. When you make those adjustments in the physical world, DOF changes inherently. In the virtual/3D world, however, you get to "break" these indirect "mechanical connections" between DOF and exposure "with impunity"-- which, for creative purposes, can be pretty dang cool.
As a result of having these two "independent" controls over f-stop (aperture width) -- i.e., one in Tone Mapping and the other in Camera Parameters -- you can arrange just about any DOF regardless of how much (or little) light is in your scene, whether that level of light is for creative or render-technical reasons, etc.
Thank you and will do, although I suspect it won't be for a while yet.
I've got all the buildings for the city generically placed and now I'm adding in the roads (which means adjusting building positionsas I go) and adding in some of the details like bridges and walkways.
After that I need to finish adjusting all the lights and glowies on the buildings to the correct look. At which point I can start setting up the scenes and adding little details as I go (such as garbage cans and what-nots)
*spoiler alert!*
The setting of the story is: It's a bit into the future and revelations (from the bible) has already happened but no one realizes it. The demons who have taken over earth are hidden in human form and are actually a kind of alien life form that goes from planet to planet feeding off the souls of the wicked until there is nothing left. (The 'dirtier' the soul, the more appatizing it is for the creatures) so it's their goal to corrupt as many people and as much as possible before feeding and moving on. The uncorruptable souls have all been released by the aliens (aka-the saved being taken to heaven) so as to not interfere but are gathered, reformed and trained by another alien race that visits worlds and tries to warn them before the demons arrive (where our current religion comes from) so that they might fight against the demons. The problem is, the process leaves the warriors without memory and they must discover who and what they are before they are discovered and killed by the demons.
The idea of "exposure poles" is fantastic! I'll need to start doing that.
I agree, I never even thought of something like that. It's always annoying when you get your scene set up, start the render and something you wanted people to be able to see suddenly has a shadow or not very well lit.
After thinking about my story setting description above, I probably should have noted that no, this is NOT going to be a porn comic LOL. While there may be some nudity, blood and lots of swearing I'm keeping that other stuff out of it.