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.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
Yeah I hear you but for non organic objects, say buildings ( http://itiseyemeeszark.deviantart.com/art/Saba-a-AlKair-New-547416465 ) that are already in the store we really can't expect, nor should we, (and I am not saying that is what you are saying LOL) that the PAs should redo their maps for PBR. So doing a conversion for hobbyists is the best we can hope for. But yes for new things I would have expected new PBR materials to be provided. I suppose that is what we should expect from a hobby POV.
I also believe DAZ3D is hoping that PAs will make better and more advanced shaders but from past experience how long it takes with no documentation I am not holding my breath.
Personally I do this for fun but I did expect more from an unbiased render engine. And what's with releasing an engine that doesn't do volumetric atmosphere. It seems a norm these days to release things incomplete. Hey it sounds like I am having a moan but I am not really as I love having Iray integrated in my favourite program. I just expected MORE.
oh soory as for that link yeah I have both parts but being dyslexic I need to read it a few more times before it sinks in.
A quick search of this thread on Google did not turn up anything resembling this tip, so if I'm repeating anything, I apologize.
I had one of those "ah ha!" moments this morning, and wondered why I didn't think of this before. Thought I'd pass it on... If you are having issues with DS crashing due to low memory, this might help.
I had a fairly large scene with a lot of items in it, (working with AtmoCam,) and Windows Task Manager showed a decrease in memory usage of over 3 GB.
The down side is you have to reload the file if you want to make changes to the scene and render again. But that's a small price to pay to keep the program from crashing when you're right on the edge of running out of memory.
I hope some of you will find this useful/helpful.
I have a question about something I read, versus what it is actually doing, so wondering if this doc is wrong. It says, "If your scene is too bright or too dark to suit you, you can use one of the settings under Tone Mapping. If you are not a photographer, "Exposure Value" may be the easiest to understand. A lower number is less light, a higher number is more light."
Well, uh uh. When I lower the Exposure Value, the scene gets LIGHTER. (If a lower number is LESS light, the scene would be DARKER.) However, I haven't found a way to lower the Exposure Value without the Shutter Speed going lower too, whether it's by moving a slider, or trying to type it in- the Shutter is impacted when the Exposure Value is changed.
So am I misunderstanding or should it say "A lower number is more light, a higher number is less light." ????
Don't use the exposure value. There's a bug (in the view of several of us: http://www.daz3d.com/forums/discussion/62381/oddness-in-iray-ev-setting) in how D|S calculates the EV when involving the ISO setting. You can get yourself thrown off if you meddle with both the EV and ISO settings in combination.
The article is a bit confusing, and depending on perspectives, is correct or not correct. The lower the EV, the more light that is let into the scene, which makes it brighter. But if you look at it from the perspective of light available to you, you'd use a lower EV if you have less light.
Changing the EV has to alter at least one of the three camera settings that affect light gathering: ISO, shutter speed, or f/stop. In the case of D|S, it's shutter speed. So you might as well just change the shutter speed from the get-go. It works in a logical way: the longer the exposure, the more light. The more light, the brighter the scene.
Thanks Tobor. I'll do that.
Can it be done with a night sky as well
Actually, looks like the doc you mentioned is wrong. In real world, as well in iRay (and other unbiased 3d render engines), Evaluation Number or EV value is calculated from 3 parameters - Shutter speed (film exposure duration), f-stop (aperture) and film ISO (sensitivity to light).
So, to be clear in the beginning, none of the above parameters will affect the lighting conditions of your scene (it will remain the same) but they will change how the virtual camera, or how the "film" or "image sensor" will respond to light, and in return, the rendered image will change, but not the scene itself.
Shutter Speed - is actually duration, representing how long the shutter is open, and it's represented in fractions of a second (1/x s). Simplified, higher shutter speed value will shorten exposure duration, returning darker image and reducing motion blur (in case of animation). To avoid confusion, while you increasing the value from 50 to say 250, you're increasing the actual number, but shortening the exposure (0.02s – 0.004s). Values lower than one will give exposure longer than second (0.5 will be 2s long).
F-Stop – is a value representing aperture, or how open or close it is, controlling how light hits the “film” and how much of it comes through the lens. Low number (F 1.4) lets more light pass inside (aperture is “more open), allowing faster shutter speed (less motion blur) but in return you have shallow DOF, or depth of field. Higher values (F 8, F16 etc.) will “darken the rendered image” because aperture is more closed, and less light hits the “film”, but you get sharper image (deeper DOF). To get proper exposure, you can change Shutter speed (extend it), with possible negative effect of motion blur (in case of moving objects), or change ISO, with possible negative effect in image noise (this is an issue in real photography, with unbiased 3D render engines, it will generally not generate more noise).
ISO – is sensitivity of “film” to light. Standard values are between 100 and 400, but you can go a lot higher to “lighten the image”. This is generally used in low light conditions (in real photography), when you can’t use longer exposure duration. So basically, low values give darker image with less noise, while higher values give lighter image with more noise.
Mix of these 3 parameters will always change the Evaluation Number, but when you change EV in return, it will only change Shutter speed, leaving the f-Stop and ISO unchanged.
I’ve noticed around, that people also increase intensity of lights in the scene unrealistically, to compensate for not changing camera settings, putting lights that emit 10000+ watts of power. Standard incandescent 100W bulb, has efficacy of around 14 lumens/W, radiating around 1380 lm, on 2700K (kelvins) temperature, which is a yellowish-orange light color. Daylight color temperature is around 6500K, which gives “white light”, while higher values give blueish light color.
Newer technology light sources have different consumption/efficacy ratio, like energy saving bulbs and LED lights, using less power, while giving more illumination, but light is a whole other topic.
I can make some image samples if anyone's interested.
Increasing lighting unrealistically is only a problem if you are inconsistent with it and have emitting surfaces.
For example, if you have a robot with glowing eyes, those eyes are going to look too bright/too dim in successive renders if the lights are all over the place with no inherent logic.
I typically use brighter lights than is realistic because renders do better with more light than less, but then try to keep consistent with it -- if you have weak lights and increase exposure, there's a good chance you'll have very slow renders.
@sale_D "I’ve noticed around, that people also increase intensity of lights in the scene unrealistically, to compensate for not changing camera settings, putting lights that emit 10000+ watts of power. Standard incandescent 100W bulb, has efficacy of around 14 lumens/W, radiating around 1380 lm, on 2700K (kelvins) temperature, which is a yellowish-orange light color. Daylight color temperature is around 6500K, which gives “white light”, while higher values give blueish light color."
I have a feeling that Studio's Iray implementation doesn't use these parameters in exactly real world values and defintions. I haven't dome any experiments yet, but I think for instance luminous intensity, luminous flux, and lumens are used diffently on different tabs and occasionally treated as synonyms. For instance, adding a spotlight with a real world lumen value doesn't give the same results that I see in an actual photo studio (from the same distance of course). I may just be using the various params wrong, but I am trying to use them as if I had Studio lighting.
The luminosity values aren't unrealistic; instead, there's a misconception on how they work. For emissive sources, both Watts and Lumens are spread over the ENTIRE surface. If you take a 100 watt light bulb, then make it 100 times larger -- as people often do with emissive geometries for lighting a scene -- it's not going to be very bright. So they crank up the luminosity, and those values appear nutsy. They're not -- it's just math at work.
There are other luminosity units that take into account surface area. They are easier for most people to use, though are not based on simulating real-world lamps. As D|S's default scene unit is centimeter, a good unit is cd/cm^2, which is candles per centimeters squar. The luminosity for any given square centimeter will remain consistent if the geometry is resized.
Area lights only use lumens, so if the area is increased, or a 3D shape is used for the emitter instead, then those values have to get quite high in order to produce realistic and natural light output, even if you leave Tone Mapping at its default, which approximates a cloudy day. Again, it's just math.
Finally, not all of the luminaires used in D|S are measured the same. The distant light, which approximates parallel rays from an infinitely distant source, is measured as light incident on the scene, not light intensity at the source. Noon sun is about 9-12 lumens per square centimeter. Unforunately, the default value in D|S is much, much higher, so people struggle with using this light. Dialing it down to 10 or so solves that.
Really wish distant light allowed some shadow softness analog in Iray, would help alleviate some of the limitations of environmental map.
@Tobor "The luminosity values aren't unrealistic; instead, there's a misconception on how they work"
I think I just got it. I've been looking at these values and comparing renders results to real life for what my EYES see. If I instead think of a camera with the Tone Mapping used, then the render results make more sense.
Sound about right?
I can tell you I've tested a lot of render engines over the years, especially unbiased ones, and iRay for DS will give you approximately the same results in terms of lighting the scene and camera response as other engines, like LuxRender or Maxwell.
I personally dont use generic lights (like dome, spotlight, directional etc.) because they are not physically correct.
If you want to "copy" the real world conditions, forget about those lights and use geometry with emissive surface. You don’t have to place fully rigged lighting setup with stand in the scene for this, you could use a simple panel for most diffuse light sources, like softboxes of any shape and size, diffusers etc.
you can also use HDR maps for this, to crank up the realism. Most of the time, I actually make HDR maps I use for renders.
Tobor is right, saying these values are widely misinterpreted by users.
For instance, you want to use a softbox 20x20 inches to light the subject. In real world, you’d use a softbox fixture with flash or CFL light source. Let’s say you use pro CFL’s for photography, 32W, daylight color, approximately 2000 lumens per bulb (efficacy approximately 62.5 lm/W) and your fixture takes 4 of them. You have approximately 8000 lumens over the emissive surface (glass) of the 4 bulbs. This light hits the front diffuser and goes farther to light the subject softly.
In virtual 3D, you’d probably use just the square polygon for this, and set it’s output to 8000lm, or 128W with 62.5 lm/W efficacy. But there are other values like Lux (lm/m2) Candela (cd), luminance (cd/m2), or in case of Daz Studio – cd/m2, kcd/m2, cd/ft2, cd/cm2, and finally lm and W.
Now, if you pick W, your light source will emit given value of Watts (lumens) whether it is a small bulb, or big light panel, meaning the light source will be brighter in smaller size and give sharper shadows. Other values represented by emission per square surface will stay consistent with emitter size change, meaning the emitter surface will be the same brightness, no matter the size, but when you enlarge the light panel/source, it will introduce more light in the scene.
The key is to keep it all, as much as possible, in realistic sizes and values, if you're going for realistic outcome.
Of course, material and shader creation and concepts are different across render engines, but behind it all, there's math.
...For outdoor scenes I usually use an ISO of 64, Kodachrome 64, which I used to use in photography as it offered nice colour saturation. I actually have to bump the saturation a bit to get the Kodachrome "look" (LuxRender and Reality actually have "real film" brand settings).
As to photometric spotlights, they are still very useful, dependent on the need of the scene (like my recent RRRR submission) as they do create hard shadows which a large primitive with teh emissive shader doesn't.
William, I hear you on shadows and the photometric distant ligh. I like using different sjy backgrounds but am disappointed that the shadows cast by a distant light cannot have the softness adjusted like in 3DL.
As to HDRIs I am rather disappointed in most that I have downloaded as they are either too low res (becoming blurry) and often include ground scenery, thus requiring working in Iray viewport mode for scene setup which on my system is incredible sluggish (I only have a 1GB GPU). Doing so for prolonged sessions eventually causes Daz to crash to the desktop as well. I really would like to see a set of just high resolution HDRI skies (similar to the "sky only" domes from Azure Skies for LDP 2) with an accurate sun at different elevations as I usually use geometry for my settings.
@sale_D "I personally dont use generic lights (like dome, spotlight, directional etc.) because they are not physically correct."
That's incorrect for the Studio implementation. The Iray Photometric lights allow geometry and all the other associated params. You just have to consider them as a unit.
Using your example, Create a spotlight and select it in the Parameters tab. From there select Light.
Select your light geometry, say rectangle for example. Now pick a height and width. These are units of cm (Studio default) so pick 20 for the height and width. Now pick your luminous flux > 8000.
(Here's my point of contention, flux is a unit of "somethings" per unit area. So the 8000 has to be scaled somewhat for the increased surface )
Finally select the color temperature. Your results will approximate photo studio lighting, You have basically just recreated your mesh light, it's just faster rendering.
http://www.cgskies.com/ has both free and paid ones...the free ones generally don't have the resolution needed to serve as a clear/non-fuzzy backdrop AND lighting one, but the paid ones usually are good enough.
I stand (partly) corrected. “Generic lights” or Studio’s Distant, Spot, Point and Linear Pint Lights are somewhat connected and used with iRay. However, they appear to be internally converted to emissive geometry for rendering with iRay.
Photometric term if often misused. Photometry is the science of the measurement of light, in terms of its perceived brightness to the human eye. Term Photometric Light is most often used in 3D software packages to refer to IES files. These IES Light files are a specific type of light that uses actual light energy and pattern descriptions from manufacturers of actual light systems. You can surely use them with iRay, but only Spotlight generic light can load them. Of course, geometry lights can also do that.
With above out of the way, generic DS lights are converted at render time for iRay, so it can “understand” what they are. You do have additional parameters for default lights in DS 4.8 for iRay, and they have dedicated size and values. You could change those, as well as emissive “geometry” of them, with options like point, rectangle, disc, sphere and cylinder. The small problem is, for some of them, you will get what you expect, but for some, you just won’t While they might be easier to use, with real geometry, you have more options.
My comment about these light’s being physically incorrect was related to emission parameters in percentage (100%) as they always were until now. Also for all of them you have Lumens and temperature values, for iRay, but you can’t switch to Watts, or any other value.
Notice on the attached image how first and third row of renders are similar but not the same, when they should be.
Luminous flux is not is a unit of "somethings" per unit area but unit of energy over time.
“Luminous Flux (Φv) is energy per unit time (dQ/dt) that is radiated from a source over visible wavelengths. More specifically, it is energy radiated over wavelengths sensitive to the human eye, from about 330 nm to 780 nm. Thus, luminous flux is a weighted average of the Radiant Flux in the visible spectrum.” Source – HyperPhysics
Yes, pretty much. Folks get too tied up with the technology, and forget this is just another expression of art. Art is what you intend it to be.
In my mind, PBRs really aren't about realism per se, they're about applying physically-based algorithms to achieve predictable results. Those results are often (but not guaranteed to be) similar to what we'd expect in real-life. That cuts down guesswork. But "realism" is not a requirement in the end result. Depending on the objectives, it doesn't really matter if you've used a "50 watt light bulb," and that Iray is producing the correct watts into the scene. What matters is that the scene is properly lit for your objective.
The problem is that the D|S stage is, in itself unreal, so before we even begin it's all fake. There is no natural place on earth that is devoid of materials that neither reflect nor absorb light, and these properties also contribute to the quality of light. So just rendering without a "proper" sky, ground, or atmosphere is already starting out as "unrealistic." To compensate, Iray provides a virtual sky and ground, and fakes haze in the atmosphere. These shortcuts are necessary or else renders would take forever. At some point art has to be practical.
The Tone Mapping controls let us bracket the exposure to a small subset of what your eyes could see if the scene were real. That alone is fakery, in the same way even the best film can't capture all the dynamic range our eyes can discern. At best we are simulating a *slice* of real life, and choosing what that slice is. (This, in essence, is the definition of art.) Personally, I couldn't care if a light is emitting 250 watts with this or that efficacy. Our brain register the scene as "realistic" if the elements are plausible -- these days, a car light could be incandescent or halogen or LED, and our brain accepts them all, despite that effiacy and light power is different for all of them. Even in real life, these have vastly different outputs for the same function. Is one more "real" than another as a car headlight?
What our brains DO register as fake are things like lack of correct shadows (or no shadows when there must be some), or physical attributes that should be there -- a roughness to a pair of jeans or a gleam to a bar of gold -- but aren't. Fortunately, this is where PBRs like Iray excel.
What D|S did was cleverly apply the existing "spotlight" and "point light" types to connect with their Iray primitive equivalents. These light types are actually internally defined within Iray, and are algorithmically produced. Iray internally supports point, infinite, area, and emissive geometry as light sources.
Emissive geometry is handled by a specific set of MDL functions, and some of these functions are actually not supported in certain Iray rendering modes (e.g. Interactive). The geometry appears to have illumination, but does not produce scene lighting. All of the primitive light types are supported in all rendering modes, though in Interactive there are limits to the number of such lights you can have in the scene.
Lights based on geometries tend to slow down Iray, especially for those without CUDA acceleration, so while these are certainly useful as lighting sources, the caveat about slower renders should always be mentioned. Iray must calculate the exitance for each triangle on the geometry, and on large or compound shapes, that can be a lot of facets, with the light going in every which direction. The average D|S user doesn't have a 1500+ core GPU to cut complex scenes down to size -- I myself am still using a PC with only 500 CUDA cores, so efficiency in scene setup is a primary concern for me.
One way around it...or at least to lower the impact. Use the smallest number of polys. That would be a plane with no divisions (that ends up as 2 triangles).
...yeah, but 30$+ a pop for a single HDRI is pretty bloody steep for us hobbyists. I already wrote DreamLight about this.
Again this is why I feel we need to have an adjustment for shadow softeness (like in 3DL) for the Photometric Distant Light.
...yeah, a scene that I created which used 8 mesh lights took over three hours longer to render (at a smaller size) than one that used a cloudy day HDRI, wet surfaces, and volumetric fog. Both scenes had 8 Genesis/G2 characters.
Did any of you already know a trick that will make Iray start to render faster at the beginning.
If you just want to make little adjustment like positioning light, making change to material, pose, etc. Render for the 1st time, then stop the render midway, leave the render window open.
Make your changes, then start to render again by hitting Render / CTRL R. (Not resuming the 1st render)
If you are planning to make another change to your scene, close the 2nd render window, leave the 1st on open. Rinse and repeat.
The downside is, you'll need lot of RAM and VRAM to do this, since we don't close the 1st render window and leave it there. Instead of making Iray preparing temp file for render every time you close and start new render window. It won't make your RAM usage double up.
It's like telling Iray "Hey look, i'm not discarding what you've build for rendering scene, use that instead rebuilding them everytime I want to render"
Use this method at your own risk since it will stress out your hardware.
I have found the same
Thank you for that tip!
Here's a tip for Iray. Forget about making any animations. With render times often stretching into several hours for a single frame, or unless you're wanting the animation the size of a postage stamp, Iray is not for you.
not true
i did some animations with a render time of 12 seconds a frame
this is esspecially handy for image series with gpu usage
render one frame/ let it stay
then start
saves a lot of time
Animation frames have different priorities than single still images — it doesn't matter so much if each render is perfect, the sequence of frames blur together so imperfections in any one frame are less visible. This means you can adjust the Render Convergence limit downwards and each frame will be finished faster. This is especially true if there's a lot of fast motion, the motion blur counteracts the speckles of a low-convergence render.