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
Great work, Elvis! I like them both. The fall off of light towards the left looks great in the first one, but the warmer colours in the second one are also very nice.
Horo and Slepalex thanks for sharing your experiments with us.
Rashad - Easter Egg
Electro-Elvis - both your renders are well lit. I like the second one more
Electro-Elvis, very good lighting. The left picture looks more convincing to me. I think that you should apply an inverse gradient in order to avoid these light spots from point sources.
Long post.......
Lots of very good things happening here. Thanks grandly to all of you for sharing your insights on this intense lighting challenge. Bravo!!!!!! I'm very pleased with what you all are coming up with.
So as I consider the approaches that each of you have taken over the past couple of pages, I can see some areas of thought that are similar between several of you. Its a great opportunity for me to demonstrate how my own thinking varies from most Bryce users, and the reasons why. The primary purpose of this thread is to hopefully alter and change for the better the way we Bryce users evaluate our work, especially in how it compares to other render engines, to see issues we might have previously overlooked in our Bryce-centric thought processes, and to develop a sharper sense for what makes a render appear realistic and for what diminishes the realism regardless of rendering engine. I want to show you how to see what users of more advanced rendering engines see when they look at our Bryce works, to illustrate the areas we might need to improve upon to remain competitive. Working primarily with unbiased engines nowaways, I've become very sensitive to shadows and other aspects of rendering that I was not so sensitive to in previous years. Its really easy to overdo the shading.
It's important to note that I am EXTREMELY technical in everything I do. I cannot help you to become a better artist, but I can help you to become a sharper technician. All my notes today are accuracy based, not beauty. I can find a scene fully beautiful that has many flaws in its accuracy. It all depends on the content. So my advice to readers is to only continue reading if you have an undeniable desire for realistic results. Speed has no bearing here. And simplicity is only useful if it works. Crazy though it sounds, I consider the original study to be quite simplistic. So will you once I've broken it down.
For my tastes, the thing that I feel each of you has missed in your admittedly abreviated studies compared to mine as the original artist, is that each of you might be skipping to the "results" of the indirect a little too quickly, and not giving enough consideration to "processes." I'll explain
All of these studies from Horo, ElectroElvis, SlepAlex have good aspects. I see no point in covering those again. You all "get it." So to save time I will try to address what I consider to be possible shortfalls in the approaches I've seen so far. Please do not hesitate to help me better understand if I have in any way misunderstood any of the notes provided so far. I will also try to use specific considerations that each of you has brought up, to better tie this all in. Thanks for your understanding.
First question you all seem to have....Why so many Domes and 3D Fills? Its because of a bug. I will in a subsequent post explain the "centering" bug that creates a slight directional bias to clustered light forms such as Domes and 3d Fill lights even when no bias has been purposefully introduced by the user. This centering bug makes it necessary to use at least two Domes or 3D Fills rotated 180 degrees from one another to correct the poor centering issue and to get a more balanced spread of light in a scene. But more on that later. For now, I want to concentrate solely on lighting theory.
Lighting tells its own Story apart from the story of the scene itself/ Make sure you're telling a believable story with your Light
Do one thing at a time. Never Multitask. Light should have a beginning middle and end. Always conserve energy
This scenario has two primary light influences. 1. Skylight (blue). 2. Overhead lights (orange). This means that technically there are two distinct lighting challenges in this scenario. To approach a truly realistic result, you need to handle each of them individually as I have done. I'll explain. Below I have detailed my EXACT process for the original file, and how I came up with a result that I feel is more realistic than the other offerings so far. I might be wrong of course, which hopefully someone will point out. But here it goes anyhow.....
Image 0: This is a photorealistic unbiased render that is very similar to our current study. Please review as it will help to qualify the rest of this lengthy post.
Image 1: Always model your lights accurately. This already rules out the Obscure Light approach as realistic, since nothing about its implementation relates in any way to real world physics. Focus first on modeling the incoming skylight accurately. That means looking at the window sill in your bedroom and noticing the way the light pools in the areas closest to the window. You will notice that the light along the ceiling and side walls of a room are bright near the window and get dimmer with distance from the window. But the point here is that the window sill itself still remains far brighter than any other surface in the room... that's how we as observers know that this window sill is providing all the thermodynamic energy to fuel the scenario. Also you will observe, that the wall opposite the windows is actually brighter than the corners where that same opposing wall meets the side walls. Why is this so considering how far away it is from the window? This is because the incoming skylight rays travel a shorter distance straight ahead than they would if traveling off to one side or the other, making the corners darker. For this reason, I do not believe Slepalex's choice to place part of the cube dome within the far wall is as plausible as it could be. We actually need that wall to bounce some of the light it receives back into the scene, so we need the dome to remain inside.
Image 2: Once the basic skylight has been introduced, I then use a twin pair of Dome Lights to represent the assumed "bouncing" of that blue light off the walls and into the remainder of the room. From the examples provided I suspect you will agree that this is a fairly plausible and conservative representation of the scene if it had been lit only by indirect blue tinted skylight with no overhead lights. I think that because many of you have not clearly perceived the influence of the overhead lights in the original render, you were expecting a dim result similar to this one. But alas, skylight is only half the challenge here.
Image 3: Now it is time to turn off all the Skylight stuff, and to now illuminate the room as if the only light input was from the overhead lights. Like a night time scene. Not so pretty, but this is what it looks like.
Image 4: Now that I have the direct overhead lighting in place, I need to indirectly "bounce" that warm colored light around the rest of the room. In this case I have used a twin pair of 3d Fills.
Question: Why would I use both Dome Lights and 3D Fill Lights in the same interior? Wouldn't one of them do the job?
Answer: Short answer is to say just trust me, I know this stuff and I am certain that it actually takes both to get the job done realistically. But if my word isnt good enough, here's a more technical explanation:
In any room, all surfaces are "reflectors." The larger a surface and the more light it receives, the more its "reflections" will affect the rest of the room. So for most interiors, the primary reflectors are the vertical walls and ceiling and floor. However, most rooms are not empty, they have stuff in them...chairs....tables... and tons of other stuff. Just because these surfaces arent as large as the walls doesnt mean they have nothing to contribute to the scene GI. It is for all the stuff other than the walls that we need to use a 3D Fill as well as a wall hugging Cube Dome to cover all the angles.
Another way to think of it: If I only use Cube Domes then the center of the room will be the darkest area, since it is the farthest point from all of the walls. The problem is that real life doesnt operate that way. Most rooms have a failry even distribution of indirect lighting. 3D Fills by comparison tend to pool most of its light at the room center, which is also wrong but in the opposite way from Domes. Its a classic example of how two wrongs actually DO make a right. Each of them corrects the shortcomings of the other one.
Image 5: This image is what results when I simply combine my skylight rig and bounce with my overhead light rig and bounce. I think this result is fine, but if I'm really picky I must admit that it still doesnt remind me of the "feel" I used to get as a child when sitting in a classroom. Classrooms never seemed like dimly lit places. So what is missing?
Image 6, 7 and 8: Here are three different "exposure" levels, needed to bring the image well into comfortable view. All enclosed interiors have this issue. To do this I simply activated the non shadow casting light from EGDLS called Bounce Light Dome whose entire purpose both indoor and outdoor is to give users a better control over the shading threshholds of the scene. I chose the second one as my comfortable viewing range, but your choice could differ. The point of this is to demonstrate that even after all the other stuff I did there was still a bit of light missing if I compared the image to real life. The result before adding the Bounce Dome was too dark, too heavy with shading and low light levels to have the "feel" of lightness and learning that I remember from childhood.
Two major things were missing. First was a very basic fact that when presented with low light conditions the human pupils will dialate, becoming wider, allowing much more light into the eye. With photography, when in a low light situation we would increase the exposure length and F-stop to allow more light to reach the film.
The result of this human eye and camera settings adjustment is that we can now view this interior comfortably. This also means that by comparison the outside world will appear extremely overexposed when we look out of the window. In fact, the outside world might even appear fully white. Therefore, if you ever observe a "photo" that features a comfortable view of an interior without the overexposed outdoor, then there is a very strong likelihood that the image you are viewing is not a real photograph at all. An HDRI could indeed pull it off, but then again we also know that most hdri are made of composited images where each of the elements indoor and outdoor are represented in their most easily viewed color ranges.
The second thing that is still missing is the multitude of additional bounces of the indirect light causing the colors to mix toward white. Those bounces tend to brighten the appearance of the overall render.
For this reason, I chose image 7 (full depth exposure 2) as my final. Some of you said you found this to be too bright. Do you still find it to be too bright after I've explained things a bit?
Here is another view of this interior. This time with a horrible wallpaper I created similar to one I found online years ago. In this case I have increased the quality of all the light rigs, so the shadow banding should be minimal enough not to be noticeable until you start to look for it. Notice as well with the increased quality of the lights and the wallpaper that there is no banding visible on the vertical walls. Active patterns can go a long way to hide shadow bands from fake GI rigs. Fun fun!!
Wow this is a long post Rashad, very interesting, I need to copy and paste it in order to re-read. Thanks
Mermaid- Thanks for taking the time to read all of that. I hope this is useful.
Alexey- To you specifcially I want to point to two ideals. In your explanation you demonstrated that you had placed the Cube Dome outside the walls in certain situations, to avoid making the walls in that area too bright. I agree with this observation. As I stated above this tendency for light to collect along the walls with Cube Domes is why I don't put all the responsibility for the indirect light within the room onto the Domes alone. I use only half the power for the domes and assign the rest of the work to the 3d fill.
I also notice you have had to use a very high quality setting, above 2000. In the most recent wallpaper example I uploaded, you can see that the lighting is smooth with very little banding except in places where we'd expect it due to the window panes. My quality increases were modest, in all cases. You experienced very long render times because of your settings. I think that rather than to increase quality to avoid banding it is often best to employ some degree of Randomness. This also means that you do not want to scale your Cube Dome to fit the walls too closely, otherwise you can lose some virtual point sources behind the wall once randomness is enabled.
You already know this, but for other readers I want to explain a bit about randomness and Dome Lights. Randomness with Dome lights works in three dimensions. With randomness at 0, all of the virtual point sources are resting on the other edge as expected. However, when randomness is enabled, the lights take on positions that can place them outside of and also deep within the area of the dome. in fact, at full randomness a Dome light essentially converges with a 3d Fill.
ElectroElvis- Aside from the issues I discussed above, I wanted to point out a couple of observations with your study. In this case it has to do with the shadows at areas such as the doorway, teacher's desk and the like. When I look at the doorway, and I take into consideration that there is no visible point-like light source at the center of the scene, then the shadow the door casts onto the wall behind it is far too "hard." It looks almost like a soft shadow light source, but one that still very much eminates from a point-like origin. I'm not sure how you scaled your 3D Fill light, or how you adjusted its settings. But for my tastes those shadows seem harder than expected.
Rashad, thanks for your reply. I read your long message and was going with my thoughts to answer why I placed the far edge of Dome Light behind the board. But you yourself have already answered this question. In addition, I have to drag the text back and forth to the google translator several times until the translation into English is more authentic. This is the biggest difficulty. Therefore, I often do not take part in the discussion.
As for your renderers, I choose Image 6 (full depth exposure 1). In addition, I believe that you need to reduce the brightness of the sun at least twice to avoid overexposure. In images 7 and 8, there are too many overexposures, especially on the elements of the window. In addition, the floor and countertops are still too bright. Our main task, like the task of the photographer, is to make the details in the shadows and lights read well and be distinguishable from each other. To do this, they came up with HDRI in photography.
Your last big render is very good except for too bright a floor and furniture elements.
Another consideration. Still, your lighting scheme is too complicated, although it is understandable for me. Then, this scheme is applicable only to this room. To another configuration of the room and furniture, you need to do another scheme.
I look with envy at renderers in other programs, where there is such a thing as radiation transfer (radiocity). There it is enough to place two or three light sources, the rest is done by the program. Moreover, there are sources of scattered light, which can be seen in any photo studio. And the render of pictures lasts a couple of hours. TA does not solve such problems. If you transfer DTE, TE, SkyLab, TreeLab from Bryce to such a program, then I am ready to work in such a program. Or vice versa, speed up the render in Bryce several times and make normal lighting. But alas, this is only a dream...
Alexey,
Thanks for reading through it all! I think I agree with you, Exposure 1 is probably the best balance of the lighitng effects. By the way I think you are doing an excellent job of making translations bewtween Russian and English. The statements you are posting feel natural and welcoming. I want you to know that the extra effort you're putting in is fully worthwhile.
Wow, Rashad,
What a lot of explanation. I think I can follow what you are saying, which is not to say, that I would be able to come up with such a lighting rig myself. Your explanation about 'surfaces reflecting light' is quite relevant. You would want to be able to create lights at these surfaces that radiate light in the way the should have reflected it. Or, of course, there should be some form of light reflection without actual 'reflective' look. Perhaps that is exactly what unbiased renderers do?
Thanks for reading it all. Glad to know that it was comprehendable. Reflections, indeed. About 8 years ago Horo and I along with several others were participating in a thread somewhere discussing the way light behaves as it interacts with surfaces. Horo used the phrase "its all reflection" and I remember at the time being surprised. It had been a little too long since I'd studied physics vocabulary. Of course, Horo was fully correct...native German speaker correcting me in English. He's an overachiever in just about any language!!! Anyhow, in physics the term reflection is a bit of an umbrella term. For our CG purposes there are two forms of reflection and it is all based on the specific nature of a surface.
Perfectly smooth surfaces, such as mirrors and water surfaces, provide sharp "Specular" reflections. Notice however that if you scratch the surface of that mirror, disturb its perfect smoothness, that its specular reflection disappears, and you can no longer derive sharp images. But what you will also notice within those scratches is thatt the scratches always appear white. This is because the mirror technically still reflects the same amount of light, but the bumpiness of the surface doesnt allow the rays to depart along the complimentary angle of their approach as they do on smooth surfaces. Bump surfaces tedn to redirect the rays into all directions, creating a scattering effect.
Polish anything solid to an adequate degree and it will reflect the environment around it to some degree. Create microscratches on any previously smooth surface and it will scatter light in many directions providing what is considered a Diffuse reflection. Matte is another good word for it.
But it doesn't end there. The nature of the reflection isn't only affected by the smoothness of the surface. It just so happens that metalic surfaces will tint the light they reflect with their own color. Dielectic surfaces on the other hand have perfectly white reflections. The conducivity of metals causes them to participate in light interactions differently thatn non metals. Bryce is already keen to this, and offers the awesome Metalicity slider to allow the user to color tint the reflections as needed.
Human skin, being dielectric, covered by oils also being dielectric, mean that human skin can only provide white specular reflections. A woman wearing makeup however, that could be infused with bronze or other types of metals, could easily take on a colored reflection if designed to do so.
True Ambience already does the type of light bouncing we need in most cases. This classroom study was a particular situation purposefully designed to demonstrate that the currently limited implementation of True Ambience (by unbiased standards anyway) means it would have difficulty reaching an accurate result with this test. Most rendering engines cheat and shortcut in one way or the other. In the case of True Ambience in Bryce 7, it is actually well on its way to being unbiased. In fact Bryce 7's TA is far less biased than TA in Bryce 6 and earlier, but still biased none the less. The main drawback is that we cannot increase the samples beyond 256. Trust me, most Octane renders with only 256 rays per pixel look quite noisy, exactly as does TA with boost light and scattering correction. TA is best when it has the benefit of alarge light producing blue sky or when it is helped with standard radials and similar items. TA just needs a few tweaks to be in the big time, in my opinion naturally.
Nevertheless, using hypertexture enhanced ambience as material on a flattened square parallel light and rendering TA is an alternative to dome and fill lights to create the ambient light in a room. It's not the better method, it's an alternate method and works a treat in one room but not so good in another. I've investigated the possibilities, the flaws and work-arounds.
I put my findings in a 8 page, 32 pictures PDF on my website: Bryce & 3D CG Documents > Mine > Light > Hypertexture Obscure Light for Indoors. There is also a pure text file for translation. I removed the source file offered on the previous page.
Horo, can you give a direct link to both documents?
Slepalex - yes, here they are.
https://horo.ch/docs/mine/pdf/HTOL4Indoors.pdf and https://horo.ch/docs/mine/txt/HTOL4Indoors.txt
Thanks also, Horo...I'm currently doing a room over the last week or so, so the PDF will prove very useful.
I have some fall-off lighting going, that is, light coming through a window is diminising in intensity as it extends further into the room - looks nice. Though, I haven't got so far as to include furniture, figures etc., yet, I'm expecting this fall-off effect will also be applicable to shadows produced from such included objects etc.
Jay
Horo- thanks for the pdf.
Jay - looking forward to seeing your completed scene
The awesome/great thing about Bryce is that despite the fact that I don’t really understand much of the theory/technical aspects, it has not stopped me from using the program and ending up with cool renders.
After re and re-reading Rashad’s explanation, I decided to just have some fun and here is the result. I used one of David’s clouds from his www.daz3d.com/bryce-pro-cloudscapes-5-example-one changed the wall material, decreased the diffusion on some materials and the diffuse for some of the lights in the scene. Render time was just over 4 hrs on my laptop.
Well, mermaid010, good! However, there is a small overexposure.
Jamahoney - you're welcome, perhaps it helps.
Mermaid - you're welcome. Your room looks very nice. Yes, as Slepalex says, there's an overexposure but that one is in Rashad's original as well.
Slepalex - by the way, if you still have issues to open my https site, you can still access it with http. If you click on the text link, you get an absolute link to https. However, if you click on the image, you get a relative link and stay http. The upper half is German, the lower half English, just like the text link. Once out of the home page, all links are relative.
Thank you, Horo. I downloaded both files without problems and even translated the text into a translator.
Thanks Slepalex and Horo, yes there is some overexposure,
Horo- Fascinating to say the least! In my opinion Obscure Light does exactly what it is intended to do, which is to get some indirect light moving around the room in a way that provides at least some illumination on all surfaces. It also provides very favorable render times. And it utilizes the not quite fully developed TA Optimization tool....supporting with real evidence the claim on the Bryce 7 product page that TA Optimized lights are fast rendering.
As the rest of us can see from Horo's scene and documentation this is a solution that can be applied to many scenarios. In some instances it will look completely plausible. Very clever solution.
Mermaid- I love it, overexposures and all. Great wallpaper!! I do not see these overexposures as problem, so long as the overall dynamics of the scene relate well to one another. Just like a shadow might reach near blackness if occluded from receiving light equally so can a highlight reach near whiteness. Unlike manipulated artwork, realism never apologizes. If indeed we were truly standing in this room, we'd see the room in its full dynamic range which far exceeds the output of our monitors. We would not see the overexposure as white shifted, we'd still see more if its underlying color so we wouldnt even perceive it as an overexposure. What matters most in my opionion is that surfaces dont appear to "glow" without being influenced by a specific bright lightsource or by a brightly reflective nearby surface. Another way to think of it is to say that while overexposures may be ugly, they are not unrealsitic. Real life can have its "ugly" moments.
Rashad- thanks for the nice comments and explanation, I’m still having fun playing with this file, The neat thing about working with your files is that we all have the same starting point, so for someone like me it’s much easier to relate to the comments/explanations. Horo has also provided a cool file together with the pdf. Thanks Horo, I will have a look at both this weekend.
2 renders - I started both with my previous file, using David’s sky and the screenshots Slepalex posted. 1st image - render Regular almost 4 ½ hours. For the 2nd render I tried to make sense of the randomness explanation, reduced the size of Dome light from Slepalex’s example, Diffuse 325, Quality 600, Bias -25, Randomness 35 – render Regular 1 ½ hours. I like the 1st render.
Looking forward to the comments.
Mermaid, the second render I like more. And in general, I believe that this is the best render of the presented here. True, there is no sunlight...
Thanks Slepalex.
mermaid - I like the first one more but the light in the second one is more natural.
Thanks Horo.
Mermaid- Very nice! The first one feels very warm. The second one seems more to the point. But I think the real factor of importance is that they look similar enough that it proves that Bryce can actually be fairly quick when the settings are properly optimized for your specific needs. In these studies you seem to have made very good sense of the Cube Dome cluster form lighting tool. The next step would be to try the exact same settings with a 3D Fill. T=You'll find the 3D Fill isnt perfect either, but it is still plausible.
Your wallpaper is much less offensive than my own. You've made several wise decisions. It was a wise decision for example to shrink the size of the Cube Dome so that it didn't fit the walls too tightly, as when you apply the randomness settings some of the virtual light sources can get lost within and behind the geometry of the walls. Fewer virtual sources within the boundary of the room consequentially lends less total light to the scene than previously.
Thd first render does probably do a better job of covering multuple angles of the models than the second render. However there are ways to make the second one look more like the first one and still keep these reduced render times. Cube Domes can run the risk of illuminating the walls they are supposed to be faking reflections from a bit tioo much. This is another benefit of putting a little room between walls and the Dome boundary. There is also a setting that allows you to tweak a Dome to only shine light in the direction toward its center. This matters more in situations where ranges are used. But I admit to not having tested it in a while. I suspect at this scale even if only firing toward the center the light would still end up hitting the walls, but only after having traveled across the entire room to get there which is more physically accurate.
Great studies Mermaid! Thanks for sharing your incredible progress and do keep having fun, which I also suspect you are doing.
Thanks Rashad, yes no doubt I am having lots of fun mainly because I just play with the controls and see what works. I'll give the Fill Light a try.
No, this is not the best lit classroom but definitely the cheapest: 1 (one) radial near the camera. Renders premium with soft shadows and 64 rpp in less than 3 minutes. The radial has no falloff but an elaborate gradient.