Post Your Renders - Happy New Year yall
This discussion has been closed.
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
Ah no, you'll never get such a wonderful render with one light (well I wouldn't!) I think a minimum of four would be the least, and still not produce such a great result.
Though ( not on the surface) specifically 3d related I'd highly recommend
http://www.amazon.com/Color-Light-Guide-Realist-Painter/dp/0740797719
http://www.amazon.com/Imaginative-Realism-Paint-Doesnt-Exist/dp/0740785508
Because in our work we are 'painting' what doesn't exist. Hence they are more relevant than they seem.
It's all about the suspension of disbelief rather than emulating reality. Emulating reality just turns you into a drudge.
I'm not enamoured of Gurney's style but his insights into the nature of reflected light etc are well worth the price of admission.
here is his blog site, where he lets more than the occasional gem slip out into the world http://gurneyjourney.blogspot.com.au/
great work you are doing
As I have said before, gamma correction is not a "recipe for realism", it is a calibration setting that you only need to apply if your system does not automatically correct for certain nonlinearities in your images. If used when it is not necessary it can adversely affect your renders.
And as stringtheory's image of the guy's face clearly shows, "realism" is, especially in this case, mostly a function of the hyper-high-resolution head scan and associated textures and bump/displacement maps.
No need to repeat yourself Joe.
That is a great quote to keep in mind. I think with the compromises I need to make for the sake of speed, having something not looking fake is way more a priority than having something looking completely realistic.
Well, when discussing a medium (3D computer graphics) where most of the practitioners, and certainly many who frequent this forum, are highly interested in "realism", then that kind of sentiment seems a bit misplaced.
And I suppose some might consider learning a new skill as somehow inhibiting creativity, but personally I don't. I suppose there's a case to be made that the finest painters manage to understand the core of how people are affected by images and the real world around them, and extract those key principles and apply them in magical ways. And since our perceptions are based upon a lifetime of observation of the real world around us, I find it hard to believe there's no merit in understanding how people perceive and react to real world light and texture.
But anyway...
And from the following discussion about the contents of one of the recommended books, it seems like the author tends to agree with my point:
"There are lessons on sources of light, light and form, understanding the color wheel, how light reacts with surfaces and other elements, and visual perception. You'll learn interesting things like why the moon looks blue, when happens to light when it travels through green canopy, tips on mixing colors, and a whole lot more."
Learning about the real world is a good thing.
That is a great quote to keep in mind. I think with the compromises I need to make for the sake of speed, having something not looking fake is way more a priority than having something looking completely realistic.
one interesting psych paper I read last year talks about the emotive quality of an image is usually related to suggestion than photorealisim. If you tell the viewer everything they won't engage their brain/emotion as much as if you leave areas where they fill in the blanks. Hence the dark monster you glimpse in the corner of your eye is sometimes more scary than the mouse that is throwing the shadow - courtesy of your emotive input.
I absolutely agree that the realism in the image of the old man is all due to the high res texture and detail. But it is so easy to screw up those great textures with bad lighting. Essentially most of what we do in 3D graphics is based on lighting some kind of texture along with bump/displacement. When I look at a Pixar movie, even though the characters and scenery cannot be deemed as "real", the lighting and textures still look real.
I think this is the important take away regardless of the type of render that is being done (ultra realism or cartoon or somewhere in the middle). I myself am not going for ultra real. I am trying to achieve a fantasy art type of look which often has higher saturation and more contrast than what we would experience in real life. But, even with that said, I find the look that I am happiest with is one that looks more realistic which I then tweak to look more fantasy art looking. Otherwise my renders just seem to look fake.
the vending machine stole aiko's change :lol: poor aiko
When animating one should definitely count the minutes. Often quality will suffer for the sake of speed, but the real issue is the final balance of trade offs. So far so good based on the example you've uploaded.
To my eye you could afford more light within the room than you have now. You've matched well enough the light coming in from the outside window. What I am missing is the light that has bounced around and returned to the wall where the windows are placed. As bright as those windows are, we can still see clearly the yellow painted wall these window sills are are placed within.
It looks to me like you are "Spot Light Mapping" the surfaces of some of the items to appear to reflect some light back toward the center of the room. Truthfully, this approach can produce very nice results but it usually requires that you carefully map every single surface in the scene, which is very tedious.
And that is how we got the incredible 3D Fill Light in Bryce 7.It is a BRILLIANT light source. When David Brinnen approached the fake GI study he had seen my generalized success using spotlight mapping, but he felt there HAD to be an easier way. He soon realized that instead of mapping each surface in the scene with spotlights, he could produce a better looking indoor GI by creating a network of floating radials within the volume of the room itself, kind of like air molecules randomly floating around. We asked the programmers to make this hand made light trick into an official lighting tool in Bryce 7 and Daz kindly did so.
What this network of floating radials did that looked so good was that it got both light and shadow rays flying around in almost every direction, creating a nearly bias free ambient occlusion. It did a great job of carving out the geometric details without making the image appear flat and 2d the way scene ambience and glow channels would have.
Below are a few links. First are a few links to the Fake GI studies we conducted all those years ago, where the ideals discussed above were discovered.
This is the original Carrara 6 Cornell Room reference scene rendered with Full GI
http://www.bryce5.com/details.php?image_id=3135&mode=search
This was my attempt at recreating those Carrara 6 full GI results in Bryce using Spot Light Mapping
http://www.bryce5.com/details.php?image_id=3131&mode=search
Here is wireframe view of the spotlights I used to produce the appearance of the color bleed.
http://www.bryce5.com/details.php?image_id=3132&mode=search
Needless to say, the more dim lights I use in the simulation the smoother the resulting illumination.
Lastly is a link to a recent render I did using the 3D Fill Light, which as I explained above, is an elegant solution available in Bryce 7 only to aid users in producing plausible fake GI for interiors.
http://www.bryce5.com/details.php?image_id=5259&mode=search
Fun fun
Rashad,
Thank you for a very informative post. The technical explanation of the multiple source vs. point lighting is great and makes a lot of sense. And thank you for the clear and helpful feedback on my scene lighting.
The 3D fill lighting in your room render look amazing. What is the impact on rendering time?
I have been wanting to test out a technique I've been thinking of (or maybe it's already being done). The Baker plugin lets you bake all shading domains into one texture map. You can also bake the rendered lighting into the texture. My thinking is that an indirect GI light render on a white surface would produce a gradient based spot light map. This can be baked into a single texture. The model could then be used as a surface replicator source and spot lights could then be replicated using the baked light map. Essentially a quickly way to emulate level of light emanating from dark and light areas of a baked GI surface. I have no idea if this is something that in practice will work; I guess I should give it a try.
Rashad, I think your understanding of CGI lighting is quite deep. Thank You for bringing this awareness to our attention.
Here's another little track render. I'm sure you're all getting quite bored by now!
Gonna play some more with the fake GI dome - more lights at lower intensity, as per the discussion here. Be interesting to see what that does (and how long it takes - with full GI, this render took about 4 ½ hours at AA-OA-SA settings of Good-0.5-1, gamma 1.8)
Tim, I love every render of this set so much - it makes me want to go outside for a hike. Too bad it's 5 degrees F right now here. :(
About the discussion about realism and lighting - my totally amateur two cents is that we should consider the final audience when balancing "realistic" v. "good enough." I have several recent renders I like on my desktop at work; none of them are perfect, in fact I think any one of us would find numerous flaws in them. But I get people asking all the time "where were those photos taken?" One image is a WIP mountain scene that I posted upthread - it has A LOT of issues still, but a woman who grew up in a mountainous region of Poland and saw real scenes just like that every day for the first 30 years of her life thought it was "amazing" and looked just like a real spot she knew.
So if you are stressing out over the realism of the lighting and textures of your images, try showing them to someone who knows nothing about 3D or even about art - you might be surprised by how many of the issues that are bothering you are not even noticed by someone who doesn't look at renders every day. :)
It's not necessary to perfectly emulate a photo - in my experience the human brain will fill in a lot of details based on their own experience and a lot of that extra time spent getting the tiny details perfect is wasted unless your intended audience is a group of artists.
Well, considering the refrain "Did you try use gamma of 2.2 to improve your render" is repeated here almost on a daily basis, I think a couple times a year mentioning a different perspective can't hurt can it? Or maybe it can... :) :)
My latest Desktop Wallpaper.
Just a quicky made for my girl's New Year Celebration.
The text in the sky is Vietnamese and it means :
May myriad things go according to your will
(or that is what Wikipedia tells me - and she hasn't protested )
To my eye you could afford more light within the room than you have now. You've matched well enough the light coming in from the outside window. What I am missing is the light that has bounced around and returned to the wall where the windows are placed. As bright as those windows are, we can still see clearly the yellow painted wall these window sills are are placed within.
It looks to me like you are "Spot Light Mapping" the surfaces of some of the items to appear to reflect some light back toward the center of the room. Truthfully, this approach can produce very nice results but it usually requires that you carefully map every single surface in the scene, which is very tedious.
And that is how we got the incredible 3D Fill Light in Bryce 7.It is a BRILLIANT light source. When David Brinnen approached the fake GI study he had seen my generalized success using spotlight mapping, but he felt there HAD to be an easier way. He soon realized that instead of mapping each surface in the scene with spotlights, he could produce a better looking indoor GI by creating a network of floating radials within the volume of the room itself, kind of like air molecules randomly floating around. We asked the programmers to make this hand made light trick into an official lighting tool in Bryce 7 and Daz kindly did so.
What this network of floating radials did that looked so good was that it got both light and shadow rays flying around in almost every direction, creating a nearly bias free ambient occlusion. It did a great job of carving out the geometric details without making the image appear flat and 2d the way scene ambience and glow channels would have.
Below are a few links. First are a few links to the Fake GI studies we conducted all those years ago, where the ideals discussed above were discovered.
This is the original Carrara 6 Cornell Room reference scene rendered with Full GI
http://www.bryce5.com/details.php?image_id=3135&mode=search
This was my attempt at recreating those Carrara 6 full GI results in Bryce using Spot Light Mapping
http://www.bryce5.com/details.php?image_id=3131&mode=search
Here is wireframe view of the spotlights I used to produce the appearance of the color bleed.
http://www.bryce5.com/details.php?image_id=3132&mode=search
Needless to say, the more dim lights I use in the simulation the smoother the resulting illumination.
Lastly is a link to a recent render I did using the 3D Fill Light, which as I explained above, is an elegant solution available in Bryce 7 only to aid users in producing plausible fake GI for interiors.
http://www.bryce5.com/details.php?image_id=5259&mode=search
Fun fun
Thank you for this rather 'illuminating' explanation Rashad, I was already sufficiently intrigued by StringTheory's approach (and thanks to him for keeping this on the front burner) to try it for my entry in this month's Carrara challenge, I found it to have a lot of promise, now I understand how and why it works even more, very interesting logic behind it too, I'm amazed at how fast it renders given that with two or three lights in your scene, there's little to calculate, yet the replicator delivers the power of a thousand lights for the same effort! An invaluable lighting technique for animations and stills too for that matter....
Nice
Yes, thanks to all concerned for such an "illuminating" discussion. :cheese:
I think that looks great Tim. Makes me want to take a stroll down it as well, but like MDO2010 it is very cold here, although I think I have him beat as mine is in the negative and not positive. ;-)
-2ºF (approximately -19ºC) here in north central Wisconsin. I can't wait for spring.
Okay here's the results of a back-to-back GI/Fake GI render, using the lots of very dim (1%) lights suggestion from higher up the thread. Gamma for both images was 2.2.
The shadows in the GI render are almost totally black. bear in mind that these are big trees - 60ft typically, so not a whole lot of light will get through. I don't think you'll get a lot of detail in the shadows without burning out the bright areas (much like real life I guess). I might try a 'bright' render and see what happens.
The fake GI setup had about 60 lights on a hemisphere at 1%, plus a green uplight set to ignore the ground. Much more detail in the shadows, but it's all kinda muddy andy lacking in contrast.
In both images, the Gia figure is lit by a separate distant light set to 'Only Gia 6' (Ringo Monfort shader on Gia, but I need to tweak it to look less like fake tan and add a sweaty sheen. Guess I need to hit the skin shader threads!)
The most surprising thing though is the render times. The GI render took 55 minutes - much less than other renders (4 ½ hours typical). I guess that's because it's a tighter shot, looking out of the scene and so there are only a handful of trees visible. The fake GI shot (2 hours typical on the other renders using 20 lights @ 7%) took a whopping 6 HOURS!!
I don't think I'm in a hurry to try that again!
Edit: Cranked the sunlight up to 800% for the last shot. 90 minute render, but looks a lot better I think.
I think that looks great Tim. Makes me want to take a stroll down it as well, but like MDO2010 it is very cold here, although I think I have him beat as mine is in the negative and not positive. ;-)
-2ºF (approximately -19ºC) here in north central Wisconsin. I can't wait for spring.
It's going to be getting pretty freakin cold here on Monday night too!
43F? That was our high today - barmy warm!
Again, I think you're mistaken. Reflections are solely a function of the reflecting surface. I could be photographing a shiny chrome object with a soft box, and the “specular highlight” (aka, reflection) will be VERY clear, like a mirror, even thought the light source is very diffuse and generates soft shadows.
Now, all of that being said, what causes soft shadows is lots of diffuse bounce light, or non-directional light coming from multiple sources OR surface reflections. Diffuse light that surrounds the object you are lighting. The reason why sunlight can cast soft shadows, even though it is highly unidirectional and collimated, is because it bounces off all the diffuse, bumpy surfaces around you, including the clouds and the atmosphere. And it is converted into diffuse light. And if you have a bright point source light in a highly reflective environment, with lots of diffuse surfaces, the bounce light will generate softer shadows.
Think of it this way: let's say we're in a spaceship that is relatively close to the sun. And the sun is a HUGE light source, but its light rays are very unidirectional. So even we're standing 100 miles from the surface of the sun, and it is a HUGE light source, the shadows it generates are extremely harsh. SOLELY because it is not a diffuse light source.
So the challenge in faking GI is to generate what appears to be diffuse light in your scene. And depending on whether you want and can afford a 100% solution, or a 90% solution or can settle for an 80% solution, there are a variety of methods to achieve your goal.
Personally, I'm of the belief that those with a basic understanding of lighting and surfaces can obtain an 80 or 90% solution with only a handful of lights in their scenes. Especially if your scene is outdoors, or is illuminated by a single main light source, like the sun. When you move indoors, where most of your illumination is by diffuse bounce light, and non-linear man-made lights of varying colors and intensities, it becomes more difficult.
Here's an example, what I consider an 80-90% solution which rendered very fast, and used only 5 spot lights and soft shadows. Certainly not for everyone, but at least it is an example of how you can get reasonable results with only a few lights and a decent understanding of how light and surfaces respond.
Us Southern Californians are big babies. :cheese:
Hello Tim
I definitely liked the 3rd shot as well, less render time is even better.
I'm still getting my head around how to best balance scenes with lights with render times.
Guess that will happen with the more I use Carrara, but it's nice knowing other peoples experiences !!!!!!!
Regards, Bunyip
In Howie Farkes scenes, set up the way he advises (his defaults) have this translucency added along with an artificial GI which works very well with his scenes. They are an excellent study course in natural realism rendering in wooded areas. In my EnvironKits, I weigh more heavily on the scene being a background for something much more important - something which the user adds to the scene. They still work using the scenes as the focus, but were designed to render much faster - allowing for animated outdoors scenes. So I've used a three (or more) light artificial indirect light rig in order to eliminate the need for translucency in the leaves - something that is a MUST, or must be cheated in order to get good shadows/light passage through dense forest - otherwise... yeah... black shadows prevail.
With a cheat light system, even if it's only one light, it's important to find a happy medium to get it right, It's easy to add too much, and just as easy to have too little. The thing is, this happy medium will often be different for different people with different situations - the reason why I include separate presets for different conditions - helps get a start on the lighting. Howie also does some of this. His more closely resemble yours... highly details landscapes of immense clarity and realism.
I think that you are doing a magnificent job with this piece! I love seeing your progress - and am always looking forward to seeing the next images! Bravo Tim_A!!! Bravo!!!
EDIT:
After looking at that last image with the high intensity light, I really love that one, and would love to see what just a tiny bit of leaf translucency would bring to the table. If you've never tried leaf translucency, what I do is set the translucency channel to "color" and set that color chip to what I imagine the leaf would do to light - what color would glow on the underside of the leaf. After that, to reduce the effect a lot (which is often very much necessary) drag the color almost all the way to the black corner of the triangle, more toward the color end than the gray scale side - if you get what I mean.
By the way, I hesitate to recommend non-Carrara-specific reading materials here, but in case there's anyone who wants to go a little deeper into understanding lighting and surfaces and how we perceive things, there's a very good book entitled "Visual Perception from a Computer Graphics Perspective"
I'm a bit biased cuz one of the authors is an acquaintance of mine, and I haven't read the entire book (just skimmed it...), but I was pretty impressed by what I read. So if you want a deeper understanding of lighting and shadows and surfaces and how we perceive them, and how that affects our expectations when viewing CG images, you might want to take a look at it.
On the other hand, they tend to agree with much of what I've been saying here over the years, so some might want to look elsewhere for reading materials :) :) :)
A "cheat" light from underneath isn't supposed to mimic leaf translucency, it is meant to simulate the light from the sun or sky, reflected off the ground back up towards the trees or structures. Basically it is supposed to mimic Indirect Light. I think Tim's issue with it looking flat is that it may be too bright or the shadows may not be as close to 100% as they could be. Or maybe it is the Gamma settings.
The Skylight, if it is faster than the simulated GI rig, can still benefit from the ground light, because the Skylight doesn't reflect like full IL, so the terrain itself will block the 360º "atmospheric" light that the Skylight provides, only letting the light in from the top and sides.
You are right of course that adding translucency to the leaves will add to the realism of the scene by allowing more filtered light through the trees illuminating the ground (but again, not reflecting upwards). As you mentioned, it will slow the render down considerably.
There is also the ability to add a very, slight glow to the leaves of the trees to simulate the effect of translucency, but it won't allow the filtered light and could also compound the "flatness" of the lighting...
Which brings me to a general observation of the whole Gamma thing and lighting "cheats." The method to get the best results from the Gamma settings is to not use the cheats as they are not part of a truly linear workflow, as outlined by Jeremy Birn in his Digital Lighting and Rendering book. Even when Phil initially brought up the concept of the 2.2 Gamma setting, he mentioned (if I recall correctly) that it was best used with GI, which is what most of the "cheats" are meant to simulate. So, if you're using lighting tricks to simulate GI, then it may be best to dial back the Gamma a bit.
As to what constitutes a cheat that may not work well with Gamma, I would have to say that less than 100% shadow intensity could cause issues with it. Glow channels in the shader could also cause some problems. There are more I'm sure, but I'm getting tired and should probably go to bed.
Hi Tim
Thanks for posting those renders, interesting the difference.
Just thought you would like to see them combined with hdr program - all the details are present just a matter of getting them in one image!