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
beautiful work Varsel
Like Joe says, there's some really nice reading available on the topic. I've never seen 'that' book - but I'll certainly add it to my want list.
Here is a very quick example of what I was talking about.
I began creating Woodlands due to some requests for some more expansive presets for the browser. I also needed this particular sort of thing myself, so it was a great starting point. The original name was Environment Construction Kit and was meant to be a fast and easy way to drag full environmental elements into the scene in chunks and move them around to create simple backgrounds. Woodlands provided a particular challenge as I needed to find a way to allow the user (like myself) to have the ability to really bunch things up and still have it work. This little laptop sure renders slowly compared to my bigger render machine, obviously, which caused me to make my point with a much smaller image.
In a forest on a sunny day, there is a certain glow to the woods - especially with a fairly full canopy of leaves. Pine forests are really fun in that respect, too - but that's a bit of a different effect of light than deciduous leaf translucency. Anyways, several Carrara artists had some really cool, intense discussions here. I think it was in the old forum. One of the biggest banes of shadows in forest situations was a lack of the user using translucency on deciduous leaves without using some sort of correction. There are ways to fake it. One of them being the same technique used in artificial Indirect Light.
So I've done a ton of work setting up the environmental illumination rigs for Woodlands. Using GI with or without IL would be easy to set up from here, if the user wishes to. But coming up with an artificial helper system can be a bit more of a challenge, so I put a great deal of effort into this aspect to provide a reusable rig that remains consistent, yet editable and, even more importantly to me, to be really easy to use in as little time as possible.
Here's what I've come up with:
The render is a scene I've literally just tossed together using Woodlands. I began with the main blank scene in the Realistic Sky catalog, went to Objects and brought in a Large Woodland Block and dragged it into the background a bit, then plopped in a Forest Block (medium sized chunk) and pulled it up toward the camera. I also used the Forest shader on the ground terrains.
The render is unedited and shows the result of the artificial lighting rigs in the kit. Translucency has been performing very slowly on my animations, so I've resorted to this method instead.
The screenshot image shows the artificial ambient light system. The lights underneath are not "Shown in 3D View" by default, so I've made them visible here, and turned everything else in the scene off, so we can see this simple rig.
There are three lights parented to a light that doesn't work. The uppermost light is just a control knob for aiming the rig. It has constraints applied to only rotate along the Z axis, making it less simple to mess up the effectiveness because it uses careful angles meant to assist the view from the camera with artificial indirect lighting effects - real or otherwise. By default, I've set up the presets to appear more on the realistic.
There are three lights to the rig.
High
The center light is the most prominent intensity of the three and aims upwards toward the direction of the direction control dial.
Medium
Slightly off angle to the right, along the Z axis, which is less intense than the High light.
Low
The same amount of angle along Z, but to the left instead, and is even less intense than Medium
These three lights together help to fill the indirect angles - but really only from the camera's point of view.
Each of these three lights has three versions to choose from. Neutral (Gray Scale), Warm (More Orange), and Cool (More Blue).
Simply make visible whatever temperature you like and, perhaps, adjust to taste. For even more indirect effect lighting, we can easily turn on more than one light from each or any of the angles - even change the colors and/or intensity.
But the default values are set for a reason after a lot of testing. So if you do make changes that work great for you, it would be best to save a base preset of your edits.
beautiful work Varsel
Agreed... I was coming back to say the same thing!
I tend to call this the "Upward Light." Yes, this can be very useful but it does have some shortcomings which I will explain in a bit. It's better than nothing, but it leaves several things to be desired as I will demonstrate.
Tim...forgive the nit picking, but in this business if you don't want your viewers to spend time wondering why your image just doesn't look right you need to focus on details... :) :)
The first thing that jumped out at me is the pose of the jogger. In fact, when we run we all make pretty much the same choreographed movements, and anything else sticks out like a sore thumb. It's all about maintaining balance whenever we move.
If you do a google search on animating run cycle or something you'll probably see images of the various poses that comprise a typical run cycle.
Your jogger seems to have his left leg in a passing pose, while the right leg is in the "spread" pose. It just looks unbalanced. There isn't enough time for the right leg to come forward and keep the jogger from falling forward. Both legs will quickly be behind his center of gravity and...kaboom.. :) :)
Rashad,
That's an interesting treatise on lighting and shadows. But at the risk of inducing glassy-eyed disinterest and even, I imagine, some anger and dismissal from some forum members, I think you clearly have some basic misunderstandings on the topic.
Joe,
First I want to say that I LOVE these types of detailed, and deeply theoretical discussions. Thanks for going back and forth with me. It might be that no one but you and I even care about this and if that is the case we will then continue this discussion in another thread altogether or we can discuss things in private, Either way, this is all quite fun.
We probably do not disagree. Our vocabulary and modes of description might greatly differ, but the physics of this sort of thing make it impossible for us to disagree too much. Indeed, I may have misunderstandings, but I don't think you will still feel that way after I explain in more detail what I am meaning to say. But before we go any further I need to make it clear that what I am talking about may well be a subject that you have not yet had time to give as much thought as I have. I say this because Bryce unfortunately requires the user to dig quite deep into theory to make sense of it and produce results that are on par with other more powerful applications. This fake GI stuff is what I've been studying for a long time, so I've had lots of time to think and re-think. In this instance, I will be very glad to learn more than I have already surmised.
You are surely right. The post I made did not describe all types of light propagation. I was talking about the standard "radial" type of light, not a light beam. Indeed, light can be phased, polarized and focused into discrete beams like lasers just as you describe. And indeed if all of the photons are traveling in the same direction coming from a similar origin then they will produce a hard edged shadow. In CG we might see this with Parallel Lights. But that is not the meat of my argument. More on that in a moment.
Just to dispel any notion of misunderstanding about real world shadows, I will explain how my thoughts on shadows work. Basically there are three primary considerations;
1. Relative Scaling. Absolute scaling is meaningless, but relative scaling is essential. The larger the physical size of the light source relative to the physical size of the target object, the softer the resulting shadows will be, and vice versa.
2. Distance. Distance is the reason why absolute scaling isn't important. Though the sun is technically much much larger than the Earth itself, it's incredible distance makes it seem like a small dot in the sky relative to us Earth inhabitants. Relative to the Earth the sun appears small in the sky and produces for that reason fairly hard shadows.
3. Shadow Propagation. The greater the distance the shadow itself travels before it arrives on a surface, the more it will soften. Hold your hand an inch away from the wall and the shadow it casts will be crisp compared to the shadow that is produced if you move your hand a twelve inches away from the wall.
So, if a light source is extremely large, but very far away, it will appear to behave as a single point, kind of like the sun, casting relatively hard shadows. If that same light source was closer then the shadow it casts would appear softer because its relative size would appear larger. As well, it is important to note that younger shadows are sharper than more mature shadows. Yes, sunlight can produce very sharp shadows but typically that occurs when the shadows don't travel very far, such as when the sun is high up in the air at high noon. Shadows fall straight downward, and do not appear to soften nearly as much as shadows cast by the sun when it is at a lower angle such as during the evening. Shadows cast in the evening get extremely stretched out, making the soft edge more visible. Though the size of the Earth and the Sun and their relative distances has not changed due to time of day, its still important that the lower angle still produces softer looking shadows because of the longer propagation range.
Now the point of my post you probably had trouble with was when I described a direct relationship between shadow softness due to light source size and specular halo softness due to light source size. But indeed, both of these functions are based on "line of sight" principles, so the link should be obvious to you after looking at my diagram below.
Which brings us to why my post about using more point lights instead of less is still correct, even if it is hard to accept. Below is a visual diagram of what I wrote in words in the previous post. I would be very surprised if you find any misunderstandings within this diagram.
The second image is a re-upload of your example image. While from a diffuse light standpoint the scene looks more or less acceptable, from a specular standpoint it gives itself away as false. This is exactly what I would have predicted.
Realize, that I have a very sharp eye for inaccuracies in renders, it's almost annoying even for me. So what I am going to point out may not have ever jumped out at you as an error before, but it jumps out that way to me immediately.
I have circled the area of highlight along the back wall that should not be there. This highlight is a perfect example of what I mean when I say that specular and diffuse highlight still behave as points even though you have clearly enabled soft shadows, it doesn't solve the exaggerated highlight problem.
Realize, that the white spot on that wall holds some of the very brightest pixels in the entire render. Knowing that this wall isn't a light source, the only light it can contribute will be light that is reflected from some other source. Even the floor in front of the window receiving direct sunlight isn't as bright as the back wall. So if this isn't a reflection of the sunlight's specular, then what is it? Where is all that light coming from? The scene overall doesn't look blown out, but that one spot on the wall looks very burned out. For that degree of brightness on the wall, I would expect the entire outdoor to appear overexposed. But as it is now very much within the visible range, I just don't see a real world justification for that pool of light on that wall.
You only used 5 lights in this simulation, and it is obvious to me as being too "coarse" of a lighting scheme. If you had used 15 lights, you'd see that the exaggerated specular highlight on that wall would soften into something much less obvious as incorrect. If you had used 150 lights, it would look even better.
Sometimes in life, less is more. Other times, only more is more. This is one of those times when only more is more.
Stringtheory,
When using these "Light Clusters" as you are, do not enable soft shadows. I can imagine a hundred lights all casting soft shadows WILL take an eternity to render. The whole point of these clusters is as I describe in the diagram above, their "disagreement" on the position of shadows creates smooth shadow gradients automatically even though the individual lights still cast hard shadows.
And I do think the hdri of the scenes combined looks the best. I will get on with describing some issues for EvilProducer that relate to your current study. Back in a bit.
Fun fun!
This sounds like it could well work. I do not have experience with baking, but he theory you describe sounds good to me.
In truth, the full process to achieve that render is a little more involved than just a single 3d Fill, but the 3D Fill is the trick that performs the major magic. Suffice it to say the render was several hours, but it was still much less time than it would have been for a full GI render. I am still trying to find way of rigging in Carrara some of the clever light tools we have in Bryce. I will keep you posted on my progress. A 3D Fill type of device should be doable in Carrara.
Thanks for your comments guys :)
Dartan: The renders so far are *with* translucency on the leaves. I'll show you a pic without, but that'll take some time to set up and render out. These are big trees - 60+ft tall (actually the variation is 55 - 65ft on the models and 10% scale variation on the replicator (shame the replicator only allow + or - and not +/-). The translucency is done the same way that Howie Farkes does his, i.e. the diffuse bitmap with a multiplier. Anyhow, the green leafy glow on the tree trunks definitely comes from the translucency.
I did try a couple of variations on the cheat light underneath (not published), but having it bright enough to make a noticeable difference also added an eerie green glow to the underside of the branches - not what I wanted!
I'd have a play with your Woodlands set, but sadly it was lost in the great unnecessary update fiasco, and despite raising a ticket and badgering support, it's never installed since.
Headwax: I love your HDR mix, but the thought of all the extra hours of rendering to achieve it gives me the shudders ;)
Joe: Interesting point about the pose on the jogger. It's just a standard shop pose that came bundled with one of the figures. Since it was sold commercially through the store, I had assumed - wrongly it would seem - that the vendor knew what they were doing when they created it. I'll follow up your suggestion and investigate creating my own set of running poses, but in the meantime I hope you'll forgive me for continuing to use standard ones.
Since this is a render thread, here's a render. ;)
Well, that took almost no time, comparatively. Sorry it's not quite the same camera angle, but there should be enough commonality to get a fair comparison. This then is the no translucency render.
Rashad,
Yeah, you're right. I think the terminology you're using, like "younger" or "mature" shadows, and shadows "travelling" and "falling", and some fairly non-scientific or non-physics-based concepts, makes discussing this kind of impossible for me. I've been doing this for many, many years, and have taught a number of classes on much of this stuff, and the terms you're using are totally foreign to me.
Yeah, your basic premise is that you can use multiple spotlights to simulate a diffuse light source, which will soften shadows. Of course. Nobody would argue that. It's just a matter of deciding how many, what intensities, what colors, and what soft shadow settings.
But honestly, without a common scientific terminology, I'm kinda scratching my head trying to understand the details of what you're talking about.
But you did indeed argue against just that, Joe, whether you realized you were arguing against it is the question. That's why I was confident that I had no misunderstandings on lighting and shadow behaviors, that my treatise was indeed sound, and that you and I did not disagree on any of the relevant points.
You wanted to explain to us why we didn't need as many lights as we had assumed, but I wanted to explain to you why we indeed do need these lights. It's not just shadows, Joe, it is highlights as well that are incorrect with point lights and there is no way to fix it except to dither it by using multiple point lights.
But at least some progress has been made. Now that we can both acknowledge the benefits of using multiple lights, we can move on to deciding on the best way to use those lights and just how many of them to employ.
Now one last thing. Above you mention needing to determine soft shadow settings for the cluster of lights. If you fully follow the concept you realize that you don't need to ask the soft shadow settings question at all once you reach a certain critical number of lights in your cluster. The more lights you use in your cluster the more softened the shadows will be without the need to employ the render expensive feature of soft shadows. Don't waste the render time, go ahead and leave the hard shadows.
The Lesson here is:
20 lights casting hard shadows renders faster and ultimately looks better than 5 lights casting soft shadows. I call this a win win.
When devising indirect light for your fake GI scenes keep in mind that we are better off using a cluster with more points lights within in it while casting hard shadows than to use a cluster of fewer lights with soft shadows enabled. The reason for this is I argue is that the simulation that features very few lights remains crude in terns of diffuse and specular highlights, despite the soft shadows it won't look as real as it could with more lights. Multiple lights gives you soft shadows and soft highlights both at the same time.
Scratch no more. There's nothing particularly interesting or controversial about any of the concepts I've discussed so far. We are already on the same page.
Not sure what this is - just playing around. :) About 10-15 minutes of modeling work and 20-25 minutes to setup the scene and render.
I made this by in the Vertex Modeler by inserting a 2D circle with 6 points, extruding an inner circle, deleting the inner poly, duplicating the whole thing a couple times, selecting all, welding with the default tolerance, removing duplicate points, adding thickness to get a flat grid of two rows of hexagons. Then I rotated that so it was vertical and used Edit>Replicate to create a cylinder of my grids. I welded that and removed duplicate points again and ta da.... these things!
The longest part was getting the replication settings right - I had to control-z about ten times. I started with 36 replications and radius of 4 inches and then lowered the number of copies until the edges of each section were just barely overlapping, then adjusted the radius until they were just touching. That took about five or six minutes to figure out. The shape on the left is just a flat section of the base hex grid with a bend/twist modifier applied to it.
Mark
Hey that looks really cool! Nice work.
Nice work!
Here's a simple 'toon cat I've been building for an animation I'm going to do. I still need to rig it, but I have set up a couple simple morphs for it, plus set up some constraints for the eyes and eye lids.
My 'toon cat model. Posed and rendered using the 'toon scene filter as a test to check weight mapping and constraints. Surprisingly the influences were pretty good with only a couple areas that I needed to fix.
I wanted to thank Stringtheory, Rashad, and Joe for the excellent discussion over light and specular effects and soft shadows. I'm very taken by the 'multiple low intensity spotlights on a plane' approach and keen to give it a try. I'm particularly curious about the nature of the '3D Fill Light' approach you made reference to, Rashad. You mentioned it was likely something that could be done in Carrara just as you had done in Bryce but didn't really give any details (unless I somehow missed it upthread) as to what it was or what was involved in it?
Thanks for taking the time to read all of that. I too find this type of study to be fun.
In Stringtheory's situation I'd say that he should map the major reflectors with spotlights first. Mapping precedent should be given based on surface area. For an indoor scene, the major reflectors will be the five walls and the floor because of their greater surface area than anything inside of the room. Map the walls and the floor with spotlights first, then the tables and the like. In the end you will end up using many more spotlights than Stringtheory is currently using if you want to have good coverage.
In theory a full GI is a lot like spotlight mapping, its just that the virtual spotlights are only a pixel wide so there are "thousands" of them in a scene. Manually, we would never map the surfaces in a room as well as a full GI would.
And while mapping the surfaces with spotlights seems like the next natural step to faking GI, it turns out there can be more clever ways to get the job done.
You haven't missed anything, I have yet to describe in detail how the 3d Fill works. I hate to continually reference Bryce during a Carrara discussion, but in doing so it might make certain concepts easier to understand.
Bryce 7 has several parameters for control over lights that Carrara currently does not have. Carrara finally received negative lights, a feature Bryce has had since the beginning. But Bryce (starting with version 7) also has separate sliders to control the diffuse and the specular output of light sources individually. This means you can easily create lights who only output specular information and no diffuse at all. and Vice- versa.
Having the ability to increase or decrease the specular output of a light source independent of the diffuse output turns out to be a major game changer. You will wonder how you ever used lights with a hardwired (locked) diffuse/specular ratios in the past.
Bryce 7 also has elegant light Clustering Dome tools, allowing you to create Domes both in spherical and cubed shapes. The cube shaped domes are good for interiors. If you build a room and place a cube shaded dome in there and size it so that it rests just along the interior edges of the walls, it will behave similarly to having mapped spotlights onto those walls. This will cover your major reflectors as we talked about above using only 1 "light source" technically speaking. In Carrara, because there are no official Dome features, we are stuck using surface replicators to custom build our domes. The only issue with surface replicators is that they randomize the locations of the lights more than I would like. But more on that some other time.
I am going to test it tomorrow. But in theory this is how you'd construct a 3D Fill Light in Carrara.
0. Turn off the Ambient Light that is on by default. It is too soon for me to know how much of an impact gamma 2.2 would have on this lighting technique, but I assume it would be useful so its probably best to set this up now.
1. Create a Radial Light Source. (No spotlights or parallel lights or other, only radials at this point). Give this light a very dim output setting, and make sure to leave the default hard shadows in place.
2. Create a Replicator. Double Click to open the Replicator dialogue. Select the radial light as your replicated object.
3. At the bottom left are tools for Random Positioning. Make sure you apply the exact same setting to all three axis. In this case lets assume 25ft x 25ft x 25ft
4. Type in 150 as the number of replications
5. For now, lets also Enable the option to turn these replications into Real Instances. Carrara will generate the real instances which can take a moment.
6. Don't worry about random scaling or rotation, for radials this isn't important.
7. Exit the Replicator dialogue.
8. Rename this Replicator as 3D Fill Light 1. What you should be seeing in the wireframe is what appears to be a bunch of random floating boundary boxes which represent the floating radials within the volume of the replicator space.The lights should be dispersed at all different random heights within the volume, the lights should not be confined to planes. The combined output of all of these lights should be tempered so as not to overpower in your test renders.
You will notice that all of the lights can still be edited at once to some limited degree. You can still control color and brightness for all of the lights at once which is all that really matters.
In the future, you will not need to convert the replications into instances as we did today, I'm just adding that step for this tutorial so that people can see visually if they've built the 3d Fill properly or not.
9. Now create a room model, or import a room you've already modeled. Momentarily turn off all of your key lights affecting this room including sunlight and any interior lights. This is only temporary.
10. Place the 3d Fill so that it is centered within the room. Make certain you scale the 3d fill so that it is just slightly smaller than the volume of the interior, this prevents lights getting lost inside of the walls or under the floor.
11. The 3D Fill is intended to represent the indirect light bouncing within the room. If there are any major colored walls in the room you can give the 3D Fill a tint of that color. If you have a red brick wall, give it a red tint.
12. Make a test render of the interior lit only by the 3D Fill Light. It should look a lot like a scene rendered on a cloudy day. You can play around with the depth of the shadows as needed. What you "should" be seeing is what looks like Ambient Occlusion, objects should shade one another a little bit and yet we should also still be able to see light arriving on the furniture in a very uniform looking way.
Roughly 50% of the total light in the scene should come from the 3D Fill. Most often when I see fake GI renders the artists tend to underestimate just how bright the indirect lighting in most scenes really is, so they cheat it and leave their indirect too dim. Brighten it up.
13. Now you can begin to re-introduce your key lights. The key lights will provide the accents needed for the scene to pop. You will have to play around with it to ensure that there isnt too much or too little indirect light for a scene lit by the current key light sources. This is the part that is a judgement call and the more experience you have with gauging by eye what looks correct the better off you will be in this stage.
14. You can always create new 3D Fills with greater or lower quality (the higher the number of replicated lights the higher the overall quality of the resulting illumination).
15. If you encounter hot spots (these can result from the radials floating very close to the surface normal of an object) there are ways to remove those but I will leave that for the next post.
16. Please please please do upload examples of your tests. Also take note of the rendering time. It is also a good idea to re-render the scene with traditional full GI (removing the 3d fill) to compare render quality and render times. In most cases the full GI render will look better, but not so much better that it justifies the much longer render time. But only you can be the judge of that.
I look forward to your experiments. Best of luck.
Rashad, thanks for taking the time to explain, I think I understand correctly the method because you were so detailed. One quick question, which is related to terminology I'm not familiar with: Radial light = bulb light?
Yes, I think so. So long as it shines light in all directions then it will work.
Ok thanks, I thought it had to be a bulb, since from the description it seemed like it would have to shine light in all directions, but I wanted to make sure I was correctly understanding. I'm very keen to try this out now just to see what happens (although it's late and I have to get up early to bed, so will likely have to wait til tomorrow), thank you very much for describing the technique in such detail, Rashad :)
Rashad, I'm not sure why you would need to create "real" instances of the replicated light, unless you somehow need different colors or brightnesses. It seems a very inefficient way to do what you want. If you need a different colored array of lights for instance, create another bulb and another replicator, and position it in the area it needs to be. The beauty of a single instance in a replicator is that there is one bulb to adjust.
Now, assuming for some reason you still want to have real instances of the lights, you have a 150 lights (or however many you replicated and converted). My suggestion to manage those lights is to group them and use the Master Light control to adjust them.
Also, are you talking about a surface replicator or just a regular replicator? The regular replicator has different controls and settings. See the screen shots.
The there's the environment light, which I haven't played around with much, but it acts similar to an Anything Glows light.
Geez, amazing how much stares one in the face in the nooks and crannies of Carrara and you don't even realize it till someone points it out! Thanks all for really taking this lighting subject apart. I sure am taking notes.
P.S. EP, Rashad actually mentioned that he only created real instances of the replicated lights to show how they were distributed, he didn't intend for real instances to be created in a 'real' project. But I'm glad you missed that because the master light control and environment light options you 'highlighted' are new to me.
I guess I'm still confused on this. If I needed to see the lights, I'd just show the mesh in the replicator's general tab, then adjust the replicator's parameters as needed.
If I needed absolute control of the light placement, the duplicate command also works well.
The nice thing about the regular replicator is that you can randomize the positions if you want, but you can also define the grid spacing, such as 2' or 10' or whatever you need. You can also define the number by the x, y and z coordinates.
I thought so too, but then assumed maybe Rashad was not aware of this? Sure he'll drop by soon to clarify.
ep and I (as well as Garstor and Wendy) have had many wonderful discussions behind the scenes about lighting in Carrara. It's a wonderful thing. Although popular lighting guides and other software may use different terminology, Carrara impressively includes a vast assortment of ways to light a scene. I've just bought Phil Wilkes Realism Rendering course and he has some very nice translations in there from Jeremy Birn's Digital Lighting and Rendering guide for use in Carrara. But they're not straight cross-overs from the book into Carrara. Phil has gone far out of his way to get it right, and show some really "Real" techniques for lighting via Global Illumination with Indirect Light, techniques to do so without GI/IL, also with GI and the Ambient Occlusion form of Indirect Light. Pretty cool.
But evilproducer has pointed out many methods to me over the years. Carrara (and evilproducer) never ceases to amaze me with new ways to get awesome results. Carrara truly can produce incredible realism with its native render engine, which remains to be my favorite way to render.
This is a variation on one I've posted before. I've included it here since I used it in a comparative render test - 3.2GHz i5 (4 buckets) vs 2.2GHz i7 (8 buckets). The i5 took 4:45, the i7 took 5:42.
Render settings: Gamma 1.8, AA Good, Object 0.5px, Shadow 1.0px, GI "Sky light". Image size is 1600x1000
Nice composition! Would have expected the i7 to whip the i5 though. Wish someone could explain about the GHz as I never managed to wrap my head around that.