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
looking through my render folder for some raw renders that are in jpg format (I normally render & work with png files and convert to jpg to share) - DS 4.5 render...
i've been saving up to but Cinema4d for a couple years, (not there yet), and i have no idea if c4d is bias or unbias?
the orange in the fruit bowl looks good enough to eat.
Cinema4D uses a biased render engine, but as I'm sure we can all agree, the results are still pretty spectacular.
I've been reading a book about VRay lately. Not because I have VRay. Not because I plan to buy VRay. I'm reading it because it talks about universal principles common to all rendering engines like lighting, shading, and textures. Until one understands the "why" it won't matter so much about the "how." Btw, that's why I use Blender. I can learn the principles of CG before attempting to master a "professional" application like Maya. When I started out on Blender a couple of years ago, I couldn't afford Maya. Although I probably could now, I've got other things I need to spend that money on. Nevertheless, I like Maya, having played with a demo version. But I'm not going to wait to learn CG until I can afford the application I think is best.
Likewise, with render engines it is not so important whether it is biased or unbiased. I've played around with Lux. If you've got the time to wait, and your scene requires it, then it is a great tool. If you need speed and understand how to do the required fakery and compositing, then a biased engine based on RSL or some such can do a great job too. I've no doubt of the power of 3Delight. I've still got a book on Renderman. At some point I'll get around to learning it. Until that time I'm learning Blender Cycles, which I understand operates like the Arnold Renderer used by Sony Imageworks.
I know this thread is just for fun, so please pardon my seriousness! :)
Another important distinction to be made in CGI is that mostly people are looking to achieve perfection not reality.
The distinction here being that reality isn't perfect.
In the example below, the only way to make my models and scene look real would be to actually physically make them, compose them, light them and photograph them. The end result would have been an accumulation of all the imperfections inherent in model construction, constraints with lighting (rogue reflection and stubborn shadows for instance) and finally photographic artifacts such as bloom flare and grain.
It's not practical in the vast majority of cases to invest the time in overcoming these real world issues in order to achieve perfection so a compromise on realism is required. As most of my paid 3D work involves illustration for packaging and product design, that has to be the hero of my renders, I regularly sacrifice realism in order to make the product look as perfect as possible.
As anyone who has been involved in CGI for long enough will know, the best way to add realism to a render is to make it less perfect, introduce scratches on surfaces, bloom and grain etc.
And BTW: I don't mind being called a 'biased renderer' myself as I know I'm biased toward Bryce. :lol:
You do make some good points elsewhere in your post, but unless I'm misunderstanding the context of this part, then I'd have to disagree here.
Surely the only important thing is the end result? Whether someone is a hobbyist or professional, the end result is what matters. Outside of the 3D community, where we like to share techniques, no one cares which piece of software or render engine you used as long as the end result is fit for purpose.
I admit this ideal is counter-intuitive, but the logic is sound. The destination itself is not the meat of the argument, it is the journey, and the particular means utilized to reach the desired result. Biased rendering methods can produce accurate images in certain isolated situations.
Consider a scene comprised of an enclosed fully white room, all six sides fully white. Within that room there is a fully Red (R255 G0 B0) colored ball placed onto the center of the floor. The materials on the walls are fully matte, and have no specular reflection. The same would be true for the red ball, it too has a matte surface and therefore does not produce specular reflections. There is a single light source in the scene, fully white in color. What would we expect?
Well, in the first place there are no transparencies, no reflections, no specular reflections, and no refractions in the scene as described above. The entire model fits safely within the diffuse inter-reflective realm of calculations, a realm basic radiosity can calculate accurately already. Radiosity has biases but the circumstances of this particular scene did not display them. Still, even though accurate as the final result may be visually, one still could not consider the radiosity render as unbiased because if there had been specular components in materials the resulting radiosity render would have had errors because radiosity skips specular calculations.
Here is a short .pdf from Caltech I think EVERYONE should read regarding the true meaning of unbiased rendering.
http://multires.caltech.edu/~keenan/bias.pdf
It is nice to see the Bryce renders here! I knew Bryce always had the shader power, but I never knew how well it does with realistic lighting... I suppose that's dumb on my part, since all those outdoor scenes have "realistic" lighting... I mean I never realized how good it could look creating "photostudio" lighting.
Again I am liking mostly the Carrara images but I still attribute that to using Carrara my whole 3D life and being "biased" myself, but the steamroller pic for example looks very "real" as far as lighting goes. Aside from the Mini Cooper, I am not seeing these unbiased renders as "real"
What program were the chaos diamonds done in...? I tried diamonds a few years ago in Carrara... (since lost the scene files). My surviving animations use a post effect tinted glow set to trigger above a certain brightness thresholdin AE - the effect was good for my needs but I won't post it since it's against the rules.... I remember staring at my engagement ring trying to get the prism colors to happen with all sorts of lighting tricks (colored spotlights etc), but finally just using an AE filter because it was simpler.... In my completely UNscientific opinion, the main reason diamonds (and anything glittery in general) has the "dazzle BLING" effect is because with 2 eyes each eye is seeing different facets, light dark, so your brain is interpreting this object that is sending different info to each eye and it has a dynamic effect that is constantly changing.... I have never seen any render or even a photo of a diamond that looked as good as a real diamond, especially under display lighting (omg a trip to Cartier's was a religious experience).... Diamonds do crazy lensflares and reflections to each eye separately. Even diamond commercials are not doing justice to the real thing....
**WARNING: RAMBLING PHILOSOPHICAL POST AHEAD!!!***
There may not be limits to what mathematical simulations can do, but there will ALWAYS be limits to what computer monitors or printed photos can display. The math might be there but you can't reproduce what a real diamond does on a monitor, period. For all the dream calculations in the world, at some point it has to be rendered and displayed. The "best" you will ever get is looking like a photograph - and in the 21st century a "photo" is no longer considered the unbiased "truth" it once was....
There were a great many items in this post that I disagreed with, especially the part where the results don't matter all that matters is the elaborate math.... I'll summarize because this is a great thread and I don't want to sound antagonistic... Newbees to Carrara sometimes argue that they want these superdooper lighting sim atmospheric whatziss. I try to remind people that Carrara is an ART tool, not a science tool -- some people get very offended by this (i don't know why, it seems obvious). Once you can accept this you will LOVE Carrara.... If you cannot accept this you will move on because you will never take control of the creation process.... Give an artist some paint and he might be able to create the Mona Lisa.... But a scientist would demand self-programming DNA and a genetic replicator to create the Mona Lisa.... It's just apples and oranges. Anyone who believes they are indulging in math or science when they push the render button is deluding themselves. You're making art, none of it is real no matter how hard you squint or attempt to explain how fancy-schmancy the calculations are.... It's the endless search for the MAKE ART button, but in this case that button is (misleadingly) labeled "MAKE REAL"....
Rashad's statement that in the future everything will be unbiased... When I started learning graphics I was all about the Photoshop. It was so "real". Meanwhile I was also "Illustrator...? Pffft, who would want THAT? Drawing... and vector? It doesn't look real." I was totally ignorant to what that kind of art is.... I was also pursuing this "realism" with Carrara/Ray Dream because the lights could look like a photo... I thought that was all there is... It wasn't until I came here, and saw what Beth "Capsces" was doing stylistically and I was just blown away. 3D doesn't have to imitate a photo. 3D doesn't have to imitate biology or science. That kind of dogma is so LIMITING.... I admire people like Sockratease and Wendy and Beth because they don't seem to be bound to Earth... Their imagination is free -- and it's a mental state, not the software that creates these limits.... I am a recovering photorealist, LOL!
But saying the end results don't matter because it's the path you take to get there... I mean, this sounds exactly like religious dogma. It really is a religion. The Church of the Photoreal..., always falling just short of the divinity.... Looking at the pics in this thread (and links) I am more convinced than before that it really doesn't matter the tool or the process - sometimes it looks awesome and sometimes less so, and that's even from the same artist.... Win some / lose some. From what I am seeing, the tool is just a choice. It does not guarantee you better looking renders, that's for sure. Maybe you feel more "honest" with yourself LOL, but it's still all about building a good scene. I suppose as a "biased" renderer (oh please, what a LOADED term) I would continue to feel that way. This thread hasn't changed my opinion, and I doubt the "unbiased" people will change their minds either...
Oh yeah, something about all 3D being unbiased in the future.., I sure hope not. Cloning *is* the future, but does that mean we will have no appreciation for paintings? I don't really want replicated flesh/art hanging on my wall. Could you imagine, In the future all portraits will be made from a DNA sample and a computer simulation will extrapolate what you look like. No need for an artistic eye. Your portrait will be more genetically accurate than you are!!!
Greetings,
Just for fun, I'll put these in here... I did a pair of renders where I tried to manage the lighting pretty carefully, and ended up with a VERY clear distinction between the DS and Lux rendered versions.
They're suggestive, but that's all...
Fwiw, here's what I said about the lighting when I originally did them:
Enjoy!
-- Morgan
Holly,
For my post "Unbiased" has a specific meaning in math, not a specific look regarding art. You might have missed the link but there is a fascinating ans short pdf from Calthech that explains why the final look isn't the point. Carrara's engine likely is considered a photon mapper. There is a section of this pdf that compares differing rendering methods and why they are biased and how they can be improved, and what the limitations of those improvements are. Photon Mappers suffer from several inherent biases, but at no point is it said that they are not extremely capable.
As far as all rendering being unbiased, it mostly comes down to calculation time and processing. Soon I suspect processors will be able to provide unbiased rendering at comparable speeds. This matters a great deal, because as you have noted yourself, we come to learn to like certain types of looks. It hasnt been all that long that unbiased rendering has existed, so likely we have not yet trained our eyes to expect that level of perfection all the time. But as time passes the standards will increase, surely.
Will there always be a market for biased rendering, most certainly. Paint and brushes are also still popular, but when it comes to most professional environments biased rendering may eventually become outdated.
A very good point you've made is about the dynamic range of displays and pixel printing placing a limit on what we see. This is absolutely certain. I'm convinced we wont always be limited to monitors, we may have access to holograms and other types of mediums that can access a greater portion of the full dynamic range of visible light. The basic math is still underneath it all.
It's not a religion, but it is a bit of nerd speak. The research I've done has led me to understand bias in rendering as having little to do with the visual end result, since many errors will be too subtle to observe with the naked eye. Biased vs Unbiased is a technical distinction, is all I'm trying to say.
I'd love to hear what you think of that pdf.
http://multires.caltech.edu/~keenan/bias.pdf
I suspect there is a minor typo in there, I think you mean to say that the unbiased renderer makes the unnecessary calculations just in case, but biased renders will skip unnecessary steps?
David Brinnen, likely a genius honestly, but who knows, made a clever observation a few weeks ago about noise filtering. His observation is about the size of the actual light probes aka rays...that they are infinitely small points, much smaller than the pixels of an image. This is not so bad, as in real life photons are extremely small. But in real life there are infinite numbers of photons flying around, and they are behaving as waves, not as points necessarily. I do not know if unbiased renderers are treating lights a propagating waves or as distinct points flying around. Most likely, someday there will be a new generation of unbiased renders that truly get to the physical nature of light and materials. For now we are still stuck in simulation mode.
This is the most realistinc image I've seen yet. Take long to set it up?
the whole realism thing is kinda overrated anyway.. pretty much everything "Professional" goes through some sort of postwork process before its displayed to the masses.
Ahhh... Ok this does help me see what we are talking about...
Someone JUST posted a thread where they say Carrara is kicking out a different GI result for every frame in an animation. I think this is the main criticism the "biased" rendering is referring to. Carrara's solution is the option to save the Irradiance Map for the first frame and then keep using that for all subsequent frames....
I stay away from Carrara's GI functions... they are poo. I was forced to get better at my Jurassic 3D lighting skills (where you use actual lights)..... After reading this I realize I am using a combination of Radiosity (for the highlights) and Rasterization - since I end up pulling as much as I can from the lighting calcs to improve render speed, and also because once I find an effect I like I start trying to "bake" it in with shaders... I never really had words for this.
These pics are a set where I tried a render with Carrara's GI first, then replicated it with area lights instead. Carrara's GI is super ashy, and super slow (with longer render times it kets less ashy, obviously).... It has a softness to the light quality that I like, but with the flaws it wasn't worth the render time (2.5 hrs). The second pic with area lights rendered (from memory) in about 3-5 min. I also varied the color of some of the lights so it would be less monochomatic - that was a choice. I wasn't literally trying to duplicate the GI render....
When I am advocating Carrara it is NOT because of the GI situation, it's because of the original RayDream renderer and shader tree. I can see how that is a big hurdle for someone starting out who is understandably just interested in getting a nice image, not learning a program's quirks....
I used Luxrender a while back to test volumes and subsurface scattering, and the results came out quite nice. The one below, however, uses the exact same lighting and scene setup but with 3Delight and UberSurface2 for SSS. The result came out quite nicely, though I'm working on another version which will utilize the Bounce GI UberEnvironment light.
Unfortunately, Bounce GI in 3Delight takes an eternity to render, so don't expect results any time soon.
Actually...it's the Bounce GI implementation in DS that takes an eternity to render. I've noticed that all the UberEnvironment settings are a lot slower rendering in DS than in the standalone 3Delight...and I mean A LOT slower. And I'm not entirely sure why that is...
Actually...it's the Bounce GI implementation in DS that takes an eternity to render. I've noticed that all the UberEnvironment settings are a lot slower rendering in DS than in the standalone 3Delight...and I mean A LOT slower. And I'm not entirely sure why that is...
I ended up cancelling the Bounce GI render. After 18 hours, it had only completed a fraction of the total image, and if I'm honest it didn't seem to look much different. I'm sure it has its uses, but it's crazy slow to render so I might have to fake bounce lighting in future.
AMEN! ;) Well said, Holly!
CG software is just a tool for the artist.
AMEN! ;) Well said, Holly!
CG software is just a tool for the artist.
There is no Church. As I said before, its just a bit of nerd speak and it has nothing to do with whether the image looks realistic or not. Did you read the pdf?
http://multires.caltech.edu/~keenan/bias.pdf
The idea of a religion is exactly why I joined the thread, because there is a technical meaning to the term Unbiased that most people in these discussions are not aware of. Unbiased doesnt mean perfection. And the look of a biased vs unbiased render could be visually extremely similar. This is not an artistic argument, this is a mathematicians argument.
Please, review the pdf and let me know what you think.
Thank you for posting the link. It makes for some interesting reading.
Hey, Rashad, it's all good. :) I've enjoyed reading your posts as well. In fact I've got that pdf downloaded and ready to read, but I've been a bit busy with other things, so I'll get back to you.
I think HeraldOfFire has demonstrated quite well in his images that it is more about the artist than the tools. The tools can produce variations in results, but that can be a good thing too.
Certainly. Take your time. I just wanted to make it clear that I'm not pro unbiased in any way. I don't mean to give the impression that the car can drive itself. Just that when the task is stark accuracy, then unbiased has the clear advantage. To me there is a clear difference between "convincing" and "accurate." I consider myself a build your own GI kind of guy. I have some of the most elaborate and effective lighting setups in Bryce developed from years of testing. Most everything I do is with fake GI even the render included below. I think the render below is fairly convincing, it has geometric complexity, and challenging materials. I'm happy with it for now especially for a fake GI. Still, I'm sure it's far from accurate. Admittedly though, Bryce 7.1 Pro has some seriously sweet new lighting powers, the image below would not have been possible in previous versions. Anyhow, I ramble.
Generally, I don't think Unbiased rendering lends itself much to art at all in the paintbrush sense. Unbiased rendering seems to provide a sort of "realm" where you can render near perfect realism, and not much else. A virtual photo studio. It takes the guess work out of it and makes less talented individuals competitive with more gifted artists working from scratch. Unbiased levels the playing field a bit, making professional results easier to achieve. Put it this way, though it is always about the artist, with unbiased rendering much individuality is taken away from the artist because the software is doing so much for you and doing it in its own internally consistent way. Most everything looks the same.
http://www.luxrender.net/wiki/LuxRender_Volumes
Ugh...you had to get difficult didn't you? Lol, just teasing. The problem with diamonds is not so much the shaders or light but the actual mesh that is used. Of the free meshes out there this was the best of the bunch I could find. This took about 3 minutes to render in Daz Studio. I was attempting to use inderect lighting via shader mixer but it took 7 hours to get to 50% and looked like crap. I might look into RIB rendering in future for real fancy stuff.
Anyhow, one order of diamonds coming up...
Ugh...you had to get difficult didn't you? Lol, just teasing. The problem with diamonds is not so much the shaders or light but the actual mesh that is used. Of the free meshes out there this was the best of the bunch I could find. This took about 3 minutes to render in Daz Studio. I was attempting to use inderect lighting via shader mixer but it took 7 hours to get to 50% and looked like crap. I might look into RIB rendering in future for real fancy stuff.
Anyhow, one order of diamonds coming up...
Point me to that Mesh Harry. I'll give it a go if you do not mind.
Sure mate. It's a free mesh over at TurboSquid so you'll have to register an account if you don't have one already but it's really quick.
http://www.turbosquid.com/3d-models/free-perfect-model/675252
I ended up cancelling the Bounce GI render. After 18 hours, it had only completed a fraction of the total image, and if I'm honest it didn't seem to look much different. I'm sure it has its uses, but it's crazy slow to render so I might have to fake bounce lighting in future.
Be sure to set the 'max ray bounces' to 1 in the advanced render settings. Anything more than that is asking for super long render times... unless of course you really need it, but most of the time you don't.
Sure mate. It's a free mesh over at TurboSquid so you'll have to register an account if you don't have one already but it's really quick.
http://www.turbosquid.com/3d-models/free-perfect-model/675252
here is a good source for gems
http://www.3dlapidary.com/Index.htm
here is a good source for gems
http://www.3dlapidary.com/Index.htm
Thanks for the link!
...this. Well said.
I often strive for 'realism' but it's with the sole purpose of being able to sell you on something else. I want the skin on my elfmaid to look believable so you'll buy into the elfmaid. I light the dragon or the alien spaceship as realistically as I can so that you'll skip right past the lighting and buy into the 'reality' of the fantastic scene I'm presenting to you.
Actually, I have to admit I find the opposite is true. Hear me out.
When working with unbiased render engines, you need to know some real-world physics to get started. Since, as you so eloquently put, they try to react in a more 'real' manner, the materials used naturally have to be modeled on the real world items that they're based on. The problem with that is that there are literally thousands upon thousands of different materials in the real world, and no engine will have shader models designed for every purpose.
Take Reality for an example. There are literally only 7 material types, 8 if you include volumetric fog available only via the prop. Luxrender does support more material types if you edit the file externally, but Reality is the interface the majority of us will know and love, and therefore be most comfortable using. But, is that surface to be considered glossy? Is it matte? Translucent? Trying to model a real world material based on such seemingly limited parameters is a far more difficult task than, say, using a purpose built shader made to emulate cloth or stone in 3Delight.
I used Reality as an example for another reason. It has an auto-conversion which takes Daz's shader materials and tries its best to convert them into a Luxrender format. Unfortunately, for the majority of surfaces (read: virtually every surface it converts) it will become a glossy surface. In many cases with a very high sheen as well. Try to do a simple render out of the box, and you might find that your stone ruin has become a glistening shiny monument.
Subsurface scatter is a wonderful example of how things can really get complicated. Doing it using 3Delight or Firefly is as simple as applying the shader and hitting the render button. Matching it in Lux requires a fairly in-depth knowledge of your texture colours, and the maths involved in calculating the refracted light. And that's before I get into the problem of having the right type of mesh for the calculations.
What I suppose I'm saying is that render engines like 3Delight and Firefly tend to have a 'catch-all' solution for the majority of surfaces you'll require. Whether that's UberSurface or PoserSurface, it encompasses virtually every material you could require by including every desirable aspect. Glossiness, opacity, diffuse, specular. It's all wrapped in a neat bundle. You don't find many unbiased engines which have that degree of control. Moving from one to the other can be jarring at best, and at worst utterly confounding.