Show Us Your Bryce Renders! Part 3

1272830323350

Comments

  • bighbigh Posts: 8,147
    edited December 1969

    thanks guys - enjoying all the renders

  • Analog-X64Analog-X64 Posts: 110
    edited December 1969

    Another practice in bryce boolean.

    Orb_Sphere2.jpg
    1920 x 1080 - 383K
  • RarethRareth Posts: 1,462
    edited December 1969

    chohole said:
    OK maybe a silly question, and you will all look askance at me for having to ask, but is the Bryce render engine a biased or unbiased render engine. :question:

    this is about the thread over in Commons isn't it? the one regarding biased vs unbiased renders, lux vs octace, vs carrara, vs DAZ, (and I threw Bryce into the ring...)

  • Rashad CarterRashad Carter Posts: 1,803
    edited December 1969

    chohole said:
    OK maybe a silly question, and you will all look askance at me for having to ask, but is the Bryce render engine a biased or unbiased render engine. :question:

    Excellent question. As David already answered the Bryce engine is most certainly biased. In fact most all render engines are biased unless they specifically are designed to be otherwise.

    I've been studying this in detail for several years now and have a good basic understanding. I will try to explain it clearly as possible.

    In CG light is treated as a set of "pathways." Pathways can be literal angles of traveling light, but they can also represent the reasoning behind a given lighting effect.

    I want to be clear because lots of people think that the look of the final image result is what determines whether a render was biased or not. Just because a render is convincing doesnt mean it is unbiased...not at all! It has less to do with the final look and much more to do with the mathematical approach being taken at the start. It's all in the mathematics.

    Unbiased render engines are built to model light and surfaces within a scene to behave exactly as light and surfaces would behave in real life. In real life light is being absorbed, emitted, reflected, and transmitted by all of the surfaces in a scene. There are simple equations that determine how light behaves in nature, and how and why light is always moving, it never stops. For this reason unbiased renders are almost always much slower than biased renders. More on that in a second.

    Now lets consider radiosity and why radiosity is itself a biased process even though it produces good results. Radiosity is a good thing, based on the ideal of black body radiation, that when a black colorless body gets warm enough thermally it will begin to radiate red colored light and glow. But with surfaces that are not fully black but in some way colored even if they are white or silvery, these surfaces will not absorb all light but will instead "reflect" some of that light onto other nearby surfaces. Radiosity is very good at predicting the amount of color bleed or energy transfer from one object to another. If you have a bright white light in a white room and add a red colored ball you should see some of that redness "reflected" by the white floor underneath the ball. But radiosity is only one of many possible "light pathways" or "reasons for receiving light." Radiosity is by definition a diffused inter-reflection. That means that all specular-reflection considerations are omitted. That equals a form of bias, as steps are being skipped for speed purposes. Specular bounces are very subtle and for that reason are easy to omit without viewers noticing.

    So it is usually in the way of indirect lighting that biases if present make themselves most evident. Bias simply means that shortcuts are taken. Unbiased means that no short cuts are taken. What would a shortcut represent? Well in Bryce for example, light does not bounce from surface to surface unless TA is enabled. So enabling TA already reduces a lot of bias from a Bryce render. But TA has many hidden shortcuts in the way it handles bounced light...shortcuts save on render time...which is why only the newest applications based on modern processors can handle unbiased rendering. Bryce was built long before the modern processors, so you can bet it is highly biased to save time even in TA mode. Example: Old TA from bryce 6 had a faulty scattering algorithm, causing rays to be fired along preferred angles only instead of along all angles fairly. The new Scattering Correction tool in Bryce 7 solves that problem removing a visually obvious and distracting form of bias. There are other biases more deeply hidden, so there is still a long way to go before Bryce can be considered unbiased.

    Unbiased render engines do certain things for you whether you want them to or not. Here is a quick list of things Bryce would do automatically if it were unbiased:

    1. Caustics: Caustics are those strings and arcs of light created when a bright light has its direction changed sharply by either a curved transparent surface or a curved shiny object usually metallic. You need caustics for realistic water, lewelry, and many other things. Bryce doesnt do caustics properly, it is better under TA but still not right.

    2. Prismatic banding: Unbiased render engines process light along three separate channels individually. One for Red/Green/ and Blue. Unbiased render engines refract each channel independently, allowing for a natural splitting of white light into bands of color to occur when striking a diamond or other transparent surface. We can fake something similar in Bryce by creating white lights by combining a red. green, and a blue point light source of equal strength. Still the result though better is still not physically accurate.
    :
    3. SSS: Sub-surface scattering occurs when light enters a medium from one angle, is absorbed into the material where it is refracted several times before it then exits the medium traveling in some random direction. Marble, glasses of milk, human skin all need SSS to look right.

    There are other things I could mention as well but the point isnt to show shortcomings of Bryce. Carrara has an excellent render engine as well, but its indirect light tool doesnt account for specular bounces, doesnt do caustic automatically, doesnt do prismatic banding at all, and takes other shortcuts as well many of them very subtle and not noticeable by the human eye in most situations. The Daz Studio render is highly biased as well, but it works with a well developed biased Renderman compliant platform insofar as the cheats are well hidden and not always obvious.

    Unbiased render engines are Kerkythea, LuxRender, Octane (by far the best of all) and Blender has a new unbiased engine called Cycles.

    Unbiased render engines are more rigid. There are no cheats like ambient channel glow sliders to "fix" the shadows, as there is no real life equivalent to the ambient channel slider. Such a glow would shine onto and affect nearby objects, not just fill in the shadows of the target object. Real light doesnt work the way Bryce, Carrara, or Daz Studio can treat it. Yet, you would find that in most unbiased render situations that you wouldn't need to add ambience anyhow because the bouncing of the key light from the walls and other surfaces of the scene have already filled in those shadows for you the correct amount.

    Hope I've clarified it a bit more.

  • David BrinnenDavid Brinnen Posts: 3,136
    edited December 1969

    Ideally what's you want is the option of using unbiased rendering, but you don't want to be lumbered with it. As I see it, rendering is all about having the control to choose what you want, not to be hemmed in my artificial limitations brought about by the choices of others with a different vision of what you are going to use the software for.

    Face_up4.jpg
    1700 x 850 - 1M
  • Analog-X64Analog-X64 Posts: 110
    edited December 1969

    Holiday Monday in Canada :) Time for a new image upload.

    Orb_Sphere3.jpg
    1920 x 1080 - 735K
  • HoroHoro Posts: 10,646
    edited December 1969

    @Rashad - thank you for your elaborate explanation about biased render engines. I'm with David here: let the artist chose what level of naturalism he/she needs. Obviously, an abstract doesn't have the same requirements as a photo realistic render. A pure landscape render is again different from a close-up of human skin. Though there is much to be desired in Bryce, we do have a lot of options to tailor the render engine to our needs - speed vs accuracy is mostly what it comes down to.

  • mx42bmx42b Posts: 25
    edited December 1969

    The Song is Over...
    Take 2
    Fixed the too glowy-ness of the vase and adjusted its shape somewhat.(thank you Mr. Brinnen)
    Alterred the intensity and specularity of the lights, added a small amount of reflection (between 1.6 and 3.3 depending on body part) and increased the specularity of each of her skin meshes.(thank you Mr. Horo)
    also made a few changes to some off camera elements (ceiling ambiance and reflectivity, panelling on far walls, shoved the moon over a bit, that sort of thing)
    This is as good as she's going to get. Not because I can't tweak anything more but because I'm tired of looking at the poor girl. After all, it's not my fault that her best friend/lead vocalist/girlfriend of four years chose to run off with the replacement drummer they picked up three weeks ago (after their regular drummer showed up for the performance so drunk he fell off the stage and broke his hip). Nope, not my fault.
    Anyway, Hope you like this version.

    thesngisovrfff.png
    836 x 622 - 1M
  • HoroHoro Posts: 10,646
    edited December 1969

    I've downloaded both pictures and compared them side by side. The difference is huge. All changes you made helped improve the result. Oh yes, we can always tweak here and there but it is a question of effort and gain. I really do like how this render has improved.

  • HoroHoro Posts: 10,646
    edited December 1969

    I planned to resume working on the HDRIs I acquired two days ago, but couldn't help experimenting with the extreme wide-angle lens.

    ST.jpg
    1000 x 1000 - 196K
  • David BrinnenDavid Brinnen Posts: 3,136
    edited December 1969

    Horo said:
    I planned to resume working on the HDRIs I acquired two days ago, but couldn't help experimenting with the extreme wide-angle lens.

    Very interesting indeed! I'm trying to figure out what I'm looking at... or even if I am looking up or down.

    I too have been playing with the lenses.

    GWL_test_scene5.jpg
    700 x 700 - 306K
    GWL_test_scene4.jpg
    1250 x 500 - 426K
  • HoroHoro Posts: 10,646
    edited December 1969

    Looking good, David. The tunnel view, I like how the side walls get bent like on the inside of a barrel. This is another example.

    Stu.jpg
    1000 x 1000 - 124K
  • mermaid010mermaid010 Posts: 5,494
    edited December 1969

    Lovely work by everyone. Analog’s abstracts and Bob’s etching renders are very nicely done.

    Thanks David --- more tutorials to work with.

    Horo and David – very interesting results with the wide angle lens. Horo the tunnel view is awesome.

  • RarethRareth Posts: 1,462
    edited December 1969

    still tweaking this as to my eye its not quite where I want it to be.. but the cloud slab pushes the render times into 5+ hours

    someplace-cold2.jpg
    1404 x 700 - 253K
  • David BrinnenDavid Brinnen Posts: 3,136
    edited December 1969

    Looking good! Don't forget Rareth, you can always "hide" the cloud slab until you are happy with your lighting setup and just employ that for the final render.

    Here's some more testing of wide angle lenses and filters...

    EWL_test_scene5.jpg
    1400 x 700 - 622K
  • Dave SavageDave Savage Posts: 2,433
    edited December 1969

    Rareth said:
    still tweaking this as to my eye its not quite where I want it to be.. but the cloud slab pushes the render times into 5+ hours

    As David says, you can always hide the volumetric slab.
    What I usually do is to render the sky separately at a much lower RPP because clouds rarely suffer from noise.
    Then hide the slab, ramp up the RPP and render again. Then render an object mask to comp the two renders together in a photo editor.

    It can save hours of render time.

  • HoroHoro Posts: 10,646
    edited December 1969

    Beautiful image, David. I thought it's a photograph if it were not for the filter, which I know quite well. Excellent example.

  • JamahoneyJamahoney Posts: 1,791
    edited December 1969

    Incredible wide-field images through this Horo/David camera - what would an actual 'figure/character' look like in these...too distorted, too unrecognisable???

    Jay

  • Dan WhitesideDan Whiteside Posts: 500
    edited December 1969

    Excellent David, the illumination is spot on...

  • David BrinnenDavid Brinnen Posts: 3,136
    edited December 1969

    Thank you Dan, Jay, Horo! I will add this one in to the examples, but I'll need to document some of the modifications I've made to get the vignetting effect to work around such a wide angle.

    As for characters, people, yes it works, but obviously with exaggerated proportions the results are stylised.

  • mermaid010mermaid010 Posts: 5,494
    edited December 1969
    space-abstractjpg.jpg
    960 x 720 - 101K
  • David BrinnenDavid Brinnen Posts: 3,136
    edited December 1969

    Worked a treat! Or should that be warped a treat? The fact that Bryce cannot inherently render a 360 x 180 spherical or angularly mapped HDR scene natively need not be a bar to exporting in hdr format and importing it back in and using it as a source of backdrop and light in your render. Good going! Thanks for showing us your render.

  • JamahoneyJamahoney Posts: 1,791
    edited December 1969

    Yeah...very nice, Mermaid...must give David's tut a try, too...later on today.

    Jay

  • mermaid010mermaid010 Posts: 5,494
    edited December 1969

    Thanks David and Jay.

    I enjoy doing your tutorials David, although a 10min one takes me 2-3 hours to do. But for me its time well spent. Thanks for the cool tutorials.

  • AtlantisAtlantis Posts: 133
    edited December 1969

    Bad times for Bigfoot...

    Experimented with David brinnen's street light's tutorial to get forest fire effect

    fire.jpg
    1200 x 575 - 91K
  • RarethRareth Posts: 1,462
    edited December 1969

    after following one of Davids "Beginner" tutorials... I have produced the following.

    Lake-scene-Bryce3.jpg
    1754 x 875 - 1M
  • HoroHoro Posts: 10,646
    edited December 1969

    @@tlantis - looks very convincing, Erich. Excellent!

    @Rareth - very nice. What is that dark area in the water at lower right?

  • RarethRareth Posts: 1,462
    edited February 2013

    Horo said:
    @@tlantis - looks very convincing, Erich. Excellent!

    @Rareth - very nice. What is that dark area in the water at lower right?

    shadow coming off the terrain off camera to the right, the sun is low to the right

    hmmm I should probably tweak the camera a bit so thats not there...

    Post edited by Rareth on
  • RarethRareth Posts: 1,462
    edited December 1969

    ok camera tweaked..

    In restrospect I could have just cropped the picture..

    bryce-landscape-2.jpg
    1754 x 875 - 1M
  • HoroHoro Posts: 10,646
    edited December 1969

    Looks much better. That shadow somehow spoiled the image.

This discussion has been closed.