3Delight Laboratory Thread: tips, questions, experiments

11819212324100

Comments

  • Mustakettu85Mustakettu85 Posts: 2,933
    edited June 2015

    mjc1016 said:
    There seems to be something funky going on with your box scene, Takeo...

    Is it one you built or a downloaded one? If built, how?

    The main 'box' has normals pointing 'out'.

    Thanks for checking!

    So the glass shader loaded fine for you? Weird, because it didn't recognise itself on this office machine of mine.

    I also noticed Takeo set "FixNormals" to "on" on every surface.
    A word of caution to everyone (this will be in the docs): this is not a "magical fix for all bad geometry". It's a last resort fix, only meant to be used for one-sided polys you can't avoid seeing from the wrong side or for surfaces that can't be fixed in your modeling app or when it is unfeasible to invert normals manually in Geometry Editor: generally, for dynamic cloth that will often partially intersect itself when draped, or for clothing with heavy smoothing modifier when normals can get twisted on some details.

    Normally, you don't need this. It's only there for a few specific cases, and even then, it's not likely to give 120% perfect results because, well, bad geometry is bad geometry.

    PS We are used to all the shaders doing a general and generous faceforward() flip in DAZ Studio, but it's not good practice actually. Moritz Moeller warned against using it a long time ago already.

    Post edited by Mustakettu85 on
  • Mustakettu85Mustakettu85 Posts: 2,933
    edited June 2015

    And the difference between UE2 "Indirect Light" and "Bounce Light" modes:

    I got them mixed a bit in my mind in my latest posts (stupid headache!).

    The Bounce Light mode will need its colour set to white to actually work.

    The indirect Light mode will need to have its colour set to black to act as a pure GI light without extra light being poured from outside the max trace distance.
    Any colour in the IDL mode is the "environment" colour, even if there is no map. It's like using a solid colour map.

    I think this is important to remember.

    Post edited by Mustakettu85 on
  • Mustakettu85Mustakettu85 Posts: 2,933
    edited December 1969

    Rogerbee said:

    Don't know, but, I don't really plan on returning to that render now anyway. As I said before, the backlighting was to bring out the backscatter in Bjorn's ears and I just changed the figures and background colour and not the lights for subsequent renders.

    I personally find backlighting to be artistically very important. Even in outdoor scenes, a subtle backlight can really help.

  • Takeo.KenseiTakeo.Kensei Posts: 1,303
    edited December 1969

    mjc1016 said:
    There seems to be something funky going on with your box scene, Takeo...

    Is it one you built or a downloaded one? If built, how?

    The main 'box' has normals pointing 'out'.

    This first image is showing it rendering as it loads directly from the scene file. The second is showing as it renders after flipping the normals with Geometry Editor. I'm rendering without caustics, because I'm having a crashing problem when trying to render any refractive caustics (don't ask, it's driving me nuts, because I can't figure out any possible causes...reflective caustics work with no problems). There are no other changes between the two.

    Yep made it inside DS and didn't want to fire up blender to invert the normals and I used the "fix normals" in the shader.

    Didn't think there would be no faceforward() in the light shader

    Works better once I inverted the normals


    Try it for yourself. Put a genesis with your shader inside a sponza atrium or something in the likes

    I'm sorry, but you aren't being very helpful. You see, I did render my presets in interior scenes. Published examples... One of them is the 'cover' of the ODT doc file you should have.
    This is another: http://mustakettu85.deviantart.com/art/Why-488890119

    Could you please point out what is wrong? Or what could possibly be wrong? What should I be looking for?

    Sorry but I thought that anyone would see what I saw easily.

    Subsurface need a lot of light for the effect. You'll get weird things with a 100% SSS unless you always put a lot of light on the character like you all seem to do. Example below with the scene from far and then a close up. One with your preset on which I used the "replace SSS color" in order to get the skin less dark already. Second close up is bree standard preset
    There is one Physdistant from the left and DelightGI.


    Look at the ceiling. It gets illuminated by bounced light inside the box and that is the goal. If you get it to work, you know your GI is good

    Quadratic or not the conclusion is the same. No light bounced off the walls

    Last note : I also have a GI light with trace () and I do have bounce light like with UE.

    Again: if you actually paid attention to the renders in my previous post, you'd have seen I am not getting any light on the ceiling with UE2 in bounce mode and the light colour set to black.
    Correct me if I'm wrong, but even a solid colour in UE2 light colour swatch will kick in as environment color when the max trace is reached.
    Is that not so?

    You can see that DelightGI handles the bounce onto the primitives inside the box well.
    The only issue is the ceiling. Which is strange. It's as if the normals are reversed, but wouldn't UE2 complain then in any case as well?

    If you truly want to help me, then could you please compare the source code for your trace()-based light with that of DelightGI and tell me what I might be doing wrong? It shouldn't take much time, the shader is tiny.

    As you mentionned a faceforward. would have worked.Don't know what Moritz Moeller said and in which context but you don't always have a good geometry and not everybody has the know how to correct them (I'm thinking about old DS or Poser stuff). Your choice


    Now going on with reorganisation.
    About the diffuse map. Why not a switch ON/OFF "Use diffuse as diffuse SSS map" instead?
    And also put the two SSS effect one after the other?

    Effect_order.JPG
    254 x 424 - 26K
    FROMCLOSEWITHDEFAULTBREE.jpg
    691 x 691 - 335K
    FROMCLOSE.jpg
    691 x 691 - 345K
    FROMFAR.jpg
    691 x 691 - 506K
  • RogerbeeRogerbee Posts: 4,460
    edited December 1969

    I personally find backlighting to be artistically very important. Even in outdoor scenes, a subtle backlight can really help.

    I suppose that's fine if you want it to be artistic, but we don't have people walking behind us with backlights in real life.

    CHEERS!

  • RogerbeeRogerbee Posts: 4,460
    edited December 1969

    I decided to try one of Mec4D's textures with RSV and Byson came to hand. What I found interesting about morphing this one was that, in his default state, Darius is 100% him and 100% M6, you'd think Darius would just be him and not a variant of M6.

    CHEERS!

    Byson_RTK.jpg
    577 x 750 - 151K
  • mjc1016mjc1016 Posts: 15,001
    edited December 1969


    Now going on with reorganisation.
    About the diffuse map. Why not a switch ON/OFF "Use diffuse as diffuse SSS map" instead?
    And also put the two SSS effect one after the other?

    That may be easier said than done...although with ShaderBuilder it is easier to do. DS does have a habit of rearranging things like that...

    Although I've found with SB, you can pretty much end up with predictable results if you pay attention to groups AND rearrange the parameters manually in them. Makes for some nasty looking spaghetti on the connections, but they tend to stay put (hopefully).

  • Takeo.KenseiTakeo.Kensei Posts: 1,303
    edited December 1969

    mjc1016 said:

    Now going on with reorganisation.
    About the diffuse map. Why not a switch ON/OFF "Use diffuse as diffuse SSS map" instead?
    And also put the two SSS effect one after the other?

    That may be easier said than done...although with ShaderBuilder it is easier to do. DS does have a habit of rearranging things like that...

    Although I've found with SB, you can pretty much end up with predictable results if you pay attention to groups AND rearrange the parameters manually in them. Makes for some nasty looking spaghetti on the connections, but they tend to stay put (hopefully).

    A text editor works pretty well for that. No headache

  • mjc1016mjc1016 Posts: 15,001
    edited June 2015

    I experienced something very weird tonight...

    I've been trying to do a render that just was not co-operating. After suffering through some extremely long render times (2+ hrs), I looked at the geometry...it had some odd normals and other weirdness. After I fixed the geometry and reapplied the shaders, I got this from the standalone in about 5 minutes...and the same thing in the Studio included took about 15 minutes. I've never had geometry so affect the render time (one render, earlier and smaller sized, was only at 50% in that 2 hrs...so it was cancelled).

    One of the other problems with the model is crappy UV mapping...so no decent tires. It would probably be easier to remodel them instead of trying to get decent UV maps out of what's there.

    mustang2_11.png
    1280 x 720 - 1M
    Post edited by mjc1016 on
  • RogerbeeRogerbee Posts: 4,460
    edited December 1969

    Is that a Daz Mustang? If it is then they are ages old and you really can't expect much from them by today's standards. DS1 was probably what it was built for.

    CHEERS!

  • Mustakettu85Mustakettu85 Posts: 2,933
    edited December 1969


    Didn't think there would be no faceforward() in the light shader

    Faceforward() in lights can do weird things to RT SSS. I've tried using a "ShadingNormal" type of function for DelightGI long ago, and I decided against it. Moritz warned against it in general, but in particular because SSS would never be really correct and predictable with faceforward(). Even in the 'older' days. It's just true. Even AoA IIRC had a anti-faceforward fix in his lights for his shader network.

    Transparent surfaces will also show edge artefacts with faceforward(), there were examples in some SIGGRAPH courses. It's a fairly old convention for good coding, staying as far from it as possible.

    you don't always have a good geometry and not everybody has the know how to correct them (I'm thinking about old DS or Poser stuff).

    Half the reason why I'm ever investing my personal time in sharing stuff with the community is because I believe in free education for everyone. Hobbyists included.
    Since the Geometry Editor arrived, normal fixing has become trivial. Everyone can do it once shown the steps. It will be a yet another how-to in the docs.

    Subsurface need a lot of light for the effect. You'll get weird things with a 100% SSS unless you always put a lot of light on the character like you all seem to do. Example below with the scene from far and then a close up.


    That's actually the first time I see this with GI. Looks like severe undersampling from a multifaceted area light.
    Are you sure you posted a finished render, not the ray cache preview? Is the box the one with the normals fixed already? Are you using an IBL map to make sure the scene doesn't open into vacuum?

    If you look at the dA post I linked to and read the description, you will see the girl is lit by IDL. The PhysicalSun delta is coming from behind her. Same with the G2F and cat in the hallway (I also should have a couple more test angles in my folder of the G2F up close against the window = in IDL; and other low-light tests on my PC).

    It's not lack of GI light intensity but something else.

    RT SSS will exhibit more noise with non-delta lights, GI/IBL included, but not generally that high with my default sampling values.


    Now going on with reorganisation.
    About the diffuse map. Why not a switch ON/OFF "Use diffuse as diffuse SSS map" instead?
    And also put the two SSS effect one after the other?

    That may be easier said than done...although with ShaderBuilder it is easier to do. DS does have a habit of rearranging things like that...

    Although I've found with SB, you can pretty much end up with predictable results if you pay attention to groups AND rearrange the parameters manually in them. Makes for some nasty looking spaghetti on the connections, but they tend to stay put (hopefully).

    A text editor works pretty well for that. No headache

    I do rearrange manually, Mjc, it is not a 100% save against shuffling either.

    Takeo, I'd prefer to keep installation via shader builder files only, no custom param def scripts. Even then, again, DS can shuffle channels.

    Now, why backscatter so deep down: because it's a cheat, really. A more-physically-based-than-most one, but a very specialised trick nonetheless. Only for those artists who really need it in very specific scenes. It has a long list of non-intuitive params.
    You clamour for pulling diffuse up, and then you want this to get in the way? One of the two least generally useful features?

    As for the switch: I'll need to think about and weigh the usability concerns with my less-tech-oriented focus group.

  • Mustakettu85Mustakettu85 Posts: 2,933
    edited December 1969

    Rogerbee said:

    I suppose that's fine if you want it to be artistic, but we don't have people walking behind us with backlights in real life.

    Unless it's a professional photoshoot, where you will have folks setting up reflection cards and even some extra artificial lights around the model, even outside. It's one of the reasons the city street 'magazine photos' look different from your generic 'fashion blogger' backyard selfies.

  • Mustakettu85Mustakettu85 Posts: 2,933
    edited December 1969

    mjc1016 said:

    I've never had geometry so affect the render time (one render, earlier and smaller sized, was only at 50% in that 2 hrs...so it was cancelled).

    I guess the way the renderer code is optimised today can get thrown off by bad ( = 'anti-physical') normals.

  • Takeo.KenseiTakeo.Kensei Posts: 1,303
    edited December 1969

    That's actually the first time I see this with GI. Looks like severe undersampling from a multifaceted area light. There is no arealight


    Are you sure you posted a finished render, not the ray cache preview? Not the cache preview and not a finished render. I don't see the need to render further at this point.
    Is the box the one with the normals fixed already? Yes you see the light bounce off the walls Are you using an IBL map to make sure the scene doesn't open into vacuum? No but I don't see why you'd need an IBL. Adding an IBL would add an additional light source. That is not what I want . Just IDL and that should work like it is working with the default G2F preset. No cheating please

    If you look at the dA post I linked to and read the description, you will see the girl is lit by IDL. The PhysicalSun delta is coming from behind her. Same with the G2F and cat in the hallway (I also should have a couple more test angles in my folder of the G2F up close against the window = in IDL; and other low-light tests on my PC).

    It's not lack of GI light intensity but something else.

    RT SSS will exhibit more noise with non-delta lights, GI/IBL included, but not generally that high with my default sampling values.

    Stop looking at the noise. That is not what I'm showing. It's about the shading especially in the part in shadow which btw is less noisy
    The scene is pretty simple to reproduce. Try it
    I may be one of the few who would be running on these case who knows. And after all that is just a preset which can be changed.You can also ignore what I'm showing as it should be easy to get a workaround by adding lots of lights or adding some diffuse. At the end, it's your choice. It just tickles me because your startup preset is not physically correct

    I do rearrange manually, Mjc, it is not a 100% save against shuffling either. I never got any shuffling issue. Am I lucky?

    Takeo, I'd prefer to keep installation via shader builder files only, no custom param def scripts. Even then, again, DS can shuffle channels.

    Now, why backscatter so deep down: because it's a cheat, really. A more-physically-based-than-most one, but a very specialised trick nonetheless. Only for those artists who really need it in very specific scenes. It has a long list of non-intuitive params.
    You clamour for pulling diffuse up, and then you want this to get in the way? One of the two least generally useful features?

    As for the switch: I'll need to think about and weigh the usability concerns with my less-tech-oriented focus group.

    It seems logical to me to pack related parameters together. It's my opinion. At the end it's you to choose. You know I can do all these (mostly done already in fact).

  • mjc1016mjc1016 Posts: 15,001
    edited December 1969

    mjc1016 said:

    I've never had geometry so affect the render time (one render, earlier and smaller sized, was only at 50% in that 2 hrs...so it was cancelled).

    I guess the way the renderer code is optimised today can get thrown off by bad ( = 'anti-physical') normals.

    It was coming out translucent in parts, black in others and all sorts of weirdness. Also, every surface shader behaved differently, but none of them were completely 'correct'...the results were always 'off' somehow. And this was a case where the GeoEditor didn't help. It just flips their current direction. I had to use Blender and repeated 'recalculate' runs to get them all in same direction, and even after that I had one door that was backwards from everything else. I'm thinking that it was constructed as half a model and the other half was 'mirrored'...and in the original application, that was fine, but the conversion to obj didn't 'make real' the mirrored parts correctly. I still haven't figured out why the tremendous time difference...

    No, Roger, that's from Turbosquid...which even makes it more surprising that it was such messed up normals. But I think it was automatically converted from max file, through 'squid's converter, instead of saved out from Max as an obj. Overall, it's got nice clean geometry and is rather high poly.

    If I was thinking I probably should have exported without writing the normals and then tried it.

    Also, I discovered a very nice trick. If you are doing glass...like window glass and it is giving odd results, especially with a 'real' glass shader...like the Kettu's, make sure the glass has actual thickness. It's especially true for PBRs and plausible shaders, but is applicable to just about any shader/renderer that isn't using transparency to mimic glass.

  • Mustakettu85Mustakettu85 Posts: 2,933
    edited December 1969

    Not the cache preview and not a finished render. I don't see the need to render further at this point.
    Stop looking at the noise. That is not what I'm showing. It's about the shading especially in the part in shadow which btw is less noisy

    Sorry, Takeo, I don't understand where to look for wrong stuff on a render that noisy. My eyes only see the noise, and my brain refuses to process anything else.

    Here's a render in which I tried to reproduce your scene. It's the G2F preset (with only transmapped new brows overlaid) sitting in a 4m diameter DS primitive sphere out of which one poly has been deleted to make a window. There is my ground prop, a DelightGI and a SimplePhys distant.

    I did kill Edge Strength in the preset because the new ones will come with it off by default.

    The pink background is with normals flipped (two versions: SSS only and diffuse only). The black background is SSS-only in an unflipped sphere ("sitting in the void"), for reference to see only the light the SSS-only version bounces on herself.

    Flipping the SSS vs diffuse back and forth, I basically don't see any meaningful difference. There is less interbounce in the ear, but I think it's logical because SSS does not reflect all the light straight back, it absorbs it and reemits (=scatters), so more energy will get "eaten".

    We need a photo reference to clear out the ear thing.

    If there is something I don't see, could you please draw arrows on the image to show the wrong places?

    No but I don't see why you'd need an IBL. Adding an IBL would add an additional light source. That is not what I want . Just IDL and that should work like it is working with the default G2F preset. No cheating please

    No cheating, on the contrary - mimicking the planet Earth. Unless there is a sphere around your box, it's opening into the void. Light energy is lost in the void, it does not return. The distant delta only comes from one infinite plane, the remaining directions of the shading hemisphere aren't covered.

    If you don't like the idea of a map, put a sphere around the scene, just make sure to include it into trace distance. Although the map should be faster.

    From the surface shader POV, the light trace() gives to it is all the same, map or bounce. Remember, no distinction between light paths.

    g2f_void.png
    700 x 800 - 470K
    g2f_pink_diffuse.png
    700 x 800 - 951K
    g2f_pink.png
    700 x 800 - 941K
  • Mustakettu85Mustakettu85 Posts: 2,933
    edited December 1969

    mjc1016 said:

    And this was a case where the GeoEditor didn't help. It just flips their current direction. I had to use Blender and repeated 'recalculate' runs to get them all in same direction, and even after that I had one door that was backwards from everything else. that's from Turbosquid...which even makes it more surprising that it was such messed up normals.

    I never encountered anything that bad in Poser/DS content. Random flipped normals in a few spots, in freebies mostly, that's that. Wasn't Poser very sensitive to wrong normals very early on?


    Make sure the glass has actual thickness. It's especially true for PBRs and plausible shaders, but is applicable to just about any shader/renderer that isn't using transparency to mimic glass.

    Hear, hear!

  • Mustakettu85Mustakettu85 Posts: 2,933
    edited December 1969

    PS Here's a diffuse-only version with a better Oren-Nayar roughness: 0.58, taken from here:

    http://www.yafaray.org/documentation/userguide/material

    Under the heading "diffuse reflection" there is an archive of the measured Oren-Nayar roughness values for various real-world stuff.

    As you can see, by increasing roughness the Oren-Nayar diffuse becomes more "even" - subtly darker in the better lit areas, but with more retroreflection along the edges.

    When very rough, an Oren-Nayar sphere will look like the Moon.

    g2f_pink_diffuse_roughness0-58.png
    700 x 800 - 940K
  • RogerbeeRogerbee Posts: 4,460
    edited December 1969

    Unless it's a professional photoshoot, where you will have folks setting up reflection cards and even some extra artificial lights around the model, even outside. It's one of the reasons the city street 'magazine photos' look different from your generic 'fashion blogger' backyard selfies.

    All well and good if that's the type of thing you're into. I'm more interested in it looking good to my eyes and not looking good through a camera.

    I tried out a V4 map and that glowing nostril effect still persists, even when you turn off that edge setting completely. Something else is awry and I'm not sure what. The texture is Farissa by Virtual World, it's one I really like as it comes with textures for different brow colours which makes it a bit more versatile. Once we have all the lighting and shaders worked out I can move onto what I know I'm good at which is creating characters. I got bogged down in testing lights and shaders and haven't done any for ages.

    CHEERS!

    V6_Farissa_RTK.jpg
    577 x 750 - 218K
  • mjc1016mjc1016 Posts: 15,001
    edited December 1969

    mjc1016 said:

    And this was a case where the GeoEditor didn't help. It just flips their current direction. I had to use Blender and repeated 'recalculate' runs to get them all in same direction, and even after that I had one door that was backwards from everything else. that's from Turbosquid...which even makes it more surprising that it was such messed up normals.

    I never encountered anything that bad in Poser/DS content. Random flipped normals in a few spots, in freebies mostly, that's that. Wasn't Poser very sensitive to wrong normals very early on?

    Yeah, I think it was.

    This was a Turbosquid model that was originally a Max or LWO one that was run through TS's autoconverter, by the guy that uploaded it. My guess is it was just a bad conversion. It really did look like it was 'half' there...like one fender was perfect, but the one on the other side was translucent, black or some of both. To get GeoEditor to work I probably would have had to do a poly by poly flip...and at almost 400K I was NOT about to do that. But it's all fixed now and in DS native...

    I like it because it has a pretty nice interior...no engine, though, so if I wanted I could make it so the doors open. I have a lot of other cars to convert/set up...I just hope there aren't many more that are that troublesome (I know that most of the Sketchup ones will be troublesome).

  • mjc1016mjc1016 Posts: 15,001
    edited December 1969

    Rogerbee said:

    Unless it's a professional photoshoot, where you will have folks setting up reflection cards and even some extra artificial lights around the model, even outside. It's one of the reasons the city street 'magazine photos' look different from your generic 'fashion blogger' backyard selfies.

    All well and good if that's the type of thing you're into. I'm more interested in it looking good to my eyes and not looking good through a camera.

    I tried out a V4 map and that glowing nostril effect still persists, even when you turn off that edge setting completely. Something else is awry and I'm not sure what. The texture is Farissa by Virtual World, it's one I really like as it comes with textures for different brow colours which makes it a bit more versatile. Once we have all the lighting and shaders worked out I can move onto what I know I'm good at which is creating characters. I got bogged down in testing lights and shaders and haven't done any for ages.

    CHEERS!

    Glowing nostrils on Gen4 has always been a problem. And I'm not sure it was all down to the mesh...

  • RogerbeeRogerbee Posts: 4,460
    edited December 1969

    It did it on V6 as well, there's something up with Kettu's shader, she might have said where but I can't remember.

    CHEERS!

  • mjc1016mjc1016 Posts: 15,001
    edited June 2015

    I'm going to paste those Oren-Nayar values here...there's more of you want to dig through the data from the O-N database...

    
    
        Brick_b: 0.275990
        Concrete_b: 0.308956
        Concrete_c: 0.461930
        Cotton: 0.482679
        Felt: 0.414686
        Human skin: 0.579386
        Leather: 0.179776
        Linen: 0.514593
        Pebbles: 0.443289
        Plaster_b: 0.543788
        Quarry tile: 0.360574
        Roof shingle: 0.819147
        Rough paper: 0.311376
        Rough plastic: 0.278057
        Rug_b: 0.613889
        Slate_b: 0.309590
        Sponge: 0.872413
        Stones: 1.107168
        Tree bark: 0.293226
        Velvet: 0.751002
        Wood_b: 0.351271
        Wool: 0.978133
    Post edited by mjc1016 on
  • Takeo.KenseiTakeo.Kensei Posts: 1,303
    edited June 2015

    I spot the problem

    In fact that is not the 100% SSS which causes the problem. It seems you have something wrong in the ambient

    Put a delta light and something that will make some shadows between Genesis and the light

    Rendered three pics

    First with a sphere that is making shadows on G2F. No GI. The part in shadow gets some specular somehow (top of the head for ex)

    Second render. Put a delta spot on the left of G2F. Nothing between genesis and the light.What is in shadow from the spot is completely black (not rendered but trust me). I then added a DELIGHTGI with no map. I get some IDL from the wall. That works. I played with shadow softness and shader hitmode and shadow sample. I kept the softness low because I want very sharp shadows. If you make them soft you may not see the problem

    Third render. I saw that G2F Top and short get colored in shadow and no specular. And I thought that must be the ambient channel. So I put 0% ambient on the top and bingo. It gets black.

    I didn't have a look at your code. There is no control of the ambient channel on your shader so there is no way to control that but there is something

    * Edit : added a screencap of the scene

    ** Second edit :

    No cheating, on the contrary - mimicking the planet Earth. Unless there is a sphere around your box, it's opening into the void. Light energy is lost in the void, it does not return. The distant delta only comes from one infinite plane, the remaining directions of the shading hemisphere aren't covered.

    If you don't like the idea of a map, put a sphere around the scene, just make sure to include it into trace distance. Although the map should be faster.

    From the surface shader POV, the light trace() gives to it is all the same, map or bounce. Remember, no distinction between light paths.

    Imagine, it's the night or you are in a cave with just a lamp torch. What is the environment color? Black. If you put a black map into your GI light it's the same effect as not putting any map. Try it

    *** Last edit : added an Iray render for reference. You see that you get some IDL from the wall and the floor that tint a bit G2F in the part in shadows.

    Iray_spotlight_reference.JPG
    711 x 898 - 82K
    directlightkettu.JPG
    686 x 847 - 58K
    G2F_DELTA_spot01_top_no_ambient.jpg
    699 x 846 - 119K
    G2F_DELTA_spot02_no_obstacle.jpg
    699 x 846 - 401K
    G2F_DELTA_spot01.jpg
    699 x 846 - 125K
    Post edited by Takeo.Kensei on
  • Takeo.KenseiTakeo.Kensei Posts: 1,303
    edited June 2015

    Other problem : the default glass transmission hitmode parameter and shadow color being inverted or something like that

    Create a scene with a plane as ground. Create a cube that will get the glass absorption shader and resize it to be thin enough to simulate a window glass. Add a PhysSpotlight and a camera to see what happens with shadows

    First pic : transmission hitmode = primitive and shadow color = black
    Second pic : transmission hitmode = shader and shadow color = white

    * Added a screencap of the scene

    glass_scene.JPG
    708 x 652 - 32K
    glass02.jpg
    699 x 846 - 44K
    Glass01.jpg
    699 x 846 - 29K
    Post edited by Takeo.Kensei on
  • RogerbeeRogerbee Posts: 4,460
    edited December 1969

    mjc1016 said:
    I'm going to paste those Oren-Nayar values here...there's more of you want to dig through the data from the O-N database...

    
    
        Brick_b: 0.275990
        Concrete_b: 0.308956
        Concrete_c: 0.461930
        Cotton: 0.482679
        Felt: 0.414686
        Human skin: 0.579386
        Leather: 0.179776
        Linen: 0.514593
        Pebbles: 0.443289
        Plaster_b: 0.543788
        Quarry tile: 0.360574
        Roof shingle: 0.819147
        Rough paper: 0.311376
        Rough plastic: 0.278057
        Rug_b: 0.613889
        Slate_b: 0.309590
        Sponge: 0.872413
        Stones: 1.107168
        Tree bark: 0.293226
        Velvet: 0.751002
        Wood_b: 0.351271
        Wool: 0.978133

    Interesting, so where on Kettu's shader would we input those?

    CHEERS!

  • wowiewowie Posts: 2,029
    edited December 1969

    mjc1016 said:
    I'm going to paste those Oren-Nayar values here...there's more of you want to dig through the data from the O-N database...

    
    
        Brick_b: 0.275990
        Concrete_b: 0.308956
        Concrete_c: 0.461930
        Cotton: 0.482679
        Felt: 0.414686
        Human skin: 0.579386
        Leather: 0.179776
        Linen: 0.514593
        Pebbles: 0.443289
        Plaster_b: 0.543788
        Quarry tile: 0.360574
        Roof shingle: 0.819147
        Rough paper: 0.311376
        Rough plastic: 0.278057
        Rug_b: 0.613889
        Slate_b: 0.309590
        Sponge: 0.872413
        Stones: 1.107168
        Tree bark: 0.293226
        Velvet: 0.751002
        Wood_b: 0.351271
        Wool: 0.978133

    What are those values taken with? Is that with energy conservation? Clamped or normalized? Both generally looks very different with the same roughness.

  • RogerbeeRogerbee Posts: 4,460
    edited December 1969

    Thought I'd try RTK out with an HD subject, in this case Boris by Smay. Normally HD slows down renders, which is why I tend not to use it much, however, with RTK there is no noticeable difference in render time and the detail is all there. Once we really get going with this shader kit, I might invest in a few more HD things if it doesn't slow up renders like it used to.

    CHEERS!

    Boris_HD_RTK.jpg
    577 x 750 - 299K
  • RogerbeeRogerbee Posts: 4,460
    edited December 1969

    I got to thinking 'What would I get if I crossed Boris with Darius!?'. The answer is this guy!

    CHEERS!

    Borius_HD_RTK.jpg
    577 x 750 - 272K
  • wowiewowie Posts: 2,029
    edited December 1969

    Rogerbee said:

    Thought I'd try RTK out with an HD subject, in this case Boris by Smay. Normally HD slows down renders, which is why I tend not to use it much, however, with RTK there is no noticeable difference in render time and the detail is all there. Once we really get going with this shader kit, I might invest in a few more HD things if it doesn't slow up renders like it used to.

    It's not specific to the shader.

    DS have improved HD morphs support for quite some time now. Although it's slightly slower when rendering (depending on what subD level the morphs were built in), there's no more long load times with HD morphs.

Sign In or Register to comment.