The Official aweSurface Test Track

1232426282966

Comments

  • Sven DullahSven Dullah Posts: 7,621
    wowie said:

    Hmm problems...I can't get opacity to work with the awePT arealight shader, it ignores the transmaps. I try to make a burning hair, have replaced the hair diff maps with a flames texture, and applied the PT arealight. The areas that should be transparent render black and solid. Any ideas why? Couldn't get it to work so applied the OmUberlightshader and it works, only it doesn't allow for tiling adjustments:(

    I'm looking into it. It has to do with values passed from DS to 3delight. Light shaders don't use the same naming conventions as surface shaders. I got around this limitation for PolyShape and Textures by adding my own parameters.

    yes Much appreciated!

    wowie said:

    The hair shader is basically 90% done. Just finished testing all the features. I've also added extra ones like saturation adjustment, a 2nd specular/reflection lobe and sheen. Render times with zero optimization - 22 minutes. With 100% optimization (the default value), 5 to 6 minutes. When the 2nd spec is enabled, render times goes to 7 or 8 minutes, depending if you're doing just specular or specular and reflection.

    Test renders below. The variations are all done with just playing around with the shader controls (melanin controls, highlight, root to tip ramp). From testing, the shader works best if the model comes with pretty detailed opacity maps. You can still plug in specular maps long with proper bump, normal and or displacement maps.

    Tks for sharing:) Can't wait to get my hands on it. Sheen sounds very interesting. Is it an on/off feature or slider controlled?

  • Sven DullahSven Dullah Posts: 7,621
    edited May 2019

    ...a WIP, still not happy with the skin, think I overdid SS slightlyblush, and that darn in ear thing wasn't adjustable, have to find another one, only she barely has ears, so I may have to find another approach to the whole thinglaugh

    image

    Control Center2pp awe.png
    1920 x 1080 - 3M
    Post edited by Sven Dullah on
  • Sven DullahSven Dullah Posts: 7,621

    It's kind of funny that wowie is about to release a shader for strandbased hair, at the same time DAZ releases a beta version with built in strandbased hairsmiley

  • Sven DullahSven Dullah Posts: 7,621

    I get a severity 2 error message when trying to use specularity/reflection on the environmental shader: The specified distribution ashikhmin-shirley was invalid. (in shader wowie/aweEnvironmentSphere on object shapematerial xxx) And spec/reflection doesn't seem to render on either the environmental shader nor the awePT areashader. I know about using geoshells as a workaround, just a heads up;)

  • Sven DullahSven Dullah Posts: 7,621
    edited May 2019

    So I found a working headset and worked with the skin a bit, now I need to get her a better outfit, looks terrible close upangry Any tips on a formal looking uniform, maybe for G3F, would be most appreciated;)

    image

    Communications Officer awe.png
    1920 x 1080 - 2M
    Post edited by Sven Dullah on
  • wowiewowie Posts: 2,029

    I get a severity 2 error message when trying to use specularity/reflection on the environmental shader: The specified distribution ashikhmin-shirley was invalid. (in shader wowie/aweEnvironmentSphere on object shapematerial xxx) And spec/reflection doesn't seem to render on either the environmental shader nor the awePT areashader. I know about using geoshells as a workaround, just a heads up;)

    A stupid typo mistake on my part. Fixed a while ago though.

    Annotation 2019-05-20 004528.jpg
    1365 x 735 - 192K
    Annotation 2019-05-20 004553.jpg
    1365 x 737 - 193K
  • Sven DullahSven Dullah Posts: 7,621
    wowie said:

    I get a severity 2 error message when trying to use specularity/reflection on the environmental shader: The specified distribution ashikhmin-shirley was invalid. (in shader wowie/aweEnvironmentSphere on object shapematerial xxx) And spec/reflection doesn't seem to render on either the environmental shader nor the awePT areashader. I know about using geoshells as a workaround, just a heads up;)

    A stupid typo mistake on my part. Fixed a while ago though.

    Great, tks:)

  • Sven DullahSven Dullah Posts: 7,621
    edited May 2019
    wowie said:

    angry

    Suddenly having big problems with emitters. Was about to make a render of a dude playing the flute when I noticed a bright spot on his neck inside the shirt that should have been in the shadow. Stripped down the scene to one emitter plane, no other lights:

    Don't know what to make of it or what to do, never noticed anything similar...helpcrying

    That looks like a raytracing bias issue. I think I've set it way too high in the release build, mostly to (wrongly) account for displacement. I've revised the code and it should be tighter now. Just to be on the safe side, I'll add a dial for raytracing bias (which is different from transmission bias).

    @wowie

    Any news on this problem? It's getting slightly annoying...running into the same issue over and over againindecision

    image

    And how's that new hairshader progressing? It's been a while, hope you're doing finesmiley

    Avala awe.png
    1800 x 1350 - 3M
    Post edited by Sven Dullah on
  • Sven DullahSven Dullah Posts: 7,621
    edited May 2019

    Some stuff I picked up during the sales...the Alpha Base...

    image

    The Alpha Base awe.png
    1920 x 1080 - 3M
    Post edited by Sven Dullah on
  • wowiewowie Posts: 2,029
    edited May 2019
    Any news on this problem? It's getting slightly annoying...running into the same issue over and over againindecision

    No longer a problem on the new build. But there were other issues I ran into that took a long time to figure out.

    And how's that new hairshader progressing? It's been a while, hope you're doing finesmiley

    wink Basically, it's done.

    hair2.jpg
    400 x 600 - 127K
    hair3.jpg
    400 x 600 - 127K
    hair4.jpg
    400 x 600 - 131K
    Post edited by wowie on
  • khorneV2khorneV2 Posts: 147
    wowie said:
    Any news on this problem? It's getting slightly annoying...running into the same issue over and over againindecision

    No longer a problem on the new build. But there were other issues I ran into that took a long time to figure out.

    And how's that new hairshader progressing? It's been a while, hope you're doing finesmiley

    wink Basically, it's done.

     

     

    surprise

  • Sven DullahSven Dullah Posts: 7,621
    khorneV2 said:
    wowie said:
    Any news on this problem? It's getting slightly annoying...running into the same issue over and over againindecision

    No longer a problem on the new build. But there were other issues I ran into that took a long time to figure out.

    And how's that new hairshader progressing? It's been a while, hope you're doing finesmiley

    wink Basically, it's done.

     

     

    surprise

    surprisesurprise

    Basically, I want it now!

    laugh

  • wowiewowie Posts: 2,029

    Have to make sure there's no bugs out of left field.

    Actually encountered one several days ago. So, does anyone have experience of DS just crashing when starting a render with a particular prop and AWE Surface?

  • Sven DullahSven Dullah Posts: 7,621
    wowie said:

    Have to make sure there's no bugs out of left field.

    Actually encountered one several days ago. So, does anyone have experience of DS just crashing when starting a render with a particular prop and AWE Surface?

    Not with a particular prop, but I can't render two scenes in a row with scripted pathtracing, without DS just shutting down, no warning. I have to restart DS to be able to do next render. Vanilla renders, no problem.

  • wowiewowie Posts: 2,029

    OK. I think i've found something interesting. At times, 3delight seems to be 'stuck' when rendering a particular bucket. If you do experience this, try checking if the material/surface zone visible in the bucket have both bump and displacement enabled. Current workaround seems to be to just enable either bump or displacement. The render only seems to get stuck when both are enabled.

  • Sven DullahSven Dullah Posts: 7,621
    wowie said:

    OK. I think i've found something interesting. At times, 3delight seems to be 'stuck' when rendering a particular bucket. If you do experience this, try checking if the material/surface zone visible in the bucket have both bump and displacement enabled. Current workaround seems to be to just enable either bump or displacement. The render only seems to get stuck when both are enabled.

    Tks, I'll keep that in mind for the next time a render stalls;) But in my experience it only happens when using  BRDFs other than the default AshikhminShirley one. And it's not consistent, sometimes using the classic ones works just fine;)

  • wowiewowie Posts: 2,029
    edited June 2019

    Well, just finished testing out the newest build. AWE Surface now employs smarter, stratified sampling for most raytracing operations (reflections, GI and subsurface). As a result, render times have improved and are down to 25 to 30 % less using the same samples settings.

    Also, Hue, gamma and saturation correction of base/diffuse textures.

    You can use the filter value to selectively mask parts of the texture you want to adjust. The filter is based of luminance/brightness. Technically, I can even add an optional texture mask so you can customize the selection even further.

    I found using the gamma and saturation settings together can help raise the brightness of textures. Plus it's technically the correct way to do it, rather than pushing diffuse strength above 100%.

    Capture.JPG
    311 x 208 - 20K
    Post edited by wowie on
  • Sven DullahSven Dullah Posts: 7,621
    wowie said:

    Well, just finished testing out the newest build. AWE Surface now employs smarter, stratified sampling for most raytracing operations (reflections, GI and subsurface). As a result, render times have improved and are down to 25 to 30 % less using the same samples settings.

    Also, Hue, gamma and saturation correction of base/diffuse textures.

    You can use the filter value to selectively mask parts of the texture you want to adjust. The filter is based of luminance/brightness. Technically, I can even add an optional texture mask so you can customize the selection even further.

    I found using the gamma and saturation settings together can help raise the brightness of textures. Plus it's technically the correct way to do it, rather than pushing diffuse strength above 100%.

    Wow this looks really usefulyes And cutting rendertimes againsurprise, where will it endlaugh?

  • wowiewowie Posts: 2,029
    edited June 2019

    Wow this looks really usefulyes And cutting rendertimes againsurprise, where will it endlaugh?

    Still needs to test it out a bit more. But here's another sneak peek of the revised subsurface. 10 min 38 secs at 8x8 pixel samples. Used 2048 irradiance samples, but I forgot to use higher samples on the hair.

    Claudine4.jpg
    667 x 1000 - 275K
    Claudine5.jpg
    667 x 1000 - 248K
    Claudine6.jpg
    667 x 1000 - 257K
    Post edited by wowie on
  • Sven DullahSven Dullah Posts: 7,621
    wowie said:

    Wow this looks really usefulyes And cutting rendertimes againsurprise, where will it endlaugh?

    Still needs to test it out a bit more. But there's another sneak peek of the revised subsurface. 10 min 38 secs at 8x8 pixel samples. Used 2048 irradiance samples, but I forgot to use higher samples on the hair.

    Tks for sharing! So, if I may be so curious, what are the major changes you've made to SS? Any idea when the new build will be available?

  • wowiewowie Posts: 2,029
    edited June 2019

    Tks for sharing! So, if I may be so curious, what are the major changes you've made to SS? Any idea when the new build will be available?

    Rather than just adding the sum of SSS and diffuse with weights, the shader also now takes into account index of refraction. There's also some weight shaping so there's more saturation on parts that are less lit and texture saturations gets piped into subsurface absorption values used by subsurface.

    Still tweaking it and render time/quality. The hair shader also still has some noise, so that needs additional work.

    Roughly 9 min, 8x8 pixel samples and just 512 samples.

    8 min 57.75 secs.jpg
    667 x 1000 - 257K
    Post edited by wowie on
  • Sven DullahSven Dullah Posts: 7,621

    I seem to have trouble getting HDRI light to work with skin. Generally I've started to use arealights combined with HDRI more and more, as that concept seems to yield better results. Here I once againg tried to use pure HDRI light. Started off with shadow- and SS- samples at 2048. Very grainy result. 4096/4096, still grainy. This was rendered with 8192/8192 and 10x10 pixelsamples, non progressive. Still very grainy if you look at her legs and shaded areas. Any tips? Am I doing something wrong? Should I use 16k samples. This render with just a single character already took a good hour to render. When using arealights 2048/2048 usually is enough to get a clean render.

  • Mustakettu85Mustakettu85 Posts: 2,933

    Started off with shadow- and SS- samples at 2048. Very grainy result. 4096/4096, still grainy. This was rendered with 8192/8192 and 10x10 pixelsamples, non progressive. 

    "Shadow" samples? Like, the shadow samples in the render settings? They are not used for pathtracing at all.

    Or do you mean irradiance samples in the surface tab? These are the samples that control actual diffuse lighting. Since all SSS starts with a diffuse pass internally, these are very important for the end result.

  • Sven DullahSven Dullah Posts: 7,621
    edited June 2019

    Started off with shadow- and SS- samples at 2048. Very grainy result. 4096/4096, still grainy. This was rendered with 8192/8192 and 10x10 pixelsamples, non progressive. 

    "Shadow" samples? Like, the shadow samples in the render settings? They are not used for pathtracing at all.

    Or do you mean irradiance samples in the surface tab? These are the samples that control actual diffuse lighting. Since all SSS starts with a diffuse pass internally, these are very important for the end result.

    Irradiance samples;) Used 8192/8192 for skin surfaces, 2048 on the hair and clothes.

    Post edited by Sven Dullah on
  • Sven DullahSven Dullah Posts: 7,621

    Started off with shadow- and SS- samples at 2048. Very grainy result. 4096/4096, still grainy. This was rendered with 8192/8192 and 10x10 pixelsamples, non progressive. 

    "Shadow" samples? Like, the shadow samples in the render settings? They are not used for pathtracing at all.

    Or do you mean irradiance samples in the surface tab? These are the samples that control actual diffuse lighting. Since all SSS starts with a diffuse pass internally, these are very important for the end result.

    Irradiance samples;) Used 8192/8192 for skin surfaces, 2048 on the hair and clothes.

    Some more info:

    The HDRI is from HDRI Haven. Those HDRIs are usually of good quality but maybe this particular one was not, so will try another one. The awe Environment exposure was set to 5, no other changes. The environment sphere exposure was set to 1 with a gamma of 1.6. I usually end up using gamma values of 1.4-1.6 because values of 1 will create a slightly washed out look.

  • Mustakettu85Mustakettu85 Posts: 2,933

    Started off with shadow- and SS- samples at 2048. Very grainy result. 4096/4096, still grainy. This was rendered with 8192/8192 and 10x10 pixelsamples, non progressive. 

    "Shadow" samples? Like, the shadow samples in the render settings? They are not used for pathtracing at all.

    Or do you mean irradiance samples in the surface tab? These are the samples that control actual diffuse lighting. Since all SSS starts with a diffuse pass internally, these are very important for the end result.

    Irradiance samples;) Used 8192/8192 for skin surfaces, 2048 on the hair and clothes.

    Aha, then these are really high.

    I'd suggest first disabling SSS and rendering her with just diffuse. If the issue persists, try to figure out if it's related to bump or specular (disable these one by one and see).

    If it's not any of these, the HDRI itself may be a problem, especially if you had to mess with its exposure etc. 

    Basically why area lights tend to sample better: they are huge, they are uniformly bright and they are solid colour. HDRIs have all those colour gradients and sharp transitions of variable brightness. Inherently problematic, and might be even more so if you sample them from geometry when you have already applied some math to the values.

     

  • Sven DullahSven Dullah Posts: 7,621

    Started off with shadow- and SS- samples at 2048. Very grainy result. 4096/4096, still grainy. This was rendered with 8192/8192 and 10x10 pixelsamples, non progressive. 

    "Shadow" samples? Like, the shadow samples in the render settings? They are not used for pathtracing at all.

    Or do you mean irradiance samples in the surface tab? These are the samples that control actual diffuse lighting. Since all SSS starts with a diffuse pass internally, these are very important for the end result.

    Irradiance samples;) Used 8192/8192 for skin surfaces, 2048 on the hair and clothes.

    Aha, then these are really high.

    I'd suggest first disabling SSS and rendering her with just diffuse. If the issue persists, try to figure out if it's related to bump or specular (disable these one by one and see).

    If it's not any of these, the HDRI itself may be a problem, especially if you had to mess with its exposure etc. 

    Basically why area lights tend to sample better: they are huge, they are uniformly bright and they are solid colour. HDRIs have all those colour gradients and sharp transitions of variable brightness. Inherently problematic, and might be even more so if you sample them from geometry when you have already applied some math to the values.

     

    Ok I'll look into it, tks;)

  • Sven DullahSven Dullah Posts: 7,621
    edited June 2019

    Started off with shadow- and SS- samples at 2048. Very grainy result. 4096/4096, still grainy. This was rendered with 8192/8192 and 10x10 pixelsamples, non progressive. 

    "Shadow" samples? Like, the shadow samples in the render settings? They are not used for pathtracing at all.

    Or do you mean irradiance samples in the surface tab? These are the samples that control actual diffuse lighting. Since all SSS starts with a diffuse pass internally, these are very important for the end result.

    Irradiance samples;) Used 8192/8192 for skin surfaces, 2048 on the hair and clothes.

    Aha, then these are really high.

    I'd suggest first disabling SSS and rendering her with just diffuse. If the issue persists, try to figure out if it's related to bump or specular (disable these one by one and see).

    If it's not any of these, the HDRI itself may be a problem, especially if you had to mess with its exposure etc. 

    Basically why area lights tend to sample better: they are huge, they are uniformly bright and they are solid colour. HDRIs have all those colour gradients and sharp transitions of variable brightness. Inherently problematic, and might be even more so if you sample them from geometry when you have already applied some math to the values.

     

    Ok I'll look into it, tks;)

    The HDRI is fine. Seems like it was an issue with specular settings and/or maps, possibly also some DoF issues. Ended up removing the specular maps and tweaking roughness and strength to compensate. Settled for 4096 Irradiance- and 2048 SS samples, increased bump strength and adjusted some SS settings. Used 12x12 pixelsamples and set specular bounce depth to 16, still rendered twice as fast as yesterday's version. I think it looks a bit cleaner but something got lost in the process, maybe I need to look at the SS settings once more.

    Tks Kettu for educating me once again=) Most appreciated!

    image

    Lakeside evening 2.png
    1800 x 1350 - 4M
    Post edited by Sven Dullah on
  • wowiewowie Posts: 2,029

    Yes. In a HDRI lit scene, SSS is rather prone to noise at the default samples. You need at least 1024 subsurface samples, with at least the same amount of irradiance samples.

    I'm planning to add additional controls for AWE Environment so you'll have the ability to separate exposure/gain/gamma/saturation for specular(what's visible in reflection) and diffuse(what's visible to global illumination). It will still be using the same texture though.

    Should be much more convenient thatn using a secondary environment sphere.

  • Sven DullahSven Dullah Posts: 7,621
    wowie said:

    Yes. In a HDRI lit scene, SSS is rather prone to noise at the default samples. You need at least 1024 subsurface samples, with at least the same amount of irradiance samples.

    I'm planning to add additional controls for AWE Environment so you'll have the ability to separate exposure/gain/gamma/saturation for specular(what's visible in reflection) and diffuse(what's visible to global illumination). It will still be using the same texture though.

    Should be much more convenient thatn using a secondary environment sphere.

    Nice! You can never have too much control IMHO=)

Sign In or Register to comment.