The Official aweSurface Test Track

1454648505166

Comments

  • Sven DullahSven Dullah Posts: 7,621
    edited July 2020

    As a reference, deleted the RR3 stuff, turned on the 1 poly "sun" emitter and increased the ambient light, rendertime 12 min.

    image

    Genaro2 awe.png
    1280 x 720 - 1M
    Post edited by Sven Dullah on
  • Sven DullahSven Dullah Posts: 7,621
    edited July 2020

    ...so I created a simple test on my superslow laptop, a snail with the RR3 emissive shader in an enclosed space, and the reflective light with 2048 occlusion samples. The floor, walls and ceiling use aweSurface with 1024 adaptive samples. Total rendertime with progressive on, 2 days - 1 sec. laugh.

    image

    Then I realized that, since every ambient surface basically will emit light, the reflective light is unnecessary, so deleted it and re rendered. Rendertime 2 hours. There is a slight difference in light intensity, but...conclusion: Don't use the reflective lightlaugh.

    image

    Snail awe RR3.png
    1280 x 720 - 1M
    Snail aweRR3 no refl light.png
    1280 x 720 - 2M
    Post edited by Sven Dullah on
  • Sven DullahSven Dullah Posts: 7,621
    edited July 2020

    ...another one, applied the RR3 emissive to the DNA Greyling, no other lightsource except the aweEnvironment with exposure 5, progressive render, Total Rendering Time: 1 hour 3 minutes 19.77 seconds, on my main rig, DS 4.9

    image

    Greyling awe RR3.png
    1600 x 1200 - 3M
    Post edited by Sven Dullah on
  • Sven DullahSven Dullah Posts: 7,621
    edited July 2020

    @wowie or anybody

    What is the correct "decay" value for the DS standard spotlight to achieve a physically corrrect light fall off? It should be 2, right? For it to work you apparently have to set light intensity scale to 10000 or thereabout, when working at normal 2- 10m distances? Am I missing something? Fact is I NEVER use any standard lights, that's why I'm asking:) Also noticed something new to me: If you set light spread to 180 degrees you lose the soft shadows. If less than 180, you can turn limits off for shadow softness and bump it up to whatever you need, but at 180 there is only prefectly sharp shadows.

    Light settings, light intensity scale is 20000, shadow softness is 1000000%.

    image

    180 degree spread:

    image

    179,90 degree spread:

    image

    standard spot.png
    424 x 678 - 59K
    1000000.png
    1280 x 720 - 172K
    179.90 1000000.png
    1280 x 720 - 181K
    Post edited by Sven Dullah on
  • Sven DullahSven Dullah Posts: 7,621
    edited July 2020

    ...and regarding the RR3 tests...

    There's nothing magical about the RR3 emissive shaders, as I suspected, the magic is done with the reflective light, which seems to be a modified UE2. The shader infact is the DS default shader with the ambient channels renamed to emissive:)I imported it to shadow mixer and had a look. I get the same results with the regular DS default or UberSurface, just need to up ambient strength to thousands or millionslaugh.

    Post edited by Sven Dullah on
  • Mustakettu85Mustakettu85 Posts: 2,933

    Yes the intensity scale is about that. I think awe area lights use similar multipliers, but under the hood. 

    As for RR, I've long suspected its main light uses specular bounce algorithm ("reflection") instead of diffuse bounce; there are a few things in the manual that make me think this way (I obviously don't have it; it came out when I had already had my system working). This would be why the results look kinda "fairytale illustration" in most renders I've seen.

  • Sven DullahSven Dullah Posts: 7,621

    Yes the intensity scale is about that. I think awe area lights use similar multipliers, but under the hood. 

    Ah ok, tks for confirming!

    As for RR, I've long suspected its main light uses specular bounce algorithm ("reflection") instead of diffuse bounce; there are a few things in the manual that make me think this way (I obviously don't have it; it came out when I had already had my system working). This would be why the results look kinda "fairytale illustration" in most renders I've seen.

    That makes a lot of sense.

  • wowiewowie Posts: 2,029

    Sorry for not posting in awhile. I got lost in Ghost of Tsushima. blush

    @wowie or anybody

    What is the correct "decay" value for the DS standard spotlight to achieve a physically corrrect light fall off? It should be 2, right?

    Yes. The formula is basically pow (length of incoming light, num_value). So, 0 = no falloff, 1 = linear falloff, 2 = quadratic falloff (physically correct).

    For it to work you apparently have to set light intensity scale to 10000 or thereabout, when working at normal 2- 10m distances? Am I missing something? Fact is I NEVER use any standard lights, that's why I'm asking:)

    It depends on how the shader is written. For example, I think Mustakettu's original point/spot lights just multiply the intensity with the value. For awePoint/Spotlight, I use exposure (EV) or pow (2, intensity_scale) instead . It's shorter to code and more concise in the UI. Rather than 1000, you can use 10 to get 2 ^ 10 or 1024 (which is what you're doing with pushing up ambient channels to those high amounts). It also means every incremental decrease/increase corresponds to half/double the original intensity. Last, that's what 3delight for Maya uses and the convention used by real life photography (when talking about light in f-stops or EV).

    Also noticed something new to me: If you set light spread to 180 degrees you lose the soft shadows. If less than 180, you can turn limits off for shadow softness and bump it up to whatever you need, but at 180 there is only prefectly sharp shadows.

    Interesting, I don't get the same behavior on my end. Though, I don't think I'm using the same light. Maybe newer DS have updated point/spot lights?

    Anyway, shadow softness is a misnomer. In most cases, it's basically the amount of desired "blur" you want to have. Going by shader examples in Shader Builder, it's the 'samplecone' parameter of transmission () shadeop. It is used to determine how much incoming light is blocked by something between the surface point and where the light is coming from.

    As for emissives, although I can see use cases for it, in my opinion area light shaders fit that need better. Area lights is properly importance sampled so you'll have less noise and less chance of fireflies compared to using ambient. I've also seen cases with iray where people are continually abusing emission on materials that shouldn't be emissive at all (ie skin). Proving my point that people will abuse emissive if it's integrated into a surface shader

  • Sven DullahSven Dullah Posts: 7,621

     

    wowie said:
     

    As for emissives, although I can see use cases for it, in my opinion area light shaders fit that need better. Area lights is properly importance sampled so you'll have less noise and less chance of fireflies compared to using ambient.

    Yeah the old quality vs speed trade off:) PT area light looks so much better and is so much easier to control, ambient renders very fast but is noisy. I'm thinking ambient is still useful for light panels, space ship/SF interiors etc. The glowing snail that rendered in 2h on my crappy laptop using ambient would apparently take days using the PT arealight shader.

    wowie said:

    I've also seen cases with iray where people are continually abusing emission on materials that shouldn't be emissive at all (ie skin). Proving my point that people will abuse emissive if it's integrated into a surface shader

    Yup, just like the ambient channel is abused in 3DL mat presets, probably to compensate for using a non linear workflowindecision.

  • Sven DullahSven Dullah Posts: 7,621
    edited August 2020

    Just found this free planet texture resource https://www.solarsystemscope.com/textures/:) Nice 2k and 8k textures, but no height maps. Tested the 2k Mercury, simply used LIE to create an inverted copy of the texture with gamma set to 1 as a displacement map.

    image

     

    Mercury2k awe.png
    1920 x 1080 - 2M
    Post edited by Sven Dullah on
  • Sven DullahSven Dullah Posts: 7,621
    edited August 2020

    The 8k moonmap from the same site

    image

    8k Moon awe.png
    1920 x 1080 - 3M
    Post edited by Sven Dullah on
  • Sven DullahSven Dullah Posts: 7,621
    edited August 2020

    ...8k Mercury using GGX specular, rendertime non progressive 13min...one emitter with linear falloff...

    image

    Mercury 8k awe.png
    1920 x 1080 - 3M
    Post edited by Sven Dullah on
  • wowiewowie Posts: 2,029

    A bit of news update. The next entire (updated) AWE Shading Kit will be free for both personal and commercial usage, first on my Google drive folter (and other sites as well) within the week.

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

    A bit of news update. The next entire (updated) AWE Shading Kit will be free for both personal and commercial usage, first on my Google drive folter (and other sites as well) within the week.

    Tks wowie, can't wait!!!

  • Sven DullahSven Dullah Posts: 7,621
    edited September 2020

    ...while (patiently) waiting, bumping this thread with a WIP...

    image

    Mats awe.png
    1600 x 1200 - 3M
    Post edited by Sven Dullah on
  • 3Diva3Diva Posts: 11,470

    ...working on the dropship (by Nightshift3D) interior...non progressive 10x10ps, 11h, can't really see a way to significantly cut rendertimes while still maintaining realistic lighting...8 emitting 1 poly planes and the environment with a jpeg, oldschool opacity on the windows.

    image

    Wow... That environment looks amazing! Most impressive!

  • Just found this free planet texture resource https://www.solarsystemscope.com/textures/:) Nice 2k and 8k textures, but no height maps. Tested the 2k Mercury, simply used LIE to create an inverted copy of the texture with gamma set to 1 as a displacement map.

     

    That's fantastic, thank you for sharing!

  • 3Diva said:

    ...working on the dropship (by Nightshift3D) interior...non progressive 10x10ps, 11h, can't really see a way to significantly cut rendertimes while still maintaining realistic lighting...8 emitting 1 poly planes and the environment with a jpeg, oldschool opacity on the windows.

     

    Wow... That environment looks amazing! Most impressive!

    Thank you very much:) Yeah quite happy with the conversion but a true pain to render;)

    Just found this free planet texture resource https://www.solarsystemscope.com/textures/:) Nice 2k and 8k textures, but no height maps. Tested the 2k Mercury, simply used LIE to create an inverted copy of the texture with gamma set to 1 as a displacement map.

     

    That's fantastic, thank you for sharing!

    You're very welcome:)

  • @wowie

    Sorry but I just have to ask: what's the word on the awe update?

  • Sven DullahSven Dullah Posts: 7,621
    edited September 2020

    A (not very) quick test, made a SM network that surprisingly seems to atleast partly work with aweSurface/pathtracing. Used the edge blend brick and connected it to opacity. The diffuse- and ambient components also work as expected but I haven't yet found a way to make it accept shadows, and I suspect it's impossible. Applied to the exhaust props here. Also created some controlmaps from FB:s Lunar crater (big crater) diffuse texture, which seem to work ok. Now I just have to sort out the Stingray materials;)

    image

    Still dreaming of a true PTvolumetric shader...

    Stingray Night awe.png
    1920 x 1080 - 3M
    Post edited by Sven Dullah on
  • Sven DullahSven Dullah Posts: 7,621
    edited September 2020

    ...working on the Stingray mats...still doing progressive rendering so a bit grainy...made some metallicity- roughness- and specular maps...

    image

    Stingray Night2 awe.png
    1600 x 900 - 2M
    Post edited by Sven Dullah on
  • Sven DullahSven Dullah Posts: 7,621
    edited September 2020

    ...right, added some flares in post...

    image

    Stingray Night3 awe.png
    1920 x 1080 - 3M
    Post edited by Sven Dullah on
  • wowiewowie Posts: 2,029
    edited September 2020

    A little bit of update.

    Due to some bugs and other real life issues, I had to postpone the release. Took a bit of time to figure out, but I've ironed the issues out. I'll probably upload the update(s) in a few days.

    Take care and stay safe.

    Update:

    Here's the link:

    https://drive.google.com/file/d/1i3P5yVxiS4wZzhdZu6FeLNuiNuIPOAVe/view?usp=sharing

    Post edited by wowie on
  • Tks wowie and likewise! Will dig in ASAP:)

  • wowiewowie Posts: 2,029

    Tks wowie and likewise! Will dig in ASAP:)

    The character presets maybe broked though. If figures turned to white, then I've included the wrong ones. Let me know.

  • Sven DullahSven Dullah Posts: 7,621
    edited September 2020
    wowie said:

    Tks wowie and likewise! Will dig in ASAP:)

    The character presets maybe broked though. If figures turned to white, then I've included the wrong ones. Let me know.

    Ok, downloaded and did a manual install. The character presets are updated, and I tested the G1, seems to work, no white ladies:) However, after a very brief test, I'm a bit confused. The shaders UIs look identical to the previous version, I'd expected some new features etc. Am I missing something or are all the changes under the hood?

     

    new awe env?.png
    381 x 656 - 76K
    Post edited by Sven Dullah on
  • wowie said:
     
    • Since I've found a way around the double shading override, Use Face Forward will be removed. Surfaces will always be rendered as double shaded. Unlike before, there is no performance penalty with such scenarios.
    • Because everything now renders as double shaded, I've added extra back shading overrides. Before it was limited to just diffuse colors, now it extends to diffuse strength, transmission strength/color, specular strength/roughness (for both specular lobe). These overrides accept textures. Cases when this is useful are one way glass (metalness/transmission), stained glass (different textures for each side), fabric with metal adornments on one side and the like.
    • There's also a way to use different metal colors (by using the first lobe specular color/edge tint for the second lobe. When this is enabled, the front facing will use the second lobe settings and the back facing side will use the first lobe settings.
    • New mask options such as transmission/metalness mask filtering, plus the ability to use a color filter for the base/diffuse texture as metalness.
    • Color correction for base/diffuse textures are now done so the correction gets applied when you enable the use of base/diffuse textures as metalness/transmission inputs.
    • Added a convenience control to control roughness for metalness in case you want metal areas to be less rough than non metal areas. There's also one for transmission as well. These don't accept textures though.
    • Added a new 'Thin Mode' switch. This should only be enabled on single sided objects (sheet of paper/glass/fabric etc). When enabled, this renders refraction properly when transmission is enabled on both sides.
    • When Thin Mode is off, diffuse translucency will now have absorption. Best used with translucency shadow strength enabled.
    • The coat layer now have separate bump layer options. WIthout a texture plugged in, enabling Coat Bump allows you to completely disable bump for the coat layer. With a texture plugged in and coat bump strength > 0, both the base bump and coat bump will be added together.
    • Complete rewrite of subsurface implementation. Render with and without subsurface should have roughly the same brightness now and thus, energy preserving. There's no more need to dial back diffuse or subsurface strength.
    • SSS Translucency boost will be removed. In it's place, there's now separate deep/shallow absorption color which will be enabled when 'Use Diffuse Texture with SSS' is above 0. Shallow absorption color have more influence on areas where SSS is more translucent, while deep absorption color controls areas that gets most light.

     

    Yeah, for example I still have "use face fwd", translucency boost, can't see the new features listed in your quote, like a thin mode switch etc.

  • wowiewowie Posts: 2,029
    edited September 2020

    Yeah, for example I still have "use face fwd", translucency boost, can't see the new features listed in your quote, like a thin mode switch etc.

    OK. I've reuploaded the file. Let me know if the new one has the updated shaders.

    The relevant DAZ Studio app shaders should be inside the 'DAZ Studio_4.5;4.x Private Build;4.x Public Build' folder inside the .zip archive.

    Post edited by wowie on
  • Sven DullahSven Dullah Posts: 7,621
    edited September 2020
    wowie said:

    Yeah, for example I still have "use face fwd", translucency boost, can't see the new features listed in your quote, like a thin mode switch etc.

    OK. I've reuploaded the file. Let me know if the new one has the updated shaders.

    The relevant DAZ Studio app shaders should be inside the 'DAZ Studio_4.5;4.x Private Build;4.x Public Build' folder inside the .zip archive.

    Nope I get the same thing, unfortunately!

    Edit: Hmm the PT arealight shader seems to be updated, found out the intensity range is extended.

    aweshaders.png
    1029 x 608 - 183K
    Post edited by Sven Dullah on
  • Sven DullahSven Dullah Posts: 7,621
    edited September 2020

    Testing the updated G1 character preset on a more or less "out of the box" Genesis figure, along with some legacy clothing, also using the included presets. Just a quick 512 samples progressive render. Will probably work as a testscene for the new shader update...

    image

    ToryG1 awe.png
    600 x 971 - 558K
    Post edited by Sven Dullah on
Sign In or Register to comment.