The Official aweSurface Test Track

1121315171866

Comments

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

    Converted the Sawmill https://www.daz3d.com/sawmill, testrendered with normalmaps on at 10x10 pixelsamples, thinking it would be cool to have a normal strength slider...

    Like this?

    By the way, the iray uber conversion script is finished. It does need the updated shader though. It won't work with older builds.

    Yeah just like thatlaugh Excellent stuffsmiley

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

    New Sawmill version;)

    image

    The Sawmill 1 awe.png
    1920 x 1080 - 4M
    Post edited by Sven Dullah on
  • Sven DullahSven Dullah Posts: 7,621
    edited January 2019

    Was about to add the crane to the sawmill environment but got carried away...

    image

    A Crane awe.png
    1280 x 720 - 2M
    Post edited by Sven Dullah on
  • kyoto kidkyoto kid Posts: 41,040

    New Sawmill version;)

    image

    ..better, though there is still "banding" on the outside of rhw logs in the foreground pile 

  • Sven DullahSven Dullah Posts: 7,621
    kyoto kid said:

    New Sawmill version;)

     

    ..better, though there is still "banding" on the outside of rhw logs in the foreground pile 

    Yup infact they are all identical, just scaled differently, and only come with diffuse and normalmaps, so should replace them with something a little bit more believableyes

    On the other hand this set is not a resource hog, no problems loading it and moving around the scene etc;) Two sides to everything, eh?

  • kyoto kidkyoto kid Posts: 41,040
    edited January 2019

    ...I wonder if there is a bark shader you can use, that would do the trick, Or maybe just changing the tiling density/ratio.

    Post edited by kyoto kid on
  • Sven DullahSven Dullah Posts: 7,621
    edited January 2019
    kyoto kid said:

    ...I wonder if there is a bark shader you can use, that would do the trick, Or maybe just changing the tiling density/ratio.

    The logs have only one materials zone, so I'd have to create separate zones for the bark and the wood, then play with tiling parameters;) Totally doable, and maybe create bump and displacement maps from the diffusemap too, I'll look into it because you're right, they look pretty bad...or I'll just replace that pile with something elselaugh ...nah a sawmill without logs is like coffee without milk, I'll have to deal with it, tks for pointing that out!

    Post edited by Sven Dullah on
  • kyoto kidkyoto kid Posts: 41,040

    ...the best way to create separate material zones  would be to use the Geometry editor. 

  • wowiewowie Posts: 2,029
    edited January 2019
    kyoto kid said:

    ..better, though there is still "banding" on the outside of rhw logs in the foreground pile 

    These might come handy for those types of surfaces. http://docs.daz3d.com/doku.php/public/read_me/index/16382/start

    Always wanted for AWE Surface to have something similar.

    This is me playing around with UV projection mapping. Planar, Ball and Spherical mapping. Haven't gotten around to doing cube and triplanar yet. Should also help with stuff that has no UV. Thinking about adding the ability to randomize the tiling with rotation and flipping.

    planar.jpg
    338 x 600 - 75K
    ball.jpg
    338 x 600 - 78K
    spherical.jpg
    338 x 600 - 76K
    Post edited by wowie on
  • Sven DullahSven Dullah Posts: 7,621
    edited January 2019
    wowie said:
    kyoto kid said:

    ..better, though there is still "banding" on the outside of rhw logs in the foreground pile 

    These might come handy for those types of surfaces. http://docs.daz3d.com/doku.php/public/read_me/index/16382/start

    Always wanted for AWE Surface to have something similar.

    This is me playing around with UV projection mapping. Planar, Ball and Spherical mapping. Haven't gotten around to doing cube and triplanar yet. Should also help with stuff that has no UV. Thinking about adding the ability to randomize the tiling with rotation and flipping.

     

     Mmmh that's interesting for sure:) I've seen the product but obviously missed the point totallyblush

    Would love to see those features in aweSurface!

    About randomizing; I've tried using the coat with its independent tiling to hide repeative patterns on for example ground planes with some luck, but the possibilities are somewhat limited;)

    Post edited by Sven Dullah on
  • wowiewowie Posts: 2,029
    About randomizing; I've tried using the coat with its independent tiling to hide repeative patterns on for example ground planes with some luck, but the possibilities are somewhat limited;)

    Naive tiling will actually always produce a very repeatable pattern. One way to break it up is just to offset the UV coordinates of each tile. One tile may have the original pattern, others would have different coordinates, though its still the same texture.

    Something like this :

     

     

    mirror.png
    500 x 208 - 55K
  • wowiewowie Posts: 2,029
    edited January 2019

    OK. Got triplanar mapping working.

    Had to tweak the code so it will relatively have the same UV tile if you use the original UVs.

    On props/objects without UV, the textures won't show up correctly in the viewport. Only in renders.

    720x1280, RaytracerFinal (8x8 pixel samples), 2048 irradiance samples. 8 min 25 secs and 8 min 33 secs.

    1.jpg
    338 x 600 - 65K
    2.jpg
    338 x 600 - 79K
    3.jpg
    338 x 600 - 42K
    5.jpg
    864 x 693 - 128K
    8 minutes 24.73 seconds.jpg
    720 x 1280 - 455K
    8 minutes 32.57 seconds.jpg
    720 x 1280 - 414K
    Post edited by wowie on
  • Sven DullahSven Dullah Posts: 7,621
    wowie said:

    OK. Got triplanar mapping working.

    Had to tweak the code so it will relatively have the same UV tile if you use the original UVs.

    Only now noticed the ears:) Sweetsmiley  I figure this opens up a whole bunch of new possibilities...less is not always morelaugh

    wowie said:

    On props/objects without UV, the textures won't show up correctly in the viewport. Only in renders.

     

     

    Limitations of openGL?

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

    Had another look at the sawmill..there are actually 3 log props and the pile of logs which is one object, everyone sharing the same texture. I chose one of the logs and created a new mat zone for the ends, made it possible to slightly alter the tiling for the bark, but not enough to make it work properly:( I don't have any bark shader presets, so will have to come up with something else...oh well...replacing them seems to be my best betfrown

    Newest version of my "savage" character, tweaking  continues...seem to have lost his lashes in the process, LOLblush

    image

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

    Question: When rendering with only HDRI lighting, I've noticed that I generally have to make gamma correction in postwork (increasing contrast typically). So lately I've tried to increase gamma for the environmental sphere from default 1 to maybe 1.3 to see if that would bring me closer to the final result. Yeah it affects the appearence of the background image but I'm not sure if it also affects global illumination? Or are there any other settings either with lighting or surfaces I'm missing?

     

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

    ...aaand another update, just some minor tweaks to the skin and fixed the lashes, think he can be let out into the wild nowsmiley

    image

    Hravnin close 1 awe.png
    1920 x 1080 - 3M
    Post edited by Sven Dullah on
  • kyoto kidkyoto kid Posts: 41,040

    ...he looks really good

  • Sven DullahSven Dullah Posts: 7,621
    kyoto kid said:

    ...he looks really good

    Tks, yeah old tech (G1 & Gen4 textures)but still kicking:)

  • wowiewowie Posts: 2,029
    edited January 2019
    Limitations of openGL?

    More like limitations of the OpenGL shaders used by DAZ Studio's viewport.

    Question: When rendering with only HDRI lighting, I've noticed that I generally have to make gamma correction in postwork (increasing contrast typically). So lately I've tried to increase gamma for the environmental sphere from default 1 to maybe 1.3 to see if that would bring me closer to the final result. Yeah it affects the appearence of the background image but I'm not sure if it also affects global illumination? Or are there any other settings either with lighting or surfaces I'm missing?

    It should affect GI, though the amount will vary. Remember that gamma alters the curve, not the maximum intensity. If you use gamma and gain (or exposure), the final intensity will be higher.

    Contrast is basically similar to gain, except you add an offset to the minimum point. As you might have guessed, offset simply adds or subtracts a given amount of brighness from the input.

    To get around the overblown background, use two environment sphere. For the first one, disable Visibility - Camera & Visibility - Reflections & Refractions Vso it does not appear in the render, but make sure it contributes to GI by enabling Visibility - Occlusion & Indirect Light. For the second one, enable Visibility Camera but make sure it does not contribute to GI by disabling Visibility - Occlusion & Indirect Light. Only enable Visibility - Reflections & Refractions on this environment sphere.

    An alternative would be to enable GI contribution on both sphere, so they'll add up.

    Edit:

    Just tested out the new build. AWE Surface 1.3 is almost ready. New features are normal mapping, vector displacement (big thanks to Mustakettu), camera based exposure limit controls and UV projection mapping. It also on average 20% faster than the 1.1 version. But that's not the best part. laugh

    Test render with 128 samples (default) - 1 min 10.78 secs

    Test render with 2048 samples -   1 min 36.89 secs

    If you want to get rid of the noise completely with 8192 samples - 3 min 31.19 sec

    All with 8x8 pixel samples.

    Test3.jpg
    338 x 600 - 94K
    Post edited by wowie on
  • Sven DullahSven Dullah Posts: 7,621

    Tks very much wowie!

    Great news one the new build! Please tell me there will be an update through DIMblush

  • wowiewowie Posts: 2,029
    edited January 2019

    Tks very much wowie!

    Great news one the new build! Please tell me there will be an update through DIMblush

    This update will be released through DIM. It also comes with some further bug fixes. Subsurface should render in high quality and less noise (if any). I'm thinking about RAMWolff's request about more translucency. I think I've found a way to add additional translucency controls, but I haven't tested it out yet.

    Edit:. Ah, sorry my idea didn't work. So no change in that front.

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

    Tks very much wowie!

    Great news one the new build! Please tell me there will be an update through DIMblush

    This update will be released through DIM. It also comes with some further bug fixes. Subsurface should render in high quality and less noise (if any). I'm thinking about RAMWolff's request about more translucency. I think I've found a way to add additional translucency controls, but I haven't tested it out yet.

    You totally made my day, can't wipe that silly smile off my facelaugh

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

    Revisited the Misty River Gorge by Stonemason. I only used normal maps this time, no bump or displacement. Found out you need to enable the "use face forward" for every surface using normal maps, to avoid black spots. Rendered at 12x12 pixelsamples/1024 irradiance samples.

    image

    Misty River Gorge (norm. maps) awe.png
    1920 x 1080 - 4M
    Post edited by Sven Dullah on
  • Sven DullahSven Dullah Posts: 7,621
    edited January 2019

    Using the environmental shader and motion blur to create a shortcircuit:)

    Post edited by Sven Dullah on
  • wowiewowie Posts: 2,029

    Did some proper testing with the final update build. I reworked some of the optimizations seems they're way too complicated and rather prone to noise. With this new build, most of the noise should be gone at 2048 samples. You can still use higher values if you want. I've also made some changes to how translucency is handled when UseFaceForward is enabled, so it renders a bit faster.

    Used my test scene with two nude figures, one with hair. Mostly to see render times with SSS and opacity. Used just a single emitter and no HDRI. The environment sphere is set to diffuse.

    First, rendered at 4x4 (RaytracerDraft) at low resolution (338x600). 128 samples - 1 min 48 sec, 512 samples - 2 min 30 sec, 2048 samples - 5 min 6 sec, 8192 samples - 6 min 58 sec.

    With 8x8 pixel samples (Raytracer Final) at 720x1280 or 4 times the amount of pixels. 128 samples - 10 min 30 sec. 2048 samples - 22 min 10 sec. I didn't test 512 samples, but it likely will be in between. With 8192 samples, render time is 34 min 40 sec.

     

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

    Did some proper testing with the final update build. I reworked some of the optimizations seems they're way too complicated and rather prone to noise. With this new build, most of the noise should be gone at 2048 samples. You can still use higher values if you want. I've also made some changes to how translucency is handled when UseFaceForward is enabled, so it renders a bit faster.

    Used my test scene with two nude figures, one with hair. Mostly to see render times with SSS and opacity. Used just a single emitter and no HDRI. The environment sphere is set to diffuse.

    First, rendered at 4x4 (RaytracerDraft) at low resolution (338x600). 128 samples - 1 min 48 sec, 512 samples - 2 min 30 sec, 2048 samples - 5 min 6 sec, 8192 samples - 6 min 58 sec.

    With 8x8 pixel samples (Raytracer Final) at 720x1280 or 4 times the amount of pixels. 128 samples - 10 min 30 sec. 2048 samples - 22 min 10 sec. I didn't test 512 samples, but it likely will be in between. With 8192 samples, render time is 34 min 40 sec.

     

    Tks for the update, very much looking forward to testing the new build:)

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

    The Nirvana Chess Cafe...an old scene...took on the task of converting it to awe. 18 Genesis1 characters, all except one fully dressed and many with hair:) I made the original using UE2 and the RR3, rendertime IIRC 15 hours or something like that. This took about 35 min. at 10x10 pixelsamples and 128 shadowsamples. This is very much a VIP, just a rough conversion, but very happy that my shaderwork with the grid dude seems to translate well, and render fastsmiley

    image

    The Kosmic Chessgame awe1.png
    1800 x 1013 - 3M
    Post edited by Sven Dullah on
  • Sven DullahSven Dullah Posts: 7,621
    edited January 2019

    Team Buddha...still a lot of adjustments to make, but totally blown away by rendering speed (and still on the old build), 59 min

    image

    Realized SSS scale is absolute, not relative, correct? Have to look into that, SSS is slightly overdonesmiley Buddha is natural size, the game characters scaled down.

    The Kosmic Chessgame awe2.png
    1800 x 1013 - 3M
    Post edited by Sven Dullah on
  • Sven DullahSven Dullah Posts: 7,621
    edited January 2019

    rendertime 50 min

    image

    The Kosmic Chessgame awe3.png
    1800 x 1013 - 3M
    Post edited by Sven Dullah on
  • kyoto kidkyoto kid Posts: 41,040

    ...that's really cool.

Sign In or Register to comment.