The Official aweSurface Test Track

191012141566

Comments

  • WillowRavenWillowRaven Posts: 3,787

    That looks VERY real ... Worth the 1h 40min.

  • kyoto kidkyoto kid Posts: 41,057

    ...yeah on the CPU. that would take about 3 - 4 hours on my system in to render in Iray and that looks every bit as good..

  • WillowRavenWillowRaven Posts: 3,787

    I'm not understanding a lot I'm reading on this thread, but that looks great.

  • Sven DullahSven Dullah Posts: 7,621

    Tks guys:) Need to work on some characters for this scene...browsing my runtime to find a couple of nice suits;)

  • ...and a WIP, 5 arealights and HDRI...rendertime with 8x8 pixelsamples 1h 40min.

    Very nice, it's quite tricky doing noise-free enclosed scenes and keeping the render times down. 

  • wowiewowie Posts: 2,029
    edited November 2018

    But when I made a couple of pool scenes using RR3 by Marshian I was looking at 12 h IIRC:)

    Really? 12 hours? Unreal.

    Managed to squeeze out more performance from the shader. Specifically, render with high irradiance samples ( >512) now take 1/3 less render time than previous development build. Overall, render times with the updated shader should be around 67% less time compared to the released version.

    I did discover a bug somewhere else though. Thankfully, the bug doesn't seem to affect the uplift in performance.

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

    ...and a WIP, 5 arealights and HDRI...rendertime with 8x8 pixelsamples 1h 40min.

    Very nice, it's quite tricky doing noise-free enclosed scenes and keeping the render times down. 

    Tks, yeah I dare say awe handles low light scenes better than Iray:)

     

    wowie said:

    But when I made a couple of pool scenes using RR3 by Marshian I was looking at 12 h IIRC:)

    Really? 12 hours? Unreal.

    Haven't used it for a long time, problem was grain, you needed to set the reflective light shading rate pretty low;) Basically a modified UE2

    wowie said:

    Managed to squeeze out more performance from the shader. Specifically, render with high irradiance samples ( >512) now take 1/3 less render time than previous development build. Overall, render times with the updated shader should be around 67% less time compared to the released version.

    I did discover a bug somewhere else though. Thankfully, the bug doesn't seem to affect the uplift in performance.

    That's really good news;)

  • wowiewowie Posts: 2,029
    That's really good news;)

    Might have another trick up my sleeve. winkIt might work, it might not.

  • khorneV2khorneV2 Posts: 147

    laugh

  • wowiewowie Posts: 2,029

    OK. I think I've chased down what's causing the bug. I've also updated the latest optimization further.

    The test renders I posted now renders in around 20 to 30 minutes for 8x8 pixel samples with 8192 irradiance samples. So basically, the same render time as 2048 with the older developmental build. As I said earlier, I won't put a limit on irradiance samples, but 8192 should cover extreme cases (low light scenes). For well lit scenes, I think 2048 is more than enough. Even with that, you should see definite improvement there as well.

    Outside of iray uber support, I think the shader is almost good to go.

  • Sven DullahSven Dullah Posts: 7,621
    edited November 2018
    wowie said:

    OK. I think I've chased down what's causing the bug. I've also updated the latest optimization further.

    The test renders I posted now renders in around 20 to 30 minutes for 8x8 pixel samples with 8192 irradiance samples. So basically, the same render time as 2048 with the older developmental build. As I said earlier, I won't put a limit on irradiance samples, but 8192 should cover extreme cases (low light scenes). For well lit scenes, I think 2048 is more than enough. Even with that, you should see definite improvement there as well.

    Outside of iray uber support, I think the shader is almost good to go.

    Very good news! What's the word on the displacement ratio thingy?

    Post edited by Sven Dullah on
  • wowiewowie Posts: 2,029
    edited November 2018

    Very good news! What's the word on the displacement ratio thingy?

    Still working on that. It kinda looks like DS is applying gamma correction to the bump/displacement textures, though it shouldn't. I did check the image gamma applied to the displacement and they're all at 1. It shouldn't be doing any gamma correction at all with those settings. But I was getting a lot of non optimal displacement bound warnings. Then I wrote in gamma correction into the shader for the displacement input with 2.2 as the gamma and all the warnings went away.

    It's kinda easy to fix now that I know what the problem is. But it does mean adding switches to the bump/displacement sections to offset what DS shouldn't be doing in the first place. Anyway, with this implemented, you should be able to have proper displacement and there's flexibility so you can turn it on/off, just in case DS decides to 'fix' the problem.

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

    Very good news! What's the word on the displacement ratio thingy?

    Still working on that. It kinda looks like DS is applying gamma correction to the bump/displacement textures, though it shouldn't. I did check the image gamma applied to the displacement and they're all at 1. It shouldn't be doing any gamma correction at all with those settings. But I was getting a lot of non optimal displacement bound warnings. Then I wrote in gamma correction into the shader for the displacement input with 2.2 as the gamma and all the warnings went away.

    It's kinda easy to fix now that I know what the problem is. But it does mean adding switches to the bump/displacement sections to offset what DS shouldn't be doing in the first place. Anyway, with this implemented, you should be able to have proper displacement and there's flexibility so you can turn it on/off, just in case DS decides to 'fix' the problem.

    laugh

    Great!

  • wowiewowie Posts: 2,029
    edited November 2018

    laugh

    Great!

    It's in. The additional dials control both bump and displacement textures, since they're calculated roughly the same way. Running some test renders to make sure they're working properly. Looks good, although I did still have to add some padding to the displacement bound.

    There will be cases where the bounds isn't optimal, mostly because the displacement bound can't read texture values. In such cases, simply lower the displacement maximum value.

    Also related, GI and SSS should behave properly now when displacement is used.

    Edit:

    I also think I've found a way to get f-stop (from the camera) and shutter time (in the renderer settings) from the scene. Haven't tested it yet, but in theory it should work regardless of which one you're using. When the camera isn't used, the default values on the AWE Environment light will be used instead. Would have loved to consolidate f-stop and shutter time in one place, but haven't the slightest idea how to approach it. Unless there's a way to create your own camera and push the parameter value to the render script.

    Post edited by wowie on
  • Sven DullahSven Dullah Posts: 7,621
    edited November 2018

    Speaking of cameras, I just need to vent my frustration a bit. It seems now as soon as I use DoF I'm running into trouble. Don't know what to do? And why am I the only one having problems with the raytracer & DoF? I mean this bug has probably been around for a long time, yet I can't find any comments/questions about it in the forums. Here is yet another example, I converted the Tuscan Villa https://www.daz3d.com/tuscan-villa to awe and created a camera with a focal length of 50 and enabled DoF. HDRI lighting, render pixel samples 12x12, Irradiance samples 512. There are diagonal lines on the walls and a shadow from nowhere on the gravel just in front of the house to the left.

    image

    Same scene with DoF disabled:

    image

    This is really getting to me now, aaaaaargh (please ignore this little outburst)frown I will contact Parris and ask if he can help contact the 3DL devs, this thing must be sorted out!

    Not using DoF is not an option IMO. As Mustakettu pointed out, and it was discussed earlier in this thread, the raytracer will calculate bump and  some other things quite differently when DoF is enabled. The gravel with bump in these renders being a good example;)

    Edit I sent Parris a PM in the hope he has some helpful hints on how to progress with this matter;)

    Tuscan Villa awe DoF.png
    1800 x 1013 - 3M
    Tuscan Villa awe no DoF.png
    1800 x 1013 - 3M
    Post edited by Sven Dullah on
  • wowiewowie Posts: 2,029
    edited November 2018

    I think it's generally caused by your use of focal distance. In that test scene you sent me, I got around the problem by lowering focal distance, and adjusting frame width and focal length to compensate for the shorter focal distance.

    Here's how the DOF settings look in the viewport.

    Mine. Frame width 27 focal length 300 focal distance 978

    Yours. Frame width 36 focal length 400 focal distance 1024

    Renders were comparable but without the artifacts.

    27 300 978 50.JPG
    410 x 662 - 75K
    36 400 1024 50.JPG
    405 x 660 - 75K
    Post edited by wowie on
  • Sven DullahSven Dullah Posts: 7,621

    Tks wowie, yeah I obviously have to start doing some things differently:) In the above example the focal point was set at the front door, maybe I should have set it a few meters in front of the camera then?

  • Sven DullahSven Dullah Posts: 7,621
    edited November 2018

    Tks wowie, yeah I obviously have to start doing some things differently:) In the above example the focal point was set at the front door, maybe I should have set it a few meters in front of the camera then?

    Hmm so tested lowering the focal distance to 350 and increasing F/Stop enough not to get a blurred render. The artifacts are hardly noticable, so yes it's a workaround. However I still call this a bug. Do not have these issues if not using DoF with the raytracer, no matter camera settings;)

    Edit: Ok so here the artifacts are reduced to these areas, with a focal distance of  300. However I found that they go away if I disable bump on the wall surface. Changing the tiling offset doesn't move the artifacts around.

    image

    Tuscan Villa awe FD 300pp.png
    1800 x 1013 - 3M
    Post edited by Sven Dullah on
  • wowiewowie Posts: 2,029

    Hmm so tested lowering the focal distance to 350 and increasing F/Stop enough not to get a blurred render. The artifacts are hardly noticable, so yes it's a workaround. However I still call this a bug. Do not have these issues if not using DoF with the raytracer, no matter camera settings;)

    Why change the F-stop? I only recommend changing the focal length and frame width. In my example, I simply multiply both by 2/3.

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

    Hmm so tested lowering the focal distance to 350 and increasing F/Stop enough not to get a blurred render. The artifacts are hardly noticable, so yes it's a workaround. However I still call this a bug. Do not have these issues if not using DoF with the raytracer, no matter camera settings;)

    Why change the F-stop? I only recommend changing the focal length and frame width. In my example, I simply multiply both by 2/3.

    Well if I take a photo of a house, I want it to be sharp and not out of focus. Having the focal point at the front door means a focal length of 2400. If I reduce the focal length to 300 the house is not in focus without fiddeling with the F/stop. Anyway, changing the F/stop won't make the artifacts go away. It seems only disabling bump or DoF makes a difference.

  • wowiewowie Posts: 2,029

    Well if I take a photo of a house, I want it to be sharp and not out of focus. Having the focal point at the front door means a focal length of 2400. If I reduce the focal length to 300 the house is not in focus without fiddeling with the F/stop. Anyway, changing the F/stop won't make the artifacts go away. It seems only disabling bump or DoF makes a difference.

    Something is not right here. Why would you use a focal length of 2400 mm for a shot like that? Or even 300 mm?

    If you want to avoid distortion, then 35-80 mm would be best.

    If you can't change the focal distance, change the actual camera distance. You can use the same f-stop.

    I suggested changing the frame width and focal length to accomodate changes to focal distance since the problem is related to focal distance. Feel free to correct me if i'm wrong. It's been a long time since I've fiddle with manual camera settings in real life.

  • Sven DullahSven Dullah Posts: 7,621
    edited November 2018
    wowie said:

    Well if I take a photo of a house, I want it to be sharp and not out of focus. Having the focal point at the front door means a focal length of 2400. If I reduce the focal length to 300 the house is not in focus without fiddeling with the F/stop. Anyway, changing the F/stop won't make the artifacts go away. It seems only disabling bump or DoF makes a difference.

    Something is not right here. Why would you use a focal length of 2400 mm for a shot like that? Or even 300 mm?

    If you want to avoid distortion, then 35-80 mm would be best.

    If you can't change the focal distance, change the actual camera distance. You can use the same f-stop.

    I suggested changing the frame width and focal length to accomodate changes to focal distance since the problem is related to focal distance. Feel free to correct me if i'm wrong. It's been a long time since I've fiddle with manual camera settings in real life.

    Oh I'm sorry, I meant focal distance, not length, my badsadblushblush

    I'll take a new look at that scene, think I'll start with looking into the materials first, it seems there's something odd with the bump settings. Tks a lot for being so patient with me:)

    Post edited by Sven Dullah on
  • Sven DullahSven Dullah Posts: 7,621
    edited November 2018

    ...a WIP of https://www.daz3d.com/sports-car-morris converted to awe + a custom character, think it's going in the right direction;) Tuscan Villa would make for a nice backdrop...

    image

    Morris1 awe.png
    1280 x 720 - 1M
    Post edited by Sven Dullah on
  • Sven DullahSven Dullah Posts: 7,621
    edited November 2018

    ...ok so I fixed the artifacts on the house walls by editing bump min-max values, or maybe it's the new lighting, or maybe camera placement/parameters. Now off to fight the new artifacts on the gravel (the dark horisontal line/shadow across the whole render). Made some efforts changing bump values, and spot rendering, no luck so far. Turning DoF off fixes it of course. Also changing the spec BRDF made no difference.

    Camera focal length 65, focal distance 695, frame width 31

    image

    Morris awe.png
    1800 x 1013 - 3M
    Post edited by Sven Dullah on
  • Sven DullahSven Dullah Posts: 7,621
    edited November 2018

    ...reduced the focal distance until the artifacts went away, ended up with 300, then increased F/Stop to keep sharpness...

    image

    Morris awe.png
    1800 x 1013 - 3M
    Post edited by Sven Dullah on
  • Sven DullahSven Dullah Posts: 7,621
    edited December 2018

    I noticed another strange thing with those DoF artifacts, sometimes spotrendering the problem area without changing anything fixes them;)

    And it really seems keeping focal distance under 500 in every situation is a good thing to dosmiley

    Post edited by Sven Dullah on
  • wowiewowie Posts: 2,029
    edited December 2018

    OK. Found a way to make all shader parameters to be animated. So, that's another old request checked off. Will apply it to the all shaders, just in case.

    Edit: Animating colors just got way easier. :D

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

    OK. Found a way to make all shader parameters to be animated. So, that's another old request checked off. Will apply it to the all shaders, just in case.

    Edit: Animating colors just got way easier. :D

    Wow... can't waitblushlaugh This is awesome!

  • wowiewowie Posts: 2,029
    edited December 2018

    Just finished uploading AWE Surface 1.1 and mustakettu's RadiumAreaPT light shader files to the Google Drive folder. I've also finished the shadow catcher shader along with new point/spot/distant light shaders. Both can be found on the freebie thread.

    https://www.daz3d.com/forums/discussion/277581/awe-surface-shader-a-new-physically-plausible-shader-for-daz-studio-and-3delight/p1

    Let me know if you need any help.

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

    Tks a lot wowie, wow didn't see those lights comingsmiley

Sign In or Register to comment.