Daz Studio Pro BETA - version 4.20.0.17! (*UPDATED*)

1101113151633

Comments

  • leemoon_c43b45a114leemoon_c43b45a114 Posts: 866
    edited December 2021

    VEGA said:

    leemoon_c43b45a114 said:

    I'm attempting to use the ghostlights I created in earlier versions of DS.

    For my ghostlight(s), I have a primitive plane with a black texture map in the cutout opacity slot with the cutout opacity parameter set to 1.0

    Emission color is 1,1,1

    Emission Temp (K) is 8725

    Emission Profile is None

    Luminance is 625

    Luminance Units is kcd/m^2

    These ghostlight settings have been working great for me.  With the updated Iray engine, no matter what I've been setting Luminance to, there's no change in intensity.  The scene is only getting illumination from my environment HDRI (set at a low intensity as general fill light/global illumination.). My ghostlights are the primary lights: one main fill light and an instance of it as a rim light.  These are parented to and set to point at a null object in the middle of the scene at approximately neck height for a standard character.

    I'm having trouble getting my head around adjusting the ghostlight(s) to get the illumination I had before.  I was getting great render speed with my scene setup.

    Anyway, I'm still poking around in the forums looking for something that I can adapt for my particular use case.

    Lee

     Revert back your DAZ or stop using them, because they don't work.

    This is on an M1 MacBook Air that I use for testing and other lightweight work.  So no worries there. :)

    I have another 2 Macs (M1 iMac and i9 16" MacBook Pro) that I use for actual rendering.  The i9 MacBook Pro is a little bit faster rendering than the M1 Macs.  So that's my preferred machine.

    I could just use my ghostlight setup without the ghost part, by removing the opacity map.  Then I'd have my emissive planes providing the same(?) illumination, just visible in the scene (depending on position.)

    We'll see, but I sure wish Nvidia would provide an accomodation for ghostlighting in a future release.

    Lee

    Post edited by leemoon_c43b45a114 on
  • DeeKay said:

    You will need to remove the black texture map. To stop the ghost light being seen in the render set the “Iray Visible to Primary Rays” property, as described earlier in this discussion. 

    Where is that discussion, please?

  • I think I have my revised ghostlights working again.  Will need to test a bit more, but it looks very promising now. :)

    Thank you to the DAZ developers for the workaround and to IceCrMn for the instructions on how to set up the "new" ghostlights.

    Here's a link to the discussion earlier in this thread: https://www.daz3d.com/forums/discussion/comment/7175736/#Comment_7175736

    Thank you all again for the awesome help!! :)

    Lee

  • DeeKayDeeKay Posts: 36

    Could a script jocky provide a script that will set and toggle the “Iray Visible to Primary Rays” property? I know it is easy enough to edit this, but it would just save a little fiddling about. Many thanks!

  • nonesuch00nonesuch00 Posts: 18,142

    With this new 4.16.1.17 public beta, I just CPU (an AMD Ryzen 7 5700G) rendered a 2500x2500 iRay scene in just under 3 hours with about 1100 iterations (max time set to infinity and max iterations set to 5000) to converge to 100% at 95% render quality which is fastest CPU render yet to my knowlege. There is a catch though, I was sitting at my computer working on other things for the entire 3 hours; if I go away from my computer and the computer goes to sleep then for some reason the convergence doesn't happen at the same rate, rendering slows to a crawl, and the number of iterations needed to converge to 100% easily exceeds 2000.

  • HavosHavos Posts: 5,367

    To get ghost lights working that use the low opacity option, eg: an opacity of 0.000001 (ie: not using the black texture in the opacity channel option), to fix the issue is it not just a matter of dividing the current lumen value by the opacity value (and this will of course multiply the lumens significantly) and thus this cancels out the fact that the new Iray is multiplying the lumens by the opacity to get the light being emitted.

    In theory someone could write a small script that would do this correction automatically, so converting old scenes to cast the same light as before would be as trivial as selecting all the ghost lights and then applying the script.

  • kyoto kidkyoto kid Posts: 41,070
    edited December 2021

    nonesuch00 said:

    With this new 4.16.1.17 public beta, I just CPU (an AMD Ryzen 7 5700G) rendered a 2500x2500 iRay scene in just under 3 hours with about 1100 iterations (max time set to infinity and max iterations set to 5000) to converge to 100% at 95% render quality which is fastest CPU render yet to my knowlege. There is a catch though, I was sitting at my computer working on other things for the entire 3 hours; if I go away from my computer and the computer goes to sleep then for some reason the convergence doesn't happen at the same rate, rendering slows to a crawl, and the number of iterations needed to converge to 100% easily exceeds 2000.

    ..could you go into your system's power settings before rendering and disable sleep mode? 

    Post edited by kyoto kid on
  • ImagoImago Posts: 5,167

    New version, old bugs. They are here since five years and they are going for the sixth.
    Timeline, puppeteer, Frame selected, interface, sliders and a lot of other stuff related to animation are broken. The only usable release is 4.12.0.86 that's really old now.

    @ Richard: Yes, I reported these bugs multiple times in the latest years, so many times I can't even remember how many. I even sent another report one week ago, still waiting an answer. It's really that hard fix those bugs? If it's THAT hard, just put the 4.12.0.86 versions of those tools in the latest release! At least I can have the chance to try Filament and the other changes.

  • Imago said:

    New version, old bugs. They are here since five years and they are going for the sixth.
    Timeline, puppeteer, Frame selected, interface, sliders and a lot of other stuff related to animation are broken. The only usable release is 4.12.0.86 that's really old now.

    @ Richard: Yes, I reported these bugs multiple times in the latest years, so many times I can't even remember how many. I even sent another report one week ago, still waiting an answer. It's really that hard fix those bugs? If it's THAT hard, just put the 4.12.0.86 versions of those tools in the latest release! At least I can have the chance to try Filament and the other changes.

    It'snot just a matter of how hard the bug is to fix but what the cinsequences would be and whether that section of the code is receiving attention anyway, to say nothing of the actual impact (the frame bug is annoying but it affetcs mainly situations where the target is away from the origin, which reduces the number of peoplke affected, and has a relatively simple work-around, make sure the camera is pointing at the origin, so it may well be given a lower priority than another bug which affects more users or has not work around).

  • nonesuch00nonesuch00 Posts: 18,142

    kyoto kid said:

    nonesuch00 said:

    With this new 4.16.1.17 public beta, I just CPU (an AMD Ryzen 7 5700G) rendered a 2500x2500 iRay scene in just under 3 hours with about 1100 iterations (max time set to infinity and max iterations set to 5000) to converge to 100% at 95% render quality which is fastest CPU render yet to my knowlege. There is a catch though, I was sitting at my computer working on other things for the entire 3 hours; if I go away from my computer and the computer goes to sleep then for some reason the convergence doesn't happen at the same rate, rendering slows to a crawl, and the number of iterations needed to converge to 100% easily exceeds 2000.

    ..could you go into your system's power settings before rendering and disable sleep mode? 

    I haven't tried. I'd just soon leave it alone as I don't render frequently.

    Maybe it's the new "Guided Sampling" mode in the iRay renderer? I did enable that!

    Or last possibility was that I had had the morph "iris concave" dialed in a 100% but then I dialed it out to 0%, I doubt that morph was so compute intensive though.

    Where is the access and documentation to this new USD format in DAZ Studio!?

  • ImagoImago Posts: 5,167

    Richard Haseltine said:

    It'snot just a matter of how hard the bug is to fix but what the cinsequences would be and whether that section of the code is receiving attention anyway, to say nothing of the actual impact (the frame bug is annoying but it affetcs mainly situations where the target is away from the origin, which reduces the number of peoplke affected, and has a relatively simple work-around, make sure the camera is pointing at the origin, so it may well be given a lower priority than another bug which affects more users or has not work around).

    If you know a way to make the Puppeteer and the Fitting work correctly, please tell me. I can find a different method for the timeline but these two are really bad bugs.

  • wsterdan said:

    olegov.nsk_e159069e49 said:

    Ouch... I mean Windows in the Parallels virtual machine on a mac os

    Sorry, not at my machine at the moment, but select "Filament" in the Renderstyles drop down menu  in the Viewport, and in the Render Settings choose "Viewport".

    Hope that helps.

    -- Walt Sterdan

    Unfortunately it doesn't work :( 

  • takezo_3001takezo_3001 Posts: 1,987

    Imago said:

    New version, old bugs. They are here since five years and they are going for the sixth.
    Timeline, puppeteer, Frame selected, interface, sliders and a lot of other stuff related to animation are broken. The only usable release is 4.12.0.86 that's really old now.

    @ Richard: Yes, I reported these bugs multiple times in the latest years, so many times I can't even remember how many. I even sent another report one week ago, still waiting an answer. It's really that hard fix those bugs? If it's THAT hard, just put the 4.12.0.86 versions of those tools in the latest release! At least I can have the chance to try Filament and the other changes.

    This is why I despise workarounds so much, as it's so easy to ignore/bypass fixing bugs when you could just use workarounds... in reality, those unfixed bugs may contribute to other bugs and hidden instabilities just like the current one that they have with the viewport being as slow as molasses making this beta (Like the last 2) impossible to use if you have a scene larger than having just one figure in it!

    I may be guessing here but there is a distinct possibility those bugs that you mentioned may never get fixed as we're going to be transitioning over to DS5 soon enough, so what would be the point of fixing bugs in an obsolete program?

  • The RED CrownThe RED Crown Posts: 247
    edited December 2021

    Imago said:

    Richard Haseltine said:

    It'snot just a matter of how hard the bug is to fix but what the cinsequences would be and whether that section of the code is receiving attention anyway, to say nothing of the actual impact (the frame bug is annoying but it affetcs mainly situations where the target is away from the origin, which reduces the number of peoplke affected, and has a relatively simple work-around, make sure the camera is pointing at the origin, so it may well be given a lower priority than another bug which affects more users or has not work around).

    If you know a way to make the Puppeteer and the Fitting work correctly, please tell me. I can find a different method for the timeline but these two are really bad bugs.

    Don't make headache to yourself , just ignore creating more tickets for nothing . just put your hope on DS5 .

    Post edited by The RED Crown on
  • skboaskboa Posts: 61

    Okay, I've found some very odd bug with "Guided Sampling" activated.

    I've made some ghost lights and used a combination of "Iray visible to primary rays" and the oldschool "cutout to 000001" and this is what happens when I activate guided sampling (see screenshots).

    It doesn't happen when it's disabled.

    screen1.JPG
    3838 x 2085 - 501K
    screen2.JPG
    3838 x 2105 - 541K
  • I have a very weird issue: The skin colors of any character in 4.16.1.17 completely went crazy (see attachment, both with V8.1 and the G8F Chanel character). Is anybody experiencing this issue, too?

    Any idea what could be done against it?

    What caught my attention was this log entry:

    "

    2021-12-19 22:59:41.803 Prepare asset load (merge): /People/Genesis 8 Female/Characters/Chanel.duf
    2021-12-19 22:59:41.804 Locking viewport redraw...
    2021-12-19 22:59:41.804 Viewport redraw locked.
    2021-12-19 22:59:42.323 Native format content directories: 11
    2021-12-19 22:59:42.323 Poser format content directories: 1
    2021-12-19 22:59:42.323 Other import format content directories: 0
    2021-12-19 22:59:42.323 Begin asset load (merge): /People/Genesis 8 Female/Characters/Chanel.duf
    2021-12-19 22:59:42.360 Determining missing assets...
    2021-12-19 22:59:42.540 Setting textures...
    2021-12-19 23:00:07.574 JSON interpreter error: Too many values for color.
    2021-12-19 23:00:33.173 JSON interpreter warning: Unexpected boolean member: hidden

    "

    Any help appreciated.

     

    2021-12-20_weird_skin_shaders_4.16.1.17.png
    1024 x 633 - 394K
  • prixatprixat Posts: 1,588

    skboa said:

    Okay, I've found some very odd bug with "Guided Sampling" activated.

    I've made some ghost lights and used a combination of "Iray visible to primary rays" and the oldschool "cutout to 000001" and this is what happens when I activate guided sampling (see screenshots).

    It doesn't happen when it's disabled.

    I get the same, but with just a spotlight and low light conditions (and possibly intersecting geometry). Caustics 'ON' seems to fix it, but I have not done any time comparisons, and is using Caustics just switching off Guided Sampling anyway?

  • skboaskboa Posts: 61

    prixat said:

    skboa said:

    Okay, I've found some very odd bug with "Guided Sampling" activated.

    I've made some ghost lights and used a combination of "Iray visible to primary rays" and the oldschool "cutout to 000001" and this is what happens when I activate guided sampling (see screenshots).

    It doesn't happen when it's disabled.

    I get the same, but with just a spotlight and low light conditions (and possibly intersecting geometry). Caustics 'ON' seems to fix it, but I have not done any time comparisons, and is using Caustics just switching off Guided Sampling anyway?

    I still have it with caustics on, but it's a lot less noticeable. Biggest problem for me is that I can't use the ghostlights with caustics enables :/

  • prixatprixat Posts: 1,588
    edited December 2021

    Another problem with Guided Sampling (and a bright light), it's finding the subdivision edges!!!!

    The cube is subdiv 4, and we also get some square artifacts for free. laugh

    subdiv4.png
    1500 x 1500 - 722K
    Post edited by prixat on
  • takezo_3001takezo_3001 Posts: 1,987

    Well it took a virus and a reinstallation of windows but it seems that it was just some random setting/prog that screwed up my viewport, it's still not as smooth as the general release, but it's a lot smother than 1-2 fps!

  • GLEGLE Posts: 52

    Imago said:

    tanutanu said:

    With Google Translate

    If you have a known problem, skip it.

    Launch daz studio.
    Load G8F.
    Launch Morph loader Pro.
    daz studio crashes.

    I tried reinstalling, but it was the same.

    Latest G8 releases are broken, try using an older package, perhaps it helps.

    I'm on the package previous the launch of the Face Transfer and everything "works fine", chars loads in seconds with lots of morphs and I can add more whenever I want.

    I tested this, with a G8 package released Sept 5th, 2019 (Face Transfer was released Sept 24th, 2019). There's no change compared to the package released June 17th, 2021.

    For me, both yield about 40 seconds of loading time in the logfile, from the line Begin asset load (open): /People/Genesis 8 Female/Genesis 8 Basic Female.duf to the line Loaded image: G8FBaseEyelashes_1006.jpg.

    There are about 10000 files in the Morphs subfolder. The Library was set to a SATA Samsung SSD, and new never used instances of Studio were created (I'm still using the PowerShell -instanceName switch, is there a more user friendly way to manage instances?)

     

    The phantom Daz Studio process in Task Manager doesn't take ages to close anymore, very good. It's not istantaneous, but it doesn't take 15 minutes, more like 2-3 so about 5 times faster it seems.

  • marblemarble Posts: 7,500
    edited December 2021

    GLE said:

    The phantom Daz Studio process in Task Manager doesn't take ages to close anymore, very good. It's not istantaneous, but it doesn't take 15 minutes, more like 2-3 so about 5 times faster it seems.

    I still do not understand why (and I'm still on 4.15), when I close DAZ Studio - especially if I have G8F in the scene - it takes minutes but if, instead, I just click "New" scene without closing, that works instantaneously. No delay. I'm so suspicious that there is something not happening that should happen that I don't take that option very often and rather close DAZ Studio but maybe I'm just paranoid? 

    Post edited by marble on
  • jbowlerjbowler Posts: 794

    marble said:

    I still do not understand why (and I'm still on 4.15), when I close DAZ Studio - especially if I have G8F in the scene - it takes minutes but if, instead, I just click "New" scene without closing, that works instantaneously. No delay.

    The close time goes in unlinking the stuff in the scene.  If you delete the scene yourself ('new' does this) the scene gets copied to the undo stack and Daz's algorithm for deletion of stuff on the undo stack works much faster.  There was some speculation that it was a known issue in Qt, but for certain the work round was to delete first then close.

    This does seem to be fixed in 4.16.  I'm not sure about G8F (etc) load times; they do seem faster but the DSON cache building seems to be slower.  This is an acceptable trade off because the G8F cache is only rebuilt if something G8F can referred to gets changed (typically added, e.g. adding a character will, I believe, cause the DSON cache for the underlying base figure to be invalidated.)

  • And so. I tried to bring back the ghost lights in the new beta version.
    The method described on page 9, hides only the visibility of the light source. 
    But it does not hide reflections in mirrors.That is, now the ghost lights work the same way as the spotlight with the setting "visible in render - OFF".
    I used ghost lights only because they do not give reflections in the mirror. Now it doesn't work. Is there any other way to fix this? Or will I have to go back to the previous version of Daz3d?

    Translated with www.DeepL.com/Translator (free version)

  • takezo_3001takezo_3001 Posts: 1,987
    edited December 2021

    Jovanni said:

    And so. I tried to bring back the ghost lights in the new beta version.
    The method described on page 9, hides only the visibility of the light source. 
    But it does not hide reflections in mirrors.That is, now the ghost lights work the same way as the spotlight with the setting "visible in render - OFF".
    I used ghost lights only because they do not give reflections in the mirror. Now it doesn't work. Is there any other way to fix this? Or will I have to go back to the previous version of Daz3d?

    Translated with www.DeepL.com/Translator (free version)

    The solution is pretty simple... just use the advanced iray shaders and apply it to a primitive, enabling emissions while blacking out the diffuse and glossy channels.

    EDIT: The solution does not work in the beta, I mistakenly tested it in the general release...

    Post edited by takezo_3001 on
  • marblemarble Posts: 7,500

    jbowler said:

    marble said:

    I still do not understand why (and I'm still on 4.15), when I close DAZ Studio - especially if I have G8F in the scene - it takes minutes but if, instead, I just click "New" scene without closing, that works instantaneously. No delay.

    The close time goes in unlinking the stuff in the scene.  If you delete the scene yourself ('new' does this) the scene gets copied to the undo stack and Daz's algorithm for deletion of stuff on the undo stack works much faster.  There was some speculation that it was a known issue in Qt, but for certain the work round was to delete first then close.

    This does seem to be fixed in 4.16.  I'm not sure about G8F (etc) load times; they do seem faster but the DSON cache building seems to be slower.  This is an acceptable trade off because the G8F cache is only rebuilt if something G8F can referred to gets changed (typically added, e.g. adding a character will, I believe, cause the DSON cache for the underlying base figure to be invalidated.)

    I hope they can also do something about that other cause of long delays - LIE. If I load a character with unbaked LIE I just sit and wait .... and wait.

  • Iray Blended Dual Lobe 发丝着色器现在呈现为完全黑色。

  • RobotHeadArtRobotHeadArt Posts: 917
    edited December 2021

    The new ground fog reder options give some pretty good results.  I wish there was some documentation that explained exactly what the parameters do.  I think it would look even more realistic if Nvidia added some options for multiplying perlin noise against the density to give it a more wispy and dynamic look.  I think it renders faster and looks better than the old cube with volumetric SSS shader trick for fog.  I did briefly try to mix matte fog with ground fog and got some really ugly results so I am not sure if these two are supposed to work together or are mutually exclusive.

    ground fog.png
    650 x 500 - 493K
    Post edited by RobotHeadArt on
  • nonesuch00nonesuch00 Posts: 18,142

    RobotHeadArt said:

    The new ground fog reder options give some pretty good results.  I wish there was some documentation that explained exactly what the parameters do.  I think it would look even more realistic if Nvidia added some options for multiplying perlin noise against the density to give it a more wispy and dynamic look.  I think it renders faster and looks better than the old cube with volumetric SSS shader trick for fog.  I did briefly try to mix matte fog with ground fog and got some really ugly results so I am not sure if these two are supposed to work together or are mutually exclusive.

    Be great if they added different types of precipitation and clouds too.

  • RobotHeadArt said:

    The new ground fog reder options give some pretty good results.  I wish there was some documentation that explained exactly what the parameters do.  I think it would look even more realistic if Nvidia added some options for multiplying perlin noise against the density to give it a more wispy and dynamic look.  I think it renders faster and looks better than the old cube with volumetric SSS shader trick for fog.  I did briefly try to mix matte fog with ground fog and got some really ugly results so I am not sure if these two are supposed to work together or are mutually exclusive.

    The highlights thread does give the parameters https://www.daz3d.com/forums/discussion/comment/7219956/#Comment_7219956

Sign In or Register to comment.