Daz Studio Pro BETA - version 4.11.0.335! (*UPDATED*)

1242527293032

Comments

  • Wow I didn't expect to come home to a new Beta build to play with!

  • barbultbarbult Posts: 23,118

    Oh, that's exciting. It's been a long time.

  • randy88randy88 Posts: 16

    Wow, I want to go back early and try a new beta version.

  • TheKDTheKD Posts: 2,674
    edited April 2019

    Damn it, I just installed the new version, and it keeps fatal error crashing while opening the scene I been working on in the old beta. As far as I can tell it has something to do with the LAMH wolf 2.0, as this is the last lines in the log file before it shits the bed

    2019-04-18 01:18:02.862 WARNING: Object::disconnect: No such slot lookAtMyHairPane::nodeRemoved() in fileio.cpp:2908
    2019-04-18 01:18:02.862 WARNING: Object::disconnect:  (sender name:   'Wolf_Grp_0_RMan_body')
    2019-04-18 01:18:02.862 WARNING: Object::disconnect:  (receiver name: 'Look at my Hair')
    2019-04-18 01:18:02.862 WARNING: Object::disconnect: No such slot lookAtMyHairPane::nodeRemoved() in fileio.cpp:2908
    2019-04-18 01:18:02.862 WARNING: Object::disconnect:  (sender name:   'Wolf_Grp_1_RMan_body')
    2019-04-18 01:18:02.862 WARNING: Object::disconnect:  (receiver name: 'Look at my Hair')

     

     

    Oh, this is in the box that pops up

    DAZStudio.exe caused ACCESS_VIOLATION in module "C:\Program Files\DAZ 3D\DAZStudio4 Public Build\plugins\lamhPlugin.64.dll" at 0033:0000000008E42A0F, DzLine3::operator=()+723855 byte(s)

    Not sure if that's helpful lol.


    Oh balls, now it's crashing non beta version as well, this is not good at all.......

    Post edited by TheKD on
  • antaresantares Posts: 26

    Wow, big thx for this working Update!

  • shaneseymourstudioshaneseymourstudio Posts: 383
    edited April 2019
    TheKD said:

    Damn it, I just installed the new version, and it keeps fatal error crashing while opening the scene I been working on in the old beta. As far as I can tell it has something to do with the LAMH wolf 2.0, as this is the last lines in the log file before it shits the bed

    2019-04-18 01:18:02.862 WARNING: Object::disconnect: No such slot lookAtMyHairPane::nodeRemoved() in fileio.cpp:2908
    2019-04-18 01:18:02.862 WARNING: Object::disconnect:  (sender name:   'Wolf_Grp_0_RMan_body')
    2019-04-18 01:18:02.862 WARNING: Object::disconnect:  (receiver name: 'Look at my Hair')
    2019-04-18 01:18:02.862 WARNING: Object::disconnect: No such slot lookAtMyHairPane::nodeRemoved() in fileio.cpp:2908
    2019-04-18 01:18:02.862 WARNING: Object::disconnect:  (sender name:   'Wolf_Grp_1_RMan_body')
    2019-04-18 01:18:02.862 WARNING: Object::disconnect:  (receiver name: 'Look at my Hair')

     

     

    Oh, this is in the box that pops up

    DAZStudio.exe caused ACCESS_VIOLATION in module "C:\Program Files\DAZ 3D\DAZStudio4 Public Build\plugins\lamhPlugin.64.dll" at 0033:0000000008E42A0F, DzLine3::operator=()+723855 byte(s)

    Not sure if that's helpful lol.


    Oh balls, now it's crashing non beta version as well, this is not good at all.......

     

    Yeah I have experienced a few crashes and issues with scenes I had built. Mine happened when I tried to get the eyes to get out of crosseyed position where they were targeting an object in the scene. One thing I am concerned with is if I can still go back to the .196 build, if necessary, or not. The rendering time seems to be faster from what I have tested so far.

    I had a scene that I built in the new Beta version to test denoising on and had something odd happen. The scene first caused Daz to close suddenly when it got to the 200th iteration for the denoise to kick in. I reopened the scene and changed the start iteration to 50 and had the viewport in iRay and then hit the render button. This opened the new window and it fell back to CPU. I closed Daz again and then opened and reloaded the scene and verified all the settings had saved including the 50th iteration to start denoise. I ran it again this time leaving viewport in the shaded view instead of iRay in case that was causing an issue. The render started, this time on the GPU. The interesting part is that it hit 50 and kept going but no denoise. I double checked the options on right side of the render window and it had denoise still set to start at 50. Not sure why but it did not kick in. I left it running and now it is at 179 and of course there is still no denoising.

    Post edited by shaneseymourstudio on
  • areg5areg5 Posts: 617

    I've been playing around with the Beta, and like I mentioned before, I'm seeing differences in rendering speed compared to 4.10.  When I first tried it, I thought it rendered about 3X faster.  This was on an outdoor scene with HDRI lighting.  I noticed a profound slowing when using MESH lighting compared to 4.10.  I've been working now with an indoor scene lit mainly with emissives and pointlights.  Like the MESH lighting example, it renders much slower than 4.10.  Not only that, but the number of samples is markedly increased.

    I'm hoping that the developers will address this.  When 4.11 is finally launched, this slowdown in indoor renders could be really bad.  Most of my scenes are indoor scenes.

  • shaneseymourstudioshaneseymourstudio Posts: 383
    edited April 2019
    areg5 said:

    I've been playing around with the Beta, and like I mentioned before, I'm seeing differences in rendering speed compared to 4.10.  When I first tried it, I thought it rendered about 3X faster.  This was on an outdoor scene with HDRI lighting.  I noticed a profound slowing when using MESH lighting compared to 4.10.  I've been working now with an indoor scene lit mainly with emissives and pointlights.  Like the MESH lighting example, it renders much slower than 4.10.  Not only that, but the number of samples is markedly increased.

    I'm hoping that the developers will address this.  When 4.11 is finally launched, this slowdown in indoor renders could be really bad.  Most of my scenes are indoor scenes.

    Are you using the 2 1080ti's in your rig description? I ask to see if there may be some difference with using an RTX card or not. I am using an RTX 2080 alone for my testing so far. I notice faster renders in indoors outdoors and hdri only across the board. Indoors I use only spotlighting and whatever lights the scene has setup already. I will circle back around and try these scenes with my P5000 by itself and try to compare soon. I have not used 4.10 in quite awhile but will try the scenes in that at some point as well. But for me it seems faster than the prior Beta (.236).

    Post edited by shaneseymourstudio on
  • AdvintAdvint Posts: 23

    I need to do some testing, but it appears that dforce is slower with the 425.31 drivers for 2080ti's. As well I cancelled a simulation and continueing to try to simulate with the same device was an utter failure. Restarting the simulation resulted in the initialization running indefinitely (I ran it overnight), having cancelled it at 8 hours in, with no end to initialization in site. I went ahead and swapped it to using the other card I have for simulating this morning after restarting. I'll update if it is successful or not later on.

  • WendyLuvsCatzWendyLuvsCatz Posts: 37,784
    edited April 2019

    oh great I cannot use iray now at all with 417 which I need for Octane 4

    DzIrayRender\dzneuraymgr.cpp(302): Iray ERROR - module:category(IRAY:RENDER): 1.0 IRAY rend error: Cannot render: found no usable devices.

     

    Post edited by WendyLuvsCatz on
  • oh great I cannot use iray now at all with 417 which I need for Octane 4

    DzIrayRender\dzneuraymgr.cpp(302): Iray ERROR - module:category(IRAY:RENDER): 1.0 IRAY rend error: Cannot render: found no usable devices.

     

    I believe Octane can use the latest drivers.

  • WendyLuvsCatzWendyLuvsCatz Posts: 37,784

    no just installed 425.31 and Octane 4 plugin for Carrara is not working anymore

    I do have my old nvidia installers

    so will just have to live without iray I guess, I have Octane 2 for DAZ studio

  • areg5areg5 Posts: 617
    edited April 2019
    areg5 said:

    I've been playing around with the Beta, and like I mentioned before, I'm seeing differences in rendering speed compared to 4.10.  When I first tried it, I thought it rendered about 3X faster.  This was on an outdoor scene with HDRI lighting.  I noticed a profound slowing when using MESH lighting compared to 4.10.  I've been working now with an indoor scene lit mainly with emissives and pointlights.  Like the MESH lighting example, it renders much slower than 4.10.  Not only that, but the number of samples is markedly increased.

    I'm hoping that the developers will address this.  When 4.11 is finally launched, this slowdown in indoor renders could be really bad.  Most of my scenes are indoor scenes.

    Are you using the 2 1080ti's in your rig description? I ask to see if there may be some difference with using an RTX card or not. I am using an RTX 2080 alone for my testing so far. I notice faster renders in indoors outdoors and hdri only across the board. Indoors I use only spotlighting and whatever lights the scene has setup already. I will circle back around and try these scenes with my P5000 by itself and try to compare soon. I have not used 4.10 in quite awhile but will try the scenes in that at some point as well. But for me it seems faster than the prior Beta (.196).

    I have 3 1080 Ti's.  It's likely due to the emissive lights I would think.  When I light an indoor scene, what I do is change the light bulbs in the room to emissive, because it seems like more natural lighting than spots.  It takes the Beta at least twice as long as 4.10 to render these scenes.  I mean, given 3 cards it goes quickly regardless, but I do note both increased sampling and longer times with the beta.

    Post edited by areg5 on
  • jag11jag11 Posts: 885

    Error trying to open the Dynamic American Flag:

    2019-04-18 12:22:47.665 Opening file american flag pole.daz...
    2019-04-18 12:22:47.739 WARNING: QObject::connect: Cannot connect (null)::currentValueChanged() to DzSkeleton::handleDrawGLBonesChanged()
    2019-04-18 12:22:47.750 Error reading file C:/Users/jag11/Documents/DAZ 3D/Studio/My Daz Connect Library/data/cloud/1_10944/dynamic clothing props/american flag pole.daz : C Exception: Memory Access Violation - Attempted to read memory at address 0x00000000
    2019-04-18 12:22:47.750 File read time: 0 min 0.0 sec.
    2019-04-18 12:22:47.758 Building Scene...

    But works okay in 4.10

  • jag11 said:

    Error trying to open the Dynamic American Flag:

    2019-04-18 12:22:47.665 Opening file american flag pole.daz...
    2019-04-18 12:22:47.739 WARNING: QObject::connect: Cannot connect (null)::currentValueChanged() to DzSkeleton::handleDrawGLBonesChanged()
    2019-04-18 12:22:47.750 Error reading file C:/Users/jag11/Documents/DAZ 3D/Studio/My Daz Connect Library/data/cloud/1_10944/dynamic clothing props/american flag pole.daz : C Exception: Memory Access Violation - Attempted to read memory at address 0x00000000
    2019-04-18 12:22:47.750 File read time: 0 min 0.0 sec.
    2019-04-18 12:22:47.758 Building Scene...

    But works okay in 4.10

    Open in 4.10 and save as a .duf scene. The .daz format files are very problematic.

  • barbultbarbult Posts: 23,118
    I'm looking for more info on the new features. Post 1 still says "in progress ". It is difficult to infer anything from the change log. What improvements have been added to dForce? What can we do with polylines now? Etc.
  • EboshijaanaEboshijaana Posts: 437

    I see two new options in render options; 

    - In the optimization, you can now select auto from the dropdown menu.
    - In Filtering, there is a Denoise option relating to alpha (rendering atm so can't look.)

  • Dolce SaitoDolce Saito Posts: 148
    edited April 2019

    A few things that I noticed immediately:

    1- Much faster renders for me (Titan V + 980 GTX)

    2- FINALLY denoiser works correctly on Titan V

    3- Optix Prime Causes CTD on image series render after few frames.

    4- Optix Prime causes CTD on too much camera repositioning on iray preview mode when two of my cards selected. These CTDs have no log, daz window suddenly disappears in the beginning/middle of the render.

    5- DForce is incredibly more stable than the previous beta. I can't even tell how much place it tends to blow up now instead figures out without explosions and continues animation.

    Post edited by Dolce Saito on
  • areg5areg5 Posts: 617
    areg5 said:
    areg5 said:

    I've been playing around with the Beta, and like I mentioned before, I'm seeing differences in rendering speed compared to 4.10.  When I first tried it, I thought it rendered about 3X faster.  This was on an outdoor scene with HDRI lighting.  I noticed a profound slowing when using MESH lighting compared to 4.10.  I've been working now with an indoor scene lit mainly with emissives and pointlights.  Like the MESH lighting example, it renders much slower than 4.10.  Not only that, but the number of samples is markedly increased.

    I'm hoping that the developers will address this.  When 4.11 is finally launched, this slowdown in indoor renders could be really bad.  Most of my scenes are indoor scenes.

    Are you using the 2 1080ti's in your rig description? I ask to see if there may be some difference with using an RTX card or not. I am using an RTX 2080 alone for my testing so far. I notice faster renders in indoors outdoors and hdri only across the board. Indoors I use only spotlighting and whatever lights the scene has setup already. I will circle back around and try these scenes with my P5000 by itself and try to compare soon. I have not used 4.10 in quite awhile but will try the scenes in that at some point as well. But for me it seems faster than the prior Beta (.196).

    I have 3 1080 Ti's.  It's likely due to the emissive lights I would think.  When I light an indoor scene, what I do is change the light bulbs in the room to emissive, because it seems like more natural lighting than spots.  It takes the Beta at least twice as long as 4.10 to render these scenes.  I mean, given 3 cards it goes quickly regardless, but I do note both increased sampling and longer times with the beta.

    The file attached is a case in point:  2 figures, emissive lighting and 1 point light.  This image rendered in 9:02 at 5134 iterations using 4.10.  It rendered in 14:34 and 5523 iterations using the Beta. This is a simple scene. More complex indoor scenes I have found are morkedly slower using the Beta, and the iterations can be quadrupled.

    k38.jpg
    1280 x 720 - 599K
  • areg5 said:
    areg5 said:
    areg5 said:

    I've been playing around with the Beta, and like I mentioned before, I'm seeing differences in rendering speed compared to 4.10.  When I first tried it, I thought it rendered about 3X faster.  This was on an outdoor scene with HDRI lighting.  I noticed a profound slowing when using MESH lighting compared to 4.10.  I've been working now with an indoor scene lit mainly with emissives and pointlights.  Like the MESH lighting example, it renders much slower than 4.10.  Not only that, but the number of samples is markedly increased.

    I'm hoping that the developers will address this.  When 4.11 is finally launched, this slowdown in indoor renders could be really bad.  Most of my scenes are indoor scenes.

    Are you using the 2 1080ti's in your rig description? I ask to see if there may be some difference with using an RTX card or not. I am using an RTX 2080 alone for my testing so far. I notice faster renders in indoors outdoors and hdri only across the board. Indoors I use only spotlighting and whatever lights the scene has setup already. I will circle back around and try these scenes with my P5000 by itself and try to compare soon. I have not used 4.10 in quite awhile but will try the scenes in that at some point as well. But for me it seems faster than the prior Beta (.196).

    I have 3 1080 Ti's.  It's likely due to the emissive lights I would think.  When I light an indoor scene, what I do is change the light bulbs in the room to emissive, because it seems like more natural lighting than spots.  It takes the Beta at least twice as long as 4.10 to render these scenes.  I mean, given 3 cards it goes quickly regardless, but I do note both increased sampling and longer times with the beta.

    The file attached is a case in point:  2 figures, emissive lighting and 1 point light.  This image rendered in 9:02 at 5134 iterations using 4.10.  It rendered in 14:34 and 5523 iterations using the Beta. This is a simple scene. More complex indoor scenes I have found are morkedly slower using the Beta, and the iterations can be quadrupled.

    If you get the chance render the scene again in the beta and set the max iterations to be 5134 amd see how long that takes. Could also set the max time to be 9min.
  • RayDAntRayDAnt Posts: 1,120
    edited April 2019
    areg5 said:
    areg5 said:
    areg5 said:

    I've been playing around with the Beta, and like I mentioned before, I'm seeing differences in rendering speed compared to 4.10.  When I first tried it, I thought it rendered about 3X faster.  This was on an outdoor scene with HDRI lighting.  I noticed a profound slowing when using MESH lighting compared to 4.10.  I've been working now with an indoor scene lit mainly with emissives and pointlights.  Like the MESH lighting example, it renders much slower than 4.10.  Not only that, but the number of samples is markedly increased.

    I'm hoping that the developers will address this.  When 4.11 is finally launched, this slowdown in indoor renders could be really bad.  Most of my scenes are indoor scenes.

    Are you using the 2 1080ti's in your rig description? I ask to see if there may be some difference with using an RTX card or not. I am using an RTX 2080 alone for my testing so far. I notice faster renders in indoors outdoors and hdri only across the board. Indoors I use only spotlighting and whatever lights the scene has setup already. I will circle back around and try these scenes with my P5000 by itself and try to compare soon. I have not used 4.10 in quite awhile but will try the scenes in that at some point as well. But for me it seems faster than the prior Beta (.196).

    I have 3 1080 Ti's.  It's likely due to the emissive lights I would think.  When I light an indoor scene, what I do is change the light bulbs in the room to emissive, because it seems like more natural lighting than spots.  It takes the Beta at least twice as long as 4.10 to render these scenes.  I mean, given 3 cards it goes quickly regardless, but I do note both increased sampling and longer times with the beta.

    The file attached is a case in point:  2 figures, emissive lighting and 1 point light.  This image rendered in 9:02 at 5134 iterations using 4.10.  It rendered in 14:34 and 5523 iterations using the Beta. This is a simple scene. More complex indoor scenes I have found are morkedly slower using the Beta, and the iterations can be quadrupled.

    Are these numbers for the beta that was just released or the one just replaced? Because I just finished a fresh run of benchmarks on 4.11.0.335 with my Titan RTX that indicates it's significantly faster with most scenes than 4.11.0.236 was with virtually no human detectable differences in image detail.

    Also, would you mind posting that image as it rendered in 4.10 in addition to how it looks in the current/previous beta? Fwiw, pixel-peeping a scene rendered in 4.10 final vs the same identical scene rendered in 4.11.0.236 or later always seems to reveal visually subtle but technologically significant increases in image detail. Thereby explaining noticeable increases in rendering times - all other variables being kept equal.

    Post edited by RayDAnt on
  • RayDAntRayDAnt Posts: 1,120
    edited April 2019

    no just installed 425.31 and Octane 4 plugin for Carrara is not working anymore

    I do have my old nvidia installers

    so will just have to live without iray I guess, I have Octane 2 for DAZ studio

    Yeah, as a cautionary note to anyone with a driver-version-restricted system on their hands (one of my two render/test systems is a GTX 1050 based Surface Book which runs on Microsoft customized Nvidia drivers - the most recent release of which is based on 391.40) you will currently loose all GPU-bound render capabilities upon updating to this beta since Iray now has a minimum driver requirement of 418.81.

    Post edited by RayDAnt on
  • areg5areg5 Posts: 617
    areg5 said:
    areg5 said:
    areg5 said:

    I've been playing around with the Beta, and like I mentioned before, I'm seeing differences in rendering speed compared to 4.10.  When I first tried it, I thought it rendered about 3X faster.  This was on an outdoor scene with HDRI lighting.  I noticed a profound slowing when using MESH lighting compared to 4.10.  I've been working now with an indoor scene lit mainly with emissives and pointlights.  Like the MESH lighting example, it renders much slower than 4.10.  Not only that, but the number of samples is markedly increased.

    I'm hoping that the developers will address this.  When 4.11 is finally launched, this slowdown in indoor renders could be really bad.  Most of my scenes are indoor scenes.

    Are you using the 2 1080ti's in your rig description? I ask to see if there may be some difference with using an RTX card or not. I am using an RTX 2080 alone for my testing so far. I notice faster renders in indoors outdoors and hdri only across the board. Indoors I use only spotlighting and whatever lights the scene has setup already. I will circle back around and try these scenes with my P5000 by itself and try to compare soon. I have not used 4.10 in quite awhile but will try the scenes in that at some point as well. But for me it seems faster than the prior Beta (.196).

    I have 3 1080 Ti's.  It's likely due to the emissive lights I would think.  When I light an indoor scene, what I do is change the light bulbs in the room to emissive, because it seems like more natural lighting than spots.  It takes the Beta at least twice as long as 4.10 to render these scenes.  I mean, given 3 cards it goes quickly regardless, but I do note both increased sampling and longer times with the beta.

    The file attached is a case in point:  2 figures, emissive lighting and 1 point light.  This image rendered in 9:02 at 5134 iterations using 4.10.  It rendered in 14:34 and 5523 iterations using the Beta. This is a simple scene. More complex indoor scenes I have found are morkedly slower using the Beta, and the iterations can be quadrupled.

     

    If you get the chance render the scene again in the beta and set the max iterations to be 5134 amd see how long that takes. Could also set the max time to be 9min.

    Why?  Basically I go for quality, so I keep the percentage at 95% and up the max iterations to 30000 and max time as high as it will go.  It's a rare frame that takes more than 30000 ... but it happens.

  • areg5 said:
    areg5 said:
    areg5 said:
    areg5 said:

    I've been playing around with the Beta, and like I mentioned before, I'm seeing differences in rendering speed compared to 4.10.  When I first tried it, I thought it rendered about 3X faster.  This was on an outdoor scene with HDRI lighting.  I noticed a profound slowing when using MESH lighting compared to 4.10.  I've been working now with an indoor scene lit mainly with emissives and pointlights.  Like the MESH lighting example, it renders much slower than 4.10.  Not only that, but the number of samples is markedly increased.

    I'm hoping that the developers will address this.  When 4.11 is finally launched, this slowdown in indoor renders could be really bad.  Most of my scenes are indoor scenes.

    Are you using the 2 1080ti's in your rig description? I ask to see if there may be some difference with using an RTX card or not. I am using an RTX 2080 alone for my testing so far. I notice faster renders in indoors outdoors and hdri only across the board. Indoors I use only spotlighting and whatever lights the scene has setup already. I will circle back around and try these scenes with my P5000 by itself and try to compare soon. I have not used 4.10 in quite awhile but will try the scenes in that at some point as well. But for me it seems faster than the prior Beta (.196).

    I have 3 1080 Ti's.  It's likely due to the emissive lights I would think.  When I light an indoor scene, what I do is change the light bulbs in the room to emissive, because it seems like more natural lighting than spots.  It takes the Beta at least twice as long as 4.10 to render these scenes.  I mean, given 3 cards it goes quickly regardless, but I do note both increased sampling and longer times with the beta.

    The file attached is a case in point:  2 figures, emissive lighting and 1 point light.  This image rendered in 9:02 at 5134 iterations using 4.10.  It rendered in 14:34 and 5523 iterations using the Beta. This is a simple scene. More complex indoor scenes I have found are morkedly slower using the Beta, and the iterations can be quadrupled.

     

    If you get the chance render the scene again in the beta and set the max iterations to be 5134 amd see how long that takes. Could also set the max time to be 9min.

    Why?  Basically I go for quality, so I keep the percentage at 95% and up the max iterations to 30000 and max time as high as it will go.  It's a rare frame that takes more than 30000 ... but it happens.

    Was just curious how close to 9min it was at 5134 iter.
  • areg5areg5 Posts: 617
    RayDAnt said:
    areg5 said:
    areg5 said:
    areg5 said:

    I've been playing around with the Beta, and like I mentioned before, I'm seeing differences in rendering speed compared to 4.10.  When I first tried it, I thought it rendered about 3X faster.  This was on an outdoor scene with HDRI lighting.  I noticed a profound slowing when using MESH lighting compared to 4.10.  I've been working now with an indoor scene lit mainly with emissives and pointlights.  Like the MESH lighting example, it renders much slower than 4.10.  Not only that, but the number of samples is markedly increased.

    I'm hoping that the developers will address this.  When 4.11 is finally launched, this slowdown in indoor renders could be really bad.  Most of my scenes are indoor scenes.

    Are you using the 2 1080ti's in your rig description? I ask to see if there may be some difference with using an RTX card or not. I am using an RTX 2080 alone for my testing so far. I notice faster renders in indoors outdoors and hdri only across the board. Indoors I use only spotlighting and whatever lights the scene has setup already. I will circle back around and try these scenes with my P5000 by itself and try to compare soon. I have not used 4.10 in quite awhile but will try the scenes in that at some point as well. But for me it seems faster than the prior Beta (.196).

    I have 3 1080 Ti's.  It's likely due to the emissive lights I would think.  When I light an indoor scene, what I do is change the light bulbs in the room to emissive, because it seems like more natural lighting than spots.  It takes the Beta at least twice as long as 4.10 to render these scenes.  I mean, given 3 cards it goes quickly regardless, but I do note both increased sampling and longer times with the beta.

    The file attached is a case in point:  2 figures, emissive lighting and 1 point light.  This image rendered in 9:02 at 5134 iterations using 4.10.  It rendered in 14:34 and 5523 iterations using the Beta. This is a simple scene. More complex indoor scenes I have found are morkedly slower using the Beta, and the iterations can be quadrupled.

    Are these numbers for the beta that was just released or the one just replaced? Because I just finished a fresh run of benchmarks on 4.11.0.335 with my Titan RTX that indicates it's significantly faster with most scenes than 4.11.0.236 was with virtually no human detectable differences in image detail.

    Also, would you mind posting that image as it rendered in 4.10 in addition to how it looks in the current/previous beta? Fwiw, pixel-peeping a scene rendered in 4.10 final vs the same identical scene rendered in 4.11.0.236 or later always seems to reveal visually subtle but technologically significant increases in image detail. Thereby explaining noticeable increases in rendering times - all other variables being kept equal.

    I'm using 4.11.0.236.

     

    Here are the renders side by side.  k38 was done in 4.10, k38a was done with Beta.  It's a little different, but I don't know if the added time makes it worth it.

    k38.jpg
    1280 x 720 - 599K
    k38a.jpg
    1280 x 720 - 599K
  • areg5areg5 Posts: 617
    edited April 2019
    areg5 said:
    areg5 said:
    areg5 said:
    areg5 said:

    I've been playing around with the Beta, and like I mentioned before, I'm seeing differences in rendering speed compared to 4.10.  When I first tried it, I thought it rendered about 3X faster.  This was on an outdoor scene with HDRI lighting.  I noticed a profound slowing when using MESH lighting compared to 4.10.  I've been working now with an indoor scene lit mainly with emissives and pointlights.  Like the MESH lighting example, it renders much slower than 4.10.  Not only that, but the number of samples is markedly increased.

    I'm hoping that the developers will address this.  When 4.11 is finally launched, this slowdown in indoor renders could be really bad.  Most of my scenes are indoor scenes.

    Are you using the 2 1080ti's in your rig description? I ask to see if there may be some difference with using an RTX card or not. I am using an RTX 2080 alone for my testing so far. I notice faster renders in indoors outdoors and hdri only across the board. Indoors I use only spotlighting and whatever lights the scene has setup already. I will circle back around and try these scenes with my P5000 by itself and try to compare soon. I have not used 4.10 in quite awhile but will try the scenes in that at some point as well. But for me it seems faster than the prior Beta (.196).

    I have 3 1080 Ti's.  It's likely due to the emissive lights I would think.  When I light an indoor scene, what I do is change the light bulbs in the room to emissive, because it seems like more natural lighting than spots.  It takes the Beta at least twice as long as 4.10 to render these scenes.  I mean, given 3 cards it goes quickly regardless, but I do note both increased sampling and longer times with the beta.

    The file attached is a case in point:  2 figures, emissive lighting and 1 point light.  This image rendered in 9:02 at 5134 iterations using 4.10.  It rendered in 14:34 and 5523 iterations using the Beta. This is a simple scene. More complex indoor scenes I have found are morkedly slower using the Beta, and the iterations can be quadrupled.

     

    If you get the chance render the scene again in the beta and set the max iterations to be 5134 amd see how long that takes. Could also set the max time to be 9min.

    Why?  Basically I go for quality, so I keep the percentage at 95% and up the max iterations to 30000 and max time as high as it will go.  It's a rare frame that takes more than 30000 ... but it happens.

     

    Was just curious how close to 9min it was at 5134 iter.

    Ah.  Well, rather than render it again I checked the log file. It hit 5124 iterations at 12.33 and 5144 iterations at 13.19.  So I would estimate it hit 5134 at around 12.42.

    Post edited by areg5 on
  • areg5 said:
    areg5 said:
    areg5 said:
    areg5 said:
    areg5 said:

    I've been playing around with the Beta, and like I mentioned before, I'm seeing differences in rendering speed compared to 4.10.  When I first tried it, I thought it rendered about 3X faster.  This was on an outdoor scene with HDRI lighting.  I noticed a profound slowing when using MESH lighting compared to 4.10.  I've been working now with an indoor scene lit mainly with emissives and pointlights.  Like the MESH lighting example, it renders much slower than 4.10.  Not only that, but the number of samples is markedly increased.

    I'm hoping that the developers will address this.  When 4.11 is finally launched, this slowdown in indoor renders could be really bad.  Most of my scenes are indoor scenes.

    Are you using the 2 1080ti's in your rig description? I ask to see if there may be some difference with using an RTX card or not. I am using an RTX 2080 alone for my testing so far. I notice faster renders in indoors outdoors and hdri only across the board. Indoors I use only spotlighting and whatever lights the scene has setup already. I will circle back around and try these scenes with my P5000 by itself and try to compare soon. I have not used 4.10 in quite awhile but will try the scenes in that at some point as well. But for me it seems faster than the prior Beta (.196).

    I have 3 1080 Ti's.  It's likely due to the emissive lights I would think.  When I light an indoor scene, what I do is change the light bulbs in the room to emissive, because it seems like more natural lighting than spots.  It takes the Beta at least twice as long as 4.10 to render these scenes.  I mean, given 3 cards it goes quickly regardless, but I do note both increased sampling and longer times with the beta.

    The file attached is a case in point:  2 figures, emissive lighting and 1 point light.  This image rendered in 9:02 at 5134 iterations using 4.10.  It rendered in 14:34 and 5523 iterations using the Beta. This is a simple scene. More complex indoor scenes I have found are morkedly slower using the Beta, and the iterations can be quadrupled.

     

    If you get the chance render the scene again in the beta and set the max iterations to be 5134 amd see how long that takes. Could also set the max time to be 9min.

    Why?  Basically I go for quality, so I keep the percentage at 95% and up the max iterations to 30000 and max time as high as it will go.  It's a rare frame that takes more than 30000 ... but it happens.

     

    Was just curious how close to 9min it was at 5134 iter.

    Ah.  Well, rather than render it again I checked the log file. It hit 5124 iterations at 12.33 and 5144 iterations at 13.19.  So I would estimate it hit 5134 at around 12.42.

    Ok thanks.
  • Kyan001Kyan001 Posts: 75
    edited April 2019

    WARNING before update. I'ts going to destroy the Ghost Lights (if you use them for your renders). You'll have to create a new scene.

    Plus, the illegal memory access and CPU fallback still happening.

    Post edited by Kyan001 on
  • RayDAntRayDAnt Posts: 1,120
    areg5 said:
    RayDAnt said:
    areg5 said:
    areg5 said:
    areg5 said:

    I've been playing around with the Beta, and like I mentioned before, I'm seeing differences in rendering speed compared to 4.10.  When I first tried it, I thought it rendered about 3X faster.  This was on an outdoor scene with HDRI lighting.  I noticed a profound slowing when using MESH lighting compared to 4.10.  I've been working now with an indoor scene lit mainly with emissives and pointlights.  Like the MESH lighting example, it renders much slower than 4.10.  Not only that, but the number of samples is markedly increased.

    I'm hoping that the developers will address this.  When 4.11 is finally launched, this slowdown in indoor renders could be really bad.  Most of my scenes are indoor scenes.

    Are you using the 2 1080ti's in your rig description? I ask to see if there may be some difference with using an RTX card or not. I am using an RTX 2080 alone for my testing so far. I notice faster renders in indoors outdoors and hdri only across the board. Indoors I use only spotlighting and whatever lights the scene has setup already. I will circle back around and try these scenes with my P5000 by itself and try to compare soon. I have not used 4.10 in quite awhile but will try the scenes in that at some point as well. But for me it seems faster than the prior Beta (.196).

    I have 3 1080 Ti's.  It's likely due to the emissive lights I would think.  When I light an indoor scene, what I do is change the light bulbs in the room to emissive, because it seems like more natural lighting than spots.  It takes the Beta at least twice as long as 4.10 to render these scenes.  I mean, given 3 cards it goes quickly regardless, but I do note both increased sampling and longer times with the beta.

    The file attached is a case in point:  2 figures, emissive lighting and 1 point light.  This image rendered in 9:02 at 5134 iterations using 4.10.  It rendered in 14:34 and 5523 iterations using the Beta. This is a simple scene. More complex indoor scenes I have found are morkedly slower using the Beta, and the iterations can be quadrupled.

    Are these numbers for the beta that was just released or the one just replaced? Because I just finished a fresh run of benchmarks on 4.11.0.335 with my Titan RTX that indicates it's significantly faster with most scenes than 4.11.0.236 was with virtually no human detectable differences in image detail.

    Also, would you mind posting that image as it rendered in 4.10 in addition to how it looks in the current/previous beta? Fwiw, pixel-peeping a scene rendered in 4.10 final vs the same identical scene rendered in 4.11.0.236 or later always seems to reveal visually subtle but technologically significant increases in image detail. Thereby explaining noticeable increases in rendering times - all other variables being kept equal.

    I'm using 4.11.0.236.

     

    Here are the renders side by side.  k38 was done in 4.10, k38a was done with Beta.  It's a little different, but I don't know if the added time makes it worth it.

    Yeah, that's actually a huge difference in terms of the realistic lighting computations visibly going on in the 4.11 Beta. And realsitic lighting is THE processing hog of unbiased rendering engines. With that said, whether that level of detail in realistic lighting interactions is itself realistic looking in the context of anything other than an attempt at a real-life referenced exact 3d recreation of reality is pretty debatable imo.

     

    By the way, if 4.11's visual quality upgrades over 4.10 aren't worth the added processing time to you, there's a trick you can use to "dumb down" 4.11 to a visual quality level about on par with 4.10 (thereby allowing you to avoid having to maintain multiple installations of Daz Studio/foregoing the use of non-backwards compatible rendering hardware like RTX GPUs):

    1. Switch your viewport to NVIDIA Iray liveview mode
    2. Go to Render Settings > Editor tab > Optimization > Max Path Length and change the value from the default -1 to 1.
    3. Increment the value upwards one number at a time and watch what happens in your scene window.
    4. Once the scene starts looking essentially the same to you regardless of increasing the Max Path Length value (most likely <10) initiate your final render.

    The resulting render should be more than detailed enough for your needs aaaand (here's the real value of this process) take potentially orders of magnitude less time for your hardware to render to 95% completion.

  • Leonides02Leonides02 Posts: 1,379
    areg5 said:

     

    I have 3 1080 Ti's.  It's likely due to the emissive lights I would think.  When I light an indoor scene, what I do is change the light bulbs in the room to emissive, because it seems like more natural lighting than spots.  It takes the Beta at least twice as long as 4.10 to render these scenes.  I mean, given 3 cards it goes quickly regardless, but I do note both increased sampling and longer times with the beta.

    In general, the Beta is faster to set-up scenes and slower (by around 33% accordining to my tests) to render certain scenes. Supposedly this is because the new Iray version is "more accurate." Frankly, if accuracy = slower renders, I don't want it!

    These days I'm using the Beta to set up and 4.10 to render.

This discussion has been closed.