Octane 4 with AI features

2

Comments

  • I'm so hyped for this one, guys have been working hard on this release, and with stalling of 3.8 official seeing V4 launched to public was a great surprise, can't wait for plugins to follow.

    ...This also means war on Redshift :)

    While I'm sure a lot of this has been in the works before Redshift started taking off I can see some of these features and the pricing as a direct response to Redshift's success.

  • valzheimervalzheimer Posts: 519

    There are some caveats of course. Materials have to be converted manually to render well in Octane? Unless all this was so much of a game changer that PA's were to start adding that to their products.

    I read a while back that there were fewer problems with DAZ Iray to Octane conversions.

    Yes Iray shader is more friendly towards Octane than 3Delight was, but you'd still have to work around things and tweak the surfaces, especially the skin shader to give it a nice look. Let's not forget that Octane comes also with Live Database of shaders always at your service to be used with the engine for free, and after a while shaders aren't that much fo a cencern when you get used to it.

    As for lights, Octane comes with integrated Daylight settings that are good for environments and has HDRI option and also uses emissive lights imo better than Iray/faster either way than Iray emissives.

    Also, RedSpecSSS has fairly well developed skin shading for Daz figures, and they slao have a hair shader that is pretty grateful for working in Daz. Why I weighed towards Octane when Iray took some speed is mostly because I render on my laptop and Iray hogs up in scenes that I would like to use. While yes, Octane has a steap learning curve, it's still one of the best industry engines out there.

  • valzheimervalzheimer Posts: 519

    I'm so hyped for this one, guys have been working hard on this release, and with stalling of 3.8 official seeing V4 launched to public was a great surprise, can't wait for plugins to follow.

    ...This also means war on Redshift :)

    While I'm sure a lot of this has been in the works before Redshift started taking off I can see some of these features and the pricing as a direct response to Redshift's success.

    And we're also finally getting the celshading/proper toon shader in Octane :) It's all about competition out there, but in the end, that's good for us, end customers.

  • valzheimer, bluejaunte, and DustRider thank you for the info!

  • Ghosty12Ghosty12 Posts: 2,080

    Well going to the link seems to still be in beta phase, so will have to wait also looks like they are setting it up for volta based cards like the Titan V if it has AI support built in to Octane..

  • joseftjoseft Posts: 310

    It looks quite good, and the free tier and $20 sub is going to be incredibly attractive to a lot of people. Redshift had been gaining quite a lot of popularity, many of whom were Octane users before (myself included), but this is going to be a big step to bringing people back.

    I tried Redshift, and initial feeling was amazing, but it has a major flaw that affects anyone trying to render Daz figures - you can't render transmapped hair to any acceptable level of quality. I have been patiently waiting for them to fix that, but if that does not come before Octane 4 public release, i will be sticking to Octane

  • DustRiderDustRider Posts: 2,880
    edited March 2018
    ghosty12 said:

    Well going to the link seems to still be in beta phase, so will have to wait also looks like they are setting it up for volta based cards like the Titan V if it has AI support built in to Octane..

    Yes, it's still in beta, only available to current Octane owners (probably only Octane 3??). Their open beta periods have been fairly fast, I would expect the full version to be available fairly soon. The AI features are designed to work on non-volta cards, they are working on getting everything running on volta now (their AI features are independent of what Nvidia is developing). Hopefully I get a chance to download the beta today and give it a test run. I also need to get newer drivers for my gpu.

    valzheimer, bluejaunte, and DustRider thank you for the info!

    Glad to help Kevin!

     Like valzheimer noted, there is a bit of a learning curve with Octane, especially the DS plugin (though I didn't find it too bad at all). For Carrara users, the learning curve is pretty modest. The shader system is integrated into the shader room in Carrara, which keeps things quite familiar and easy to understand (no nodes - though as you advance you can use NGE to build your shaders). The Carrara plugin also supports Carrara's dynamic hair, and instancing (but nested instancing not quite as well).

    The DS interface is a bit more challenging, but extremely powerful. You can stay away from NGE for the most part for those who are node challenged (or just dislike nodes), so that's not a big worry (see attached DS material editor screen grab below). DS lights don't transfer, but it's very easy to set up your own mesh lights and/or HDRI. Some shaders transfer well, others don't. I typically tweak just about everything, or set up my own shaders. For skin I usually use a specular/glossy mix or diffuse/glossy. Difuse/glossy is what comes from the translation for skin shaders, which can work extremely well as the diffuse node has SSS. Specular/Glossy can give a bit better SSS effect, but takes longer to render. As  valzheimer noted, Redspec TGX makes an outstanding shader for DAZ figures (V4 through G8F and M4 to G3M) for those who prefer not to make their own (but it's a skill well worth the time investment). Also as noted before, the LiveDB has a lot of pre-made shaders (some from Otoy, and some generously shared by the user community) that are a great resource as a final shader or a starting point for your own customized shaders. These shaders include a wide variety of substances including skin, velvet, glass, metals, water, cloth, plastics, etc.

    You can also export your scene to render in the standalone version of Octane, which will free up your computer to work on other things while rendering. Another feature I really like is the ability to zoom and pan the image in the render view port, it really helps to see progress, and errors quickly (no lag).

    I attached a couple of sample renders. The first one is from the DS plugin with the default skin shaders as they came directly from DS, and the "cubes" are a wine shader from the Octane LiveDB, and the lighting is just an HDRI. The second image was done with the Carrara plugin, with a lot of shader tweaking, which is normal for Carrara. Great Carrara shaders will almost always be great Octane shaders, but when using DS/Poser content in Carrara, you almost always have to adjust the shaders, so the same is typically true for Octane. I really like how the Dragon came out (specular/glossy mix) but you need to zoom to full resolution to really see it. The girl (G2F) is also wearing a lot of V4 clothing, which came out pretty good (a lot of old V4/M4 vintage props too).

    Octane for DS materials/shader editor

    DS/Octane Render - Default skin shaders. Galley Link

    Carrara/Octane Render (please zoom to full resolution to see dragon skin details) Gallery Link

     

    ORDS Materials.JPG
    612 x 1038 - 142K
    Post edited by DustRider on
  • DustRider, I'm tempted even more. Thanks for the examples!!

  • kyoto kidkyoto kid Posts: 41,847

    ...so the Carrara version uses the Carrara shader room?  Sweet. I wish Daz would have built their shader mixer on that concept as I am one of those "node challenged" types due to dyslexia.

    Having a standalone, that's nice for those of us without boatloads of memory as we can shut the Daz programme and scene down once its submitted to Octane to free up more system memory for the out of core textures.

  • Leonides02Leonides02 Posts: 1,379

    This is exciting, but I won't be using Octane again until it can read MDL. I paid the $400 and now I never use the software because the time I'd save on rendering is instead spent on material conversion. 

    It doesn't translate well, in my opinion. 

     

  • kyoto kidkyoto kid Posts: 41,847

    ...not needing an overpriced top of the line GPU card and being able to shut Daz/Carrara and the scene down after sending to the standalone to free up more system resources is a major bonus in favour of Octane over Iray. I've worked with Reality/Lux, which also required converting materials so no stranger to that.  The nice part about Octane, I don't have to wait half day or more for a test render complete only to find a surface needs more adjustment.

  • bluejauntebluejaunte Posts: 1,990

    Why don't you try the demo and see how it performs?

  • JonstarkJonstark Posts: 2,738
    kyoto kid said:

    ...so the Carrara version uses the Carrara shader room?  Sweet. I wish Daz would have built their shader mixer on that concept as I am one of those "node challenged" types due to dyslexia.

     

    It's actually even better than that.  Not only does it use the Carrara shader room, the Octane shader 'surrounds' the native Carrara shader (and pulls info from it).  So the same shader works for both Carrara and Octane, and you have the ability to refine the Octane portion of the shader.  The Carrara native render will ignore the Octane portion of the shader, and render it normally within the Carrara native engine, while the Octane plugin reads all the info, both Carrara/Octane and renders from both.  Once built, you've got one shader that works for both renderers, no need to switch back and forth.

     

    I'm tremendously excited by what I see in the video, Octane was already blazing fast, but the de-noise function makes this into 'blink and you missed it' territory.  And apparently it will be a free upgrade?  Wow, pinch me, I must be dreaming!   I already have Octane and the Carrara plugin, so I wonder if I qualify to try out the beta, and if so where to go to get it...  Man this looks like a game-changer  :)

  • Silver DolphinSilver Dolphin Posts: 1,638

    Sounds like we need a script that converts Iray to Octane. I'm willing to pay for this. hint hintenlightened Oh there is time as the free version is not out yet. This is going to be a great way to get more customers. The pros will pay for the more than 2 video cards and the hobbiest will limit their renders to 2 cards. The subscription is not something I will do. I pay for what I use and don't like paying rent.

  • DustRiderDustRider Posts: 2,880
    edited March 2018

    From this post on the Otoy forums, is sounds like as long as MDL is a "black box" from Nvivdia that is designed to support Nvidia GPU rendering, it may not be a top priority, given the fact that they are trying to support more than just Nvidia GPU's in the future.

    Re: OR MDL support - Is "open source" going to happen?

    postby Goldorak » Wed Jun 07, 2017 7:16 am

    1a) Did Nvidia change their mind after April 2016?
    1b) Did Otoy change their mind after April 2016?

    OTOY and NVIDIA both want MDL in Octane in 3.x and we're still going to try to do this ASAP. 

    But - we are also working to get Octane on AMD GPUs, mac AMD, etc. and if we add a dependency to the core engine that only works (well) on NVIDIA GPUs, we may be blocked from ever getting feature and speed parity on other devices and platforms . Not ideal...

    2) Why can Otoy not support MDL if it is not open source?

    NVIDIA told us they wanted to open source MDL and make it an industry standard just like OSL, which is why we felt good about supporting it after OSL was done. The issue is that without source code to a working shader compiler to see how MDL to bytecode is being generated, it's a black box - and not an open standard on par with EXR, OSL alembic and others that provide this src as well as spec (you want both). With OSL, the full compiler source is on gitbub, and it makes a huge difference in terms of how we brought it into Octane. 

    If we had full source to MDL compiler, we could, for example, have a shot at real time MDL -> OSL/OSO (by adding a new compiler target to NVIDIA's MDL compiler) , so we would not have to do anything special to support MDL materials in Octane 3.1+ on all platforms.

    3) When will we know if NVIDIA is going to agree making MDL open source?

    They say they are looking into it, but no date has been given. After we are done with OSL, and if this is still unchanged, we may look into an whether we can get MDL output cross compiled to OSL, even without the MDL compiler source. But it's really not ideal, and shouldn't be something we have to do if NVIDIA is going to follow through with making MDL open source.

     It would be great to have MDL support, and better materials conversion (I tend to use Octane less than Iray because Iray requires less work too). But I can see from an economic/management point of view why OSL and other GPU support is the priority (OSL having greater industry support, and not being locked to Nvidia GPU's could potentially gain an increased user base .... like Mac users). But once your used to it, shaders setup can be really fast and easy, and Octane doesn't work my system as hard as Iray, so that's another big benefit for me. Of course with the AI Lighting and AI Denoiser features in Octane 4, It could very easily swing the time factor to a big plus for Octane.

     

    Edited to fix the odd formatting at the beginning of the post, sorry for being soooo loud!!

    Post edited by DustRider on
  • DustRiderDustRider Posts: 2,880
    edited March 2018
    Jonstark said:
    kyoto kid said:

    ...so the Carrara version uses the Carrara shader room?  Sweet. I wish Daz would have built their shader mixer on that concept as I am one of those "node challenged" types due to dyslexia.

     

    It's actually even better than that.  Not only does it use the Carrara shader room, the Octane shader 'surrounds' the native Carrara shader (and pulls info from it).  So the same shader works for both Carrara and Octane, and you have the ability to refine the Octane portion of the shader.  The Carrara native render will ignore the Octane portion of the shader, and render it normally within the Carrara native engine, while the Octane plugin reads all the info, both Carrara/Octane and renders from both.  Once built, you've got one shader that works for both renderers, no need to switch back and forth.

     

    I'm tremendously excited by what I see in the video, Octane was already blazing fast, but the de-noise function makes this into 'blink and you missed it' territory.  And apparently it will be a free upgrade?  Wow, pinch me, I must be dreaming!   I already have Octane and the Carrara plugin, so I wonder if I qualify to try out the beta, and if so where to go to get it...  Man this looks like a game-changer  :)

    Hey Jonstark!!

    To get the beta, just log into the Otoy/Octane forums and go to "Octane Licensed Customer Forums > Development Build Releases > Hello World: OctaneRender 4 is here" (or click this link after logging in). Make sure your video drivers are up to date before trying to run it wink

    I installed it and .... WOW!! I exported the scene with the girl above to ORBX (minus the bikini, so can't show the results), imported to Octane 4, and set to the PMC kernel, AI Lighting enabled, the render took about an hour, compared to 1hr 45min without AI Lighting. I tried AI Denoiser as well, but it doesn't work with PMC yet, so it wasn't clearing up at a decent rate with all of the specular shaders on the cubes (would have been about the same time as the original). Now I need to try something without all the specular materials. But a ~40-45% speed increase with only AI Lighting is pretty impressive!!

    Post edited by DustRider on
  • DustRiderDustRider Posts: 2,880
    ghosty12 said:

    Well going to the link seems to still be in beta phase, so will have to wait also looks like they are setting it up for volta based cards like the Titan V if it has AI support built in to Octane..

    I can confirm now that the AI functions are designed to work with Kepler and later cards, so no Volta needed. I've been testing it with a 970M (laptop), and not had and problems yet (other than remembering how to use the stand alone).

  • kyoto kidkyoto kid Posts: 41,847

    ..even sweeter.

  • ArtiniArtini Posts: 10,305

    I only hope, that they will find the time to realize all of their promises.

    Octane for Unity does not work in the new releases (2017.3.1).

    I have only managed to tests Octane in Unity 2017.0 long time ago,

    but I was not impressed with its speed on Nvidia GTX 1080.

    Iray in Daz Studio is much faster on that graphic card.

     

  • IllucidIllucid Posts: 25
    edited March 2018

    Edit: Octane requires an NVidia card to work.

     

    Post edited by Illucid on
  • agent unawaresagent unawares Posts: 3,513
    Illucid said:

    Am I missing something?

    Maybe the bit where no one in this thread is trying to convince you to use Octane?

  • DustRiderDustRider Posts: 2,880
    Illucid said:

    Am I missing something?

    Maybe the bit where no one in this thread is trying to convince you to use Octane?

    lol smiley

    Well yes, you may have missed something, like the part where future versions (IIRC Octane 2018 & the RNDR sdk) will have the ability to use just about any GPU (including AMD and integrated GPU chips on the CPU like the Intel integrated HD Graphics processors) and CPU's. So, even though it's not available now, probably some time this year, or possibly next year, Octane will be able to render on just about any hardware you have (they show it running on a phone and a tablet in the video).

    This is more about options, rather than trying to convince you, or anyone else to use Octane. In fact, I would recommend that the load/pose/render "I've never touched a shader and never plan too" crowd not use Octane, as currently you need to do a fair amount of shader adjustment to make things look their best. It's not really that hard to do, but it seems like for many people this is something they don't want to do ... ever. But for others, this may be a viable option, especially if it is free (though the DS and Carrara plugins may not be free, since they are done by independent programmers, not Otoy).

  • nicsttnicstt Posts: 11,715

    Personally, I want something that doesn't tie me to one specific manufacturer for GPU; although, Octane's implementation makes it a little more palatable.

  • kyoto kidkyoto kid Posts: 41,847

    ...I'm already sold.

  • valzheimervalzheimer Posts: 519
    DustRider said:

    From this post on the Otoy forums, is sounds like as long as MDL is a "black box" from Nvivdia that is designed to support Nvidia GPU rendering, it may not be a top priority, given the fact that they are trying to support more than just Nvidia GPU's in the future.

    Re: OR MDL support - Is "open source" going to happen?

    postby Goldorak » Wed Jun 07, 2017 7:16 am

    1a) Did Nvidia change their mind after April 2016?
    1b) Did Otoy change their mind after April 2016?

    OTOY and NVIDIA both want MDL in Octane in 3.x and we're still going to try to do this ASAP. 

    But - we are also working to get Octane on AMD GPUs, mac AMD, etc. and if we add a dependency to the core engine that only works (well) on NVIDIA GPUs, we may be blocked from ever getting feature and speed parity on other devices and platforms . Not ideal...

    2) Why can Otoy not support MDL if it is not open source?

    NVIDIA told us they wanted to open source MDL and make it an industry standard just like OSL, which is why we felt good about supporting it after OSL was done. The issue is that without source code to a working shader compiler to see how MDL to bytecode is being generated, it's a black box - and not an open standard on par with EXR, OSL alembic and others that provide this src as well as spec (you want both). With OSL, the full compiler source is on gitbub, and it makes a huge difference in terms of how we brought it into Octane. 

    If we had full source to MDL compiler, we could, for example, have a shot at real time MDL -> OSL/OSO (by adding a new compiler target to NVIDIA's MDL compiler) , so we would not have to do anything special to support MDL materials in Octane 3.1+ on all platforms.

    3) When will we know if NVIDIA is going to agree making MDL open source?

    They say they are looking into it, but no date has been given. After we are done with OSL, and if this is still unchanged, we may look into an whether we can get MDL output cross compiled to OSL, even without the MDL compiler source. But it's really not ideal, and shouldn't be something we have to do if NVIDIA is going to follow through with making MDL open source.

     It would be great to have MDL support, and better materials conversion (I tend to use Octane less than Iray because Iray requires less work too). But I can see from an economic/management point of view why OSL and other GPU support is the priority (OSL having greater industry support, and not being locked to Nvidia GPU's could potentially gain an increased user base .... like Mac users). But once your used to it, shaders setup can be really fast and easy, and Octane doesn't work my system as hard as Iray, so that's another big benefit for me. Of course with the AI Lighting and AI Denoiser features in Octane 4, It could very easily swing the time factor to a big plus for Octane.

     

    Edited to fix the odd formatting at the beginning of the post, sorry for being soooo loud!!

     

    Actually MDL support is still rumored and spoken about, and from the last time I've seen it mentioned Goldorak posted in the forum in December '17 ;) https://render.otoy.com/forum/viewtopic.php?p=327955#p327955

     

    postby Goldorak » Fri Dec 08, 2017 12:32 am

    Survive, and thrive IMO. I wouldn't bet on iRay in the long run relative to Octane. We do have MDL support on the roadmap, which will be a huge boon for Daz users. And V4.

  • agent unawaresagent unawares Posts: 3,513
    DustRider said:

    From this post on the Otoy forums, is sounds like as long as MDL is a "black box" from Nvivdia that is designed to support Nvidia GPU rendering, it may not be a top priority, given the fact that they are trying to support more than just Nvidia GPU's in the future.

    Re: OR MDL support - Is "open source" going to happen?

    postby Goldorak » Wed Jun 07, 2017 7:16 am

    1a) Did Nvidia change their mind after April 2016?
    1b) Did Otoy change their mind after April 2016?

    OTOY and NVIDIA both want MDL in Octane in 3.x and we're still going to try to do this ASAP. 

    But - we are also working to get Octane on AMD GPUs, mac AMD, etc. and if we add a dependency to the core engine that only works (well) on NVIDIA GPUs, we may be blocked from ever getting feature and speed parity on other devices and platforms . Not ideal...

    2) Why can Otoy not support MDL if it is not open source?

    NVIDIA told us they wanted to open source MDL and make it an industry standard just like OSL, which is why we felt good about supporting it after OSL was done. The issue is that without source code to a working shader compiler to see how MDL to bytecode is being generated, it's a black box - and not an open standard on par with EXR, OSL alembic and others that provide this src as well as spec (you want both). With OSL, the full compiler source is on gitbub, and it makes a huge difference in terms of how we brought it into Octane. 

    If we had full source to MDL compiler, we could, for example, have a shot at real time MDL -> OSL/OSO (by adding a new compiler target to NVIDIA's MDL compiler) , so we would not have to do anything special to support MDL materials in Octane 3.1+ on all platforms.

    3) When will we know if NVIDIA is going to agree making MDL open source?

    They say they are looking into it, but no date has been given. After we are done with OSL, and if this is still unchanged, we may look into an whether we can get MDL output cross compiled to OSL, even without the MDL compiler source. But it's really not ideal, and shouldn't be something we have to do if NVIDIA is going to follow through with making MDL open source.

     It would be great to have MDL support, and better materials conversion (I tend to use Octane less than Iray because Iray requires less work too). But I can see from an economic/management point of view why OSL and other GPU support is the priority (OSL having greater industry support, and not being locked to Nvidia GPU's could potentially gain an increased user base .... like Mac users). But once your used to it, shaders setup can be really fast and easy, and Octane doesn't work my system as hard as Iray, so that's another big benefit for me. Of course with the AI Lighting and AI Denoiser features in Octane 4, It could very easily swing the time factor to a big plus for Octane.

     

    Edited to fix the odd formatting at the beginning of the post, sorry for being soooo loud!!

     

    Actually MDL support is still rumored and spoken about, and from the last time I've seen it mentioned Goldorak posted in the forum in December '17 ;) https://render.otoy.com/forum/viewtopic.php?p=327955#p327955

     

    postby Goldorak » Fri Dec 08, 2017 12:32 am

    Survive, and thrive IMO. I wouldn't bet on iRay in the long run relative to Octane. We do have MDL support on the roadmap, which will be a huge boon for Daz users. And V4.

    It sounds like they're going to take the time to do it right instead of building it in a way that might limit their ability to work with non-NVIDIA GPUs in the future.

  • DustRiderDustRider Posts: 2,880
    DustRider said:

    From this post on the Otoy forums, is sounds like as long as MDL is a "black box" from Nvivdia that is designed to support Nvidia GPU rendering, it may not be a top priority, given the fact that they are trying to support more than just Nvidia GPU's in the future.

    Re: OR MDL support - Is "open source" going to happen?

    postby Goldorak » Wed Jun 07, 2017 7:16 am

    1a) Did Nvidia change their mind after April 2016?
    1b) Did Otoy change their mind after April 2016?

    OTOY and NVIDIA both want MDL in Octane in 3.x and we're still going to try to do this ASAP. 

    But - we are also working to get Octane on AMD GPUs, mac AMD, etc. and if we add a dependency to the core engine that only works (well) on NVIDIA GPUs, we may be blocked from ever getting feature and speed parity on other devices and platforms . Not ideal...

    2) Why can Otoy not support MDL if it is not open source?

    NVIDIA told us they wanted to open source MDL and make it an industry standard just like OSL, which is why we felt good about supporting it after OSL was done. The issue is that without source code to a working shader compiler to see how MDL to bytecode is being generated, it's a black box - and not an open standard on par with EXR, OSL alembic and others that provide this src as well as spec (you want both). With OSL, the full compiler source is on gitbub, and it makes a huge difference in terms of how we brought it into Octane. 

    If we had full source to MDL compiler, we could, for example, have a shot at real time MDL -> OSL/OSO (by adding a new compiler target to NVIDIA's MDL compiler) , so we would not have to do anything special to support MDL materials in Octane 3.1+ on all platforms.

    3) When will we know if NVIDIA is going to agree making MDL open source?

    They say they are looking into it, but no date has been given. After we are done with OSL, and if this is still unchanged, we may look into an whether we can get MDL output cross compiled to OSL, even without the MDL compiler source. But it's really not ideal, and shouldn't be something we have to do if NVIDIA is going to follow through with making MDL open source.

     It would be great to have MDL support, and better materials conversion (I tend to use Octane less than Iray because Iray requires less work too). But I can see from an economic/management point of view why OSL and other GPU support is the priority (OSL having greater industry support, and not being locked to Nvidia GPU's could potentially gain an increased user base .... like Mac users). But once your used to it, shaders setup can be really fast and easy, and Octane doesn't work my system as hard as Iray, so that's another big benefit for me. Of course with the AI Lighting and AI Denoiser features in Octane 4, It could very easily swing the time factor to a big plus for Octane.

     

    Edited to fix the odd formatting at the beginning of the post, sorry for being soooo loud!!

     

    Actually MDL support is still rumored and spoken about, and from the last time I've seen it mentioned Goldorak posted in the forum in December '17 ;) https://render.otoy.com/forum/viewtopic.php?p=327955#p327955

     

    postby Goldorak » Fri Dec 08, 2017 12:32 am

    Survive, and thrive IMO. I wouldn't bet on iRay in the long run relative to Octane. We do have MDL support on the roadmap, which will be a huge boon for Daz users. And V4.

    Wow! Thanks for the info and link, I had completely missed that one. That's great news, hopefully it's not too long before it gets done. It would make working with Octane sooo much easier, and Octane is so much easier and faster on my laptop, I prefer using it when I'm not being lazy. 

  • valzheimervalzheimer Posts: 519
    DustRider said:
    DustRider said:

    From this post on the Otoy forums, is sounds like as long as MDL is a "black box" from Nvivdia that is designed to support Nvidia GPU rendering, it may not be a top priority, given the fact that they are trying to support more than just Nvidia GPU's in the future.

    Re: OR MDL support - Is "open source" going to happen?

    postby Goldorak » Wed Jun 07, 2017 7:16 am

    1a) Did Nvidia change their mind after April 2016?
    1b) Did Otoy change their mind after April 2016?

    OTOY and NVIDIA both want MDL in Octane in 3.x and we're still going to try to do this ASAP. 

    But - we are also working to get Octane on AMD GPUs, mac AMD, etc. and if we add a dependency to the core engine that only works (well) on NVIDIA GPUs, we may be blocked from ever getting feature and speed parity on other devices and platforms . Not ideal...

    2) Why can Otoy not support MDL if it is not open source?

    NVIDIA told us they wanted to open source MDL and make it an industry standard just like OSL, which is why we felt good about supporting it after OSL was done. The issue is that without source code to a working shader compiler to see how MDL to bytecode is being generated, it's a black box - and not an open standard on par with EXR, OSL alembic and others that provide this src as well as spec (you want both). With OSL, the full compiler source is on gitbub, and it makes a huge difference in terms of how we brought it into Octane. 

    If we had full source to MDL compiler, we could, for example, have a shot at real time MDL -> OSL/OSO (by adding a new compiler target to NVIDIA's MDL compiler) , so we would not have to do anything special to support MDL materials in Octane 3.1+ on all platforms.

    3) When will we know if NVIDIA is going to agree making MDL open source?

    They say they are looking into it, but no date has been given. After we are done with OSL, and if this is still unchanged, we may look into an whether we can get MDL output cross compiled to OSL, even without the MDL compiler source. But it's really not ideal, and shouldn't be something we have to do if NVIDIA is going to follow through with making MDL open source.

     It would be great to have MDL support, and better materials conversion (I tend to use Octane less than Iray because Iray requires less work too). But I can see from an economic/management point of view why OSL and other GPU support is the priority (OSL having greater industry support, and not being locked to Nvidia GPU's could potentially gain an increased user base .... like Mac users). But once your used to it, shaders setup can be really fast and easy, and Octane doesn't work my system as hard as Iray, so that's another big benefit for me. Of course with the AI Lighting and AI Denoiser features in Octane 4, It could very easily swing the time factor to a big plus for Octane.

     

    Edited to fix the odd formatting at the beginning of the post, sorry for being soooo loud!!

     

    Actually MDL support is still rumored and spoken about, and from the last time I've seen it mentioned Goldorak posted in the forum in December '17 ;) https://render.otoy.com/forum/viewtopic.php?p=327955#p327955

     

    postby Goldorak » Fri Dec 08, 2017 12:32 am

    Survive, and thrive IMO. I wouldn't bet on iRay in the long run relative to Octane. We do have MDL support on the roadmap, which will be a huge boon for Daz users. And V4.

    Wow! Thanks for the info and link, I had completely missed that one. That's great news, hopefully it's not too long before it gets done. It would make working with Octane sooo much easier, and Octane is so much easier and faster on my laptop, I prefer using it when I'm not being lazy. 

    On the down side, I haven't seen it mentioned in 3.08 or Octane 4 version listings yet, but there's hope for what V4 will eventually bring, I think they just gave us a quick look at what they'll have for sure, V4 is supposedly the reason we are sitting on 3.07 for so long..

    By now though I just want to see 3.08 hit the OcDS plugin already and get us moving from 3.07, I so need those celshaders already! haha

  • JonstarkJonstark Posts: 2,738
    DustRider said:
    Jonstark said:
    kyoto kid said:

     

    Hey Jonstark!!

    To get the beta, just log into the Otoy/Octane forums and go to "Octane Licensed Customer Forums > Development Build Releases > Hello World: OctaneRender 4 is here" (or click this link after logging in). Make sure your video drivers are up to date before trying to run it wink

    I installed it and .... WOW!! I exported the scene with the girl above to ORBX (minus the bikini, so can't show the results), imported to Octane 4, and set to the PMC kernel, AI Lighting enabled, the render took about an hour, compared to 1hr 45min without AI Lighting. I tried AI Denoiser as well, but it doesn't work with PMC yet, so it wasn't clearing up at a decent rate with all of the specular shaders on the cubes (would have been about the same time as the original). Now I need to try something without all the specular materials. But a ~40-45% speed increase with only AI Lighting is pretty impressive!!

     

    Thanks Dustrider!  I haven't got it yet, but it's good to know I can, maybe later today I'll put together some time to give it a whirl.  Too bad the AI Denoiser isn't available yet, that's the bit in the video that had my jaw on the floor, but if AI Lighting is giving 40+% speed increase that's pretty huge too.  :)

  • DustRiderDustRider Posts: 2,880
    edited March 2018
    Jonstark said:
    DustRider said:
    Jonstark said:
    kyoto kid said:

     

    Hey Jonstark!!

    To get the beta, just log into the Otoy/Octane forums and go to "Octane Licensed Customer Forums > Development Build Releases > Hello World: OctaneRender 4 is here" (or click this link after logging in). Make sure your video drivers are up to date before trying to run it wink

    I installed it and .... WOW!! I exported the scene with the girl above to ORBX (minus the bikini, so can't show the results), imported to Octane 4, and set to the PMC kernel, AI Lighting enabled, the render took about an hour, compared to 1hr 45min without AI Lighting. I tried AI Denoiser as well, but it doesn't work with PMC yet, so it wasn't clearing up at a decent rate with all of the specular shaders on the cubes (would have been about the same time as the original). Now I need to try something without all the specular materials. But a ~40-45% speed increase with only AI Lighting is pretty impressive!!

     

    Thanks Dustrider!  I haven't got it yet, but it's good to know I can, maybe later today I'll put together some time to give it a whirl.  Too bad the AI Denoiser isn't available yet, that's the bit in the video that had my jaw on the floor, but if AI Lighting is giving 40+% speed increase that's pretty huge too.  :)

    Have fun! The AI denoiser is available for the pathtracing kernel, just not the pmc. I'm testing it with the pathtracer now, and it looks like it will give the same quality image in about 1/3 to half the time for the dragon image.

    Post edited by DustRider on
Sign In or Register to comment.