Adding to Cart…
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.You currently have no notifications.
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
Octane is quick, even without denoiser.
Next version of IRAY with denoiser whould be cool. Then Iwouldn't need anything more. The least I needed is CYCLES.
If I want that (Cycles), or a PBR render based on it, I'll use Poser's SuperFly renderer. I'm tempted to try Octane agian after V4 comes out, but for now Iray is what I'm going to use.
aside from testings, I use IRAY for DS exclusivley and mostley octane render for Poser.
That's an impressive looking Octane render you posted. @Masterstroke indeed. I do some scenes with a lot of elements, and I'm partial to TruForm's and Shaara3DMuse's vehicles and sets, which are set up for Poser and can therefore require lots of shader replacement to use in Iray. My concern about Octane (which obviously yields more aesthetically pleasing renders than Iray) is that I then will have to do texture tweaking and shader replacement for all my Iray sets as well (since I've read that Octane is not always friendly with Iray materials). Can you speak, please, to that concern (and I realise that you don't use OR for DS)?
Well: first of: thank you, but i cannot take any credit for it. Ihad nothing much to do to get that good result. I just wanted to test how fast that ORV4 can be with that new denoise function. And you're right. It can allways get better by using the Otoy shaders instead. It then requires some fiddelings as you said. Since I am quite happy with IRAy, OR will be mostley used for Poser scenes. OR has some difficulties with geometry shells anyway.
..Cycles means having to deal with Blender and it's clunky keyboard driven UI.
Octane 4 will be able to work on AMD (meaning Macs as well) and even Intel integrated graphics.
All about perspective, as it is more of an efficient UI, that spreads out tasks and streamlines workflow.
@xyer0 "Octane (which obviously yields more aesthetically pleasing renders than Iray)"
Sorry, can you back that up? A lot of people post average or worse Iray renders. A number of people post wonderful Iray renders. Which do you have in mind?
I ask because the scene you commented in this post, while good, doesn't display anything that can't accomplished in Iray.
Like Masterstroke noted, using Octane typically will require some shader tweaking. Some will convert extremely well and you may be satisfied with the results without modification. Others will just need one or two settings modified, while some may require a serious rework to get the results you want. One thing I have learned is that often the texture sets are much better (amazing details) than I thought they were from the promos, or from use in Iray/3delight. So that has been an unexpected plus from tweaking shaders in Octane.
Bottom lune, Octane will require more hands on shader work than Iray (even if your a compulsive shader tweaker like me).. So the downside is it can require more work than Iray, the upside is that you can end up with better results because you were "forced" to adjust shaders (but keep in mind you could do the same with Iray, it just sometimes you may not realize how poorly the shaders are designed for your environment).
The best way to know if it is for you or not is to give the demo a serious try. Most of the time it's just a matter of adjusting specular, ior, roughness values. Sometimes the best thing is to just use one of the many shaders in the LiveDB. Other times you may want to completely rebuild the shader. For many people this process is not enjoyable, while for others it's a part of their workflow that they really enjoy, and always want to make everything look the best they can.
As I noted earlier in the thread, I use Iray more than Octane now, and the main reason is that I spend less time adjusting shaders in Iray than with Octane. But if Octane 4 turns out to be everything I think it will be, that is likely to change due to improved speed and other features (it will definitely change if Otoy implements MDL support, then shader/material conversions would be a non-issue).
...for 20$ a month vs. nearly 600$ to purchase the software outright originally, yes that is an incentive to finally give it a try as it isn't a time limited trial version with tools and features locked out, it is the complete package.
If by "back that up" you mean cite clinical data, double blind polls, or wavelength graphs, then, no, I cannot. But my opinion that OR is relatively aesthetically superior to Daz's implementation of Iray is not based upon the render that was posted. I started researching Octane in 2014 while trying to determine if I was going to make the switch, Reality, 3Delight and Firefly being my only options then. At the time, I felt that Octane competed well against Reality for image quality, and, of course, it blew Reality away in render time. I was poised to take the plunge when Daz premiered their Iray version. A while later, either TRRazor from redspec or Laticis (I forget which) did some head-to-head tests on deviantart.com and asked opinions. It was clear to me and the tester that if money were no object, Octane for Daz had a subtle je ne sais quoi appeal to its images which Daz Iray lacked, (though the difference could conceivably be approximated in post). This anecdotal testimony and my eyes are the only back up I have, but I feel confident (and I have no dog in this fight) that if you make comparisons, you will FEEL (for it is an artistic inner impression of quality) the same.
Thanks, DustRider, for breaking it down for me. And now I understand why Octane renders look better to me than Daz Iray. Is LiveDB free, or does it require a fee?
Are you using standalone or the Daz plug-in? I remember that the gentleman who made it took quite awhile to update it, and I was wondering if that (his consistency) had changed.
My choice is production volume versus quality. I'm already well past my envisioned deadline, and learning a new system might not be too efficient timewise. Since the proof of the pudding is in the eating, I will take your advice and jump right into the demo. Thanks again for taking the time to explain your own rationale amd make suggestions.
The DAZ plug in has been taken over by the guy, who's in charge of the Poser plug in. Following the forum, he does a good job. Allways helpful.
I guess you don't make animations. If you did, then you couldn't live without CYCLES. I mean sure, you can use Iray than "only" takes 10x time (no denoiser) and 10x memory** (no micro-displacement) to do the same things. But why ??? Sure with Cycles you have to spend some time exporting and fitting assets, but you only do it just one time. Then you go light-speed compared to Iray.
Apart from Iray, DAZ Studio just does not fit animation for so many reasons anyway.
As for Octane AFAIK it doesn't support micro-displacement*** so again it's not good for animation. As for out-of-core memory I honestly believe it is a non sense. If the scene exceeds 4-8GB it only means one thing. IT IS TIME TO OPTIMIZE THE SCENE. Do you really believe a 20GB scene will render fast anyway ???
EDIT
** 10x memory is the case when a large scene is built using multiple instances of sculpted objects, that's a common case in production. If you only use small scenes with a few sculpted objects then micro-displacement may not be required. But it would still be useful for memory optimization. Micro-displacement uses adaptive subdivision to automatically generate the required geometry.
*** Of course Octane does support displacement the same as Iray. But it is not micro-displacement. That is, it does not have adaptive subdivision. That's what makes the difference between micro-displacement (adaptive subdivision) and standard displacement (standard subdivision).
Nope, don't do animations. ;-) If I did, I whouldn't go for Blender either. Powerful software for sure, but i feel lost there. Blender is worth a Sheldon Cooper award for UI design. I've downloaded allmost every version of Blender, tried it and shut it down while yelling at my screen.
LOL .. Well honestly I was in the same shoes the first time I used Blender. It was 2.4 so a loong time ago. But I was lucky since I came from Lightwave so I was used to shortcuts to optimize the workflow. And actually, as odd as it may sound, the Blender interface is optimized for production. During production when I model an object, or when I do animation, I really can't waste my time selecting menus or switching widgets. Shortcuts are just the only way to go. I couldn't live without shortcuts.
I can't agree with that. I could easily put many more than 3 characters into a scene, including an environment, and the final result would look exactly the same as one run using more resources. The real issue is whether or not you can be bothered to put in the time to optimise. If you can not be bothered, then you either need a large VRAM GPU, or use an out of core renderer like Octane.
True, IRAY render. about 2h render time.
The LiveDB is free. It has some really nice shaders, which can be used as is, our as a starting point for your own. It's not only a great shader resource, but also a great learning resource.
I use the DS plugin, the Carrara plugin, and occasionally the stadealone (like now, when the plugins haven't been developed yet for the new version, or I want to just mess around with it). Like Masterstroke noted, since Paul (face-off) took over the development of the DS plugin, things have been great. We still don't have the plugin for 3.8 , but I'm sure it will be available fairly soon (the updates are usually very fast now, this one has taken a bit longer). Paul also does the Poser plugin, and IIRC a couple of others (the Carrara plugin is already at 3.8, it's done by a different developer), so that, possibly combined with some issues associated with integrating some of the new features into the plugin has slowed things down a bit. The DS plugin is quite stable now as well, and no longer a crash fest like it was prior to the current developer taking over.
If your production volume is the primary concern, and your not doing animations, my first thought would be to stick with Iray (or 3Delight??), because the integration of DS is so good with Iray, and most products come with very good Iray materials/shaders now. But the longer I thought about it, I realized the Direct Lighting kernel in Octane is really fast (similar to using the interactive mode with Iray - but faster), so there might be some real time savings there if your doing a series of images where the actors/props are all the same (if SSS isn't needed, because the Direct Lighting Kernel doesn't have SSS). Actually, if your doing images where the actors/props are all basically the same, say for a comic or graphic novel, then there might be some time savings with Octane vs Iray, since Octane is a bit faster for similar quality. It's hard to say for sure, especially when I'm not sure what your doing, and your skill set. If Octane 4 was out, and had all of the noted features, then Octane might be the better choice in general, and definitely for image series type work (it's not that I don't think it will have all that has been promised, but over the years (30+ doing computer tech stuff) I have learned that it's best to go with the product the does what you need now, rather than "waiting" for the one that is "better", because "stuff" happens, and even the best of intentions and planning can miss small details that slow things to a crawl, or completely prevent the intended outcomes).
Hope this all makes sense, and helps a little
Edit: Just thought I'd update my post to note that the developer of the DS plugin, Paul (face_off), also does the Modo, Revit, Inventor, and Autocad plugins, and of course the Poser plugin, which was his first Octane plugin.
Whilst it may be the real issue, it shouldn't be.
The thing is, the word 'bothered' puts the blame on me; that isn't where the blame should be.
If I have to be 'bothered' to tweak (just to get it on the card), I start looking for alternatives that work how I want, not force me to work how 'they think I should'.
So yes I am very frequently bothered, but it annoys me that I have to, when they are false restrictions that other products offer different work-arounds. Sure there are learning curves, but learning is something I enjoy, fighting restrictions, much less so.
Just a couple of questions. First, could you post an apples to apples example (or link to an article) or two of an Iray render that took 10x the time of an equivalant Cycles render, that would be very helpful (I tried to find a gallery for you where maybe you had posted some examples, maybe my google-fu failed me)? I'm guessing your using GPU rendering for Iray, and cycles (systems specs would also be handy to have)? It sounds as if you might be using DAZ models to anumate in Blender, I was wondering how you are geting the fully rigged, wieght mapped, and JCM info into Blender, or how you are getting the animated models into Blender (or are you using Superfly in Poser for this)?
I'm guessing your kidding a bit about not needing more than 4-8 gb for a full scene? I seriously doubt that Otoy would have put the time and money into developing out of core textures, and soon out of core geometry, if it wasn't needed by the industry. Yes, more stuff in a scene takes more time, but it's been my experience that a fully detailed scene, will take a lot of resources, even with optimizing (which can also have a point of diminishing returns). For Pixars Monsters University, they "tried" to keep the per frame size to 20Gb (https://www.theverge.com/2013/6/21/4446606/how-pixar-changed-the-way-light-works-for-monsters-university), so I would guess that their most recent film Coco took 2-3x that per frame (it seems to be hard to find any real information on per frame memory requirements), and some frames took 100 hours to render (http://variety.com/2017/film/spotlight/coco-pixar-disney-1202637602/).
Yes, frame render times like those are beyond what us mere mortals with very limited resources can handle. But my point is, to get the best possible image, regardless of still or animated, it requires more resources, both in processing power, and the memory needed to hold the data.
"Just a couple of questions. First, could you post an apples to apples example (or link to an article) or two of an Iray render that took 10x the time of an equivalent Cycles render, that would be very helpful (I tried to find a gallery for you where maybe you had posted some examples, maybe my google-fu failed me)? I'm guessing your using GPU rendering for Iray, and cycles (systems specs would also be handy to have)? It sounds as if you might be using DAZ models to anumate in Blender, I was wondering how you are geting the fully rigged, wieght mapped, and JCM info into Blender, or how you are getting the animated models into Blender (or are you using Superfly in Poser for this)?"
There is nothing wrong with your google. You will find it very hard to find an "apples to apples" example simply because we aren't dealing with two apples, but rather with two different kinds of fruit. Cycles and iRay have different shaders and different ways of rendering them. Different artists have different ways of applying the shaders which have an effect on rendering time. What matters is how your way of working affects render times in the respective renderers and, as has been mentioned, it's not just the time spent rendering, but also the time spent preparing to render that needs to be taken into account. The complete workflow. So if I have to import a model into another application, fiddle with the rig, reapply shaders and fiddle with them in order to get similar quality as in the native application, this becomes as important a factor as the actual perceived render speed difference.
"Yes, more stuff in a scene takes more time, but it's been my experience that a fully detailed scene, will take a lot of resources, even with optimizing (which can also have a point of diminishing returns). For Pixars Monsters University, they "tried" to keep the per frame size to 20Gb (https://www.theverge.com/2013/6/21/4446606/how-pixar-changed-the-way-light-works-for-monsters-university), so I would guess that their most recent film Coco took 2-3x that per frame (it seems to be hard to find any real information on per frame memory requirements), and some frames took 100 hours to render (http://variety.com/2017/film/spotlight/coco-pixar-disney-1202637602/)."
...which is why you can't compare what film studios are doing with iRay or Cycles or Octane artists' renders. For one, they don't use a gpu renderer. There isn't a gpu renderer advanced enough nor was there a gpu card with enough memory to hold the scenes that Pixar generates, even when you take into account the out of core feature. So that 100 hours is on a cpu renderer, which is as far from an apples to apples comparison as you can get.
I would like to chime in and just state one thing: all of the render engines mentioned in this thread have their very strong arguments and sides and each and every one is useful. What subjective "which is better" falls upon is how much use a particular user can have of any. One would argue that 3Delight/Renderman is the top-of-the-industry easy-to-work-with rendering engine used in so many high production industries with very awesome results, whilst in Daz community 3Delight didn't meet majority of acceptance especially since the entrance of Iray.
To state which is better we could all individually take our *personal* experiences and put the on the table and compare endlessly. At the end of the day everyone would have a valid point, because it works for them.
Now, from technical end what one engine has and the other doesn't facts are out there, but how much of that would an average Daz Studio user use to make an output of the image that is aesthetically appealing to them (again personally and subjectively) and their targetted audience (which is again personally and subjectively oriented towards observing the art). Let's not make this a war between render engines because neither side will win or lose no matter how many arguments are stated and backed up.
Octane: has big attention right now, is very nicelly developed, gives great results with Daz Studio and Poser and variations of their figures. That's a fact. Learning Node Graph Editor is mandatory with Octane, for which I am not sure how many people would invest the time to inspect, learn and advance in, just for the output of the final image. Yes it is faster, yes it is shiny, yes it is all new.
Iray: it is as well still being developed and we cannot say what new features would bring. Iray was built by Nvidia which rings to its name quiet an echo, even if there's the other engine who can use the same specs faster. Everything in Daz store right now is Iray optimized and available within a click, as Daz has adopted Iray as its number one rendering solution, free of charge for regular users. Let's not forget this means A LOT to someone who wants to build a scene and hit render, even if time takes longer.
I have no wide experience in Cycles (or Evee also new of Blender), or Reality/Lux that is still being used, but I am absolutely sure strong positive points can be stated for these engines as well.
To sum it up, what works for you, is your best choice, and other person can only take the facts and decide for themselves, what their aim is, how much extra effort needs added and what kind of an output is to be expected. There are people who also render up to 20% and then use photoshop to get great final result. And imo what you get in the end is the only thing that matters and the whole process behind it is up to a user to build, step by step. Things that work for me might not work for someone else and they won't use the extra features or find them useful or easy to handle, and vice versa.
Great thing for everyone involved though is that we have so many options to chose from.
I like the idea of a different rendering engine as a new "flavor", but that's a lot of money and work for that new spice. Would like to see something big come down the pipe.
To sum it up, what works for you, is your best choice, and other person can only take the facts and decide for themselves, what their aim is, how much extra effort needs added and what kind of an output is to be expected. There are people who also render up to 20% and then use photoshop to get great final result. And imo what you get in the end is the only thing that matters and the whole process behind it is up to a user to build, step by step. Things that work for me might not work for someone else and they won't use the extra features or find them useful or easy to handle, and vice versa.
Great thing for everyone involved though is that we have so many options to chose from.
Oh great, here you go- posting something elegant and rational and inviting Richard Haseltine to show up and lock this thread because it's starting to go in circles. Dang-it, man, I was enjoying the back and forth.
I'm really interested in some of the potential new workflows possible with Octane. If I'm reading the documentation correctly, it looks like I could export a scene from DAZ Studio using the Octane plugin using the .ORBX file format, export a fire/smoke simulation from Houdini using .VDB or a dynamic hair using .ABC, and then import both the .ORBX of DAZ scene and the Houdini .VDB/.ABC into the standalone Octane render tool. To me this would be the holy grail of being able to get the power of Houdini and the pre-built DAZ assets/scene setup into the same rendered image. This is really not possible today because all the various exporters in DAZ Studio (.obj, .FBX, .ABC, .DAE) are bugged in various ways or don't support everything when loading into Houdini and also because of how totally different assigning materials are to surfaces in Houdini from DAZ Studio manually setting up a complex scene is unacceptably time consuming. One would still need to convert the materials to Octane from Iray but it seems like the conversion is much easier than the alternative of building the surfaces by scratch in Houdini.
So true!! That's why I was hoping for images to see how the final results compare. If they are very very similar, at 10X the speed, then Cycles might be worth investigating. I'm guessing the 10X speed is partially due to using a denoiser, however it's been my experience that if used to soon/agressively, it can reduce image fidelity/details. Not as big of a concern for animations, but a single image artist may need more render time than the animation frame. But of course, I would hope that any sample images from both render engines would show the best human skin shaders the user was able to create, as well as caustics, metals, and volumetrics.
Years ago with Poser I did both the export of static objects to another software to render, and the "hosted" figure integration (Transposer - similar to Poser Fusion). Both were less than optimal (kind of a real PITA compared to native use of the figures inside the application) and not something I would really want to repeat. That's why I was asking about Padone's work flow, as that is something I would also need to consider if attempting to use Cycles (or upgrade to Poser Pro 11, which has it's own drawbacks).
I was just countering the statement "As for out-of-core memory I honestly believe it is a non sense. If the scene exceeds 4-8GB it only means one thing. IT IS TIME TO OPTIMIZE THE SCENE". For some people this may be true, for others, maybe not so much. Take the case of someone rendering a very complex scene rendered at greater than 4K resolution (large poster, billboard, etc.), this could easily exceed the 4-8gb limit mentioned. Octane has made it's way into smaller, real production studios, thus the real need to enable the use of greater memory pools (note the samples at the end of the Octane 4 intro video below). Just like the big boys, these studios no doubt have a real need for increasing the memory available to GPU rendering.. I think though, that with out of core memory and geometry in Octane 4, that the memory barriers to rendering "Pixar" sized scenes will be gone, but there are other features that Pixar's Renderman has that Octane can't compete with.
As a side note, Pixar's Renderman 22 will even include GPU processing capabilities called XPU (https://renderman.pixar.com/news/new-rendering-technologies-at-siggraph-2017-1) so the use of GPU's for processing is beginning to be more important at the "big" studios.
Octane 4 Video
https://youtu.be/6xE3J56pabk
Or, in theory, you might be able to export the animation from DS to .orbx, then import it into Houdini using the Octane plugin, and let Houdini perform it's magic and render via the Houdini plugin?? I don't know if it would really be possible, and probably not because it just sounds too easy (didn't really research it), but it would be awesome if it did!
...I have tried Blender every time a new version was released and still gave up when I saw the same clunky UI that needed keyboard commands to get around.
...when the time to optimise results in diminishing returns compared to just letting the full scene render and you cannot afford a high memory GPU, that's when Octane has the advantage.