They both are rgb unidirectional path tracers. Arnold just got a lot more of development time and money than cycles, with a specific goal in mind (movie production) instead of being a multipurpose renderer. That’s why it’s considered ‘slow’ and lacks some features that some people consider important but are barely used in movie production (like accurate caustics). It was conceived to be rock solid and used in renderfarms with powerful cpus, and that’s why they haven’t paid much attention to the gpu side either.
Anyway, this is offtopic so I’ll stop writing about Arnold since the topic is Spectral cycles vs Octane.
It’s great that someone with Octane experience speaks up, these are still very vague statements though. Not using Octane, I have absolutely no idea what you could be referring to… I think we should avoid the sensory lexical field as much as possible : feel, seem, look, etc. What does the “physical” checkbox change in a render ? Does disabling it bring the render closer to a Cycles render ? Can we see an on/off comparison ?
You are asking for something rather impossible. The artists who use these tools for actual production work will refer to “feel, cinematic, look, vibe, etc” because this is exactly the language that we use for everything else that we do.
Meanwhile there are a number of commentators on this thread who want to focus exclusively on the science and math of it all, and if you look at their creative work it’s as lifeless and unrealistic as you can imagine because it’s simply not their focus.
So you have this weird situation where the more technical people are frustrated by the imprecise language spoken by the artists – yet ultimately it’s the artists who are the end users of this software, and if the tools aren’t useful to them, all of the effort of creating Blender is completely moot. I’m pretty sure that Ton would agree.
For once, can you share a work you did that emulated the feeling you’re describing evoking life and all the realistic attributes that isn’t possible to replicate using cycles?
This should probably be the simplest way to communicate your frustrations.
Please stop hiding behind other people’s knowledge skill and choice of toolset
Jama is concept artits,he also uses UE5.
Concept art is mostly post-processing and overpainting skills.
The results ultimately depend on how experienced you are in digital painting, color theory.
And it doesn’t matter what you use, cycles, or octane, or a viewport in a 3d software.
So maybe the artists can describe some of the words the use like:
…because the technical terms… well are defined by math and used models…
…so maybe we should not separate humankind by the domain language they speak because of their jobs / interest … but use language to speak with each other…
It doesn’t help nobody to just say: cycles is bad(which version by the way… i mentioned the rewrites like xcycles…) and of course it also doesn’t help to say cycles is the best and can do everything…
So how should we compare images if not by any measurements… ( if some image also may jist look different because of used color correction or uncalibrated monitores ) ???
–
So anyway: after my search of the original cornell box ( actually the link from:
) …the one without inner boxes… i found
Using his multi exporter (which i did know of but never used… because i have not those rendere…). But he also did some github wahnrender_comparisons and also has some scenes for comparison on his site : www.janwalter… download
…
So maybe we (?) can put togehter some setups and look if cycles specialista and octane specialist or any other can say somethign about this so everbody will learn something…
It objectively is superior since it can handle things that Cycles Principle BSDF can’t, like iridescence for example.
Now I am wondering this too and I am quite astonished that Arnold Standard shader is open source. I did not know that and I guess many others didn’t know that either.
It’s just not to be expected considering the company owning Arnold - but a pleasant surprise.
Can we please start a conversation/discussion about the validity of implementing the Arnold shader into Blender since it kind has become one of the leading Über-shaders.
It is available in Octane, Redshift and Clarisse, I think adding it to Cycles is the right way.
Rather than only digging into pixel values of Cornell-box renderers, I’d recommend looking at technical papers that describe the underlying architecture and its math.
Unfortunately I am more the artistic kind of guy, too much math and my brain recoils in horror.
I know this thread is about the comparison between Octane and Cycles, but I think we should consider widening the net and looking at more industry über-shaders.
I like the look of Octane and I respect its abilities, but I think Arnold is the better example especially if we take into account that Brecht has worked for Solid Angle on Arnold.
He is probably one of the few people that has extensive knowledge about both Arnold and Cycles.
[quote=“Okidoki, post:189, topic:1460308”]It doesn’t help nobody to just say: cycles is bad(which version by the way… i mentioned the rewrites like xcycles…) and of course it also doesn’t help to say cycles is the best and can do everything…
[/quote]
I take it you don’t work in a creative capacity and have to deal directly with clients?
I deal with this all the time. Clients (Directors, Ad Agencies, Producers, etc) speak exactly in vague terms. It is part of my job to be able to dig in, and figure out exactly what is not working. If I got frustrated every time a client tells me something vague like “Hmmm…I’m not crazy about it, it doesn’t feel right” I would have never been able to work in the entertainment industry for the better part of three decades.
So far, I have yet to see any of the more “technical” participants of this thread make any effort whatsoever to understand exactly why we feel that the images that Octane generates are superior to Cycles. Instead they dismiss what we’re saying and throw out not-so-veiled insults and putdowns, while accusing us of being unhelpful.
This is a great example of what this thread is missing- a clear-cut technical goal that one render engine can achieve that another can’t. More of this, and this discussion might go somewhere useful
I second this. I am not sure what all is involved in a project of this type, but if it was a paid add-on I would be quite willing to dig into my wallet to purchase it if it made Cycles renders closer to Arnold.
I don’t know either, but the reason I suggested it is because it already is open source. That should make it much easier.
The idea that Autodesk is spending money on RnD for something that ends up in Blender gives me a warm fuzzy feeling - that is the kind of teamwork I’d like to see more often.
On the other hand I wouldn’t recommend it if I weren’t convinced that it is a very nice shader - if I had to pick only one it probably would be my personal favorite.
That’s the reason i’m asking… for me it’s also very difficult to know why someone says that this or that is nice to have or the other way round why that and this is bad… if there is no nice explanation behind it…
And even if i think this thread is interesting… it dissipates from the original question… what is the difference… so someone know how to use it (or not)…
I already learned a bunch of stuff by just being here on BA… not everytime participating in the disccussion ( except giving )… and also not evertime understanding what’s going on…
So back to the topic:
For example @AlexeyAdamitsky mentioned the arnold shader/material before and the link was just posted ( or i’m mixing this up ??)
The problems with the sheen and metalic closeres (mentioned on the github page) are halfway solved now ?? (at least blender 3.5 doesn’t complain about metallic…)… so this is… i personally don’t know yet…
( Also do be honest: i wasn’t very much in ueber-shaders before and the Principled BRDF is just… “easiest” to use for my simple trials of oh- this look-nice … )
[ So while i typed this there where some replies… maybe i already talk crosswire ??]
As Alexey mentioned, I suspect that it’s not one thing but a combination of multiple factors. For instance, I had not considered the Camera as a potential factor in the look of the render.
I also tend to agree that perhaps the Cornell Box isn’t the best example to use, as I see some limitations in being able to evaluate what exactly is going on with just colored boxes and spheres.
I wonder if there is a environmental scene that can be created with Quixel elements which could be used as a better point of comparison?
Let’s be honest This thread hasn’t been about spectral cycles vs octane since its inception. It’s been more about what can cycles do to bring it closer to photorealism. Soo talking about Arnold is fine.
If Arnold and cycles are in theory identical in how they function, it does raise the question on why is the look soo different between them. Is it that ubershader that is making the difference?
Look, I have no idea why the sudden animosity. I haven’t seen insults anywhere in this thread ? or accusations ? there is no need for any of that. Not sure whether you were referring to me in particular, or others ? but personally I’ve been trying to dissect the images that have been posted thus far and figure out where the differences arise.
That’s exactly what I’ve been suggesting isn’t it ? If you’re good at these, no reason to be afraid to be a little analytical, especially since you have experience rendering with Octane. I personally don’t have a license and too little time right now, however I am glad to scrutinize other people’s renders if they’re able to output some.
Earlier, the camera was mentioned. In what capacity does a “physical” camera change a render ? Does it handle phenomena such as chromatic aberration ? is this any more correct than doing it in comp ? does it look good because Octane is spectral ?
I had a look at the doc. It seems to say a “physical” camera is just another way to set it up using real-world units, such as focal length instead of field of view, exposure, distortion, dof, bokeh, etc. these are all standard properties of a Blender camera, so I’m not sure what would be the added value from this, if any.
Oh, it has a couple neat things that might be of interest :
Optical vignetting
This simulates the effect in some real camera setups where the brightness and saturation decreases toward the edges of the frame. Depth of field must be on for this, and it’s impacted by the F-stop/aperture settings. There’s a vignetting option in the Camera Imager that does something similar for the Thin Lens camera, but as a post-production filter, where optical vignetting simulates a real camera vignette.
Aberration
This simulates imperfections in lenses that cause the image to soften (defocus) or distort. It can be used to simulate more unique lenses where these imperfections are seen as giving the image character. This is different from chromatic aberration which splits colors on sharp edges. Chromatic aberration isn’t currently supported.
Are these what previous posters were mentioning ? can we perhaps try renders with and without those ?
I’m also going to quote this because this is exactly what we are trying to do in this thread.
the cornell box exists to have a real world baseline. If you are just picking some quixel assets and dropping them into a scene, what is the baseline?
We need to be technical, because it isn’t a feelings problem.
You’re totally right that the artists will ask us technical people to make the renders more ‘glamourous’, and we need to be able to take those soft descriptions and turn them into hard settings, driven by science. That is why I keep banging the drum of getting more technical. We can’t wave our hands at the developers and tell them to make cycles better.
Have you ever been in a museum? what percentage of paintings are highly stylized?
You really seem to be trying to play both sides, either artists are artists and can’t be bothered with the technical aspects, or we are all supposed to be fixated on photorealism and stylization is a cancer.
We can all tell that cycles renders do look flat in comparison, Now what can we do about it? how can we pinpoint the issue? Is a more complex scene laden with photoscanned assets going to suddenly reveal that the falloff curve isn’t right, or the tonemapping is wrong?
It just seems like you want even more examples of octane being superior to cycles. Great. It is. Now what?
All I know is that the Cornell Box seems rather limited, and perhaps a better baseline is necessary. I’m not quite sure why it seems to have achieved this status as the ultimate tell-all standard. For instance, when it comes to PBR shading and how well a shader is displaying a texture the Cornell Box is utterly useless.
If myself and other artists in this thread could sit down with Brecht and converse about it, I am sure that we could develop a good communication pipeline that he would be able to translate into actionable tweaks. I suspect that he’d ask the type of pertinent questions that would allow him to narrow down the issue in a more technical way. I also suspect that he’s also quite aware that there are some fundamental issues with Cycles, given how he probably knows where all the bodies are buried.
I am starting to wonder if, as it’s been mentioned above, it might be the way Cycles lights behave?
I posted the World Creator video example since IMHO it is an excellent example of where the “is it CG or is it real?” line has been blurred. If I didn’t know anything about that video, I would just assume that it’s real footage captured by a high end film production camera – no question.
The reason why I specifically posted that video, is that in several years of following Blender artists on these forums and on Artstation, I have yet to see a Cycles render that blurs that line. I don’t consider myself a high caliber CG artist – I would have a hard time getting that level of realism out of Octane myself. But there are high caliber Cycles artists out there that surely know what they’re doing…and if so why is it that I have yet to see a single example of an outdoor environment that approaches that level of realism? I believe that it’s a legitimate question, and if I’m missing such artists – please someone link to some examples that will completely make me re-examine my conclusions about Cycles.
This thread ultimately is just a conversation and nothing more. Cycles will do what it wants to do. If Blender developers are even aware of this thread, they are certainly not chiming in with any thoughts or information. And even if they did, my guess is that they have their priorities for the next three years at least, and I doubt that this thread would change that in favor of addressing some of the things that have been mentioned here.
This leaves me hoping that between the new Principled Shader v2, AgX, and perhaps some other subtle behind-the-scenes improvements, Cycles will come around to at least play in a similar photorealistic ballpark as Arnold or Octane.
)… and also not evertime understanding what’s going on…
)