I never said there is no “decent” landscape works in Cycles, merely that I have yet to see one that I could have difficulty determining if I was looking at a photograph or CG art. When it comes to other render engines, Octane in particular, but also Unreal, I have been fooled.
But regarding reducing it to software as such – I can honestly come to no other conclusion. As I said, I refuse to believe that Blender artists are inept at photorealistic landscape work…so what other conclusion should I come to? It’s a bit like cooking, there is technique and there are ingredients. If I give a top chef who has excellent technique, crappy ingredients, the result will be limited by how good those ingredients can taste…no?
But if someone can find that photorealistic landscape rendered in Cycles which has completely fooled them, please I’d love to see it!
So you’re saying there’s no such work, so cycles is bad, not bad artists. Artists are cool, artists are good, it’s just that their work is bad. Hmm, how do you determine they are good artists if not by their work?
And cycles is used in Hollywood? No? Well, because it’s bad?
And why aren’t there renderings in Mitsuba, luxcore? Each of them if we talk about algorithms are more realistic. Following your logic, the conclusion is clear.
Cause and effect are different things. Even if cycles became the world’s best renderer tomorrow, it wouldn’t change anything, it would remain in its niche.
Why use redshift when there is octane? Why arnold, why vray, why corona?
In your analogy, Cycles or Octane would be two different ovens, not two different sets of ingredients.
The ingredients would be rather things like textures, shaders, assets, HDRIs etc.
So your claim is, if you give a top chef the best of ingredients, but force him to work with a less than excellent oven, the resulting meal would taste worse than the same cooked using an excellent oven, to a notable extent.
I doubt that would be the case. I tink there’s just a lot more going into it than choice of oven.
edit: In other words, I have a hard time believing it’s a binary matter of either the work gets screwed by the artist or by the renderengine used.
Ok, but as I said, I would love to see some convincing results then. Ideally not a plate of fruit made up of scanned models. Someone show me a convincing landscape rendered in Cycles, please!
I would not forget that accurate albedo and exposure values play a huge role (especially in cases like creating a contrast between rock and snow).
Cycles never had a tonemapping feature designed specifically to compress highlights, so it was only recently that we could now get good results using accurate values for the sun without blowing everything out and without doing any corrective mapping in the compositor or with curves. When your color transform is off, all of your shading and lighting values will be off as you compensate for it.
In fact, compensating for outdated tonemapping compared to professional solutions was one reason for the Light Falloff node, the smooth parameter was a way to prevent blowout and with AgX should more or less be obsolete unless there is a specific look being sought.
It was not my intention to offend you, nor do I want to paraphrase you. I’m trying to talk about substance, not form. So if it was unpleasant, I apologize.
Nevertheless, how can we determine why the program is bad and not the artists working with it?
Usually, a statement is proven by the person who made it, not refuted by those who disagree with it.
To prove you are right, you need to show the basis of your assertion, to prove you are wrong, someone has to do a first class job.
I am sure you’re writing with a strong desire to figure out what really differs between Octane and Cycles’ respective outputs. But I haven’t seen anyone so far take the necessary steps of comparing renders pixel-for-pixel and break down what makes up this difference in light response, apart from talking in considerably vague terms…
I mean yes, @patnard & @pixelgrip did. How much work would converting a real-world scene from one renderer to the other ? I can’t find the last thread where this was debated… we had a nice render to work with, but the comparison was flawed because some values didn’t match very well.
Yea, me too.
I believe an important bias is playing into that metric, namely that Octane and UE being commercially established for a long time, are far more likely to attract experienced image professionals, people who will inevitably demonstrate better picture making skills regardless of the software.
Thank you for posting it, yes I am very familiar with that video. It still feels to me like it might belong in a AAA video game cut scene, but I would never mistake it for real video.
Believe me, I’m equally frustrated. For one, I don’t have access to Octane since I’m on an Intel Mac and they stopped supporting the platform. Secondly, vague terms is unfortunately all that I have to go by, I can appeal by other people to confirm that indeed there are differences in perceived photorealism between Cycles and Octane, but I’m not a software engineer, nor a scientist; I am a filmmaker/creator who is evaluating the best tool for the job that I’m tasked with, while at the same time really trying to figure out if I’m nuts or if others see what I’m seeing.
I’m not trying to solve the problem, I’m just trying to understand it…and perhaps see if there are any techniques that could lessen the issue in Cycles.
Yes, I get it…not the first time I heard that one. It’s frustrating, but I can’t other than to urge you to use your own eyes. As I said, I do not believe that only Octane has first-class artists, while all Blender users are a bunch of chums when it comes to this stuff. I see incredibly impressive stuff come out of Blender artists all the time…how can it be possible that absolutely nobody has managed to produce something that is in the same ballpark as what I see routinely come out of Octane and even Unreal?
One of the only videos that I’ve seen (albeit heavily processed) which comes close to the photorealism I’m talking about (i.e. where an objective but educated observer would have a hard time determining if it’s a CG image or video) is William Landgren who incidentally has created a somewhat controversial Blender plugin called ProLens. His video Impetus has some genuinely believable moments. You can catch it here: https://youtu.be/ZJeEhfhyO5Q?si=dmOh7lKgDv6XSkpI
He is using Cycles (I think…unless it’s EEVEE), but he’s also utilizing his ProLens add-on which makes me wonder if perhaps the issue with Cycles is the camera and not the renderer?
Honestly, a lot of this boils down to selection bias. While I’m personally of the opinion that Octane can look better, it can also look worse. I can easily find you examples of Unreal landscapes that look awful. You believe that Octane always looks better, which is a self-reinforcing belief, you’re going to seek out examples that support that belief
No Joseph, I’m not saying that. I can easily find tons of examples of every render that look awful. All I’m asking for for some video examples of Cycles really selling on photorealism for landscapes. I have yet to see a single one…have you?
Those are really good, but they don’t look realistic to me. Stranded has that grey haze to it that tends to break my believability for it. The Fisherman’s House doesn’t have the grey haze, but something else breaks it for me, maybe it’s the dock? I’m not sure what the difference is, but I tend to like spectral renderers more than other renderers. Indigo is another renderer that is spectacular (Octane is based off of it) and Maxwell too. The images from them just look great. When I first started using them even rendering a simple cube looked more real to me.
