the team of redshift annonced an automatic sampling method for the 3.0
it’s @ 20min
the team of redshift annonced an automatic sampling method for the 3.0
it’s @ 20min
Regarding the idea of speeding up Cycles by introducing yet more features that introduce bias, did anyone ever consider the cumulative effect of more and more biasing in the effort to speed things up?
Add enough of them together and your images end up being no more realistic than the ones done when Mental Ray integration was the big thing in DCC app. development. Bias a little here, bias a little there, bias the shading a little, your dream of making realistic images endures the digital equivalent of “death by a thousand cuts”
Logical idea, but would need to see some test of what my images could look like without any bias and have it be ugly for me to worry.
Also realism isn’t really my aim.
That may very well be, but other renderers have been doing it for several years while Redshift 3 isn’t even out yet.
Doesn’t that really mean Eevee is all for you? Why would you even worry about cycles’ speed in the first place?
greetings, Kologe
Ok So, Redshift 3.0 does everything. Best of the best of the best. Sounds like it will cost a small fortune and its for big production studios only.
dude calm down, i was just responding to rawalanche who said that redshift is failing in term of usability ( i found it pretty fast to setup when you know what you’re doing )
It does not cost a fortune, it cost 500$ with 1 year of maintenance and free upgrade.
It’s for little and big studios, as a lot of freelancers use it too…
And there’s no renderer that can be the best of the best of the best
500€/workstation/render node/ year.
This means that for a small studio of 3 with a small renderfarm with 10 nodes you need to pay a yearly fee of 6500€.
It´s not a mega price, but is much more than what I find reasonable for a small studio for a render engine that is not used every single day in every single project.
If the price would be just for the workstations,and you get unlimited render nodes for example, or a 50€ price tag per node (maybe up to 20 nodes, so it´s kept for small studios) it could be more reasonable, but charging the same for a UI license than for a node license is what makes it unreasonable.
Just my opinion.
Cheers!
The thing is that it must be a user decision, is what something like Vray does, it gives you a wide range of options an tools so you can adapt the result to your needs, maybe you need realism and you want to avoid bias, but maybe you need a bit better time and the client is not after photo realism, just a good looking picture, it all depends on the industry and the client, having the tools is the important thing IMHO.
Cheers!
Lol, take a look at vray, redshift and so many other renderes that are by definition, more biased, and still outputs more realistic imagery than Cycles, lol x2
Nope. Vray and Redshift are just raytracers/pathtracers just like Cycles. Identical setup will give you nearly indentical image. The reason Cycles artwork quality is generally worse is due to several reasons:
#1 Can be solved by having more competent people use Blender.
#2 Can be solved by experience with Cycles’ limitations and knowing their workarounds.
#3 Can be solved by having someone create appropriate addons.
But my point is that when it comes to Cycles’ implementation of basic rendering math. Everything checks out. There’s nothing in the Cycles itself that will lead to worse quality.
Maybe one year ago and more, but from the introduction of Principled BSDF, (correct fresnel and easy material setup), Filmic LUT and Denoiser (clean and smooth renders without thousands of samples), Cycles artworks quality are nearly on par with others.
Still there are a lot more amateurs and “kids” using Blender + Cycles compared to Vray/Redshift/Corona, but that’s obvious being free.
Though, if you visit CGSociety and take a look at “Recent” artworks (not the beautiful, featured ones), you can see tens of crappy stuff made with commercial apps
This can’t be emphasized enough. People like to repeat how bad Blender’s average userbase creates bad art (I myself used to think this), but there’s an insurmountable amount of crappy artworks from proprietary software as well. They just aren’t shown on the apps ads or on the schools (that promotes those apps) reels, for obvious reasons.
Those arguments do not make much sense. Yes, people do crappy artwork with all the 3D software packages. That’s not the point. The point is that proportion of crappy artwork to the size of userbase is significantly larger for Blender than for the other 3D packages.
On Artstation, you can search artworks by software, and then you can sort them by recently uploaded. This can give you really nice unbiased random sample of average artworks:
3dsMax:
You can see that while all the 3D packages have their share of low quality artwork, Blender leads in the proportion of low quality artwork.
Bias has nothing to do with realism.
See “Five Common Misconceptions about Bias in Light Transport Simulation”:
I would be one of those who publish the considered crappy/low quality jobs. I simply use Blender and not other apps because it’s free and OpenSource. Of course, just a hobbyist here. Also, I do not have to register or wait for mail confirmations to download the software.
This has been discussed many times, and I do not see why the result does not sound logical for some people. By the way, I love the so-called crappy/low quality work from which many apparently hurt their eyes, try to avoid and discredit a software/engine. That means a lot of people are having fun with 3D software: Young, old, newbies, users who have not been endowed with talent but having fun anyway, etc.
So sorry if those of us who do crappy art are discrediting Cycles
, but that point that I have quoted above is the truth.
I have absolutely nothing about people making low quality art. My posts were simply a response to @juancarlosgzrz who wrote:
I was just saying that Vray and Redshift themselves do not produce more realistic imagery, but users of these engines do, on average. That’s not to say renderers do not matter. They do. There are in fact some renderers, that do get the rendering math wrong, such as Modo internal renderer, Cinema4D internal renderer, Blender internal, etc… Those renderers tend to output low quality art regardless of user’s skill, because they get many things wrong, but Cycles is not one of those renderers. Cycles is slow (in many, but not all cases), but not low quality.
There’s absolutely nothing wrong with producing low quality art. Not everyone does 3D for living, and even people who want to do so weren’t born with all the skills necessary. I was there too. These are some of my very first works from 10+ years ago:
Yes I know. Just quoting that point to say that you’re right about it.
I agree with the overall point you’re trying to make, but I’d honestly probably rank the Maya set as the worst of those samples. Sturgeon’s Law (90% of everything is crap) seems to hold true in all of them, though.