Seems like the Luxrender development is going well as far as features and light simulation is concerned.
Though what concerns me is that there’s still a number of WIP threads on the Luxrender forums where it’s around the artist struggling to get clean renders with parts of an image rather than getting excellent results and proceeding to add more to it. It really does seem like that there’s some people who give the impression that you need to fiddle with the settings and/or fight with the engine before you get something that really stands out.
Of course I’ve seen plenty of clean images, but the majority of them are built simply to play with the light physics rather than create a full environment. (ie. making an image catering to Lux’s strong points).
Luxrender can be hard to use but you need to read the manual/wiki and at least try and experiment/understand for yourself.
Some people have the sort of opinion that it’s a light simulator and for that reason everything should be built to real and expect it to work, IE “I can throw anything at it and it will come through clean after 6 hours”.
No, actually though you’ve really got to think what you’re doing as much if not more than any other renderer.
You can’t make layers behind layers of glass with dispersion enabled behind another layer of frosted glass, stick a 1000 triangle light behind that and expect it to produce acceptable results in acceptable time.
Relative to other “simpler” (subjective) renderers, Lux struggles with glass surfaces, IMO this is a source of a lot of noisy renders.
Eventually when VCM starts being integrated we might see a big change. Not only by being able to more effectively sample glass and diffuse materials at same time, but afaik the mode should be a more robust/versatile default in comparison with bidirectional. Currently it is, as you say, often necessary to tweak/fight the render method and settings to fit the scene, or to build the scene with the render method in mind which is of-course not fun for new users.
This sort of thing is by the far the biggest culprit of people struggling with Lux, and it’s also where the 5 day long renders come from. An engine like Lux makes people think the old rendering/comp maxim of “It doesn’t have to BE right, it has to LOOK right” no longer applies. That’s not true, an unbiased renderer is just another tool to make things look right without so much fiddling with arbitrary properties. When designing a render engine, there’s a balance between “fast, but takes work to get good results” and "gets good results, but takes work to do it quickly. A physical/unbiased renderer shoves that slider as far as it will go towards the latter.
If you want fast results out of them, you need to pay attention to what exactly you are throwing at it: Do you really need bounce light from the sun off the floor to illuminate this room, or can you shove a fill area light behind the camera? Is multiple scattering on that candle that takes up 35 pixels REALLY necessary? If you disable dispersion on this, is anyone going to actually notice? Can you fake that sunbeam in post? Is the 10000x5000 HDRI REALLY necessary, or can you just replace the background later?
Clever comments, Rhys and J_the_Ninja.
But, starting from the fact that a good faking knowledge does help, would you still advise Luxrender over Cycles?
I’m not trying to flame or anything, I’m just a curious hobbyist user trying to get te most out of his limited spare time.
Thanks for any hint, and keep up the good work Lux team!
luxrender - can give some awesome images but you it will take a lot of time to render and setup materials etc lights.
cycles - if faster but you have to know exacly what you are doing, it is more flexible as well.
vray - cost quite a while for that price you can make cycles buying hardware incredible fast. Or you can buy the best rendering engine on market
Would you advise a SUV over a sedan car? Without knowing more, you can’t answer. They fit different purpose, each one has its strengths and weaknesses.
Regarding the LuxRender UI for materials, I hope that things will be able to improve with the opening of the nodes API in Blender.
I’d say Cycles is much more tightly integrated within Blender and has a wider support of Blender features while LuxRender has more advanced rendering algorithms and is much more focused on physical correctness. Regarding speed, it’s not a clear cut, there are many scenes that would render faster with LuxRender than Cycles, it all depends on the setup, and Lux will soon have a full GPU renderer.
I think you typed yafaray but had something else in mind because yafaray is free and open source too.
What the different purposes? Animation for Cycles VS still pictures for LuxRender?
OK, let’s talk about still pictures alone: could you please try to summarize the strenghts and weaknesses of the two?
Thanks in advance for your help!
Just look at my previous post. LuxRender will give you a result comparable to a picture taken of the setup in real life and it has a set of very powerful algorithms to do so. Cycles will make special effects easier, it allows to cheat with completely unrealistic lighting situations. That might or might not be what you want.
When using LuxRender you need to have the mindset of a photographer in real life (with some added possibilities), with Cycles you can keep the mindset of a more traditional 3D CG artist.
Both are great tools in my opinion, they are both quite versatile and I think you really need to see by yourself which one best suits your needs.
Fair enough: I don’t have the knowledge nor the time to become a photographer, so I guess it’s a better bet to try and understand Cycles and cheat my way to a decent - whilst not physically accurate - picture.
Just to be understood, the main purpose I use Blender for is to make some architectural visualization shots, mainly interiors.
Though just as with Cycles and Luxrender you have to know exactly what you are doing to get a good result with Vray, which is why really good Vray users are so valuable on the market.
I don’t think that any render engine is just “jump-in-and-go” friendly. You gotta know which settings to use or you will have to wait forever to get good results or settle with less quality, and that is true for all renderers.
To the Luxrender team:
Congratulations on the new release!
This one looks amazing just as all the other releases, will take some time to play with it as soon as I can
I use Vray often and you gotta try to render more complex scenes before you can say that it is “jump-in-and-go” friendly. If you don’t have the right settings when rendering a high-resolution image with a lot of complex materials and light portals and other tricks you will watch Valve release the third installment of all their games before your render is finished (in other words, the world will end before it’s done)
Even if you are just rendering an extremely simple scene with Vray (a monkey head in a box) you can set up the render settings so that it renders a really fast clay render but add one simple texture to the ground and it will immediately take 4x longer to render if you don’t adjust the settings accordingly. Now I’m not claiming to be a professional Vray user or anything (far from it), I am reading the manual almost daily and I still learn something new all the time, I probably haven’t even explored 20% of the vast ocean that is Vray.
I got to say that Cycles is probably the most user-friendly render engine I have ever used and the one I would personally say is the easiest to learn
I’m glad that memory leak got fixed. I look forward to being able to use material nodes for Luxrender in the future. As for Cycles vs. Lux? Why not both! Both keep getting better. I mostly use Cycles right now, but I still spend some time learning and testing Lux too. It’s the best of both worlds.