why not abandon toon style renderers, the new node based shader system, with better render API … freestyle is going to be a way better option than toonshaders and edge renders gonna be.
+1. Awesome work you’re doing, thank you.
Now I finally don’t regret buying a GTX 460 
I don’t think so. I think it will be the actual renderer with many things fixed. With an unbiased renderer I think it is not possible to extract the layers (reflection, shadows, SSS, refractions,…). The actual renderer had many problems where instead maths calculating things there were tricks simulating the calculations. So these bad tricks will be fixed with real calculations but without extreme things that an unbiased renderer does (using espectral light where light is defined by a group of wavelengths). Brecht says in the text of the post that it will sit between actual blender renderer and a physical unbiased renderer. And I find it is fine, layers are needed, you don’t have refection layers or SSS layers to play with in renderers like luxrender.
I’m really impressed and excited! This video is just… WOW!
I usually see this kind of performance somewhere else, related to some other software, and usually think that although blender is cool, it is also still far from certain kind of applications. But this demo totally blew my mind!
I hope and believe openCL implementation is a target for cycles, since it is not in blender philosophy to restrict users to specific hardware. I guess Brecht is just more familiar wit cuda so fa because of his octane experience. Afterall it says “no openCL yet”.
Now, a consideration: Brecht says
I do think there’s still a place for a “small studio” production render engine, that sits between a render engine with a focus on complete realism, and a totally programmable and customizable Renderman style engine
so, what kind of quality should we expect from Cycles? Being half-way between let’s say, Maxwell and 3delight, what exactly means? Shouldn’t we get photorealistic renders? Or is he just talking about performance?
this is incredible!
Totally unexpected and amazing, i see the blender era… behold!
Nice.
I think this will kick out many many funky old features, like world settings, halos, wire material etc … , at least at the beginning …
As Brecht wanted this summer before he gets to octane team …
New branch ? 2.6 branch ?
Bright future for blender users !
delic: Wireframe would be much nicer as a node. Halos? Good riddance. The world is better without them. 
Not wanting to put words in Brecht’s mouth but I think he’s concerned with the kind of control an artist has over the render. It’s about allowing them to produce work that isn’t necessarily 100% physically correct. It gives them the freedom to embellish reality or depart from it.
Anyway talking about physically based, have you seen this?
It will be interesting to see how this all plays out.
Hi,
OpenCL is not supported yet due to technical issues, there’s an incomplete OpenCL backend, but there are some non-trivial things to solve before it will work. Don’t have a time estimate for this, but obviously it’s an important issue to look into.
Render passes are definitely planned, and they can be combined with unbiased rendering. But unbiased versus biased is not really a good distinction I think, there’s many different aspects to rendering that can vary independently.
Regarding the name Cycles, you can interpret it in different ways, pick whichever you like:
- Reference to render cycles / iterations.
- Reference to the word Blender.
- Perhaps as a Belgian, I’m a bit of a cycling fan?
The first is the most important one, the renderer keeps going as you change materials, lights and objects, it doesn’t rebuild the scene from scratch each time. This is crucial to the design. More work is needed to get this working smooth for production scenes, but that’s where it’s going to matter.
From working on open movies, I know pressing F12 and waiting minutes for the render to even start is a nightmare, so it’s one of the things I’m trying to address.
Brecht, this is the second greatest Easter present ever (of course, life eternal being the first )
Great to have you back, and this is truly amazing development that will (I believe) bring an amazing boost to Blender. Thanks!
Brecht you are a genius, your render is incredible fast even in CPU only mode!!
Thanks man, really thanks for this wonderful surprise, i am sure you really excited the whole blender community!
Yes, this makes sense. I remember reading from Brecht about the design of the shading refactor, and it sounded like different shaders for different tasks, physically based or tricky.
Nice to hear words directly from you, Brecht. My own personal thumbs up, and best wishes.
You know, you’re going to have a lot of questions here 
…My humble one: how long were you working on Cycles? When was it born?
Brecht you are the man… Seriously!
When I watched the video my jaw dropped a considerable amount and I found myself in a position that I didn’t even know was possible. This made my day, my week, my month and maybe even my year.
I don’t really understand why it is so much easier to write a whole new render engine from scratch rather than adding onto the existing code, doesn’t it take much more time to code everything again?
Also (sorry if this has been asked before or if I’m just really thick) what language(es) do you use to code this engine?
Wow - this looks like Octane!!! If this is free…I am now sitting in my chair again after dancing around my attic for a few minutes grinning.
This. Is. Awesome!! WowowoW! and welcome back Brecht 
yet another reason to finally upgrade my machine.
One question (hope) tho, and I’ll prolly get slapped for this, but is the “old school” material/texture system going to remain an option, or will I finally have to hunker down and learn to properly use physically based rendering? --hoping the former-- 
I’m not Brecht, but people who worked with the shading code in Blender Internal have reported that the code-base there is not only a mess, but also full of hacks and workarounds, if you read on code.blender.org you’ll find that Brecht found it quite difficult to refactor the existing code so that BI could shine as a good and flexible GI capable engine.
Large parts of the code would’ve needed to have been rewritten anyway, it’s at the point where creating a new engine won’t have a major time difference.
uhm those who claim Gi and interactive rendering is for noobs need to look for a new CG software.
I am excited to see this, Hypershot, VRay RT, Thea, all those engines use interactive renderings
and are very fast without the GPU already.
Finally Blender has a chance to catch up but I ask myself how long this will take.
Good engines take some years to really fine tune.