Maxwell Render | Official public forum

At this moment, We wish fast, faster and the fastest rendering time only!

I use render farms and they really are no that expensive to use.

fast, faster and the fastest rendering I wish
fast, faster and the fastest computer I need
much, more and the most money I hope...

sandykoufax wrote:fast, faster and the fastest rendering I wish
fast, faster and the fastest computer I need
much, more and the most money I hope...

+1

GPU rendering, PLEASE

+1e100000
speed is the most important aspect now, when maxwell 2.7 is so full of features.
it's not logical to use renderfarms for every render, like small interiors, etc.

Quality must be for small projects too... for every day's work... We don't pay renderfarms for every day's tests...
So, since we are Maxwell users and we ONLY have quality, it has to be fast...

The Octane solution is also good... A really nice biased renderer for quick previews with great results and GPU speed for fast unbiased rendering...

Fire and draft rendering in maxwell is great but can't compare to GPU or biased solutions especially for night scenes or scenes with many lights

I don't want to compare the two products but since I use both... and NL has made clear that there will never be a biased Maxwell, then speed is a must have...

Wait for quality...

Or get shit now.



But yes, I wish for GPU support too.

I'm a heavy Octane Render user and also used Maxwell a longer time.
I do general 3D and no specific like products or only archviz.

One thing I learned is that all this "unbiased or nothing" rendering is more a philosophy without a practical background.
The regular client (in my experience) doesn't care if a caustic in a water pool is 100% physical correct. When I do some indutrial viz for clients they want the material look right or even tell my to tone down stuff even if I used a unbiased mode to render. So I end up with an in practice biased render and client tells me "it looks right now".

Once a client told me he wants photo-realistic, I fired up a unbiased mode and he told me it is too realistic and dont fit to the work a former 3D supplier made.
I also do regular jobs for a clients construction online-shop. We don't have the aim to be 100% realistic (like only 5 different cars for over 50+ products) but even his common every-day customers think the works are photos so we ended up adding more technical details and he gets less requests about parts which are missing on the "photos" on his web-shop.

My bottom line is that we surely can make the 3D world better with unbiased rendering but at the end of the day its the client who tells what he likes and how fast the work is done to his visual pleasure. Not our happyness about what a nice unbiased render we did.

For us we need speed and best would be GPU plus CPU support like iray or Arion. But I can also get why Nextlimit hesitates to introduce it. As I can see on Octane Development the tech Nvidia introduces is pretty unpredictable. The GTX600er series was introduced and all GPU renderer need to be adopted again consuming lots of deve-time again. Even iRay didnt work for months with current graphic cards. If NV feels to change something with a driver or cut CUDA rendering for the Kepler and shift this power to expensive Pro-Cards only the GPU Render Devs have to redo or rethink again.
For me as an end-user it also hard to make investment decisions. I render here with a bunch of 3GB GTX580 cards since 2010/2011 now and can't/don't want to upgrade because the current series is slower on CUDA. If I see the CPU market my 980X still does a great and fast job and I don't have to be afraid if after a new socket upgrade my rendering software still runs or not.

Michael Betke wrote:One thing I learned is that all this "unbiased or nothing" rendering is more a philosophy without a practical background.
The regular client (in my experience) doesn't care if a caustic in a water pool is 100% physical correct. When I do some indutrial viz for clients they want the material look right or even tell my to tone down stuff even if I used a unbiased mode to render. So I end up with an in practice biased render and client tells me "it looks right now".

yes, that's the point. It would be better to have a working and maybe partially biased solution (maybe with a big warning sign, that you're leaving the unbiased trail now at your own risk...) than having a non working unbiased solution...
And btw. the bad workarounds for these problems are baised too.
If i remember correctly, long time ago there have been announcements for an improved unbiased solution to fix these problems, called RS2 at this time - hopefully we'll see it in action at some time in the not too distant future!

I think that anyway even un-biased engines must be biased somewhere inside as the user can hide stuff from GI etc...
but even than I think maxwell needs to be either faster or combine some sort of noise-reducer INSIDE the render engine thus
avoiding the need to use 3rd party plug-ins that have no camera profile for CG renderer.

That is not bias. If the engine decided on its own that it would hide something to GI because it figured it was unimportant -- that would be bias. Simulating situations inaccurately is very different than allowing you to accurately simulate situations which do not physically exist.

noise-reducer INSIDE the render engine

It does depend on what sort of noise though, theres is very sharp specular from distant tiny lights etc, perhaps thats a candidate, esp with high DOF... But then again that can be dealt with in post, so no real need. And conceptually I dont think it can happen during rendertime.

But with bright materials, especially transparent materials, the noise is actually saying something too, the colours indicate unrealistic excesses of that colour of light,.... in some situations controlling the noise will guide you to very realistic settings, a form of reverse engineering. So the noise is essential really.. Maybe this goes without saying.

and finally I don't even like totally noise free renders, looks fake..

I disagree.
In real world you don't have hidden emitters and AGS...
So maxwell cannot be totally accurate anyway !