All posts related to V2
User avatar
By Half Life
#338650
My policy is to keep the saturation less than 225 for any color, but the problem is that Yellows (in particular) only look "Yellow" at high value and high saturation... Blues for example rarely have that problem because Blue looks "Blue" at low value and low saturation.

It's important to separate the psychological identity of a color from the spectral reality of the color in the making of Maxwell materials -- "white" being a perfect example... we think of 255 as being "white" but in reality 225 is white in Maxwell, anything higher is a noise-maker.

If you want no "flake" then you are better off just using a basic "diffuse" material with an additive layer or coating with Nd of 1.491.

Best,
Jason.
User avatar
By Tok_Tok
#338654
This confuses me a bit; in the tutorials of Mike Verta 242 was the upper limit for white (and i suppose all colors). Now you are saying that already 225 is the limit. When i use 225 as white, it comes very close to being grey instead of white. This is with most colors the case, 225 is perfect for not getting too much saturation but it looks a bit dark. And with 242 the colors keep their character but tend to get easely blown out by bright light.

I can't really make a question out of this, i'm trying to understand the subject :)
User avatar
By Half Life
#338657
The easiest way I can explain it is that even an emitter (in the output file or in Photoshop) can not exceed Value 255 (for our purposes) -- so how can we judge an emitter is "bright" or emitting light if we put it next to an object that has the same value but is not emitting light?

For the simple purpose of contrast a "white" object must be of a lower value than a "light" emitting object -- that is just a practical POV.

But the reality is much more technical -- if you take a "white" object and test it's tristimulus values you will find that at the very brightest it may go to value about 237-244 but that is a blindingly white surface... most white surfaces will be closer to 225 (eg: walls, paper, etc) for the practical purpose of not being hard to look at.

So too in MXED -- it is possible for a real world material to exceed 225, certainly -- there are many optical brighteners and fluorescing colors that seem to glow when exposed to ultraviolet light but they exceed Maxwell's current parameters (AFAIK no UV light).

For instance here is a measured value of the whitest white in the RAL line (9003) in several color spaces: http://www.perbang.dk/rgb/EBEEEC/

The bottom line for me is: it is easy to adjust contrast/saturation in post -- and hard to get a good render with high values... so why not just use lower values and just adjust in post.

BTW - I miss Mike, he has not posted here for a long time...

Best,
Jason.
By numerobis
#338660
these are the values NL mentioned in the manual:
Important: Avoid setting the Reflectance 0° color too bright. Setting it to 255 for example
means this material will reflect back almost all the light it receives, which does not happen
in the real world. Maxwell Render still keeps the amount of light reflected/ absorbed within
physical limits but the result with such high Reflectance values means the light will keep
bouncing around in your scene with very little loss in energy, which will produce noisier
renders and a washed-out look with very little contrast. A reasonable setting even for very
bright materials would be around 230-240. For example, the Reflectance value of a white
piece of paper, when converted to RGB, is around 225. Please note that this only matters for
the Reflectance 0° color. The Reflectance 90° color can be left at fully white (RGB 255).


thanks for the link Jason! very useful :D
User avatar
By fuso
#338690
Hi guys,

Just a small question, slightly out of this context which I have been wondering about for quite some time now. I have a complex
scene which I have to render at daytime, at nighttime and (unfortunately) with both mixed. I have massive problems with the
noise to clear up when using that combination of the physical sky plus emitters 'switched on'. I know this has always been an
issue but I have to do it this way (post-production is not an option in this case).

So my question is: Do I generally still need a BSDF layer in an emitter material for any of the situations mentioned above? I'm
just trying to optimise that scene as much as I can and I already went through checking if emitter geometry is as low-poly as
possible, if it isn't touching other geometry and if materials have too high RGB values. Also I used AGS glass wherever I can.

Oh, one more thing regarding high RGB values, does it only matter in the reflectance 0 slot or do I have to check all slots in all
layers, even when using a simple two layered plastic with a base and a specular BSDF? And how about transparencies??

Thanks in advance for any help.

Jost
User avatar
By Half Life
#338696
You should use 225 as a guideline for both Reflectance 0 and Scattering color (if using SSS) -- this applies to any maps occupying those slots as well.

Transmittance should be about 244 for typical of glass-type transparency.

The noise is the result of unevenly powered emitters -- the sun being much more powerful than the emitter materials. The solution is to raise the power of the emitter materials before render and then reduce them back to the desired levels with multilight... the more powerful the emitter the higher its priority in the render.

There is no need to have a BSDF with any emitter material unless you plan to see it with the "light turned off".

It should be said that real dielectrics (glass) anywhere in your scene may be noisy until about sample level 35-36... in glass heavy scenarios it may be advisable to use a render farm service.

Also worth saying: don't surround your emitters with perfectly smooth metals (reflectors) and don't put them behind glass.

Best,
Jason.
User avatar
By fuso
#338698
Hi Jason,

Thanks a lot for your input and for clarifying this once more. I'm glad to see that it's the render engine and not me as most of
the 'rules' apply to my initial set-up in that particular scene. But quite honestly, one starts to ask himself why to continue using
Maxwell, with so many 'rules', which really are more restrictions than anything else.

We are using a reasonable render farm here in my office (consists of 9 PC's, 4 are 16 GB RAM dual quadcores, and the rest are
still reasonable fast 8GB RAM quadcores all running on Windows 7). I consider that serious hardware to throw at a single still
rendering of 2400x1600 hopefully to reach SL 20. The scene is not using high-res textures (only a few anyway), the emitter
lighting is pretty much balanced as I can see in the night view, the geometry is not particularly heavy, there are no dielectrics
or SSS materials in the scene, etc, etc. This just can't be right!

If I turn on 'multilight', me and my colleagues (nodes) can just go on holiday for a couple of days so I don't even dare to try.
Anyway, thanks again and btw, I really appreciate your incredible efforts and your passionate involvement in this forum and
beyond! I have been with Maxwell render since early alpha days, but I am staring to run out of patience, so does my employer.

All the best,

Jost
User avatar
By Half Life
#338706
Let me reiterate -- the sun is much more powerful than the emitters... therefore it is stealing rendering engine priority from the emitters (no matter how balanced they may be with each other).

If you raise the emitters power (into balance with the sun) before render time then the optimization will be more biased towards clearing everything at the same time... at which point you can scale back to realistic settings via multilight.

I can render similar sized scenes on a single i7-920 machine with 3GB of ram overnight using multilight... on a proper render farm it may only take 1-2 hours.

Also with Multilight you will only need to make one render which you can adjust to all output variations you need (and then some).

I can appreciate your frustration with the network rendering system and judging by the release notes of each new version it is receiving a fair amount of attention from the dev team (bear in mind it's not a huge staff)... I would really suggest trying a proper render farm though.

Computer performance is rarely just about the guts -- it has quite a bit to do with how it is maintained and used... my machines are immaculately maintained and thus I can get the best out of them, I would expect the same performance from a professional render farm service. Office networks are far less predictable due to things like software background-process bloat, internet connection, virus scanner, etc... My main work machine has none of those things -- I daily kill the bloat, defrag, keep it off the net, don't use a virus scanner, etc.

I was talking to a local friend the other day (he's a Maxwell user too) and I had to remind him that Maxwell is only at V2.5 -- We had both started using Photoshop at V3 which was the first version with layers, such a fundamental thing for image editing now but back then it was revolutionary... there was no 16-bit, a crude brush engine, few filters and forget about 3D and the internet. Now we are on Photoshop 12 and while the core of the V3 features are still there, the program is all but unrecognizable from that humble V3 app.

Maxwell hasn't even gotten to that V3 yet -- we can't imagine what kind of awesome software this will evolve into over time.

3D is really filled with limitations on all sides -- I am convinced it is not yet a medium fit for artists to create in as it really is still in it's infancy and requires users to work around so many limitations and be so technical in their approach... but new developments in software and hardware happen at an amazing rate and I am positive that Next Limit will be responsible for some major developments yet to come in the 3D world.

Best,
Jason.
By brodie_geers
#338708
This may be turning into a discussion for another thread, but since the original question is pretty much taken care of, I'll throw in my two cents as well regarding fuso's issue.

Half-Life's setup sounds pretty awesome, but as he hints at is a bit idealistic for a typical work environment, which most of us are using and to which NL is writing the software for. What I hear fuso saying is that he's got a lot of render power (and that is a lot by any standard) and that he's running a relatively low resolution (in my opinion) and relatively simple scene which is taking a long time to render (if I read you right, if you turn on ML and render for 2 days it wouldn't reach SL 20).

There does strike me as being something wrong there. As a matter of comparison, I just ran an exterior scene at 5000x2500 px with about 8 high poly trees 40ish high poly bushes and about 6 high poly cars with physical sky and about 6 emitters (cranked up as Half-Life recommends) and after about 2 days it was just under SL19 as I recall (which was plenty for my needs). This was with ML on as well as MatID, Object ID, and Shadow channels on. And that was using 2 machines (a workstation with a 980x and a dual core xeon processor that runs at about the same speed).

It's hard to say why, with seemingly more processing power and a much lower resolution you're render would be so slow. We'd probably need to see the scene to know for sure, but based on what I'm hearing, I'd have to agree that something is wrong with that scenario.

-Brodie
User avatar
By fuso
#338709
All I was trying to emphasise is that Maxwell has already been incredible even with things like SSS in it's beta stage! To me this
current version is a reversed development, stability is being sacrificed for fancy new features in a rush and without thoroughly
testing them to bits before releasing them to unsatisfied customers, including myself. And as I've said many times here, I have
been passionate about this great piece of software since many years now, and part of me still is.

I just think calling it 2.5 is brave, where others are bullet-proof at version 1.0 and improve from there. Here it's a bit like a lucky
draw - what won't be working next time? But it normally goes under between all the oooh's and aaah's about the new features.
You have to admit that more vioces are getting louder now, many people have huge problems with their deadlines and stress
levels. It's just like being cheated on really. It happened so many times that I had to roll back to an earlier, more stable version
in order to get the problem fixed. That involves an incredible amount of working hours where clients are not happy to pay for!
With all those work-arounds I'm just not sure if even the lower prize tag is justified...
User avatar
By Half Life
#338711
I've not seen too many bullet proof softwares at any version -- Photoshop still has horrible bugs at version 12 and causes some users to stress over deadlines and whatnot as well :wink:

It's all a matter of perspective -- many people want to gripe about windows and it's various bugs as well... any particular hardware vendor can also be the subject of griping (for instance I hate Dell and HP).

The reality of it is there is always room for improvement and nothing is perfect ever -- testing is no guarantee because there are so many hardware/software variables at play and so many different workflows used that things can slip through easily.

The only real solution would be for Next Limit to drop all those free plugins (15 and counting) and focus on putting all that time and energy into improving the core functions (MXED, Studio, Core Renderer, Network)... which I would personally be OK with but I think would likely be suicide for them financially.

Best,
Jason.
Help with swimming pool water

I think you posted a while back that its best to u[…]

Sketchup 2026 Released

Considering how long a version for Sketchup 2025 t[…]

Greetings, One of my users with Sketchup 2025 (25[…]

Maxwell Rhino 5.2.6.8 plugin with macOS Tahoe 26

Good morning everyone, I’d like to know if t[…]