[rogersalph]

Hi Guys,
I am new to Maxwell and I usually model in either Rhinoceros 5 or Sketchup Pro. Our company have recently invested in renderings and I am still pretty new at this (as I am the supposed expert in this, there is no where I could go other than the forum). So please excuse me for all the questions I have developed.

1) Could you guys please kindly provide me with tips which can help me remove noise from my renders? As you can see below, I have the rendering going for 72 hours and with the lights penetrating into the interiors, some of the noise (visible on the dining chair and ceiling) seems unavoidable. As our design constantly change on a day to day basis, I need to minimize rendering time as much as possible to enable the design changes to take place during working hours. How much is the optimum time/sl for each rendering? Ex: Interior? Exterior?
2) Also, the shag pile (carpet) which I have can't seem to be applied to a white material as it would create more noise.
3) The third thing which I want to ask is the ways in which you should be balancing between your environment settings (ex: background, reflection, refraction and illumination). How do you set so that the skies are reflected on the glass of the interiors, with the northern light creating a shimmer in the space. I manage to do a bit of that in the first image, but the intent was to flood the whole space with that effect. Do we fill up all 4 slots and are there any rule of thumb as to which hdr images should fall into each category? The reason is as shown in the first image that the sky appeared to be burnt out (white) whereas it appeared fine in the second rendering.


I have read through the forums rigorously and these are the things which I have summarized.

Issues Pertaining to Noise:
1) Recheck materials such as glass etc which might cause caustics, taking longer time to render etc.
2) Make sure the value of camera and lighting is according to real life - question: how do you define the real values in real life? is there a chart or rule of thumb I am unaware of? The lighting in the scene is set at 65w fluorescent tube at the moment (for all three renders)
3) Simplify light polygons as much as possible.

Issues Pertaining to Cameras:
1) Set camera at iso 200 and shutter speed at 1/800 etc. When doing so, how do you determine ozone, sun power, etc?


http://minorityexpress.wordpress.com/20 ... attempt-1/


http://www.rogersarchitecture.com/exten ... house#/i/1


http://www.rogersarchitecture.com/exten ... house#/i/2


Any further tips on setting materials, lighting, tutorials would be great to know! I welcome any critics and comments so that I can progress in achieving the realism in our architectural design. Also, feel free to answer any of the questions as you see fit! Thanks!

[gadzooks]

They look like a great start to me. Here is a helpful Topic i started which will lead you in the right direction (Interiors Wise).
Gadzooks

http://www.maxwellrender.com/forum/view ... 15&t=36779

[rogersalph]

Hi Gaz,
Thanks for the reply. I am going through the check lists which you have provided. Any idea where I can learn how to use sky dome? I have always resorted to the physical sky function and often have vignetting appearing in my renders. It would be great if you could provide me a link to that. Thanks!

[JDHill]

Vignetting will appear in any Maxwell Render, depending on the camera setup (increasing with lens length), and the Devignetting parameter found in Scene Manager > Output > Simulens.

As far as Sky Dome goes, in the SketchUp plugin, it depends on whether the Sky Dome's Color parameter is set to Application or Custom. In the first case, it is actually Image Based environment that is used, where a pure white HDR image is used for illumination, and the Sky Dome gradient colors are set to mimic the gradient of SketchUp's background. When set to Custom, the standard Maxwell Sky Dome is used, where you directly specify the gradient colors.

Regarding your original questions:

1. While there is no optimum time/SL, in this case, I would focus on optimizing materials for the scenario, as the noise you see here is actually very improbable light paths beginning to be resolved at higher SLs. Some important factors here would be: making use of AGS where appropriate, and making sure that materials are not violating conservation of energy through the use of additive-mode layers and over-bright colors and textures.
2. I am not clear on what you are asking, regarding the shag carpet. Is this a displacement material? Regardless, a main thing to check would be that RGB values (color and/or texture) are held to 220-230 maximum values.
3. This is dependent upon the quality of the HDRIs used. Poor HDRIs can result in unbalanced results, since you will try to boost the levels to compensate for the fact that an HDRI does not actually contain enough information (contrast). It's difficult to say much more than that without inspecting the actual scene and HDRIs.

Regarding your second set of comments/questions:

1. Definitely, the main things to check are the same ones mentioned in point 1 above.
2. The most important thing is to model at realistic scale. Given that, define emitters according to their real-life values, and set the camera up as you would in real life as well. This can be as much trial and error as it would be with a real camera; however, there is a chart at the end of the plugin manual, which gives ballpark EV values for various lighting situations.
3. Yes, it is important to keep emitter geometry as simple as possible.

And regarding the final question about cameras:

1. Depending on the fStop, these values (EV=14 @ fStop=5.6) would be suitable for a bright outdoor scene. Regarding atmosphere settings, for realism, you should not alter the Sun Power or Atmosphere Intensity values, as those are set up to simulate reality at 1.0. While the other atmosphere parameters will also affect the overall brightness of the atmosphere, they should be set according to the type of sky you are simulating, with exposure being compensated for using the camera, as you would do in real life.

Many of these points are reiterating those in gadzooks' post, though.

[rogersalph]

Hi JD,
Thanks for the prompt reply. The comment on sky dome have answered most of my doubt about the difference between sky dome and constant dome. They are actually referring to the same thing. Constant dome is displayed in maxwell studio and sky dome is displayed in the sketch up plugin for maxwell. It's a little confusing when Vray's sky dome is actually referring to sky hemisphere mapped with a sky image, JPEG, HDR, etc.

In regards to the shag carpet, I have used a maxwell grass plugin to create random blades for the carpet. I have used the grass material from the link provided below:

http://support.nextlimit.com/display/kn ... s+Examples

I then use photoshop to convert the image to grayscale and white. One image for the Reflectance 0, and the other image for the Reflectance 90 slot. Each time I try to tone the carpet to white rather than gray (as appeared on the image), the noise would take a much longer time to clear/it stays there even after SL 18.

I have often used HDRI's from Ronan Beckerman's 10 free HDRI's :
http://www.ronenbekerman.com/10-free-hd ... kies-maps/

I am not sure about how you analyze good vs bad HDRI's to obtain good reflection, and mood of the environment. It would be great if you could show me some samples or how to differentiate them.

I find the chart of the EV values really useful as I can determine different camera values as a rule of thumb instead of changing the sun's power as it will sometime burn the image out too much.

[JDHill]

Yes, the name of Sky Dome was switched to Constant Dome at some point in time, and I have not yet changed it in the plugin. I have a whole internal dialogue where I argue with myself about whether it is more confusing to change the names of things, or to have them not matching anymore, when they have been changed elsewhere. Obviously, the answer is different, depending on whether the user is a new one, or an old one.

Regarding the shag carpet, I understand better now. The main thing to keep in mind here is the same, though: using unrealistically-high reflectance 0° values (i.e. whites > 220-230) is not a good idea, since those things do not generally occur in nature. When you tell Maxwell "white", you are not just saying "white" in the sense of the color white; you are also simultaneously describing how much energy returns from a surface with the color applied. If you give it a true 255 white, you are saying, "this surface reflects with perfect efficiency." That means that energy is not lost when light is reflected, and the result of this is that more energy continues bouncing around in the space. The same goes for any individual component of an RGB color, as well, such that a pure red reflectance 0° color does not describe a realistic material.

Secondly, there are two grass materials linked on that page; the first uses SSS, and may therefore render substantially slower than the other. I would either start with the non-SSS material, or create a material from scratch, which does not use textures at all, since you do not likely need such detail for creating a carpet using grass. Lastly, check the way you are generating the grass; it will be faster to render the Flat blade type, rather than the Curve or Cylinder types.

Regarding good or bad HDRIs, I can't really give you any examples. The difference between the good and the bad here generally comes down to the image's ability to cast sharp shadows (as discussed, for example, here and here), and given what you're written in the meantime, this may not be much of a factor in your particular situation. If you have to use different images for different Image Based channels, you may need to balance their outputs using the Intensity values in each channel. If not, you can use the same values for all channels, by setting the Background channel to use your HDRI, and then setting (or leaving, since it is the default) the other channels to use the Background channel for their values. You may want to ask more about this in the main Maxwell forum.

[rogersalph]

Hi Guys,
I am attempting to render one of the most complicated scene (by my standards) up to date. This is the link:
http://minorityexpress.wordpress.com/20 ... n-house-2/

I am having difficulties getting rid of the noise even though I have followed the do's and dont's provided by gadzooks.


Interior Lighting do's and dont's

1. Do not use glass in windows unless you have to have reflections. And if you need reflections it's best to use the AGS material.
2. Hear are the three best methods for lighting an interior scene.
a. place emitter planes in the windows.
b. Use skydome instead of Physical sky as skydome will clear up much quicker.
c. Use IBL (Image Based Lighting). But make sure it's a good HDR image with enough lighting information.
3. DO NOT EXCEED 240,240,240 RGB IN REFLECTANCE O IN ANY MATERIALS.
4. Turn off multilight function to save ram
5. Group emitters or combine emitters to limit the amount of emiiters.
6. Do not enclose emitters inside of a glass material.
7. Displacement causes more noise avoid it unless you have to have it.
8. Use fill lights around your scene (Behind the camera) to help in reducing noise and render times.
9. Make emitters as low poly as possible. A simple plane with one surface is much more efficient than a sphere.
10. Never have emitters pointing into metal (shiny) cones that produce caustics.
11. Avoid using coating in materials.
12. Always try and model to a scale. Not having a scale to your scene will produce problems when you try and light a scene with emitters. A interior room that is 10' x 10' foot that is lit with a 100w emitter will give predictable results.

Please add or update or correct this as i thinks it's very useful for everyone to have this especially people who are new to maxwell.


As our workflow is to model in sketchup and render in maxwell, we are under tight time constrains and I can't have the rendering go for more than 24 hours. Are there any additional tips that I can get to help with this issue?

The other thing I have in mind is that would you guys consider modelling only the bits you see and exporting it to Maxwell to render? Or would you export everything and render different scenes one after another? It might take some time to plug in the materials for each scene (repetitive process) which saves 30 minutes per image. I guess if the answer is yes then, it's justified to spend the additional 30 minutes to set up for each rendering - to save rendering time.


Any help/suggestions would be appreciated. Thanks!