Maxwell Render | Official public forum

For anyone that's been following this post I've decided to consoldate any updates and tests into this first post in this thread. I think this will make the organization and hopefully readability a little easier. I'll quote snippets from any replies or helpful informational tips in here too.

I've renamed this thread from WIP to SOAR because I'm taking the initiative and coining the acronym Story Of A Render. Why? I'm personally a little slow on the uptake (mild learning disability) and the way I've worked my way around it by using the learning by teaching, teaching by learning method. Somehow it just sinks in.

Hopefully, these will work as somewhat of a tutorial series or something like that for NOOBs (which I'll happiliy admit I am somewhat of a NOOB). So if you're a NOOB, don't like actively participating in forums, prefer to read and not to type, are afraid to post or whatever, PM me with what's buggin you and I'll post it as SOAR and take the brunt if they annoying, stupid, mundane question (most of them aren't by the way - remember ignorance is not knowing, stupidity is not caring to learn).

Consider this my attempt at a contribution to the maxwell society.


SOAR - Laboratory : Light Propagation Study

Here's the scene rendered in VRAY (Yeah still making my paces through learning that one too).

ORIGINAL

I took the file into M~RStudio and tried to make a maxwellrender out of it and this it what I got.

M~R Test-Render-1

which looks pretty bad. My lights have intensity checked (at 7200) and units are W/sr. Comes out really dark. I'm pretty sure it has something to do with the light setup because playing with iso and such just washes everything out.

Should I be using different units for studio lights? Oh it's a 6hr render (I forget SL).

Response - regarding units::

Kabe wrote:

lsega77 wrote:My lights have intensity checked (at 7200) and units are W/sr. Comes out really dark.

Luis, light units are broken in RC5, it's a known bug. Either crank up the
light a *lot* or wait for RC6 or however they'll call it.

Kabe

Response - regarding light setup::

Hugh wrote:Two thoughts:

1. I think your light fittings are too far from the walls, especially the window side.

2. I know its a no-brainer but you have got caustics turned on, haven't you.

In general this kind of lighting will always give you a dark ceiling.

With black floors and black work surfaces, seriously, the only way light is getting back to that ceiling is if you fill the room with bald white people.

Well, three thoughts then.

Response - Regarding Scene scale and testing setup

mverta wrote:Yes, first of all your scene scale is off or your camera settings unrealistic. Make sure your scene scale is 1:1, and tell me your camera focal length.

Second, I'd assign a white lambert to the whole thing and just debug the lighting using an ISO of 400, with f-stop 4.0 and ss around 125, but might have to be much lower.

Start there and I'll guide you through the rest.

_Mike

:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

M~R Test-Render-2

Note: first attempt at this render was overbright.I had to drop the ISO to 32 and raised the SS to 900.1 (which are absolutely ridiculous numbers) to have anything visible.

Response - Settings Adjustment::

JDHill wrote:Sorry lsega77, I have been quite busy today, and haven't had a chance yet to inspect your scene, but try changing:

F-Stop: 8
ISO: 100
SS: 125

...and go from there...I'll take a look as soon as I can.

~JD

mverta wrote:Also, do me a favor and drop the roughness from Lambertian to 99.

_Mike

Response - What to post for testing when overbright, adjusting test render for post, materials adjustment

mverta wrote:Ok, couple of things:

You don't have to publish blown-out renders. If your scene is rendering bright, at least for the time being, adjust the ISO in MXCL so the exposure is at least correct. We're trying to judge the relative light propogation right now, not absolute exposure. If you're watching the render and you can see it's too bright, then I'd suggest increasing the shutterspeed until it's just shy of being overexposed.

Also, what do your emitters look like, geometry-wise? I'd also recommend pulling the light fixture diffusers out for a few tests. We just want to see straight emitters and shading.

This looks like one of those scenes that will be better to test with a more real-world reflectance like 200,200,200 instead of pure white, which basically doesn't exist. Sometimes that's more useful, but it doesn't appear so in this case.

I also would like to know if your scene is completely enclosed, i.e. has all 4 walls.

You don' have to let this render for 5 hours to check, I bet you can get a decent idea in an hour, or perhaps reduce the size just for the tests.

Thanks,

_Mike

INFORMATIONAL TIP - Regarding light setup in practice

Hugh wrote:

lsega77 wrote:
With this image I don't understand why it seems that the edges are darker and the middle so overbright. I know my lights are somewhat away from the wall but I don't think that light attenuate that drastically along the sides (if that makes sense).

The illuminance in the centre of the room has a contribution from all the lights in the room (depending on the cut of angle of the louvre), so will always be brighter. The edge have far fewer luminaires contributing to the light level at any one point so to compensate, the space between the wall and the centreline of the first fitting should be no more than half the general centre to centre spacing across the room. In this case the effect is amplified by the grouping of luminaires toward the centre of the room.

The way the vray render was set up (invisible area light below louvre) will give you a light distribution more akin to a light fitting with an opal diffuser, a very wide distribution. Which is why the light cutoff is so high on the walls. A specular (read mirror finish) louvre, in general has a much narrower distribtion and will give you something much more like the maxwell images in reality.

What I'm saying is that the maxwell light distribution is more realistic given the geometry of the light. If you want the lighting to look like the vray one, you can do one of the following:

1. fake it with invisible area lights (seen that you can do this somewhere if you hide emmiters in studio)

2. Change the lighting layout (closer to walls, more evenly spaced)

3. Change the light fitting to something with a much wider distribution. Which would be fine IMHO unless the space has lots of VDU's in it.

:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::



M~R Test Render-3

ok, so I adjusted the material to 200,200,200 and left the emitter intensities at 7200. F-stop, ss, and iso still the same (8,125,100 respectively).

I think this turned out a little better and also shows me that I need to add some lights to iluminate the forground a bit (maybe some recessed cans on the back wall to brighten this area up too).

So as you can see the emitters are just simple planes, with no segments. They're raised up a bit from the face of the ceiling because they were hiding inside of the housing for the parabolic fixtures (that I removed for testing).

:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

M~R Test Render 4

I've rearranged the light fixture location to be closer to the windows and walls. I think this light setup works better.

Response- Regarding darkened edges::vignetting

Kabe wrote:Hughs answers sounds good, and it might contain some truth, but in such a
scene with so much scattering it could never lead to such darkening.

The darkening you have in your image is natural vignetting, as jpd spotted correctly.
I have summed it up here: http://www.maxwellrender.com/forum/viewtopic.php?t=2923

You can easily check that by rotating your camera.

Hope it helps

Kabe

Response - Regarding amount of light needed in real photography, testing suggestion

acquiesse wrote:Hi,

This is a really nice story of the progression, I like the format....

It seems to me the problem is that people are doing what the think are "reasonable" settings, to get the results that "look right"...

I know from my experience using a manual SLR that it can be suprising how much light you need, the experience of our eyes is not the same as the camera.

A photo looking directly out of a window will have a huge amount of contrast compared to one perpendicular to a window, so you will get 2 very different photos from the same lighting conditions.

I was wondering if the next step would be to take a photo, noting all the settings of ISO, shutter etc. and compare this to your render. I don't have access to a room using the sorts of luminaires in your photo, I presume from your work you may do?

This would give a better reference and would probably help renders in the future...

I expect people have done similar things in the past, but I haven't seen any...

Resonse:: Reference Photo provided by Ricardo (Thanks Ricardo!!!)

ricardo wrote:I had the camara on my lap while reading this, so took a reference pic:

Nikon D70s, ISO 400, f/4.5, 1/60s:



It's obvious that the camera missed the point on white balance and I could use some longer exposure, but no tripods at hand...

The kitchen is 16 sq meters, all white ceiling and walls, grey floor , with four of these tubes:



Ricardo

Response - Follow up on vignetting issue and M~R Informational tip.

Kabe wrote:

lsega77 wrote:So If I'm understanding your explanation in your post on vignetting then my camera angle reletive to the angle of my emitters is causing this vignetting effect?

No, it's even simpler: It's just the image angle at any given image point.

The more you are away from the image center, the darker the image will
be. Moving or rotating cam or emitteres won't change a thing in this regard.

Would it be safe to say that since my camera's direction of view is parrallel to the emitters orientation vignetting occurs. If I adjust my camera angle to be other than parrallel to the emitters this would compensate for the effect?

The only way to compensate would be to weight the energy based on the image angle, either by adjustments during sampling or during the image conversion.

Currently there's not much you can do beside using a tele lens.

Yes, it blows, I know, I tried to make cubis panoramas using M~R...

Kabe

::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

SOAR - Laboratory : Light Propagation / Materials Study

So here's a textured version with the most current lighting scheme.

Film ISO: 200
Shutter Speed: 125
F-Stop: 8


M~RTest-Render-5(Material-Test1)

Obviously there's a little material tweaking that I need to do (ie. ceiling, wood-should be a little more glossy/specular too flat right now). I'm pretty happy with my floor material but I think I need to make the color less black and maybe more greyish. Countertop material is going to change since I'm not happy with this.

It seems that aside from the lighting scheme the material settings had a lot to do with the way the light was 'bouncing' throughout the space.


:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::



M~RTest-Render-6(Material-Test2)

20 hours, 17+ SL, 400 ISO, FStop 8, SS 125. I'm done with this one mainly because I'm sick and tired of looking at this project.

Summary/Conclusion

It's pretty obvious that the final render needed more than just 20 hours to clean up but when you have a marginally standard laptop there's but so much waiting a man can handle. Here's what I've taken away from this exercise.


1. Light Units are broken for RC5 so my lights had to be cranked up.
2. Good way to test light propogation is by applying a white mat (lambertian uncheck -roughness set to 99 RGB 200,200,200).
3. Set the camera parameters realitically (F-STOP: 8, Shutter Speed: 125, ISO: 400 to start).
4. Make sure scene scale is set to 1.
5. Pay attention to the lighting design (especially in a space like this which has all it's light coming from recessed ceiling fixtures).
6. Vignetting - or natural vignetting is inevitable and has to be compensated for in post.
7. Material settings play a big role in how light interacts with the scene. (But that's pretty much true of all renderers I think).

I think that a good exercise at this point would be to take a picture of a space, noting the camera settings and time of day and then dry and model and render in maxwell to match.

Before I invest time in another scene I think I'm going to work a little bit more on texure/material development. In the final render the only thing I really like was my floor material so I think materials are my current weakness to tackle.

Hopefully once I do a little more mat development I'll provide a scene and reference photo for any and all to test with the next S.O.A.R. Since this post generated over 4000 hits I guess I'll do another one.

Thanks to any and all that helped me with this one. Maybe one day when I have a better machine I'll rerender it longer to clean up the noise.

Best regards to all,

Luis

Ready Aim Fire....

Well, 2 thoughts:

1) interiors seem to be taking a long time in RC5 in general
2) curious about the roughness settings on most of your surfaces

...well 3 thoughts...your signature is hilarious.

~JD

I dunno but it has this very sterile look to it, perhaps your lights are too white or something... The scene looks good, the textures as well, but I think somehow your lights are killing it... bit hard to put a finger on..

JDHill wrote: ...well 3 thoughts...your signature is hilarious.
~JD

Billy Corgan hehe, bleaky bleak

lsega77 wrote: Other thing. I don't know if this is a bug or not but in order to build this scene in studio I did it peacemeal (wall and floor as one .mxs then furniture etc). When I tried to render out the .mxs for the microscopes I built... nothing doing.

Specifically, what is happening? Any error messages in Studio or mxcl?

Is there any ground plane at all in the scene? You should add it, a little bit bigger than the lab so that light can get reflected of it and into the room.

Mihai wrote:

lsega77 wrote: Other thing. I don't know if this is a bug or not but in order to build this scene in studio I did it peacemeal (wall and floor as one .mxs then furniture etc). When I tried to render out the .mxs for the microscopes I built... nothing doing.

Specifically, what is happening? Any error messages in Studio or mxcl?

Is there any ground plane at all in the scene? You should add it, a little bit bigger than the lab so that light can get reflected of it and into the room.

Well I'm trying to create the .mxs file from 3Ds max. When I hit render the progress bar flashes for about a second and dissappears. No error, no .mxs.

Do you mean a ground plane outside the room? Not sure if I follow. The blackness of the ground is a box for the floor plan.

dutch_designer wrote:I dunno but it has this very sterile look to it, perhaps your lights are too white or something... The scene looks good, the textures as well, but I think somehow your lights are killing it... bit hard to put a finger on..

That's what bugs me about it too. How much brighter do I have to make those lights to get the room to brighten up a bit? I thought 7200 was pretty high.

Two thoughts:

1. I think your light fittings are too far from the walls, especially the window side.

2. I know its a no-brainer but you have got caustics turned on, haven't you.

In general this kind of lighting will always give you a dark ceiling.

With black floors and black work surfaces, seriously, the only way light is getting back to that ceiling is if you fill the room with bald white people.

Well, three thoughts then.

lsega77 wrote:My lights have intensity checked (at 7200) and units are W/sr. Comes out really dark.

Luis, light units are broken in RC5, it's a known bug. Either crank up the
light a *lot* or wait for RC6 or however they'll call it.

Kabe

Hugh wrote: With black floors and black work surfaces, seriously, the only way light is getting back to that ceiling is if you fill the room with bald white people.

That's just funny! Crits noted.

Kabe wrote:

lsega77 wrote:My lights have intensity checked (at 7200) and units are W/sr. Comes out really dark.

Luis, light units are broken in RC5, it's a known bug. Either crank up the
light a *lot* or wait for RC6 or however they'll call it.

Kabe

Ok, thanks Kabe. I'll see what cranking up the lights does. I'll change the floor color too.

you also might want to try messing with the scale, i know that works for me a lot of the time.

I think it was said before, but light units are wrong in RC5. You can probably get light back to the ceiling if you used a coating or two BSDF layers on your materials.

Btw, where did you get your lights from? They are amazing and what I've been looking for one of my projects.

misterasset wrote:I think it was said before, but light units are wrong in RC5. You can probably get light back to the ceiling if you used a coating or two BSDF layers on your materials.

Btw, where did you get your lights from? They are amazing and what I've been looking for one of my projects.

Thanks for the compliment on the light fixture. I modelled these a long time ago. Kinda crude actually, the trick was all in the material really. In maxwell I used the basic aluminum and in vray I just used a vraymtl with reflection set all the way to white.

I've never really shared files over the net before but if you can provide the link to one of those free file hosting services I'll be happy to post it for you and anyone else to download. Just let me know.