realistic lighting in mental ray

[lebada]

wow..dude...try to up the detail of the rendering to min-4 and max 16 at least to get those jagged edges out..

or use 16/64..but that'd take more time to render

[asomma]

I will do that if this pleases you buggy166 :)

But I am not looking for a smoother picture right now. I want to get the tones and indirect coloring right first...

I hope this thread can provide some numbers for realistic lighting in mental ray.

[JeffPatton]

There's a wonderful example scene on the training/tutorial CD that comes with max. You have to dig for it, but it's named "MR_LightGallery_Max6.MAX". Give it a look and either disect the lighting or merge your scene into it.

[sebast1an2]

hi.. i'm a maya user, so i have no 3dsmax.. but this lightgallery example sounds very interesting.. could someone show some renderings of this lightgallery? thx

[JeffPatton]

Here's the rendered scene I mentioned above. (interior lights are off)
http://www.pixelperfectgraphics.biz/cg_files/room.jpg

[asomma]

Thanks folical9! I'll have a look at it.

Do you know anything about taking real light values and putting them into 3d scenes?
I.e. If I want to test how five specific spots will affect the room??

also... Do you have any tips on how to make a lightbulb visible in a room, sp that you actually see the lightsource??

Feel free to experiment with my scene...

[Atyss]

First of all, always use decay with your lights when using mr's global illumination. mr GI model is very physically accurate, and requires a physically correct behavior of light, wich means you need decay. Standard inverse square should do it.

I think the first image you posted is going in the right direction. But you'll need A LOT more photons. In production scenes, it can reach up 2 millions! But with FG you can decrease it quite a bit. Absolutely consider using GI and FG together.

Oh also, don't use light multipliers above 1.0. Again, for light propagation purposes.



Cheers
Bernard

[asomma]

How many GIphotons do you suggest I use in combination with final gathering???

[asomma]

I forgot to say that both of the above renderings are with Gi and final gathering. I will add final gathering numbers.

[Atyss]

Try something like 50,000 to 200,000, with GI accuracy up at 500. Also use Fast Lookup, this is quicken the calculation.


Cheers
Bernard

[asomma]

Do I use decay for daylight too? If I use inverse square decay here the render becomes black, and if I cannot up the multiplier then what??

I've tried to set the energy level higher, but this only makes the colors of the scene affect eachother more. I.e. the white roof becomes brown because of the brown floor...

My problem right now is that the walls that face the light source directly becomes too bright. How can i simulate the more even light of a blue sky with no sun?

[sdp777]

Folical9 - breathtaking my friend. Good lord would I love to have your mastery of MR and materials.

You have to be in the top 10 of the cg artists I have run across over the years.

I first read your posts on the MR.mat files you posted with the sphere shapes. What a thread.

Thanks for the inspriation.

[Phantomstranger]

if you are having a problem with certain lights over saturating then you need to use exposure control on the render.

turn off the global settings on each light and adjust the gi and caustic photons for each light. this is the most accurate way to get real world light simulation. obviously the sun would give off more and the other lights would give off less.

turn down the final gather samples to around 100 to speed things up but turn it back up when you want to make production renders. you might want to turn on the "all objects generate and receive caustics & GI". turn on fast look-up also.

decay lights work wierd in Mental Ray. inverse square usually doesn't give the results most people would expect. either it's a bug or it's what happens when you mix gi with light decay. regular inverse softens the light usually but inverse square never works for me.

[asomma]

http://home.broadpark.no/~asomma/render2502.jpg
http://home.broadpark.no/~asomma/screenshot.gif

The lighting in the scene consists of two mr Area Omnis, one outside each window.


SETTINGS

multiplier: 0,67
decay: Inverse square starting at 5,85 meters

mental ray indirect illumination
energy: 450 000
decay: 1.6
GI photons: 200 000

render setting
final gather samples: 500
final gather max radius: 0,2 meters


This render is better, but I am not satisfied with the result yet.

The areas facing the two windows are too bright. It looks like the sun shines right on them. How do i simulate the more even light of a JUST blue sky outside??

The roof is, in my eyes, too affected by the indirect illumination.
Can one control how far the indirect illumination travels?

There are some errors under the nearest picture. Are these because of too few photons or too lfew fg samples?

Finally, the whole room is too dark. If I increase the multiplier to 1 (not above), the already bright areas become white.



Please help me!!

[asomma]

If anyone has a good idea whats wrong with my scene you could download my scene file, and try to make it look good by changing light settings?

download scene file here

http://home.broadpark.no/~asomma/livingroom.zip

[JeffPatton]

sdp777, thanks for the compliments but I didn't create the interior scene in my previous post. It's one of the MR samples that ships with 3DS. Just wanted to make sure I wasn't getting credit for something I didn't create.

Ligting is one of my weakpoints right now, I'm struggling with it as well.

Asomma, I like that last render, getting much better!

Jeff

[asomma]

Good to hear that you struggle with the lighting too folical9 :)

Do you use msn? I would really like to discuss some lighting issues with you sometime...

[JeffPatton]

I just added my info to the MSN thread (been meaning to do that for a while now).

http://www.cgtalk.com/showthread.php?s=&postid=1164600#post1164600

[joske]

asomma,
have a look here at my suggestions man :
http://www.cgtalk.com/showthread.php?s=&threadid=120241

this is how I do it, it might not be THE ONE AND ONLY way to do it, but it works for me...
that might help :shrug:

first make sure you have your units correct :
a 'cm' in your units should represant a cm in real life
this is very important to get your sample size under control

your energy should be something between 20.000-100.000
not more... if that doesn't do it... lower the decay (1.5 1.4 1.3 .... until the scene is well lit)

keep your photons around 50 cm (this is why your units are so important) and don't use more that 20.000 - 100.000 to start with

keep the FGather low (50-100) for test renders (when doing final renders increase, this will only improve the quality, not the brightness or anything, so you'll get a good idea of the result and still have fast testrenders)
keep away from the FG radius, this will only slow down the renders, and can be fine tuned when all looks ok

[joske]

Originally posted by asomma

The roof is, in my eyes, too affected by the indirect illumination.
Can one control how far the indirect illumination travels?



increase decay, but than compensate with higher energy level

[/QUOTE]
Finally, the whole room is too dark. If I increase the multiplier to 1 (not above), the already bright areas become white.
[/QUOTE]

this is because your energy level is 450 000

lower it, than increase the light multiplier, not the energy multpl.

good luck

[nebille]

Hi everyone, I have been reading through all the replys and have found some interesting points that i will adopt. I to have taken apart the scene with the armchair supplied by max and found some tricks in there as well.
I have had problems with lighting in mental ray eversince i bought max 6. never had problems before. I use all the correct units because i do work for architects and the scenes have to be 1:1 in scale.
I am determined to use the daylight system to recreate the lighting effect in my scenes but have had trouble doing so. Its not so much the lighting is wrong but the way the light reacts with the materials. Have any of you with problems tried this
Make sure your default switcher is set to mental ray/vis
Use a normal daylight system in your scene
create the geometry you need to illuminate and cast shadows from
check that they all geometry is set to receive and create GI
put one material on all geometry prefferably the standard material.
make sure your logarithmic exposure control is on
check for exterior if its an exterior scene
Hit render with final gather on

The lighting will be spot on

Here is where i ger into trouble

Change the materials in the scene and watch how unrealistic the scene gets

I need help in a major way with materials . The lighting seems fine when using the daylight system on its own , maybe you guys with the lighting problem could try the above workflow and see if the lighting remains a problem . Then you may like me need to work on the materials and also like me get a better understanding of the materials that i / we are using

If anyone has any advice other than telling me to shutup moaning it will be much appreciated

Kind Regards

Nebille

[JeffPatton]

nebille, have you been to the Mental Ray shader thread?
http://www.cgtalk.com/showthread.php?s=&threadid=104578&perpage=15&pagenumber=1

[nebille]

No I am going to go there know.. Thanks for the link. I am very knew to this forum . I usually go on 3dlinks.com.

Kind Regards

[joske]

Originally posted by asomma
If anyone has a good idea whats wrong with my scene you could download my scene file, and try to make it look good by changing light settings?

download scene file here

http://home.broadpark.no/~asomma/livingroom.zip

did some changes, but it will take me mabey to long to explain all of it... so here's what I could cook up in a few minutes
rendertime about 4-5 mins. to get rid of the blotches on the cieling just increase Fgather (longer render times)

with sun :

[joske]

no sun , only omnis in windows simulating skylight
(increase energy or decrease decay to get more brightness)

[joske]

make sure to get your units AND your system units 1u -1cm
this was not the case in your file
then you should get to see the photons like this to know you're going in the right direction with your light setup...

[joske]

then just slap a little Fgather on it (50-100) and voila... :eek:

I guess this should get you going man
now it's up to you to get some nice materials, furniture and textures and all

I can email you the .max file if you want to go from there, just send me a personal message or an email or something with your email contact for that.
(zipped max file to big to upload, sorry for that)

good luck with it

:bounce:

[asomma]

Hi all!

Thanks for the good tips. Especially to joske who helped me by changing my existing scene. That really helped! Here is the best render so far. I added a bump map for the walls and a diffuse texture for the couch.

http://home.broadpark.no/~asomma/render4.jpg

LIGHTS
multiplier: 0.6
decay: inverse square starting at 212 cm
area light radius: 85 cm
area light samples: 20 by 20
indirect illumination settings set to global

RENDER SETTINGS
gi photons: 500 radius: 25 cm
fg samples 2000
samples per pixel: min 1 max 16


Correcting the units especially helped (1u=1cm). Joske also added a box around my entire scene with the normals inverted. What does this do Joske?

Most important of all I think, was using ray traced shadows even for the diffuse daylight. This corrected all my previous problems of overbright surfaces. To get the shadows look soft enough the area light radius is now 85 cm. REMEMBER to boost your SAMPLES in the area light parameters if you want a clean spotless render, for this render i used 20 by 20.

All these tips are based on the changes that Joske did in my scene... He is the man!!


I also found that the number of photons and final gather rays affects the color of some areas dramatically. Here you can see a render with less of both. It also has less area light samples.

http://home.broadpark.no/~asomma/render3.jpg

Notice the yellow glow under the table? And the shadows in the tv-furniture?

Well I'm very happy with the imporvement in my scene. I'll still try to brighten the scene some more. Don't know if I'll do that with exposure control or by boosting the multiplier. What do you think?
I will also begin to add interior light to my scene, and will probably be needing help for that too!


anyway! here's my new scene http://home.broadpark.no/~asomma/livingroom2.max
Take it, play with it, and please post your results if you improve something!!


Alex

[joske]

glad i could help
:cool:

the inverted normal box around the entire scene is intended to focus the photons onto your interior
imexp it will :
1. bounce the photons on the box and back into the interior, thus prevent photons from firing away into infinity.
2. enable you because of that to use less photons to get a good coverage of your cieling and walls, just to reduce Fgather times
(more photons means longer Fgather process time imexp)

(I saw this trick in a maya/mental ray tutorial, it was a sphere to simulate a skylight, so mabey a sphere would be even better...??)
http://www.3dluvr.com/intercepto/maya/tutorials/making_glimpse/making_of_glimpse.htm

intresting to learn that improving the light samples to 20-20 could improve the shadows that much !!
I never tought of that nice one :thumbsup:

does 20-20 take alot more rendertime then lets say 5-5 samples?

[asomma]

I didn't notice that setting 20 by 20 effected rendering time at all. But as you know, the rendering time was quite long already.

I've tried turning on your spotlight as a sun, but there are some problems.
http://home.broadpark.no/~asomma/sunlight.jpg

Do you see the intense red and yellow under the table?

Does anyone have a suggestion on how to solve this problem??
Do I use exposure control or lower the multiplier or the energy level??

[joske]

I would try to get other materials for the table and floor first
in the Architectural material library that comes by default with max you will find plenty of wood and floor mats to start from...

give that a try, if that doesn't work you will have to play with the lightning settings, maybe apply a seperate energy multpl. to only the sun (reduce it), I wouldn't touch the other settings and the omnis at first.

try that, it should work imexp.

good luck

[asomma]

Can I make my own materials? What are the rules? Is there something special about the architectural materials?

[joske]

nothing special about them, except they are OK to start from.
:shrug:
So if you know enough your self about materials, don't need them, but if you haven't got a clue they are a good start.

So it's up to you to decide, ... but I would just give it a try
If you're scene turns out better, ... then why not use them?

[asomma]

I've now tried switching to arc. mat. but it doesnt do anything, the table and area between the table and the floor is still to bright or colored. It looks like a indirect lightning problem.... Like the light bounces infinetely up and down between the table and the floor.

I've also tried using a seperate energy multiplier for the spot but no matter how I shange the settings, I always get the same effect... Could it be a units problem? Or perhaps a photon radius problem?

[joske]

mabey increase the decay (less bounce)
and compensate with a little more energy

if that doesn't work try to work with smaller and more photons
(longer renders :hmm: )

[asomma]

!!!!!!

I've now found what caused this red glow under the tabletop. I tried to do a rendering with the radius set to OFF. (in the gi render settings). When I did this, the strange lighting dissapeared. When I set the radius back to 30 it comes back again. I'll show you on two renders I just made.

http://home.broadpark.no/~asomma/radius30.jpg
Render with gi and fg using 20000 photons
Radius set to 30 cm

http://home.broadpark.no/~asomma/radiusoff.jpg
Render with gi and fg using 20000 photons (same as above)
Radius set to OFF


What causes this? How does the photon radius work with final gather?

Does anyone know?

[WDABUILDERS]

Adjusting the decay, giving longer bounce, gets more light in the space. However in areas where light is reflected, like the table, it bounces an unrealistically number of times, adding lots of illumination.

I'm finding that if you use the reflectance qualities of a material, as a reality check for the material properties, you get better GI results. This can be displayed, if not already, by settings in the Preferences, Radiosity, Material Editor, Check the box. Use the information in the help files and radiosity tutorials for common reflectance of materials.

If the materials are reflecting light correctly you should not have to adjust the decay or add anymore lights than would be actually needed for the space.

Try FG only, first. - max min radius 20-50 cm and min at 1/10 of the max. Use the exponetial exposure control to get a relative brightness close, or as is possible. Then work the light multipliers. About 50 percent of the time you need to add GI phtons for better illumination. Adjust the size and so on to get the best results.

The FG first is a bit backwards, however for architectural renderings it seems to work and I have been advised a number of times to forget about the photon Gi and just use FG.

For what it may be worth
WDA

[asomma]

Thanks WDABUILDERS!

I'm reading through the reference just now, and this seem to bee exactly what I was missing. I'll test it and post my results!

Alex

[JeffPatton]

I haven't seen anyone mention the trace depth settings. I was under the impression that it controlled the number of "bounces" if you will. :shrug:

http://www.cgtalk.com/showthread.php?s=&threadid=117519&highlight=trace

[WDABUILDERS]

Lowering the trace depth settings for asomma's table would decrease the light energy at the table and in the enviroment. Max trace depth being a global setting for all photons effecting all the lighting behavior and similarly FG global illumination, probalbly is not the best choice.

It seems best to start with relatively accurate lights / placement, energy levels and using inverse decay for standard lights. Use the trace depth settings and FG Falloff to FINE tune the volume of lighting not create it.

Most of the time the luminarie/s and or the reflective material qualities are not accurate. Being photoreal, if a scene is to dark or too bright switch on the exposure controls and view it like a photographer.

We inherently know how light works in our day to day enviroment. Changing the number of bounces or extending falloff is just changing the physical properties of light. The materials and thier reflective qualities control how light is distributed in a space, light behavoir being constant. This at least from an architectural or photorealistic perspective. This scene is not percieved as an alien world, there are just to many known benchmarks or visual ques as to how it should look.



Painting with MR GI is FUN :cry: But when it's right it's awsome.

WDA

[asomma]

folical9

How does the number of bounces in the trace depth settings, affect the scene. Will the photoncalculation be more precise or will it just increase the color bleed??

[JeffPatton]

Here's two renders. The first I have set to 5,5,5 (I think thats the default). The second one I increased to 10,10,10. The only changes made to the scene are the trace depth settings.
http://www.pixelperfectgraphics.biz/cg_files/trace-005.jpg
http://www.pixelperfectgraphics.biz/cg_files/trace-010.jpg

I can't really tell any difference in color bleeding (orange on the floor). Looks like it just made the scene brighter like more GI bounces?

Note: If I used non-mental ray materials, I didn't notice a large change in the scene when I increased the trace depth settings. I even went as high as 1000,1000,1000. However in the images above, I used a mental ray material with a photon map and you can see the huge difference in the scene with only a minor adjustment in the trace depth settings.

Jeff

[rlph.]

Oh, very nice find Folical! Cheers! I just wish stuff like this would have been in the help files....

[WDABUILDERS]

The trace depth in this case is basically increasing the light exposure for the viewer. In this scene the additional calcualtion time is minimal, but it will increase with complexity. That would be why I would try the exposure controls....



http://www.wdadesign.com/links/CGtlk/123.jpg


Not against using the trace depth settings, but the exposure controls are quick easy and do not eat up cpu times.

Folical-The material to shader, another thread, option works really good for kicking up the GI light in a realistic manner. Like adding it to walls, increasing lighting levels in a room, without increasing trace depth.

WDA

[asomma]

So what is accurate light placing, and accurate energy settings? This is exactly why I started this thread. I want to be able to have a scene where all the lights settings are correct in relation to eachother. I.e. The relation betwen sunlight, daylight, and a 60 watt bulb. When set, I want to be able to switch these lights on and off without adjusting everyting.

Like on a camera I should only need to adjust exposure, right??


How should I work to get a realistic result?
I'm guessing that this would be the right approach:

I'll first set the lights multipliers in relation to eachother. I.e. The sun is 1, skylight is 0.5, a 60 watt bulb is 0.1. I would use inverse square decay for inside lights. In the indirect illumination settings I use low energy values (10000) and decay at 2.

I'll then lower the value or RGB level of all materials until they get the correct reflectance percentage.

For rendering i'll use just 25000 photons or so with a radius of 25 cm, combined with 2000 final gather rays.

To adjust the lightness of the scene I'll use exposure control. There, I set the Physical scale equivalent of the brightest lightsource, in my case the sun. (If I turn the sun off, I would set the physical scale equivalent to the sky... and so on) I'm not sure how many candelas bright sunlight is but i read here
( http://www.electro-optical.com/whitepapers/candela.htm ) that it is 3000-6000 cd. Therefore I set the physical scale to 5000.



Is this the right way to go? Please comment...

[daik]

Hello alex (Eirik, from school here). I havn't read all the posts, but your thoughts in your last post seems to be in the right direction, to use exposure control to get a more accurate result. Don't know if anyone else has mentioned it, or you already have planned to do it. But you could perhaps add a tiny bit of glare... to add to the realism.

Anyhow... yhink you're on the right track :)
-e

[joske]

Originally posted by asomma
!!!!!!

I've now found what caused this red glow under the tabletop. I tried to do a rendering with the radius set to OFF. (in the gi render settings). When I did this, the strange lighting dissapeared. When I set the radius back to 30 it comes back again. I'll show you on two renders I just made.

Render with gi and fg using 20000 photons
Radius set to 30 cm

Render with gi and fg using 20000 photons (same as above)
Radius set to OFF


What causes this? How does the photon radius work with final gather?

Does anyone know?

turning of the photon radius will have mental ray estimate the most apropriat radius, so it will never be turned 'off'.
so it's just a matter of finding the right size here.
However smaller photons will cause you to use more photons to have the entire scene covered in photons...
More photons will cause Fgather to process longer...
It's up to you to search for the best looking and fastest rendering balance...
Tuning off the radius might be the easy way out, but could give you longer process time than needed.

So it's important to see the photons like in my previous attachment, just to check if everything is well covered.

http://www.cgtalk.com/attachment.php?s=&postid=1168595

you can also 'predict' the intensity, in this example I would say the photons are 'to bright', the should look like your shining a flash light on a wall, bright in the center, allmost dark in the middle part, and then an outside 3th ring agian brighter.
Make sure the entire seen is covered with overlapping photons.
But try to do this with the smallest amount and biggest radius you can. However to big of a radius will give to little detail in the cracks and corners of the scene. Reduce the radius untill you get them right. However Fgather will take care of that also.
So You don't have to go to extreme with that.

So in this case : Smaller photons will give you more detail and reduce the glow under the table, but will increase the fgather process time.

give that a try
succes

;)

[asomma]

thanks joske :)

I have now solved this table problem by using realistic raflectance values in my materials. Setting decay (in indirect illumination) to 2 also helped.

My problem right now is how to get the correct color without going over the maximum reflectance value. Does anyone have a tip?

I'm also struggling to get enough intensity in the sunlight without ruining the rest of the lighting in the scene.

Does anyone have a reference for correct candela values??
And how do I set the light multipliers realistic in relation to eachother?

I.e. What should I set the multiplier for a 60 watt bulb at compared to a candle light, or the sun, or a halogen light?


I need an expert!

[joske]

what do you mean with 'realistic raflectance values in my materials'?

is that ambient, diffuse and specular settings? or something else...

can you post an example of your renders, it's nice to see the progress evolve. :D

[asomma]

WDABUILDERS tipped me on this:

I'm finding that if you use the reflectance qualities of a material, as a reality check for the material properties, you get better GI results. This can be displayed, if not already, by settings in the Preferences, Radiosity, Material Editor, Check the box. Use the information in the help files and radiosity tutorials for common reflectance of materials.

In the diffuse slot of a material. You simply adjust the value of the color if it's just a color, or you adjust the RGB level if it's a texture. The color or texture will normally become much darker than you want so you'll have to use log. exp. control to get it right.

As quoted above, the 3dmax reference gives standard reflectance values for common materials like wood.

EVERYBODY SHOULD TRY THIS!!!



Regarding posting my renders...
I will do this as soon as possible, just need a few adjustments first. Stay tuned!

To everybody: This thread has done a lot for me so far, and I'm still evolving. Everybody should download my scene, make adjustments, and post results with the settings that made it, so that we can compare renders.
This way we will all get better and more realistic renders!

[WDABUILDERS]

asomma

With standard lights you need to aproxiamte the light distribution for the bulb and fixture, by feel, experience, observation, realtive light energy. Fakiosity- using multiple lights to represent a table lamp for example. Or the general sum of fairly equal lighting and use area lights.

I think your using max6 correct? If so try the photometric lights. You can use .ies light distibution files to represent the luminous flux of a bulb and fixture (luminarie). This is referenced in the link you posted previously. Dialux has a site and some free software. It's oriented to the light engineering disciplines, but very useful. Part of this is a very extensive library of ies. files and related manufacturer info.

Light distribution files & photemetric lights won't really make it easier for accuracy. But a close aproximation to real light distribution is better than a single omni.

This may not be the direction you want to go, but using radiosity (max6) can give you a good reference source for using MR. You also have the Pseudo Color Exposure Control. Gives luminance and illumance values on a color scale. You can then adjust your light values to match the reference values in the link you posted.
With max6 working across the render engines yields fairly similar results. This type of work flow is great for lighting studies but not good for making money.

I will try to post some lighting reference and ies links. If it will help.

WDA

[joske]

Originally posted by asomma
WDABUILDERS tipped me on this:



In the diffuse slot of a material. You simply adjust the value of the color if it's just a color, or you adjust the RGB level if it's a texture. The color or texture will normally become much darker than you want so you'll have to use log. exp. control to get it right.

As quoted above, the 3dmax reference gives standard reflectance values for common materials like wood.



Asomma,
this is why I advised you to use the max6 architect. mats.
:wip:
:D
with these mats you can be sure that the diffuse is good to start from. Then just fine tune them or replace the textures as you wish. I'm sure this way you will not get the problems similar to the glow under the table in the future...

For the sun candela values :
Don't worry to much about that... just take your own office or room as a reference, that's what I do. Keep in mind that the sun can be blocked by clouds, reflecting glass and drapes, be behind neighbouring buildings, ...
So the real candela value inside will be totaly different many times.
Just do what you feel looks good, that's imo more important than getting the exact values in real life.

wda,
what you're suggesting is correct. Radiosity/max5-6 will give a real live, and accurate light simulation, ies and stuff...
you can get some .ies light examples here :
http://www.erco.com/
however In Mental ray to my knowledge this doesn't work.
So again, Mental ray is not working with exact and accurate values, but more trying untill it looks good in your opinion and taste.
I'm sure nobody will be able to tell the diff. if it's done in an acceptable manner...
:shrug:

[asomma]

Thanks Joske! I understand this now. But also in arc. mat you must be careful not to make the material to bright. To get the color you want, you'll have to select a color much darker than it really is. I wish there was some way of previewing the material with log.exp.control on. It's so difficult to hit the right color!


The reason I'm pushing forward with this realistic candela values:
I'm an architect student. If I am able to translate real life light data in to 3d max light settings, I am able to preview the different light situations in a room. I can there test different light types, colors, and positioning.
That's why it's so important for me to have the correct decay, multiplier, physical scale and so on...

I don't just want to make a render look good. I want it to tell me wether a specific lighting in a room will be good OR bad


How can I translate real life light data into 3dmax/mental ray data??

[joske]

you can't ! :)


forget about mental ray for that purpose.
go for Max/Radiosity, this will allow you to do exactly that.

However it is not that easy, and you will have to do your modelling according to very specific rules to prevent light leaks + it renders alot slower (once you save the radiosity solution it goes a bit faster).
But you can enter everything with real life and exact candela values + the daylight system will allow you to simulate sun, clouds, north direction, time of day,... in a specific part of the world.

you should go fo that solution in your case.
However for commercial, time and budget restricted projects it's not a good idea

[asomma]

So you can't do this with mental ray? What is the advantage with mental ray then? I personally like mental ray more...

[WDABUILDERS]

asomma,

Joske has a valid piont. The fundemental problem with MR is it's learning curve, it's very s t e E P . The documantation, well kind of rots, and tutorial information for Max is just beginning.

As far as being slow, yes but thats tied into the learning curve. With being able to save photon maps and FG info and not have to calculate every time, actually makes it just as efficient as radiosity. The MR render engine with the scanline disabled is significantly faster than with radiosity. Then there is the mental ray shaders really powerful escpecially the displacement aspect. It puts any max bump or displacement to shame for speed and quality.

If you want to stick with mental ray, I would suggest....
1. Become a student of light, not an enginner, but study how it works in your surroundings yadayadayada....
2. Always start with your lighting solution. Use the Processing>material overideand assign a 50% reflective quality material, build up your lighting, then assign materials.
3. MR allows you paint with light- accent the subject, give it emotion...... life.
4. Think of building the lighting as the sum of the parts, work on individual lights, sets of lights, areas of lighting-you don't have to do everyting at once.
5. Experiment, not on the current project but with different effects and aspects. There's no shame in Photoshop-levels-duplicate and multipy- to bring the image to life either.

Good Luck:thumbsup:
WDA

[joske]

euh... nope not to my knowledge no.
:hmm:

Advantages : where to begin?

caustics, dist of field, displacement, renders alot faster (if you know what your doing ;) ),motion blur, sub surface scattering, glare, dirt, ambient occlusion, endresult just looks better and sharper imexp. (i used radiosity for a while, when VIZ4 was released) ,...

i'm sure i forgot a lot :shrug:

only thing is that you will never get the 'exact real life' translation, like you can do in radiosity.
But imo (i'm architect myself) it's not that important unless you're working for a company that makes lights or something.
Just try to get your scene similar to good examples in architecture magazines or your immidiat surroundings. Nobody will be able to tell the diff. trust me :thumbsup:

But I'll leave that decision up to you my man...

[asomma]

I get your point. I like mental ray better too and I'll continue to work with it and post results.
Anyway, the reason I want to understand candela values is just to understand more about how the lighting units in mental ray works. If I know the realistic light numbers then I can work my way from there.

[joske]

Originally posted by asomma
I didn't notice that setting 20 by 20 effected rendering time at all. But as you know, the rendering time was quite long already.

i did some tests about the rendertime for higher sample settings
5-5 10-10 15-15 20-20
it takes almost 5 x longer in rendertime (5-5 compared to 20-20)

however the quality is accordingly...

check out my post over at cgarchitect

http://www.cgarchitect.com/vb/showthread.php?p=37948#post37948

cheers
phil

[WDABUILDERS]

Just got the New 2nd edition, "Rendering with Mental Ray", TH. Driemeyer.

The values are all relative. If you use the ies based lighting, photometric correct, you have a baseline from which to work from. If you use standard lights, it's not so easy. The key to the whole game is the inverse square setting and some physics. Mental Ray will obey the laws of physics. Just as with photography brightness, perceived is relative, but the true nature of the illuminated scene is always correct.

So dive into the Physics Text books, pick a multiplier, pick a cd level and work it from there +/-. Steridian calculation and the inverse square properties understanding is very helpful. Remember the exposure control is your friend!!!!

WDA

[revelaciones]

mental ray is incredible powerfull all of us saw the work that mental ray did on the matrix reloaded and the matrix revolutions, also the mental ray is slow, and the 3ds max integration is not the best, mayby the next version, mental ray 4 will be much faster,
anobody knows when brazil 2 will be abaible

[FFASHION2399]

Hi,all
have a look my test of mentalray

the first one I want to express a strong light in the noon

the second one withoutdog is morning

give me some critics
http://www.27cn.net/wh/bbs/topic.asp?l_id=17&t_id=433


another link
http://www.27cn.net/wh/bbs/topic.asp?l_id=17&t_id=97
this one is rendered in brazil

[joske]

try attaching the file (20kb limit)
or link it online

this way nobody can see it...

;)

[FFASHION2399]

so quick reply,:bounce:

I am thinking how to make the picture below 20000bytes
(some one post images bigger then20000bytes,how to do it?)

http://www.27cn.net/wh/bbs/topic.asp?l_id=17&t_id=433


can you guys see it?:buttrock:

[FFASHION2399]

Corner on the wall have some black photons,but it seems natural dirtys,I don't know whether I should correct it.

I have a question post in the forum,but no one answers,I have to try to post here another time

how to use "use file"setting in the final gather

I don't know where to load file,

[joske]

reduce and compress the renders in photopaint - photoshop programs to achiev 20kb
to insert a bigger image hotlink them online by clicking on the 'IMG' button in the post windows like so:

http://www.27cn.net/wh/bbs/upload/fyzw_200446132951.jpg

the "use file" setting is for a file u saved yourself
so first save a solution then load it to reuse it.

keep in mind the fg solution is calculated from the camera view
so changing the cam position or any object in the scene will need a new calculation for fg...

that aibo looks sweet:drool: i've been modeling on one myself, but gave up on it when i reached the body:blush:

to get rid of the black photons give this a go:

try higher samples in the lights
try higher fg settings
try smaller and more photons

that should get rid of them, however rendertimes will increase accordignly...

good luck

[Bercon]

I usually get fast and good results with this kind of FG settings:

Samlpes: 1000-2000
Max Radius: 5-10
Min Radius: 0,1 - 1

Always use max and min radius, always.

[FFASHION2399]

thank you joske and masterbercon,thank your quickly reply.
I will try your tips,and post my new version later.:applause:

[FFASHION2399]

photon5000 radious:3cm

finalgather
samples 2000 Max radious:10cm Min radious:1cm

global light properities
Energy:50000
decay:2.0 Gi photon:20000





http://www.27cn.net/wh/bbs/upload/fyzw_200446182719.jpghttp://www.27cn.net/wh/bbs/upload/fyzw_200446182910.jpg

[FFASHION2399]

I have increase some parameters,the picture seems not get better on the black points on the wall,maybe I have to increase more,but the time will be increase too.

I like the color tone this version.It call to remembrance somewhere and sometime I have been to.
:)


a new question: what relationship between "Global illumination:photons" and the "GI¡¡£ð£è£ï£ô£ï£î£ó¡¡£õ£î£ä£å£ò¡¡£Ç£ì£ï£â£á£ì¡¡£ì£é£ç£è£ô¡¡£ð£ò£ï£ð£å£ò£é£ô£é£å£ó£¢
when tune the parameters,what is the main job of these two seperate GI photons each other, are they equal or something?

[FFASHION2399]

I found the most important trick,before final gather it must have a photon map test,the photon map should overlap and cover the scene at least 80%,then turn on final gather will be Ok,there will not black points and glow points there.

http://www.27cn.net/wh/bbs/upload/fyzw_200446194626.jpg

[joske]

yep looking good ffashion :thumbsup:

you can have very low fg settings 50-250 if you cover the scene in photons (50-100 cm radius and about 10000-25000 photons untill the entire interior is covered) and still have a spot free render :cool: this wil also reduce the render times alot!

when using mr area omnis the sample rate can be very determing on spots
5-5 samples very quick render, but alot of spots
20-20 samples very slow render, but no more spots...

can u post a wireframe of the aibo (no meshsmooth) for me, mabey after seeing how you made it, i might pick up my aibo model again...

[FFASHION2399]

do you mean this sample£¿I neglected it,at first I think the sample parameters only work when"show Icon in Renderer"is enabled.I will have a test.

I have a question then£¬do you see the two red circle I draw in the picture,what is the upper one "photons"do?and what is the nether"GI photons"do?
are they equal? I don't know too much.


I will post a new thread named"aibo ERS-7 robot dog",wil include my wire model and some render pictures,ok, IN the 3dsmax forum.

:)

http://www.27cn.net/wh/bbs/upload/fyzw_20044622223.jpg

http://www.27cn.net/wh/bbs/upload/fyzw_200446221149.jpg

[joske]

yep those are the samples i was refering to
they allways work, you cannot turn them off!

the 1st 'photons' parameter is for accurasy (i don't change this most of the time)

the 2nd 'GI photons is the number of photons

also it is best to check if your units and system units are set to cm, this way you can check and know how big your radius will be of the photons. then make the room in the correct scale (ceeling 320 cm, door 210 cm high for instance).
then rendering only with GI (no fgather) you should be able to check that the interior is completely covered by photons.

[FFASHION2399]

I always find when the{ "1st photons:2nd GI photons"=1:20 },the global illumination works well,and the photons in the render will increase if you increase the 1st photons,if not1:20,the photons in the render will not increase.

first photon default setting is 500,the GI photons default setting is 10000,this proportion is 1:20,may be i should obey to it


the wire frame of my dog has post in a new thread,with some of my renders:beer: :bounce:

[joske]

you can allways put a sphere with inverted normal (big enough to include the whole scene, includig lights) to reduce photons from shooting into space... they will bounce back on your objects

then setting it 'invisible to camera' and you wont see it when rendering

i'll have a look at the wires, thanx
;)

[fabriciomicheli]

oh boy... this thread is just great! I learnt a lot reading this one!

A question: I never worked with EXPOSURE till now. Somebody can explain to me how to use it? or where I find it in MAX and what is it for?

:cry: buah! I look like a newbie again...

[asomma]

Rendering -> Enviroment -> Exposure Control.

Always use Logarithmic Exposure Control with mental ray.
Don't care about the top 3 settings at first, just experiment with the "Physical Scale".

In the scene you are doing, render out one image, change the physical scale, and render another. Now you can see the difference...

[MikeBracken]

Hello all. This is a very informative thread. I have a question.....when using GI and final gather I love the results ,but sometimes I get a strange "misty" looking render ,especially where the back walls and the ceiling meet.Any quick ideas on what to check?

BTW...Its a room with direct light coming through the window....no other lights.

Thanks

[joske]

hard to tell without an example dude
give us a render and it will be easier to tell...

[MikeBracken]

OK.Heres the culprit.Like I said before ,Im getting a strange "misty" appearance.Not sure why.And Im getting some washed out textures (the wine picture among others).Any help would be great.Let me know if I should post the scene settings.
Thanks.:)

http://www.geocities.com/mmikeemax/litroom12.jpg

[Matze4d]

@Mmike

did you try exposure control? If you did i would lower the brightness value increase contrast and maybe lower middle tones as well e.g. 0.5.


@all
I have 2 questions maybe it was posted here but i cant find it... I am looking for references how the photon energy value of MR is related to real world units. MR is claimed to be physically correct but without an absolute value of light energy i cant make sense out of it.
The other one is regarding the decay values, from a physical point of view not all kind of light has a decay, e.g streaks of parallel daylight shouldnt experience a decay (besides through mist or dust in the air). Am i wrong with this?

thank you

[joske]

MmikeE
this is log exposure problem for sure
turn it off ! i never use it because of the same strange effect it gives...

matze4d
don't worry about the 'correctness' of your renders.
if it looks good = it is good
think about it : nobody will be able to tell the difference...

about the decay : i use it for sky omnis and sunlight, but again : if you think the render looks better without using decay : then do it

the important thing is that your render looks good, period.
don't worry to much about the rest, you can't tell anyways :D

[X96]

I'm getting all these little fuzzy dots from my caustics how do I blend them away or fix it so they are'nt there?

[JeffPatton]

Increase the radius or the number of photons, or both. Maybe this will be helpful too:
http://www.jeffpatton.net/MR-Caustics101.htm

[MikeBracken]

Thanks for the help guys.I hadnt even considered the exposure.Ill give it another go and post the results.:shrug:

[ToddD]

I'm far from an expert, but Io agree with Joske, I never use exposure controll either, didn't care for the results...

[herbert west]

I've found that to get well-defined shadows you need around 50,000+ GI photons, and FG samples set to 500 or more. Exposure might help with your contrast problems, but there might not be enough shadow detail there to begin with to compensate. Let us know how it turns out.

[MikeBracken]

Todd.....

Loved the audi.....awesome.....just checked it out.
I turned off the exp control.......strange light gathering in blotches.
I will get the latest test render up on my site and post it here as soon as possible.
Again thanks for all the help guys.

[MikeBracken]

Cant figure out whats going on here:scream: .......Any help would be nice.I must be overlooking something.The strange shadows are killing me..lol.And the lampshades are translucent arch mats with 25% translucency and 0% shinyness.Not sure why they are so shiny.

I tried many different gi and fg settings with mixed results.But could never fully get rid of strange shadows or the "shiny"lamp shades.
Thanks in advance for any suggestions.http://www.geocities.com/mmikeemax/strangeshadows2.jpg

[MikeBracken]

Well I changed the texture on the blanket.No more 2 sided.Not sure why it was.But still have the shiny lamp shades.

Thanks for any suggestions.

[Haola]

hi i just get the living room scene , i played with lights and i get this . Rendering with FG only 100 , so it has some problem such as under the TV-set , if u increase to about 2000 - 3000 it would be fixed

http://img.photobucket.com/albums/v724/haola/Linh%20Tinh/up.jpg

[titopte]

Hello,
first i´d like to thank everybodies contribution to this thread, it helped me clarify some MR concepts that were diffuse.
I have a lighting problem too, involving FG ( i guess), i posted the images here :http://www.cgtalk.com/showthread.php?t=234449
some black "bites" keep appearing without any evident reason.
Hope somebody could help.
Thanks again

[FATdan]

does any1 know how to setup glass pains for windows in GI interior scenes? no matter what i try (or which type of glass; lume, standard raytrace or phys phen) it cancels the GI and renders the scene extremely under-exposed (if there is such a word:)
appreciate any help!! ta

heres a render of the scene im working on.

[oreutenauer]

hi ... nice post here for working mental :)

wich material should i use/are you using to test my/your lights before mapping ? this material where applied at all objects of the scene

[-Vormav-]

MR 's lighting is too dependent on your materials, so you should setup all of your materials to be close to having the physical properties that you need when you're testing the lighting solution.

[joske]

@MmikeE
looks like some sort of smoothing prob. add a smoothing modifier and play with the settings to get rid of the black areas in the blanket

@Haola
don't count on fg to get rid of the spot underneed the tvfurniture and behind the sofa
try to get rid with better photon solution (bigger photon and more to get better spread and overlapping)
you might get the same result in the end (than higher fg value) but it will be a lot quicker.

@FATdan
never use glass planes in my windows it will not work, trust me
add a scanline layer with glass and raytrace reflections and combine in photopaint wth alpha channel (will be quicker and nicer)

@ in most cases u can use the same max materials than u should be used to.
so nothing custom for mray. only for some objects u need mr materials (glass tables, canlde wax, ...)

@titopte
most of the time this problem is caused by perpendicular raytrace reflections endlessly reflecting each other
this could be the case in your scene (but that's for u to check, could be something else ...)
try turning off all reflections, if the black spots are gone you know what to look for
then add all raytrace reflections one by one, and see which one is the one causing the spots
once you know the material causing it, u could :

-consider leaving that reflection out
-or changing slighty the reflection angle of one plane.
-or change the raytrace depth from 9 to 3 or 2 in the 'render scene - raytracer tab'

all this is in the assumption that this is a reflection problem ofcourse
changing the photon size and n° might also do the trick sometimes
and also make sure all your object have correct (outword) normals and no 2 sided materials
if all that doesn't work... well i guess you're on your own then :twisted:
just kidding just send your scene and i'll check it out myself that could be a lot faster than just stabbing in the dark here...

[FATdan]

joske, superbe. thanks for the info!! much appreceated!

[voolkan]

really nice thread, very informative and really had fun going through it.
here is one of my tests with mental ray, tell me what you think.
please ignore those lines in the shadows, its probably a unnoticed crack in the mesh or something.
http://www.3dfaculty.com/fileupload/store/uno1.jpg

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.