lighting question

[bertrand]

i ve write this thread already but it don t appear so i do it one more time but shortly.
I got few questions and will be very happy if some people help mee.
1
a maya transparent blin with raytrace give transparent shadows not a dielectric one. Why? (i ve used a shadow shader but the shadow seem flat)
2
i ve read about fg secondary bounce and also this tut http://www.lamrug.org/presentations/apr2005/LAmrUG_April_2005.pdf. And ther is a great difference between the pictures but i cant get it to work, first fg trace depht got 1 number and not 4. So an explanation on the way to get fg going further in a scene and to get a brighter illumination without burning some part would be great.
thanks for this site I hope it will help to bridge the gap between beginner and expert.
bert

[Wish]

1. Don't use standart shadow shader. Use standart maya ray trace shadows with "sort" parameter(see F1) at your render globals.
2. Please, tell more about your problem. And some pics of test render woud be great!
But it seems to me that irradiance of the shader will save you.
________
MEXICOCITY HOTEL (http://mexicocityhotel.info)

[Jak_Carver]

i ve write this thread already but it don t appear so i do it one more time but shortly.
I got few questions and will be very happy if some people help mee.
1
a maya transparent blin with raytrace give transparent shadows not a dielectric one. Why? (i ve used a shadow shader but the shadow seem flat)
2
i ve read about fg secondary bounce and also this tut http://www.lamrug.org/presentations/apr2005/LAmrUG_April_2005.pdf. And ther is a great difference between the pictures but i cant get it to work, first fg trace depht got 1 number and not 4. So an explanation on the way to get fg going further in a scene and to get a brighter illumination without burning some part would be great.
thanks for this site I hope it will help to bridge the gap between beginner and expert.
bert


Ok, here comes the mini tutorial on FG multibounce :P


Ok, I have set up a simple scene with only a room and a cube inside it. I am using DGS shaders for both, though you can sue whatever shaders you want; I just like to work with physical accuracy.

I have my camera environment background color set to a light pale blue for the FG lighting. Go to your render settings and activate the PreviewFG preset. Now go down to the Final Gather section and you will find many settings, Im not getting into each one and how to work them because I will be here typing for quite a while. Instead we will focus on FG multibounce in MR 3.5 (Maya 8). Note: this works in Maya 7 as well with MR 3.4 though many people didnt know this!



Ok, back to business, there are 4 settings we care about, and they are the following:

Trace Depth
Trace Reflection
Trace Refraction
Secondary Diffuse Bounces

You must use these settings to activate and use multibounce, but use them correctly or else you will not get it to work. Alright I hope this tutorial helped, take care.

J/K! :D Seriously though, you will notice that by default your settings are:

Trace Depth=============2
Trace Reflection==========1
Trace Refraction==========1
Secondary Diffuse Bounces===off

If we render, we get something like this:

http://www.3dmercs.com/ForumImages/multiBounce0.jpg

You can see the FG but its not bouncing around, but it's only rendering at 2 seconds. So lets add our first bounce by changing our settings to the following:

Trace Depth=============4
Trace Reflection==========2
Trace Refraction==========2
Secondary Diffuse Bounces===on

We render again and we get this:

http://www.3dmercs.com/ForumImages/multiBounce1.jpg

Now we can see more light bouncing around in our scene, but it took twice as long to render at 4 seconds. Now lets change our settings to:

Trace Depth=============8
Trace Reflection==========4
Trace Refraction==========4
Secondary Diffuse Bounces===on

We render and end up with:

http://www.3dmercs.com/ForumImages/multiBounce2.jpg

We can see a bit more light distribution in our scene but the difference is very subtle, plus our rendering time is now 16 seconds. This seems like a short time, but consider rendering a highly detailed scene with complex geometry and realistic lighting and multibounce can make the difference between rendering from 2 hours to 16 hours!

The truth is that using multibounce, while pretty easy to control and awesome, you shouldnt render a scene out with it at all; its just not efficient. Instead the smart thing to do is to render with FG and Global Illumination. This allows you to use GI for handling your "multibouncing" of light in your scene while FG helps fill in smaller details and interpolate your GI solution, which in essence means you can render out the same scene with the same lighting in a fraction of the time; and all this without any hit in quality. If you want to use multibounce with FG thats ok, but I guarantee after some use you will get tired of the high render times and will end up resolving to using FG+GI to make some great renders in a relatively quick time. Good luck and I hope this helps you further your skills with MR, take care.

[bertrand]

thanks to both of you.
Even If I dont understand why trace reflection and trace refraction are used in a diffuse environnement.I think that was the point i miss previously.
So i i use the global illumination with fg. I notice that some area can be really bright . for example i got a light going throught a window hitting a wall in a close box. So i got a dispertion of the photon.but not so far in the box.Is there a parameter or a combination of parameter) so as to get the light less strong when hitting the wall but bouncing more?or it is just a question to increase the number of photon and decrease the exponent?

[Wish]

As for me, I prefer to render without GI. That's more faster, the result can be the same, but some more work with tweaking of illumination
________
BONDAGE HARNESS (http://www.fucktube.com/categories/387/harness/videos/1)

[Joj]

Is there a parameter or a combination of parameter) so as to get the light less strong when hitting the wall but bouncing more?or it is just a question to increase the number of photon and decrease the exponent?

I've done some test to find out how you could do that and it works when you increase the photon intensity and increase the exponent too.

you can keep the same intensity on the wall but more light in the whole room

[Edgeloop]

So, what if you have a scene or maybe an object with like over 500K polygons, with different material nodes assigned to it... How would you do a lighting test, using photons etc...

Is there a way to get quick renders? instead of having to wait like 16hrs... :eek:

I have a project that i am working on and i am trying to get all the necessary answers before i hit the Hours :)

And that is really what is making me backing off slightly...the Hours! I am just kinda scared of about my rendering time! :(

[Wish]

Yeap! There is the best way! Don't use GI))
________
Fix Ps3 (http://fixps3.info/)

[Edgeloop]

Then, what should i be using then...? :confused:

Sorry if my questions seem silly... :rolleyes:

[Wish]

Your question isn't silly in any case. A lot of people doesn't know what to do in such case, because they doesn't work with renderman and compositing. Faking of GI will reduce your time much more. Use point lights with falloff(in maya very cool is volume lights) to make a lot of faster and cooler color differenses. Use spots to simulate GI bouncing. If you have an exterior use dome lights... and so on. It's not difficult. And believe me, that this is very interesting. Also use an a occlusion to simulate soft shadows.
All this stuff will reduce your time and gives you a lot of flexebility to make scene perfect on the stage of compose.
________
MARRIAGE ADVICE FORUMS (http://www.love-help.org/marriage-advice/)

[bertrand]

As i understand :
you should use fg and gi.
fg with no multi bounce.
gi with high exponent and high photon intensity so as to get the light going further in the scene (are this parameter time consuming?).
after rendering time should be dependant of the size of fg points and gi points and of the quantities.
A question now. Because soft raytrace is time consuming is there a way to get scanline shadow going throught transparent windows.(i have tried light linking but i can t get it to work?)
bert

[Wish]

Is this question to me?
________
hot girls Cam (http://www.girlcamfriend.com/webcam/babes/)

[bertrand]

basically it s more a resume of the way to use global and fg.
but this one is for U. If you only use fg how do you got color bleeding. and by he way if you gant an idea for shadow?
bert

[Edgeloop]

Bert, i am sorry if i am using your thread to ask my personal questions... I just thought i'd post a question here since ur thread is about photon, shadows, fg and gi :)
Hope you don't mind that... ;)

Wish, sorry if i got you confused :D (U mind if PM you?) :o

[Jak_Carver]

As i understand :
you should use fg and gi.
fg with no multi bounce.
gi with high exponent and high photon intensity so as to get the light going further in the scene (are this parameter time consuming?).
after rendering time should be dependant of the size of fg points and gi points and of the quantities.
A question now. Because soft raytrace is time consuming is there a way to get scanline shadow going throught transparent windows.(i have tried light linking but i can t get it to work?)
bert

Actually one misconception that is out there is that you should alter your exponent on your GI photons in your light attributes to control the intensity of your photons; this is completely wrong and there is one tutorial going around that tells to do this but Im not gonna name names, though you can find their ad on this website hehe.

You should never adjust your exponent from the default 2.0 if you wish to retain physical accuracy, or else you loose that accuracy and now you will find yourself wasting hours tweaking your photon solution and getting no where; it can drive you mad. When your photon intensity is really low and not affecting your scene as much as you'd like, you need to increase the photon intensity, usually in a physically accurate scene this means using crazy numbers like in the hundreds of thousands or like I usually do, in the millions. Dont be afraid to do this, its perfectly normal for working with MR's GI photon system.

I cant show you how to light your scene up correctly with MR and Maya because this would be a loooooonng post, but I will give you some advice.

First, use physically accurate lights, in Maya this can be achieved by setting up a MR area light (I assume you know how to do this) and then give it a MR light shader, like the physical light shader. This will give your direct illumination physical accuracy.
Next, make this light emit photons, keep the exponent set to 2 so it uses the correct inverse square formula for calculating physically accurate light fall off on your photons (if not then you are giving your direct light physical correctness, but taking it away from your photons so whats the point? It would be like accelerating a car with the parking brake on...).

Now do some test renders without GI turned on and adjust your direct lighting to whatever makes you happy.

Once done, turn on GI and tweak your GI photon solution until you are satisfied.

Lastly, turn on FG to interpolate your GI solution and finish off your lighting. This will allow you to use about a third as many photons while allowing your FG algorithm to take care of the rest so to speak, meaning you can create high quality renders at a fraction of the cost; win win situation if you know what I mean. With some experience you will see how powerful this technique can be as well as how fast it can spit out renders especially if you get familiar with the FG and GI optimizations available from MR. If your renders take too long when using GI it means your doing something very wrong. The fastest way to render with MR and keep your visual quality is by exploiting FG+GI together. Combining these two will always be faster at rendering stills and animations than by rendering with GI or FG alone and is the best (and fastest) way to render an animation without flickering and artifacts. Hope this helps you somewhat. If you have any other questions feel free to start a thread or just reply with a new post and ask away. Take care.

[Edgeloop]

Hey Jak_Carver,

That is a really good explanation! :)

It really makes more sense to me reading the way you proceed on illuminating your scene...

It was rather confusing for me, reading and trying out tutorials without knowing exactly what's being done or without it being explained well...

Thanks man! :)

[Jak_Carver]

Hey Jak_Carver,

That is a really good explanation! :)

It really makes more sense to me reading the way you proceed on illuminating your scene...

It was rather confusing for me, reading and trying out tutorials without knowing exactly what's being done or without it being explained well...

Thanks man! :)


Yeah I agree, there arent many good tutorials for MR out there; sometimes I think that good techniques in MR are being kept secret or something, seriously!

The only tutorial I have seen that actually teaches things correctly but does leave alot out, is the Gnomon MR DVD. I can tell they know exactly what theyre talking about, but they do hold ALOT of information back. Other tutorials, like the one from Digital Tu-*cough* are so bad, they give out false information! I work for a college in the animation program using Maya, and a student showed me that tutorial and I was speechless because of the misleading info and horrible techniques the video was "teaching". I promptly had the student turn it off and I sat down with him for a few weeks and showed him the correct way of doing what the tutorial was misleading him on, and I filled in the gaps the video was leaving all over the place. Another strange thing is that I have never seen nor heard of any tutorial anywhere that teaches how to use MR's shaders in Maya, Max, or XSI, which is really strange. This helps me believe that this information is being kept under wraps for some reason. I would make a tutorial on all this but then maybe some men in suits would show up in the parking garage at night and no one would ever see me again lol. :eek:

[CTZn]

lllol Jak, you are a sick paranoïd alien ! There is some tenderness in these words, dont misunderstand me !

I will try to give my opinion on why there are not many good tutorials on mr (btw there are some very good ones, most of them free). Then I'll add my two cents on how to setup quickly GI. Wish's method may be good if you know already what the behaviour of light should be. You may know that without prior knowledge of cg techniques, its a matter of observation I believe. And then you'll forgive me for my average-not-so-correct english ;)

Jak, your job is to teach, isn't it ? Well, that pretty much explains why, take care.

Joking :D Most of the people learning mr hard do so because they earn what they need that way. They learn and they apply knowledge immediately, in order to satisfy their clients even more. It results that they have no time, and no interest (to bring more competition over the market and) to teach people who are lacking the basis to understand what they could teach. So they learn and work and at the end they are very good at that. They just don't care about teaching, and maybe you're right, in some cases they care not to teach.

In the other hand, Jak, your job is to teach, even if it has to be for weeks (should I say for months ?), you told it... Your concern IS that people understand what you know, as a teacher you are very sensitive to that.

Besides that, most of us here (I may be wrong though) are using mental ray for Maya, wich I believe does not allow as much control than mental ray Standalone does. That is where the other part of missing infos are, according to me. You'll get that point if you read all the e-docs shipped with Maya (I guess you did already). That's it I think... btw, where's your parking garage ? XD

Next part !
Regarding Jak's workflow wich I believe is the one to follow, I'll talk shortly on setting GI up.

At the very beginning you don't want to have many photons at all, almost all of it reside in their intensity, not density. So start with an accuracy of 1, and a number of light-emitted photons at a point where you can see some tens of them, or say some hundreds. If you see none of course you'll need to cranck their energy up. But not as much as the are just white dots, you want the biggest part of them to be smoothly evoluating from light values in their center to dark values on their edge. When that part is done thats it for the tough one. Then increase number of photons/accuracy so they just cover all of the surfaces, then just double the values and you're done with GI, pretty high settings in hand ;)

[edit] Jak, what's your gamma screen setting ? All your images look very dark to me !

[bertrand]

thank s jacks .
It s a point that there is some tutorial that work but you don t know why you do things.The trouble i think is that for maya there is no good documentation about mr and all the node.I will really enjoy to get some explanations, and examples, on how to use shading mental ray node properly!
about the gamma i think that mental ray has a gamma of 1 so too adjust it you can do it in a retouching programme (or in Fcheck) if you image is a 16 bit. I thinkt it s not recommended to do it in maya. but...
bertrand

[CTZn]

I'm not sure there is a counter indication using gammaNode within Maya... Jak ?

[Jak_Carver]

I'm not sure there is a counter indication using gammaNode within Maya... Jak ?


There are many ways to address this, but my technique is to render out my animation or still in openEXR format and then use the miDisplay utility from MR to correct my gamma to whatever I want. I use openEXR because it gives the best full high dynamic range images; even better and more accurate than regular.hdr format. In my workflow I tend to use XSI Illusion to do this quickly and easily.

[Wish]

2 Edgeloop, Bertrand: Ok, guys. Look at this. How much time do you spend to wait for GI calculations? Try to use color volume lights to simulate color bleeding. That's all! There is no need to wait for GI. Use your art skills and make a candy without GI.

PS: Or you, guys, have a lot of free time?
________
Asian live (http://www.girlcamfriend.com/webcam/asian-girls/)

[CTZn]

my technique is to render out my animation or still in openEXR format and then use the miDisplay utility from MR to correct my gamma
Thank you for the tip, I never used openEXR but gotta check that !

[Jak_Carver]

OpenEXR was developed by ILM and is a fantastic and powerful high dynamic range format for images. MR is great because it can render your images out in this format.

[Wish]

Renderman olso can render out in this format)))
________
red headed Cam (http://www.girlcamfriend.com/webcam/redhead-girls/)

[bertrand]

to wish . i agree with you about the fact that gi is not always usefull. but i m teaching maya at part time. i try to properly understand how work MR .
About open exr i think that jashaka (the free compositing software) can composit it.
jack could you please see my radiance thread?
thanks