PDA

View Full Version : Some questions for you experienced cg lighters!!


manuqc
01-12-2012, 03:40 AM
Hi There,

I ve found lately in me a growing interest towards cg lighting and shading.
As a short description, I always liked a bit of everything from modeling, texturing to final retouching..etc
So I guess I am an environment generalist as you might call.
What I like from lighting and shading is that I feel it is the part of the "process" where things
start to tight up and take life if you can say so...

Although I find my lighting workflow a bit disordered imo and thats why I am coming here to ask for your help!

I might say that I have production experience modeling, texturing & shading but not lighting, yet. Although is what I practice
the most at home lately (also shading&comp) as I find it very technical, but still really like it.

Ok, so "my problem" is mainly how to deal with a lighting workflow and how to approach it (I am referring to the
vfx industry) I ve been following tutorials from master zap and jeff patton and they conseil to use always a
photometric workflow that pair well with mia materials to use MR as an example (ok..at least thats what I understood!).

also I might add that I read quite a bit about linear workflow and tried to get my head around it. I use softimage, 3ds max
and I am learning maya lately on my freetime (I find it closer to SI than max workflow wise). For the sake of simplicity
you can answer me focusing in SI and maya (as they are more used for this tasks in vfx than max)

Ok, so to be more punctual I'll ask some actual questions!


1)photometric vs non-photometric. How do you start the lighting for a scene?
(I am assuming you guys always work in a metric scale?) When I said that my questions are focused to vfx is because I know
in archviz you can get precise data about real lights to be used (alsp..ies profiles...etc) from the architects (so lets leave that appart
for now).But in film? How do you properly start lighting a scene "photometrically" without that information. Do you actually need it?

Or is the artistic eye enough? (This is a point where I am having a bit of trouble, because there is a part of my brain saying
go full photometric and another that says trust your eye. But technical vs artistic is like oil and water, well I guess thats the job of
a lighter, so how to achieve a good balance and have a steady workflow will be my question.


2) How do you deal with all this photometric, non photometric stuff across render engines? From my experience I know mental ray
and vray. But I ve been in work environments using arnold for example which I am not that familiar yet and not sure if same rules apply.
I dont expect lighting knowledge to be completely transferable, but is there anything in specific to keep in mind?


3)if lets say tomorrow I start a job using renderman not having use it at all. Is that even possible? I mean lets say you are a senior
lighter using solely MR. Can you still apply for a senior lighter job using another render engine?


4)Do you ever mix both photometric and non-photometric? I know master zap would melt in sadness if he knews about it...but do you?


5)How do you deal with photographic exposure. Now I know this is more a software related question, but in general terms what is your workflow.
Meaning, do you use photographic exposure in your 3d app as a reference and then recreate it accurately in your comp package?
(yes using render channels) Or do you set the exposure only in your comp package, if so what is your workflow (using nuke for example).


6) Do you use SIBL lighting technique in vfx, in production. If not, why not?


7) Something that mixed me up a little bit with maya and softimage when comparing it to 3ds max is that 3ds max has native "photometric lights"
whereas from what I ve seen SI and Maya have standard lights that need a photometric shader. I think this gives more liberty, but does
that mean that any light in Maya or SI can be used as photometric? (This might sound obvious but I better ask..anyways)


8) When lighting a scene with live-action footage were there are other "real" light sources involved, besides the sun itself, do you work in
an additive light workflow taking in consideration the lights used at the moment of shooting? or is cg light based solely on hdri for this?


9) Finally, This is really a software related question so if you want to skip it, no prob, hope its ok. Is there a way in SI to output
"MR ad output, levels & raw" render channels other that with a third party plugin at a shader level?. I am ok with this but
just wondering if there is a native SI way for this (yes like render elements in max).



I want to thank you all in advanced for taking the time to read and hopefully some you experienced lighters will like to share
some thoughts in regards to my questions. I'll be pleased to know your opinions.


Regards,



-Manu

Array
01-12-2012, 05:19 AM
Not sure about how "experienced" I'd consider myself, but here goes...

1) Trust your eye. Rendering software is a tool, plain and simple. Don't feel like you need to work within constraints which are defined by programmers. The mentality is oftentimes "this feature is computationally intensive and is being touted as the latest and greatest in rendering technology". Who cares? It's useless if it's not as flexible as your other lighting tools are and won't give you the look that you and/or your client/boss needs. If a certain kind of light is fine tuned for giving you a specific look, then great, use it when you need that look, but don't feel that you have to light in a way which is considered "correct" by someone who wrote the software by shoehorning that feature into everything that you do.

3) You might have a bit of a steep learning curve going from MentalRay to PRMan, depending on your pipeline and how much of the under the hood stuff is hidden from you, but you'd be able to pick it up. People move from different render engines/pipelines all of the time when they switch jobs, and studios take training periods into account when they hire people. Yes, you can apply for a senior lighting position at a studio which uses a renderer which is different from the one that you're accustomed to. The main thing is understanding the fundamentals, not which dials to turn.

Goranimation
01-12-2012, 06:32 AM
Your approach to the field seems a bit backwards. You seem very focused on technical this, photometric that, which is great to know, but that's not what a lighter does.

A lighter in film/animation is someone who uses light to tell a story the same way an artist uses a brush to paint. What lights you use all depends on what you're trying to do. Photometric lights are great for architectural contracts that require accurate lighting for clients, but in my limited experience is rarely used in film/animation as the main lighting tool.

What you need to understand first is how light works, and how to use light to tell a story. Anyone can grab lights, plop them in a scene and press render, a real lighting artist is able to craft the light and make it accent the scene elements.

Studying live action lighting is a great way to start. Look at how film lighters try and light their scenes, understand what a usual lighting setup works out, then apply that to 3d. There is very little between live action lighting and 3d lighting besides the tools.

Also understand the limitations of the tech, this is where understanding light comes in handy. Not all studios can afford the render times of full GI/FG renders so you will have to understand enough about light to fake the effects as best as you can, keeping render times in control.

There are a lot of books on cg lighting, Essential CG Lighting Techniques with 3dsmax is a good read if you're starting out, lot's of information to start from, but your studies won't stop there.

When I'm lighting a scene I usually Start with understanding the motivation of the shot, then taking that into account, figure out my key light placement, sometimes you need to take into account the natural lighting in the scene and figure out how you can play up the lighting to support the motivation of the scene. From the key light I figure out my fills (no shadow lights with spectacular turned off) to touch up the darker areas of the scene where I feel necessary (taking into account bouncing light and bleeding colors), then if I want to accent the form more throw in some rim lighting. That's a pretty basic break down of the process, and it obviously can get a lot more complicated then that.

Trust your eye, the final product is what matters, if it looks good and it looks right then who cares if you are using photometric lighting with the settings on max.

As for Exposure settings, I usually have this on and follow basic photography principles when it comes to setting exposure levels. Exposure settings let you play up the contrast on the lighting. If you want to understand exposure, get a camera that lets you play with exposure levels to understand how the settings affect the output.

Hope that helps, there is a lot to study, so good luck on your journey into cg lighting!

manuqc
01-12-2012, 03:49 PM
Hi! First off I want to thank you for taking some time trying to help me out! I really appreciate that.

Ok. I think that you both agree that artistic (eye for composition, approach) should prime over
purely technical. I also agree with you on this. And I am not saying forget about the technical
part but trust more your eye as you seem to propose.

"You might have a bit of a steep learning curve going from MentalRay to PRMan, depending on your pipeline and how much of the under the hood stuff is hidden from you, but you'd be able to pick it up. People move from different render engines/pipelines all of the time when they switch jobs, and studios take training periods"

@Array
This is good to know. Although I imagine that shall also depend on the studio. Meaning that probably a bigger studio have a better structure approching this. Whereas a small studio might need someone to produce, or try to produce faster without necessarily some time for training. (I am just assuming here)


Your approach to the field seems a bit backwards. You seem very focused on technical this, photometric that, which is great to know, but that's not what a lighter does.
I agree with you. And this is the main reason I asked for you guys help. I can say that I have a fair understanding of mental ray as an engine. But more towards archviz and product lighting.
(read, photometric)

@Ripshot
I guess my problem is that after following training from MR gurus (zapp, patton.. and I am using MR just as an example because is the engine I have spent the more time with) I assumed that in order to be "accurate" (i.e achieve photoreal look) I needed to stick with a photometric workflow.

And that sort of blocked me lately when trying to light some concepts at home not knowing if I should've simulate a real lighting scenario (sorry to say it again, but a photometric light rig) or to build light progressively adding fill light, key light...etc (and still have real accuracy, without necessarily being 100% technical)

When I'm lighting a scene I usually Start with understanding the motivation of the shot, then taking that into account, figure out my key light placement, sometimes you need to take into account the natural lighting in the scene and figure out how you can play up the lighting to support the motivation of the scene.


Yes understanding the motivation/mood is definitely the "most" important aspect I agree. And yes this might get lost/relegated if focusing too much on the technical aspect


As for Exposure settings, I usually have this on and follow basic photography principles when it comes to setting exposure levels. Exposure settings let you play up the contrast on the lighting..

I also follow camera principles when using exposure, but I was asking whether to apply exposure while test rendering or straightly in comp. And what is the common workflow in an environment studio between the lighting and comp departments to keep things tight.


Also understand the limitations of the tech, this is where understanding light comes in handy. Not all studios can afford the render times of full GI/FG renders so you will have to understand enough about light to fake the effects as best as you can, keeping render times in control.


Yeah I sort of thought about this but wasn't sure to which degree lighters use workarounds to
achieve nice stuff without crazy render times. I have solely lit stills up to know. So I guess
my goal has always been "photoreal as possible" without taking (thaat much) consideration in crazy render times, if the quality justified. But yeah in a production environment dealing with animation things might be a lot tighter as you said...

So once again thanks for your time, I find your answers helpful, and I hope more people want join and share some extra thoughts about their workflow.

Regards.

earlyworm
01-12-2012, 10:02 PM
1)photometric vs non-photometric.


One of the key requirements in feature film (in both vfx and animation) work is "art-directability", directors and supervisors will often ask for things which are not physically correct (and occasionally not even artistically correct). Using the real world is normally a good starting point, but when your asked to deviate from real world you need tools which are able to make these changes (and lot's of them). Typically your working in the in-between world of "physically-plausible" where things just need to look physically correct rather than be physically correct (if that makes any sense?).


3)if lets say tomorrow I start a job using renderman not having use it at all. Is that even possible? I mean lets say you are a senior
lighter using solely MR. Can you still apply for a senior lighter job using another render engine?


Depends on the company. Larger studios are likely to have more specific methods in place for lighting and there will be some initial training in bringing you up to speed with their way of doing things - so you'll be judged more on your ability to make nice images rather than anything else. Having familiarity with the particular software package they use is certainly a plus through.


6) Do you use SIBL lighting technique in vfx, in production. If not, why not?


Yes. Image based lighting is a very common technique, although it's also becoming more and more common to apply HDR images to 3D geometry (rather than spherical environment maps) in order to get spatially correct lighting environments.


8) When lighting a scene with live-action footage were there are other "real" light sources involved, besides the sun itself, do you work in
an additive light workflow taking in consideration the lights used at the moment of shooting? or is cg light based solely on hdri for this?

Depends on how direct or indirect the lights in the scene are - often you can get away with just a keylight/sun and an HDR, but other times you'll need to place a spotlight or more in order to get light in the right place. Also the sun will get occasionally get painted out of HDR images so that the sun can be repositioned to get more control over the lighting.

playmesumch00ns
01-13-2012, 10:28 AM
1, 2, 4 & 7) Don't really care about photometric lights most of the time to be honest. I've only ever needed to use them once. There are much more important things to worry about. The key to being a great lighter is knowing how thing *should* work in the real world, so when you break reality to "follow your eye" you understand what you are doing and why, rather than just pushing things around until you create an unusable light rig that just happens to work for the shot in question.

3) I wouldn't say you could jump into a senior role with no renderman experience (I certainly wouldn't hire you for such), but you should be able to pick it up reasonably quickly if you spend some time reading the docs, siggraph courses etc.

5) You bake exposure into the lighting. Your backplates will be balanced across the sequence by the comp team, then you just match your lighting to that. Further grading and exposure tweaks will be done in comp.

6) IBL is used on pretty much everything here.

8) You match the lighting in the plate as close as you can. If you're lucky, any hdri you have will have those extra sources in (because it'll have been shot on set). If you're unlucky they'll have changed the setup and you'll have to rebuild the rig by hand. Either way, as somebody else mentioned you'll need to chop up bits of the hdri to stick on area lights and/or geometry to get the correct parallax.

manuqc
01-13-2012, 06:28 PM
@ Will Earl & Playmesumch00ns, Hey thanks for taking the time contributing with your opinions I do appreciate it.


As a general consensus, I seem to understand that lighting should not be approached as if building a desk from ikea, but with an artistic approach, flexible & adaptable. Based on real world lighting knowledge.

Also, thats a good point, referring to art directability and the need to adapt to almost any demand
from an art director... So yeah I guess it would drive me nuts trying to use a full photometric workflow (that I understood its not explicitly necessary) and keeping up with constant changes.

About HDRI Lighting, thats good to know, I like using HDRs as well. Do you guys use the sibl plugin from hdrlabs or hand build your rigs?

--

And ok well, just to wrap my previous thought about photometric, I think my confusion started as I currently try to be fluent with three packages, 3ds max, softimage and maya,
(more specifically softimage and maya for vfx) and as we might expect there are some differences in light choices (naming), although these 3 packages have spot, infinite (directional) and point light types, I got to say that I learned 3ds max first so used it as main reference as I was more comfortable with.

What I mean is, master zap said there are some "legacy" lights in 3ds max we should never use. I do apologize if I seem to focus a lot on technical side for this but I am just curious to know what you think. I did a quick comparison image with some questions:

http://dl.dropbox.com/u/4138955/drop3d/lights_comparison.JPG

wanted to know if you have any thought about switching between apps.

You bake exposure into the lighting.
How do I do this, can you point me to a reference to read?

apply HDR images to 3D geometry (rather than spherical environment maps) in order to get spatially correct lighting environments.
I didn't know about this, where can I learn more about?

Thanks again for your time guys, I know I am asking a lot of questions and some might seem
very basic to you, but I better ask to learn a proper way of doing things... ;)

Bitter
01-13-2012, 08:43 PM
1. Honestly, I would rather work in radiometric than photometric terms. Radiometry is a bit more straightforward.

2. Similar things apply between packages. But keep in mind Arnold is restricted to BSDF materials and path tracing. Sometimes a bit harder to art direct, not unlike iRay. Vray and mental ray can use multiple methods. But when all else is equal, you can work in similar ways.

3. If you are familiar with how rasterization works then you have a background understanding of renderman, just not the specific tools. Here in LA at small to medium VFX houses you will find yourself using mental ray, Arnold, or Vray much more often than Renderman. And yes, some people will go from senior in one software tool to another somewhere else. Once you learn the basics, your knowledge will transfer.

They are hiring an artist, not a fleshy software package.

4. Try not to. playmesumch00ns pretty much answered that in line with my opinion and that of a few others.

5. Stop using your 3D package as your entire pipeline. Those packages are not designed to do what Nuke/Photoshop/Lightroom are designed for. Save yourself much pain by using the right tool for the job. :-)

6. We use HDRs constantly. sIBL, the specific tool? Not been anywhere using that yet but I hear good things.

7. 3DS Max lends itself more towards Arch Viz. That's why the tools tend to differ. You have a different core audience and the fringe audience tends to be mixed. I would not worry about the "why's" details between packages but more how I can get a desired result from different packages.

8. HDR when used with something like FG is good for fill lighting. It does not generate actual shadows etc. Using something like native IBL in mental ray, etc then you can use it to also generate the set lighting and shadows (if the data is good). Then you can use other lights for artistic purposes.

9. You should use an additive workflow, avoid "level" and "raw" passes. They will eventually (hopefully) go away. You cannot match your beauty render using those in a post package.

-diffuse
-indirect
-reflection
-refraction
-additional color
-scattering
-translucency
-data passes

Rely more on that (indirect has your AO and diffuse has your shadow)

playmesumch00ns
01-14-2012, 03:07 PM
I think we might be talking about different things when we say "exposure". I simply mean whether the image is bright or dark overall, like a photographic exposure. I don't know much about mental ray but I understand it has some kind of "exposure" thing which is essentially a tone-mapping thing to mimic a filmic look. If that's what you're talking about, then what your'e really talking about is grading, and this is normally done in compositing, in nuke or a similar tool.

As for the geometry/hdri workflow, there was a nice video of a workflow using mari kicking around. I think it's this one, but youtube doesn't seem to be working where I am so I can't check:
http://www.youtube.com/watch?v=1d8ypguQjFw

manuqc
01-14-2012, 06:29 PM
Hey guys, thanks again for your answers.

5. Stop using your 3D package as your entire pipeline.
I am realizing that maybe one of the problems I have to help me understand the "whole thing"
is my lack of knowledge about compositing. I am still learning it but I am relying more on my 3D software atm. Hopefully I'll start compositing my first images pretty soon, and post them here, then things will get more clear I believe.

Regarding passes (render channels) I was a bit doubtful on which channels to choose for compositing.

-diffuse
-indirect
-reflection
-refraction
-additional color
-scattering
-translucency
-data passes

Thanks David, this will help. I mean that list is a generic base for any render engine I imagine. For mental ray I use only AD & design output (result).

But what I asked about softimage (question #9) was because I could not find a channel output for AD as in max. (without using an external shader)

http://dl.dropbox.com/u/4138955/drop3d/imgchannels.jpg

work in radiometric than photometric terms
Yeah this makes sense!!

I simply mean whether the image is bright or dark overall, like a photographic exposure.
Yes I was referring to the same thing (the other are some extras in the MR exposure control, which I avoid, things such as saturation, vignette, shadow, contrast...etc. I do those in PS but will start to do them in nuke instead)


I can't check:
http://www.youtube.com/watch?v=1d8ypguQjFw

Hey this is amazing, I mean I did some photogrammety modeling some time ago but wasn't aware about this ptex-mari workflow... I mean it looks awesome... sigh* man too many things to learn... Ok thanks for sharing.. I guess one thing well done at the time.

Again thanks for helping me out. I learn a lot from your answers.

I guess the best way to keep learning is to start posting some images for feedback.

earlyworm
01-16-2012, 10:21 AM
The first example on this page gives a brief rundown on the idea of using HDR images on geometry. It also shows the difference between using geometry and a spherical environment map in a scene - notice where the buddha passes the tv and moves towards the lamp.

http://www.glyphfx.com/examples.html

You don't always need to go that full on. You can also A) project the plate onto scene geometry (which matchmoving or modelling occasionally provide, but you may have to do yourself) or B) extract the lights from the HDR and place them on cards in your scene.

Worth a read is the course notes from Ben Snow's talk about lighting at ILM. It should give you a reasonable idea of what the lighting tools are like at some of the bigger studios.

http://renderwonk.com/publications/s2010-shading-course/snow/sigg2010_physhadcourse_ILM_slides.compressed.pdf
http://renderwonk.com/publications/s2010-shading-course/snow/sigg2010_physhadcourse_ILM.pdf

manuqc
01-16-2012, 02:21 PM
Hi Will,

Those docs look right on spot, I ll make sure to look those in detail. Thanks for sharing!!

-Manu

CGTalk Moderation
01-16-2012, 02:21 PM
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.