Max 5: Image Based Lighting with HDRI?

[Dave Black]

1. No, and some.

2. That's right.

3. You have to use a skylight to do it. Skylights are not actual lights but ratraced pixel lighting. If there is a map in the light, it will project. You have to blur you map a bit too using the blur feature inside the bitmap map. The tech behind a skylight IS how HDRI lighting can be achieved. Also, you have to have a very high sample rate and/or a wide sample filter(advanced lighting) to really get a good HDRI based render.

4. Create 2 maps. One HDRI with an appropriate exposure level(from the source light probe). One copy of that HDRI map with a slight blur. You place the blured one in the skylight, and the non-blured one in the enviroment channel of your raytrace material.

5. The best maps are from debevic's site. Ususally, there is no black area at the bottom of these maps, though it is possible through your coversion to FPTiff files, that you either transformed the image wrong, or that you are seeing the camera/sphere/ tripods.

Whew...Oh, and don't forget that these images contain many exposures, so you need to tweak that according to the scene. You can do so by adjusting the RGB intensity amount in the output section of the HDRI map in max.

I know there's gonna be more about this, so lemme know what made no sense.

-3DZ

:D

[sledgeweb]

As always, you're very helpful. Thanks once again for your time and attention.

Most of what you said makes sense to me. I was glad to see that some of what I was doing was right. I hadn't found a good tutorial for default MAX rendering. So... I kind of made it up as I went along. Sounds like I'm on the right track.

I do have a couple of follow up questions though :) !

1) What do you mean by "and some". Just... you REALLY don't need it. Or... you don't really need it, but it has some helpful features.

2) Thanks for confirming.

3) I'll have to play with this when I get home. I've never messed with the sample filter and rate really... what's a good ballpark figure? I'll run some tests... but a good starting place would be helpful.

4) Makes sense to me. I read somewhere that image based lighting wasn't possible in MAX 5. Is this not true? If I place a large plane in a scene, and use image based lighting with a skylight to light it, will I be able to see the dark/light contrast in the lighting of the plane? (I'll try this when I get home)

5) So... the black area is unecessary? Crop it out? Or should it be there if you are creating a ground plane? Seems to me perhaps the black area represents "below the horizon". Therefore, say your hdr is of a forest, you wouldn't want the bottom of your sphere reflecting the bottom of the trees. I'll post a pic of the HDR I'm using in a bit.

Again thanks for your help. After your help with the Sub-D, I'm working on something that is far better than anything I've previously done. What seems to happen over and over is that I learn one thing, and as I progress with it, I run into another area where I need help. In this case, my model is looking good, but I wanted to render it with HDRI to get an idea of how the reflections on it look. This opened up a whole new can of worms...

OK, I'll post more details later, but you've already helped my tremendously. Thanks so much.

-Sledge

[sledgeweb]

Example of HDR image:

http://www.cubit.net/3dwip/ststefan_innen1.jpg


Appears to me, the black area represents the horizone and below. You wouldn't want, for example, the columns to be reflecting on the bottom of the sphere... unless your sphere was placed near the ceiling. Assuming the sphere is resting on the floor, you would want the base of the columns to reflect around the middle of the sphere. Now, if I crop the black off and place the sphere at the horizon line, the base of the columns would reflect on the bottom of the sphere, which wouldn't be realistic.

Perhaps these are just bad HDR maps? Because they don't really have the floor below the sphere in them? Although, if you have a ground plane in your scene, your sphere should relfect that and not the black...

-Sledge

[Dave Black]

Ok, first off, Max 5 can do image based lighting out of the box. Anyone saying it's not possible is full of poo. It does take some messing about, but it's completly possible.

Second, a good HDRI light probe will have the ground in the image. It's part of the reason you use a chrome globe. It picks up 360degrees on both axis. That black should go.

Let me clear this one point up for you. Image based lighting using HDRI is not just a spherical map of pics of a chrome globe. When you go out to take an HDRI image, you take multiple pics of the image with different exposures. This allows the image to contain more lighting information than can be contained with a normal image. If you use HDRI shop to make you image tranlations to FPtiffs, open a HDRI image. Press shift- + and shift - to increase or decrease the exposure of the image. You can see that the image holds more data than meets the eye. HDRI based lighting(or image-based lighting/rendering) is when you take a high data rate image and apply it to a sphere. The sphere then uses the pixels of the image to broadcast light rays. These rays can be allowed to bounce around, but ultimatly, it reproduces the light of the HDRI image. The skylight is an image-based lighting system, but at it's default the image is colored 100% white, or partially blue. That's what make things look like GI. GI is image based lighting.

If you want light to emmit localy instead of globally, like you mentioned with the plane with an image on it, you have to use Radiosity. This is where objects selectively and locally emit light based on color temp and other factors...which means an image on a plane will emmit that image as light particles, much like GI, but a bit different...actually much different, and that will have to wait for another topic.

As for learning new stuff as you work out other things, it's how we all learn, so that means you are making progress! Yay!

I'll check back soon.

:edit: you really don't need the plugin, but it's got some nice features. Just check it out, it's free. I personally don't use it, but that's because I'm rather happy with just using the functions in max.

-3DZ

:D

[sledgeweb]

Thanks for the response. I've read up a lot on HDRI images, so I understand how they are made and how they allow ranges far greater than 0-255 using floats instead of integers. I think I was confused on the image based lighting...

I thought, using image based lighting, the light areas of the image would cast more light than the dark areas of the image. If this isn't the case, what is the advantage of using image based lighting in a scene? If the brightness of the light is going to be the same from every direction, global I suppose, then why use an image at all? Why not just use the 100% white default?

I suppose the difference would be color? So my next question would be, if you had a HDR that was... say half yellow and half blue, and you used it for image based lighting on a solid plane, would you get a globally lit GREEN light, or would have the plane be lit in yellow and half in blue? If the skylight just mixes all the color and creates a global green light, it seems simpler to just change the hue of the skylight to green rather than use an image.

Does this make sense? I thought the purpose of image based lighting was to get a light setup based on your image map. So if you had some bright lights in your map, those would project light into the scene. Same with color... if you have a green tree, I would assume it would bounce green from it, and blue from the sky if you have a blue sky.

I have a better understanding now, but one more resposne should clear everything up.

Thanks,
Sledge

[sledgeweb]

The following is where I read that MAX can't use image based lighting on it's own:

At this point, the only thing you can do with the HDR TIFF image in 3ds max, without a third-party renderer, is use it as an environment map. It will not, without a 3rd-party renderer, be a light source that will illuminate your scene. 3ds max r6 says it supports .hdr images, and I’ll have to wait and see what is all supported. In 3ds max with the RenderPipe systems, you can use the HDR TIFF file to illuminate your scene using the RenderMan shader “Skylight.sl”, which is supplied by ART VPS. In order to look correct, do not use the Light Probe versions of HDRI. They only include information from one camera position, and hence do not have light and image information from the back. A panoramic HDR image is required.

This comes from:
http://www.cgarchitect.com/news/HDRI_IBL.pdf

-Sledge

[Dave Black]

What that person said is somewhat false. I don't know how else to say it. The HDRI in max is a bit of a fake, as instead of being based on photons, it's based on raytracing. The technology in R5 can reproduce the effects you are looking for, but it's not really true GI. It's "Light-Tracing". Eh. I'm confusing myself now...hehe.

Your second remark was also based on a incorrect assumption, so I won't address it. GI is taking into consideration the lighting of the enviroment map. Try taking a FPtiff file, and converting it to a standard .tiff file. Once you have setup a scene with a normal .HDRI image(of the same source) try swapping it out and you will see the difference.

It's a hard call. HDRI in max works. Some of the higher level functions may not be apparent, but all in all, the output is the same.

If you are currently unhappy with max's scanline HDRI, then perhaps brazil or VRay will catch your fancy.

-3DZ

:D

[sledgeweb]

Well... it's not a matter of happiness with the renderer, it's more just me trying to wrap my brain around something I've been studying and trying to understand how it is implemented. I don't have a NEED for it right now, it's just another bit of knowledge I'm attempting to add to my arsenal. I guess I'm just trying to get a good feel for how MAX works in this area and what it's capabilities are. I won' bother with a 3rd party renderer right now, as I'm trying to focus on learning MAX. I do have access to a machine with final render on it, so eventually I may explore that. Right now, I just want to work with MAX and figure out how to do the best with what it offers. Perhaps MAX 6 has some upgraded feature useful in this area... have read up on it yet.

I'll run some test tonight if I have time and post some results.

Thanks again for your answers. I certainly have a much better grasp on things now.

-Sledge

[Dave Black]

I guess It's kinda hard for me to answer some of the more technical apsects of your questions because I simply have not figured out exactly how the new system works, but I have gotten good results.

Lemme stop talking about that now, and maybe we can talk about those samples you were asking about before.

Simply put:

Samples = Accuracy

Accuracy = better image, less blotchy, etc

More samples = longer render

So, if you decrease your samples to say a bare minimum of 15(from the default of 250) you will get a fast but very ugly render.

250 is probably the max I'd do. It's pretty darn high. I usually start with 5 samples and no AA with a filter size of anywhere from 1 to 5. It's a balance. More samples there are, the smaller the filter can be. The less samples there are the bigger the filter. See the filter sorta blurs out the samples(practically speaking, though it's actually not really). So if you have only a few samples and a small filter, the mesh will look blotchy. If you open up your filter to say, 5, it will smooth those blotches together. I personally cannot detect a difference in speed due to changes in the filter size, but I can't be 100% sure until I run further tests.

I'd say for fast renders try this combo:

No AA(big speed improvement), 5 - 30 samples with a filter size of 1 - 5. For nice renders, to final renders, you'd want 75 - 100 samples with a filter of 1 - 2.

See the balance? You've gotta place it by eye, though, as 250 samples is often overkill. I find it hard to distinguish between 100 -150 samples and 250 - 500 samples(which again, I feel are insanly high).

Just some thoughts.

I too am currently putting max's new render tech through it's paces, so I'd be really keen to learn your take on things.

-3DZ

:D

[Dave Black]

Oh, and one more thing. Try turning on the "show samples" option while playing with the values for the first time. It will make alot more sense as to what the heck is going on with them. Oh, and the documentation(online and printed) is really great on the subject.

:edit: perhaps you'd also be interested in sharing some test files. I've got a few that might help. Let me know.

-3DZ

:D

[sledgeweb]

Great info. I'll run some test tonight and see what I come up with. What kind of test are you doing when you say you are putting the renderer through it's paces?

BTW - how are things looking for the tutorial vids?

If you want to send some test files my way, no problem. I can run some renders and see what happens.

Later,
Sledge

[TheWriter]

Here is an hdri tutorial I did back in my lightwave days... may help.

http://www.highpoly3d.com/writer/tutorials/hdri/hdri.htm

[ToddD]

3DZ covered it all very well as usual!!
Anyway here is a link to a Tutorial by Pete Draper, it addresses the exact issue you have asked about(Hdr/light tracer), and produces some great results! Pete is published in a lot of the 3D publications out there, and all I can say is I have learned alot from the tuts I have seen on his site, and in print. Good luck!

http://www.xenomorphic.co.uk/doc_hdri.htm

[sledgeweb]

Great tutorial TBonz.

Unfortunately, I didn't have time last night to do a lot of testing. I intend to set up a scene similar to that shown in the tutorial TBonz linked to... a few simple objects with different materials / reflectiveness. I did make two renders of an object I'm working on, one with the HDR in the skylight, and one with the default skylight setting of 100 and white.

Got this:

http://www.cubit.net/3dwip/fengConcept/engine_IBL_test1.jpg

The top image uses the default white skylight at 100. The second image is exactly the same, except I put a blurred version of the HDR in the skylight.

Also, if someone could clear something for me on that posted tut. How do you instance copy a material to the environment or skylight map? I thought you could drag and drop, but it won't let me drop a material into those tabs. So.. I click the tab, go to the material library, select SCENE and load it that way...

[Dave Black]

You can't drag and drop a material into these slots. You drag and drop the map instead.

More later...

-3DZ

:D

[Dave Black]

Oh, I forgot to mention, and this may be what is causing some of your problems:

MAKE SURE YOUR MAPS ARE SET TO SPHERICAL ENVIROMENT MAPPING!

-3DZ

:D

[sledgeweb]

They are set to spherical.

Haven't had time to get back to this yet. Probably not until Saturday. How about you?

-Sledge

[Dave Black]

Ok, sledge...

Myself and Equinoxx have been doing some HDRI tests, and after some rather startling dicoveries, I thought I should post what I had wrong.

My biggest mistake, was saying not to use the .HDR import plug from splutterfish. From our preliminary tests, floating point .tiff files simpley don't work for lighting...Relfections, yes, lighting not so much. I have not completed the testing, so I don't have a conclusion as to why, though it would seem somewhat obvious that a true .hdr would yield more predictable results.

I'll post more later, but try using that import plug. It will probably make a pretty big difference.

-3DZ

:D

[ToddD]

I agree the lighting seems a little "bland" when using the Tiff as opposed to the hdr from the splutterfish plug! Just wanted to verify, I had played with this stuff a while ago:beer:

[sledgeweb]

OK, this is driving me nuts. I'm using max 5. I downloaded the hdri plug in, and unzipped the contentd to c:\3dsmax5\plugins\hdr

The plugin doesn't load on startup. I can load it manually using the plug in manager. But I don't get the HDR option in my file type drop down list when I go to open a bitmap...

Any help?

SORRY, figured it out. Had to copy the right plugin file to the plugins folder... heh... ok, off to do some testing!

Thanks,
Sledge

[sledgeweb]

Here's a quick test I did....

The top image is with the blurred hdr file in the skylight map. There is some blotchiness on the plane. Sample size is 300 and filter is .5 (lowest possible).

The bottom image is with the default skylight, no image in the map slot.

My question is, does MAX actually light the scene based on the bright areas of the hdr? So that if I had an hdr with bright lights in 3 different directions, would I get three shadows cast? Doesn't look like it, but can't really tell...

http://www.cubit.net/3dwip/fengConcept/hdriTest01.jpg

How's your experimenting going, 3DZ?

-Sledge

[Dave Black]

We are going to figure this out once and for all.

I'm going to contact discreet about this one.

I'm also going to DL the VRay demo and see if a different renderer is more capable, or if it's just us...

Be back soon...

-3DZ

:D

[Dave Black]

Ok, ladies and gentlemen...

I'm currently running some tests using the *gasp!* scientific method.

I will post detailed results comparing max's Light Tracer utilizing HDRI, and Light Tracer utilizing standard 8bit maps.

I will post the results in an hour or so...

-3DZ

:D

[sledgeweb]

OH BOY, the scientific method...

So, eh, what is your hypothesis? :)

I'm eager to see your results. I've gone from thoroughly confused, to somewhat grasping the idea, to pretty sure I've figured it all out, back to thoroughly confused.... ugh...

-Sledge

[Dave Black]

HDRI Test 1

The goal of this test is to determine whether or not MAX R5 actually processes extra lighting information held within a High Data Rate Image.

The tests were conducted with the following testbed*:

Default Max Scene
1 Skylight
1 Teapot Primitive
1 Camera
1 HDRI Light Probe: grace_probe.hdr
1 LDRI Light Probe: grace_probe.hdr, coverted to 8bit .tiff using HDRIShop.

Notes:

Mapping for both probes set as follows: Environ -> Spherical Enviroment.

The Tests:

Set 1

1. Each lighting probe was placed in the Skylight's Map channel, then a default material applied to the teapot primitive and rendered.
2. Each lighting probe was placed in the in the enviroment channel of a default raytrace material and then rendered with all scene lighting off.
3. Each lighting probe was placed in the Skylight's Map Channel, and in a default raytrace material's enviroment channel and applied to the teapot primitive. Scene was then rendered with scene lighting on.

Set 2

*tests repeated as in Set 1, but with both light probe's RGB Output setting at 3.0(up from the 1.0 default).

Results:

The results conclusivly show that the .HDR Light probe's added bit data(exposure) is being utilized by 3DStudio Max R5. Further tests regarding Floating Point .tff files, as well as varing differences in the HDRI to LDRI are forthcoming.

*see images for further details and results


http://www.3dzealot.com/cgtalk_helpdesk/LT_Setup.gif

http://www.3dzealot.com/cgtalk_helpdesk/Scene_Setup.gif

http://www.3dzealot.com/cgtalk_helpdesk/Test-Setup-1.jpg

[Dave Black]

And the results:

Top to bottom on each result sheet:

With lighting only, With Reflections Only, And lastly with Both.

http://www.3dzealot.com/cgtalk_helpdesk/Test-Results1.jpg

http://www.3dzealot.com/cgtalk_helpdesk/Test-Results2.jpg

-3DZ

:D

[Thomas V.]

After reading this topic, started experimenting with HDRI too...

My problem only is this:

http://www.studio-venema.com/images/HDRI_test.jpg

One way or another I can not seem to render multiple objects with the same material applied to it, Is this the limitation of using HDRI or am I using wrong settings?

Im rendering with Brasil R/S, and using the HDRI-plugin, HDRI enviroment is set to Spherical, same goes for the blurred Skylight-map used in Brasil's Luma Server.

This is my first post, please be gentle :)

[Dave Black]

Thomas V.:

Welcome to the forums!

Yet another gent from the Netherlands I see. We've quite a few of your countrymen running about the max forum. Great guys, all.

Marcel, Magicm, and the mighty Equinoxx to mention a few.

Anyway, I think you should post your question in the Max Forum as a seperate topic. This will ensure that you get a speedy answer. This topic is more about using Light Tracer in max to get good hdri effects. Plus, that's quite an odd problem you've got there. Might need to pick at the brains of the whole max community.

Again, welcome aboard!

-3DZ

:D

[sledgeweb]

3DZ-

Good test. Looking forward to the additional tests.

However, your setup here fails to answer one question that is bothering me. Or, maybe I'm just failing to read the results correctly... I dunno...

But... one thing I'd like to figure out, is can you use an hdr with a specific light source in it, say a lightprobe with ONE window in it, and have the light cast in your scene from the probe as if it is coming from the window? That is, would the HDR cast more light from the window in the lightprobe then other areas, resulting in a shadow being cast from the window source direction?

I haven't been able to get this to work on a non-reflective object. Let's say you have a lightprobe with window in it, and a wood floor in your scene. Can you get the floor to be lit brighter towards the window than further away? Seems possible since the skylight is a "dome" around your scene...

If you could test this, let me know what you find out. I hope I make sense...

It seems to me, the HDRI works good for reflective materials, but not very well on non-reflective materials. Can you get good results using non-reflective materials?

-Sledge

[Dave Black]

Hey, sledge.

Ok, first off, study the first test result sheet a bit.

The first .HDR test is simply lighting. There is not reflections. If you notice, the handle on the teapot lid is casting a shadow. The main lightsouce in this lightprobe is a set of windows, but there is one that is much more intense than the others. This is where the main light is coming from. I'm currently searching through my probes to see if I have one with a few windows to see if we can get multiple shadows. That's the part that I'm wondering about.

The top image that I was just refering to, is 100% how an HDRI image should be lit. It has directional light, with a soft area lighting effect. I think the main problem alot of people are having is that they are not using a wide enough filter. That's what will give the subject a soft lighted look. If you use the defaults (.5) with 250 samples, you get a pointilized mess. Even at 500 samples the effect remains. Only after increasing the filter size does the lighting actually soften. I'm going to do a test right now to see if bumping the samples up to 1000 gets around this...glad I just got a new machine...hehe..

[Edit:] Don't forget too, that bluring the .hdr probe a bit will help with this effect somewhat.

BRB...

-3DZ

:D

[sledgeweb]

Good info. I'll have to bump up my filter size.

Including a ground plane to receive shadows in your test might help to better see how shadows are being cast.

Keep up the good work.

-Sledge

[JeffPatton]

I have a HDRI with two distinct light sources (windows). I tried it with light tracer and although the two light sources are not on the opposite sides of the room, they are far enough apart to see two shadow's in this render.

http://www.pixelperfectgraphics.biz/CG%20Posting/Shadow-2x.jpg

cone angle @ 33:

http://www.pixelperfectgraphics.biz/CG%20Posting/Shadow-33.jpg

[Dave Black]

Did you try closing down your cone angle? I think you should set it all the way down to 33degrees. Try that, and then let us see the results.

-3DZ

:D

[JeffPatton]

I had it at the default 88, will change it to 33 now and re-render.

[JeffPatton]

Here's a close up. I changed the material on the vertical boxes to the same matte material I have on the plane to make sure I'm not getting any kind of feedback from reflections, etc..

I wish I could make the shadows more pronounced.

http://www.pixelperfectgraphics.biz/CG%20Posting/Shadow-close.jpg

[Dave Black]

Is it possible that the shadows are how they should be?

I wonder what would happen if you increased the intensity (exposure) of the .hdr probe?

Give er a try!

-3DZ

:D

[JeffPatton]

Would i do that with HDRshop? or in Max? I tried increasing the RGB output and it just made it more "orange" and the shadows looked burnt.

[Dave Black]

It will take alot of mucking about with levels if you want to get rid of that unclamped glow in the HDRI file.

Remember that HDRI images provide only the "Fill" component of the lighting setup. A secondary "Key" light is needed to fully flesh out the render.

Notice how all GI renders, HDRI and standard, completly lack the specular component of light. This is why more lighting is needed to really make it work.

One last thing I want to point out before I leave work, is to remember that you are building layers of light components on top of each other to get the final photo-reaistic effect. The diffuse properties of a material are just as important as the light. You can light a teapot and some boxes untill the cows come home, but you are still only rendering light and shade. I made some images real quick to show the power of using all the components together. To everyone reading, stop getting discouraged when your light setup is not giving you the results you want. Remember that all the old materials stuff still matters.

HDRI is a great tool. This is where the image would be without it:

http://www.3dzealot.com/cgtalk_helpdesk/HDRI_Test504.jpg

And here it is with HDRI:

http://www.3dzealot.com/cgtalk_helpdesk/HDRI_Test503.jpg

Big difference, no? Anyway, this has been really fun. I'll be back online tomorrow morning.

Keep testing!!!!

-3DZ

:D

[JeffPatton]

"Remember that HDRI images provide only the "Fill" component of the lighting setup. A secondary "Key" light is needed to fully flesh out the render."

So when you setup a HDRI scene, you use a regular lighting scheme plus the HDRI for fill? I didnt think you wanted the "regular" lights to do anything other than affect the specular highlights of your objects.

Please ignore this if you feel this is going off topic. I know the topic was more-so how light tracer handles HDRI files. I originally replied because I was wondering like you, if a HDRI with multiple windows would cast multiple distinct shadows.

I think I need a better HDRI file than what I tested with. I have purchased a glass globe as well as a fish eye lens for my camera and plan to make my own HDRI's soon (I hope!)

[Dave Black]

Seriously look at really high quality lighting setups using HDRI. The HDRI/GI portion of the effect is mainly the fill component. You have to have a key light to provide the harsh shadowing and specular components, as well as add more directional light. You've seen for yourself how HDRI lighting affects the scene. I think you could push that shadow effect you are looking for with a good bit of tweaking of the exposure settings, but I'm wondering if the point is almost not worth the effort.

I too have been bit by the HDRI bug. I'm looking to purchase one of the spheres debevic uses too. Looks like we will be moonlighting at photon theives...hehe...

-3DZ

:D

[JeffPatton]

"but I'm wondering if the point is almost not worth the effort"

I agree. If I start putting lights in the scene to match the HDRI light positions, of course I will get shadows. So, no point in tinkering with that anymore.

I cant wait to try and make my own HDRI's. The 12" ball and lens and such will be here later this week I hope.

I will revisit some of my hdri scenes and add lights and see what I get. Thank you for your advice!

[Dave Black]

While we are sorta OT, I've got a few questions for you since you've already done alot of the leg work on building an HDRI capture system:

First off, why the fisheye lense? I was under the impression that the sphere did all the distorting, and that a standard lense was used for capture...

Also, having read about their sources for the "gazing spheres", they mentioned a 3" mirror polished steel ball from an industrial supply company. Are you going with the more conventional garden globe, or did you special order the more accurate mechanical part?

And finally, have you figured out how you will mount the ball to the tripod?

Maybe too OT, but I think I speak for quite a few when I say we are interested.

Thanks!

-3DZ

:D

[JeffPatton]

This may get lengthy....sorry.

I purchased a 12" steel (chrome finished) "gazing globe". I was not happy when I saw the fine scratches on this thing. I have tried polishing them out, but they are still there. So if I get my camera close enough to where it fills up the picture, you see a slight haze from the scratches. Maybe I'm just getting too close to this globe. I have more tests to perform with it. The good thing about the steel globe is that it has a small hole in the bottom. I was able to thread this hole and screw in some threaded rod. This makes for a great stand (small footprint). Looks like a giant chrome lollipop! Yeah...my neighbors think I'm insane I'm sure!

The reason I have ordered a fisheye lens (and filter kit) for my camera is for the sharp details. I'm thinking the gazing ball will be ok to use for the HDRI itself. But for an environment map, I will use my camera and get a nice highres 360 panoramic image for reflections.

I was unable to find a polished steel ball (industrial). I would like to try that too....but I think I will wait for the Christmas season and just find some nice chrome ornaments for cheap! That should work too.

If I can get some decent results, I plan to post my HDRI's for people to use.

[CamillaGem]

Nice job with the tests, 3DZealot! Very logical method.

[xaoei]

I have been doing some experimenting myself with HDR Images and integrating them into my scenes, and while researching I came across this thread. I must say that this thread contained some very helpful information.

However, while reading I was somewhat confused at the following statement:

Notice how all GI renders, HDRI and standard, completely lack the specular component of light. This is why more lighting is needed to really make it work.
To me, it seemed that with your HDR render, the balls still had specular light? The LDR images seemed to not have the specular highlights, due to the fact that they were washed out because of the low dynamic properties of the image. Had the image itself been brighter, it would have created the illusion of specular light that you see in the HDR image. For example, I can up the contrast of the LDR image in Photoshop, and gain a similar result as the HDR render. Ex.

http://www.sciform.com/LDRadjust.jpg

This was done by taking your LDR rendering 3DZealot, and adjusting the brightness/contrast etc over those spheres to bring out the specular light. It's not as accurate as the HDRI render, and by no means a substitute, but I'm still trying to show that the specular light still exists, but does not display correctly in this instance because of the fact that the image has a LDR.

Maybe I am misunderstanding the fundamental concept of specular light entirely. I am a bit confused. :shrug: Would be interested in any responses to this.

Here's a close up. I changed the material on the vertical boxes to the same matte material I have on the plane to make sure I'm not getting any kind of feedback from reflections, etc..

I wish I could make the shadows more pronounced.

I think that the shadows are accurate. I am guessing that in the HDRI that you used, the light source was overhead or almost overhead? We all know that at early/late times of the day, shadows are elongated and more prominent. Maybe try finding a HDRI taken just before or at dusk to test if that could be why the shadows are not as pronounced as desired? The shadows should render much longer. If it's an indoor HDRI that you are using, usually light coming from a window (or multiple windows for that matter) does not define shadows beyond what your test revealed, unless the sun is actually directly shining through the window (which doesn’t usually happen until dusk/morning) - or - there is a strong light present inside. Usually harsh, defined, shadows come from lights that are in close proximity, and HDR images do not contain spatial data, so that's out of the question, and would define an instance in which additional lighting would be needed: Harsh shadows in the day time, when the sun is either overhead or not present.

The light coming through the window, if the sun is not present, I don't think is strong enough to cast the harsh shadows you are looking for - you have to remember the sheer intensity of the sun. Having a sun in your scene is going to create a much more defined shadows, due to how bright it actually is. Just because the window looks bright doesn’t necessarily mean it is bright.

Usually anything that casts harsh shadows like you are talking about is something that will be either the sun, or another light source present in your scene (I.e. a strong light at close proximity to the object you want to have strong shadows) and will facilitate an additional light external to your Skylight + HDRI setup.

3DZealot:

I do not see the need for additional environmental lighting setups beyond HDRI lighting. Everything seems accurate to me, according to the tests I have seen. I could be entirely wrong however, as I am still trying to fully grasp HDR. (Mandatory escape clause :p)

[Dave Black]

Interesting.

The "enviromental" lighting is important. Very much so. It provides the fill component of light. The soft, multi-colored diffuse light that is bouncing off everything.

Specular light is simply the form/shape/glow of a light's source. A light bulb would produce a specular highlight. In CG, lights are an effect, not an object. Thusly, the SHADER produces the specular component of light by faking the highlight. Specular highlights are simply reflections. Period. A varied surface with relfective properties will cause specularity.

HDRI has several principles that are important. First, the images can be used to project a broader range of brightness. Second, they offer a handy source for omin-directional reflective enviroments.

In theory, with the proper software, one could use HDRI to simulate any lighting setup. In reality, usually one must use HDRI to provide the fill component of light, and for reflections. One usually still must use some form of key light to provide crisper shadows and more directionally interesting shading.

LDR images simply cannot store the data needed to offer the highest range of color intensity. Ie, the cannot make a light that is above 255 brightness. With HDRI, you are using a floating point number, instead of an integer, allowing you to have a brightness in the negatives, as well as far beyond 255, and can be a decimal value. You can adjust the actual exposure intensity by increasing or decreasing the RGB output values within max. This will allow you to see that the HDRI images have greater color information, as well as allowing you to increase or decrease the intensity of light on the fly.

Since this thread started awhile ago, I've managed to iron out most all the kinks with using this system. Make sure you go and read Paul Debevic's site to really gain as much knoledge as possible on the subject.

-3DZ

:D

[Aldaryn]

A HDR environment can create specular highlights, only not the usual way.

A specular highlight is the reflection of the light source on the object, and the HDR image havnig the exact images of the light sources can give you quite nice and realistic highlights.
All you need to do, is to set up all the shaders correctly, this time not with "specular highlight"s but with proper reflections.
This will require to use glossy reflections, but the final result will be just like using faked specular highlight, but even more accurate, compleatly matching the illumination given by the HDR environment, and thus blending your scene closer to reality.

However, if I'm right, MAX 5 lacks a sane way of creating glossy reflections, (there is something, but lets forget about that method, that turns *global* raytraced solution into blotchy somethin...) Only way would might be is to use reflection mapping... sigh. Seems to do this 3rd party renderer is needed.

So, it's like 3DZ said: The scene needs a key (or two...) light(s), positioned to match the HDR key light(s). But if you've got a good wide range HDR image, this light source will be only neccesary to fake speculars (for non 100% glossy objetcs) couse...

...couse the overall dynamic range greatly determinates how accurate youre shadows are. The wider the range, the more crisper shadows can be achieved (if there are any light sources in the HDRI, that are really crisp shadow casters...) .
Lets take a theoritical experiment:
You've got an image of a bright sky, one sun, and the deep blue sky without clouds. Pretty simple.
Now, illuminate a scene with different ranged images of this simple setup.
If you use a quite low range (like LDR RGB...) the whole scene will be blue. Almost like turning on a blue skylight, the shadows will be "skylight like...".
Now, lets increase the dynamic range. Sky and sun illumination value now gets a bit futher from each other, and now, some blurred, but clearly visible shadow can be seen in the scene.
So, lets increase tre range again, and now, take a minute to think about that light info is exponent. So the sky, and sun values are really far from each, thus producing crisp shadows by the sun, and some secondary skylight shadows from the sky's light :D
Also notiece, that the first image was blue, and and the last one was bright white, with a blue skylight tint in the shadows cast by the sun.

Hope this makes some sense.

Cheers!

- A.

//Edit: 3DZ posted while I was writin' so if something is redundant, sorry

[xaoei]

A specular highlight is the reflection of the light source on the object, and the HDR image havnig the exact images of the light sources can give you quite nice and realistic highlights.
Exaaactly what I am trying to say. The HDR Image that he was using seemed to have a very wide range, and thus was producing very accurate and believable specular highlights. This is why I was wondering why you needed a matching light to correspond with the HDRI light if you already had extremely accurate speculars.
...this light source will be only neccesary to fake speculars (for non 100% glossy objetcs) couse...
Did you mean 100% reflective objects? (Instead of glossy?)
I would assume that you wouldn't need lighting for 100% reflective objects such as the sphere image that 3DZealot posted, because extremely accurate speculars are already there from the HDRI images.

Either way, I think I am understanding it better, I think I was trying to envision HDRI as a total lighting solution when it is in fact, like 3DZealot said, more of a filler light. Even though it has the ability to fake speculars (given you have a wide range HDR Image) it cannot represent accurate speculars on all surfaces / materials, and therefore additional lighting is required for such tasks.

Correct? :D

[Aldaryn]

... meant by 100% glossy a reflective thing without reflection blur, not a 100% reflective object. Must be the fault of the Brazil or Vray sahaders expression on representing glossiness. (100 (1.0) crisp - 0 (0.0) max blur)
So just wanted to say, that specular faking vill only be neccesary with object not having perfect reflections. (no matter hov strong they are refletive)

- A.

[flavrsaver]

Is there a way to hide the hdr image that's in the environment background when I go to render my final image? I want to be able to render my object without the blurred environment map in the background. So far I haven't found a way to do this.

[Markuss]

flavrsaver: This is a pretty old post so you might have figured it out allready, but if you haven't you can remove the backround in photoshop or similar programs if you save an alpha channel with the image....

[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.