Multipass different exposure settings

[ivanjay]

Good evening all,

 

I am shooting a rendering of an interior scene. I have the exposure set via mr exposure and it looks good for the interior lighting. I know that to be physically correct the windows behind the scene should appear over exposed because of all of the light coming in.

 

However, the average viewer does not appreciate that and wants to see out those windows...

 

Is there a way to shoot multiple passes and compose to get the correct exposure both interior and exterior so all looks well? Not sure if I can set this up in effects and send off to my rendering farm or how to go about it?

 

The only way I can think of is to render twice and use a layer mask which is just a pain... Or to photoshop the windows in, which is also a pain.

[mskin]

why not render the full view exposed for the interior, and then do a region render of the window exposing it for the outside? you can save the FG and GI maps so the second render should be much quicker.

 

how big is the window? is it dominant in the scene? you could also render a separate pass (alpha? or diffuse) to get a quick selection aid for photoshop

[ivanjay]

why not render the full view exposed for the interior, and then do a region render of the window exposing it for the outside? you can save the FG and GI maps so the second render should be much quicker.

 

how big is the window? is it dominant in the scene? you could also render a separate pass (alpha? or diffuse) to get a quick selection aid for photoshop

 

There are several windows along the entire back wall so a region render would not be possible.

 

I use 3ds max and mental ray. How would I go about rendering a separate pass (alpha or diffuse) and use that for a quick selection in photoshop? This technique is foreign to me but sounds interesting enough...

[Crazy Homeless Guy]

You can always give the middle finger to the Max camera exposure settings by turning them off, and tuning the lights manually to get the balance you are looking for. Personally I think everyone is little carried away with the concept of camera exposure in the computer.

Edited August 18, 2010 by Crazy Homeless Guy

[notamondayfan]

You can always give the middle finger to the Max camera exposure settings by turning them off, and tuning the lights manually to get the balance you are looking for. Personally I think everyone is little carried away with the concept of camera exposure in the computer.

 

umm interesting. yes you have to balance lights, but i find it easier to set up the lights with real world figures, and then adjust the camera. i find it gives a better / more accurate final image.

 

as for the background, i wouldnt do anything inside max, and just do your background in post. This will give you the most flexibility.

[Crazy Homeless Guy]

umm interesting. yes you have to balance lights, but i find it easier to set up the lights with real world figures, and then adjust the camera. i find it gives a better / more accurate final image.

 

as for the background, i wouldnt do anything inside max, and just do your background in post. This will give you the most flexibility.

 

Camera exposure is very nice, but it is just that, camera exposure.

 

I am sitting at me desk, looking out the window at the building across the street. With a camera I would need to choose whether I want to see the inside of my office, or the exterior. The camera is not advanced enough to keep both properly exposed at the same time.

 

However, the tone mapping in your eye is very advanced. Not only can I see everything properly exposed in my office, but I can see the exterior properly exposed at the same time, as well as the incandescent lights in the building across the street. I am not sure whether these are all exposed properly at all times, or if your eye is just adjusting for each space so fast that you don't notice it adjusting.

 

You can create the "eye exposure" concept in many ways. One is to do it in post with two passes, and composite them. Another is to get them close in one pass and tweak them. Getting them closein one pass either requires you to pump up your interior lights, or crank down the exterior lights.

 

Lately, I have been playing with turning off expoure and cranking down the sun light. I have also been turning off camera white balance. It gives me less to worry about. I don't need to think about how the color of the light is going to interact with my white balance. I simply need to think about the color of the light, and the color of the light is what is going to directly effect my scene.

 

At least this is what I have been doing the past 3 months, and plan on working with it more and more in scenes.

Edited August 18, 2010 by Crazy Homeless Guy

[notamondayfan]

yes you are correct, a camera cannot see the full range of light, like our eyes can. A camera sees 4 f stops (i think) and the eye can see a great deal more.

But now were are going into the whole photorealism vs human perception realm. Artistically we can do anything we like to make images look how we want them to look, but in order to achieve photorealism we have to stick to certain rules, such as sun intensity, camera exposure, light tempature, etc. so in shots such as an internal, the external areas will appear blowout and over exposed.

[ivanjay]

I think we are getting a bit off topic here.... My goal is not to achieve photo realism... My goal is to make a presentation that sells a project.

 

With that being said, most people can see the interior of a space and the exterior of a space properly exposed maybe with only a split second time for adjustment. Regardless of how this is done or should it be done in CG this is the look I am desiring.

 

I am looking for best practices to get in that direction.

[ivanjay]

i agree with Ivan it's kind of getting off topic why not just use the ramplayer so you can render two images and load them up and look at the differance?

 

It isnt a matter of looking at the difference. The challenge is compositing the two. Currently the way I do this is render the full scene at interior exposure and render the exterior scene and outdoor exposure. This takes a very long time as it is two full renders. I cannot use region renders because of the amount of windows...

 

Than I have to manually do select, cut, and paste into in photoshop for each window. It is just a pain in the you know what. Trying to find a better way as I run into this often with my scenes.

[ivanjay]

thats what the ramplayer does composites the two images if you looked at my tutorial that i gave you in the first post it would help you out. You can do a lot of stuff with the RP

 

I did but that discussed animations. I am rendering stills.

[Nic H]

^ your stairs are far too steep in that model.

i ahve no idea how the RAM player could be useful for multiple exposures tbh.

[erickdt]

I'd render out your full interior as one "pass". Then setup a version of the file with the proper exterior exposurer. Then select all your window glass, click on select invert (so now you have everything but your glass selected), right click, VRay properties, check matte object box, set alpha contribution to -1.0, render out frame sequence or still image. This way you're only rendering you windows where they can see outside. Now you can then easily composite the two in post regardless of whether it's an animation or a still...

 

E

[notamondayfan]

sorry for deviating too far :-)

 

well if I were you I would render the final image (setup for interior exposure) in exr 32 floating point format, then use a mask to adjust the exposure of the window areas in a compositing programe like after effects or fusion.

rendering as 32 bit exr means you can adjust the exposure in post, as 32 bit float wont clamp colour values (unless you have colour clamping ticked in vray!) so even if you exterior is rendered as white, you can turn down the exposure to see the exterior.

 

Deano

[mattclinch]

ram player is absolutely no good for comp'ing a specific area of an image which needs a different exposure. please stop giving out bad advice.

[mattclinch]

then please enlighten us on how you plan to use ram player to adjust the exposure of a specific area of an image, which is what the OP was asking.

 

loading two images into ram player simply shows you the difference between two images. there is no compositing, no masks, no exposure control.

[ivanjay]

I have to admit in watching I am not really feeling the ram player does what I need. The entire goal here is to eliminate two high resolution images that take hours each.... With a multi pass I can do one high quality rendering and a second short pass to fill in the windows..

 

That being said I am not using vray so although I appreciate the tricks anything related to vray is not an option.

 

If I rendered to a 32 bit exr I saw a note about being able to easily select the windows and change their exposure. I have never used exr... Can you get a bit more specific with that? Maybe I will try it.

[mattclinch]

are you using light portals outside the windows? if so, i think there is a way to adjust the "transparency" of them which can help stop areas blowing out. never used it so can't say anymore than that.

[ivanjay]

are you using light portals outside the windows? if so, i think there is a way to adjust the "transparency" of them which can help stop areas blowing out. never used it so can't say anymore than that.

 

I am using light portals. I played with the transparency. I even tried them off. They didnt help, again with the sunlight coming in the windows are just overexposed as to be expected.

[Crazy Homeless Guy]

I don't think the topic of possibly not using camera exposure and why is really that far off topic.

 

I threw this scene together fairly quickly in Vray. The scene has 1 direct light, and a sky system tied to that direct light in the environment slot. The sky system is set to a multiplier of .03 to keep the intensities down. I am using some tone mapping inside of Vray, but not photographic exposure.

 

I am not going to tell you this is perfect, but it definer is not horrible. The image is still dark on the interior, but the adjustments are fairly easy from this point forward. Even dropping 1 or 2 room lights would take care of that problem.

 

I do beleive this is something that could be accomplished in Mental Ray.

[Crazy Homeless Guy]

Ok. Same scene, 2 additional ies lights added to fill out the room lighting, and a Photoshop curve adjustment layered applied to it to bring the mid tones up a couple of notches.

[ivanjay]

Ok. Same scene, 2 additional ies lights added to fill out the room lighting, and a Photoshop curve adjustment layered applied to it to bring the mid tones up a couple of notches.

 

So you just shut exposure to do this? This is pretty close to what I am trying to do.

[Crazy Homeless Guy]

So you just shut exposure to do this? This is pretty close to what I am trying to do.

 

Yes, if by shut you mean eliminated. I am working on setting this scene up in MR to test the feasibility of using it to light interiors. I have not used MR in about 10 months so I am a bit rusty.

 

Attached is the point I am at in the MR scene. The differences so far are that the light coming in through the window in the MR scene is actually a spot light instead of a direct light. I could not get the direct light to cast area shadows like I could in Vray. Also, the sky and indirect GI are created by

using a Skydome. In the Vray scene I used a Vray sky with a Direct Light attached to it. This didn't seam to work in the MR scene.

 

A few of the things that frustrate me about MR are coming back. Look at the reflection on the chrome ball floating in the room. Notice that the corners that are reflected from behind the camera do not have GI. This is a problem.

 

Currently the GI is created using Final Gather. I am thinking of breaking out Importons and Irradiance Particles to calculating the GI in the final pass. To me the light created using the CTRL.Ghost stuff is vastly superior to what you can get by default with MR, though I think there are people who would disagree with me on that. The CTRL.Ghost stuff is not as easy to use, and in past versions of Mental Ray they were quite slow in comparison to FG /Photon methods.

 

Since I am testing some things, I may also try and test a couple of different tone mapping options in this scene. There is a bit of blow out on the bed. I can probably control this with some simple tone mapping. I guess I could even use the physical camera exposure for this, but I am thinking of a different method.

 

In reality you might be able to do all of this using the recommended camera exposure, and the "physical sun and sky system." I am just feeling that those methods actually give you less control over your scene as a whole, and force you into problems. But that may just be me. I am sure there are a few pros lurking that could prove me wrong.

[notamondayfan]

I have to admit in watching I am not really feeling the ram player does what I need. The entire goal here is to eliminate two high resolution images that take hours each.... With a multi pass I can do one high quality rendering and a second short pass to fill in the windows..

 

That being said I am not using vray so although I appreciate the tricks anything related to vray is not an option.

 

If I rendered to a 32 bit exr I saw a note about being able to easily select the windows and change their exposure. I have never used exr... Can you get a bit more specific with that? Maybe I will try it.

 

exr format allows you to store bore bit depth in an image, and because 32 bit exr can be used as a "floating point" image. None floating point imags, such as jpgs, will clamp colours and depth, so if you render a landscape image with a bright sky, the sky will appear blown out and very white. Now if you take the jpg into a composting program, and turn the exposure down, the sky will become darker, but you will not see any definition, such as clouds, as when renderered, the sky and clouds and sky both appeared white.

Now with exrs (or any floation point format), if you rendered the image again, and took it into your compositing software, and adjusted the exposure, the sky would look exactly as it would if you had rendered with the correct exposure to see the detail in the sky.

Im probably not making much sense, so I sugest you search and experiment yourself. Render an image over bright, render as exr, and play with the exposure in post production.

As for masking, again a pretty simple task, just search how to do it in your prefered software.

 

Deano!

[Tommy L]

I agree with Dean, I think the best mode is to use 32 bit. I always do, just so I have exposure control in post.

Quite an elegant solution from CHG tho.

I must also mention that I thought Libertyanimation fella had been banned. Guess not.

[mattclinch]

A few of the things that frustrate me about MR are coming back. Look at the reflection on the chrome ball floating in the room. Notice that the corners that are reflected from behind the camera do not have GI. This is a problem.

 

This is avoidable if you use radius interpolation method for FG. But yes, its a big problem and a massive oversight from Autodesk.