Cg source hdr exposure issue.

[thanulee]

Thank u all for the replies and the info. Thomas, is not the blown out areas that much my issue, as it is the direct lighting that i get, which as u can see in the first image, is not corresponding to what i should get.
My setup is pretty str8 forward: linear workflow, dome 1.0, hdr spherical default settings, nothing fancy.
Care to setup a simple scene in vray 3.2 and onwards to see what im doing wrong? Cause i have one of ur timelapses at the moment, and i face the same issue
Appreciate all the help!

[suurland]

The "problem" is you are using linear color mapping. So as described by me and others you have to color map it in your rendering or when you have done your post. The reason your rendering with linear color mapping is dark is because you are trying to match the sky, if you matched the ground instead you should be able to get something similar, but that will blow out the sky which you then have to bring down again.. that is what exponential does and the reason my renderings looks as they do. As far as I remember the reason I used exponential is because I did tests where I used one of the exposures used to make the hdr map to match my rendering to, which means with exponential I got the result as shooting the sky with a camera, the final result shouldn't be linear, then it will look like when you open the hdr map in photoshop, which isn't pretty for skies with direct sun light.

/Thomas

[thanulee]

Thanks.
I insist that this is not the case here. Exponential just desaturates the whites, thats how it tackles the blown out areas, thus it flattens the image alot and is far from optimal using this cause u have no information for post.

If i just render the hdr instead of linear using exponential, the result not only will be dark, but it will also be more flat as well. Btw, im on reinhardt at .8-.85 (not that it matters) which its almost linear and just helps the engine a bit.

Overall, plugin the hdr with everything at defaut values gives me darker/less contrasty results than expected.
I can tackle it in post for sure, but i was wondering why i dont get 1:1 with the website. Check images again, and plz, i dont know what im missing here. Did u happen to post maybe ur hdrs for the website?
thanks for the help.

[thanulee]

Ofc all these can be done in post, but thats not why i posted this.

For instance now that i got the timelapse, in order to use it proper i have to: a) blow out the sky so my scene is light better, b) override the reflections cause once blown out sky, reflections follow with aliasing issues and blown out as well and c) color grade/clamp again the sky in post to match the rest, which will end up either with blown out highlights to match my overall comp, or flat and not matching the comp that well.

The other solution is to use the hdr only as reflection override and use a vray sun and a colorpick a gi value as well. This while works great for stills, doesnt for the timelapse: i will miss all the natural flickering and exposure changes that I get from the footage and i will have manually to reproduce it in post (which is a pain, i wouldnt buy a timelapsed hdr in the first place ).

I hope u see why im being picky here. If there is anything more i could try, hit me.

Have a good night!

[crazy homeless guy]

Very true, pretty much every stills or film camera tone maps images, our eyes see in logarithmic colour too so it's useful to do as a last step if you're trying to do something photographic.

Yep. I Most film footage looks like poop before it is tone mapped with a LUT. It is flat, gray washed, and overall unappealing. Then bam... It is tone mapped and the film comes to life. It is almost the exact opposite of what we do in 3d where typically we wind up with blown out whites and we have to crank them down in order to get something usable.

Take this image for example... The raw footage is on top, the image tone mapped with a LUT is on bottom.
https://lutify.me/wp-content/uploads...ts-example.jpg

[thanulee]

It is almost the exact opposite of what we do in 3d where typically we wind up with blown out whites and we have to crank them down in order to get something usable.

A debate here: do u guys tend to over or under expose? I under expose so i control my antialising/rendertimes better and crank up in post as in real photography so not to lose information in the whites. My compositor says, i should blow up stuff cause cranking in post from under exposure, gives noise in the darker areas.

[crazy homeless guy]

A debate here: do u guys tend to over or under expose? I under expose so i control my antialising/rendertimes better and crank up in post as in real photography so not to lose information in the whites. My compositor says, i should blow up stuff cause cranking in post from under exposure, gives noise in the darker areas.

What your compositor says makes sense but I tend to lean under as it is easier for me to work that way. I like to be more or less WYSIWYG coming out of the 3d app before post work.

[squintnic]

I still dont understand Thanlee
Cant you render full unclamped linear and adjust the colour mapping in post to whatever you want it to be? That way you can have and exponential look or something linear. Perhaps you colour map the background differently to the ground? Unless you are using the render direct from the frame buffer I think you are overthinking this.

I wouldnt worry it too much as a lot of arch vis guys get caught up on numbers and settings and forget to look at the actual images they are producing and comparing them with reference.
Thanks

[thanulee]

Ya i can post the shit out of it lol. Im just wondering why when i buy some images do not match with what i see as "raw". Unless it aint raw and i ve misunderstood it. Cheers guyz

[joconnell]

Yep. I Most film footage looks like poop before it is tone mapped with a LUT. It is flat, gray washed, and overall unappealing. Then bam... It is tone mapped and the film comes to life.

Well, part of the reason for that was film is generally shot in log space where they bias the details to be saved in the shadows more than the highlights - humans notice more subtle differences in the darks than they do in the brights so it made sense back in the old days when storage wasn't plentiful to come up with a format that squished the most useful data into the smallest space possible - something like the 10 bit log cineon format. Modern digital cinema cameras (save for dlsrs) all shoot in a log format so as you've mentioned they all need a lut to be applied to bring them back into a "human" colour space to make it look like what it was shooting!

[joconnell]

A debate here: do u guys tend to over or under expose? I under expose so i control my antialising/rendertimes better and crank up in post as in real photography so not to lose information in the whites. My compositor says, i should blow up stuff cause cranking in post from under exposure, gives noise in the darker areas.

Vray samples for the exposure that you set, if you're underexposing you'll get noise in the darks as vray won't do as thorough a job so in this case your compositor is right. If you render to a float format like exr with no clipping or tone mapping you won't lose any bright information so no worries there. Lele that works for chaos group used to do his renders one stop brighter I believe to give cleaner shadow info.

It kind of depends what you're going to do with the renders in post - if you're going to wildly colour correct them then you'll definitely get better quality by trying to have a similar exposure level at the 3d render point - vray's going to be aware of this when it's sampling so you'll get far cleaner results.

[suurland]

I just did some tests and I'm getting result that is quite far from yours.

I've made a simple version of the sample scene (max2014 file) that you can find here: www.cg-source.com/_thanulee/thanulee.zip just have in mind that this scene was made in 2012, today I would set the colormapping mode to "Color mapping only (no gamma)" and then switch on "Display colors in sRGB mode" in the VFB. those two are the default values now.

Image 1 : Rendered with exponential, straight out of the render with NO post, same image as used on the website.
Image 2 : Color Mapping changed to linear.
Image 3 : Also Linear but made it a bit darker to match the luminosity of the exponential image.

Also notice my shadow in the reflections are sharp, in your renderings they are extremely blurred!

[thanulee]

Thanks John nice to know this info, and thank u Thomas.
@Thomas, well, idk what is going on really. I ll check the scene u providing and let u know. I am definitely doing something wrong, or something is bugged, if u get this result and i get the one i showed u. Ya shadows blurry like shit.. I suspect that irradiance map doesnt play very well here.
I really appreciate the help here guys.

[thanulee]

Glad i sorted this out:
Ur overall mult is 2.0 instead of 1.0 and ur inverse gamma is 0.8.
And the blurry shadows seem like a bug not related to the hdr, is due to 0.99 reflection glossiness. Cheers