So I’m still in the process of learning how to use LWF. and I think I’m not there yet.
Since everyone makes his own cook, I think this would be interesting to compare each different setup you adopted.
basically, I think the most important part is the colour mapping panel.
i use exponential mode. dark 1bright .454gamma 1.
lwf not checked.
I render in max vfb.
all settings in 3ds Max gamma/LUT are set to 2.2.
after seeing my results, I think this is not correct.
what do you guys use for colour mapping modes?
Give some examples of you respective LWF settings.
i’m using those settings for LWF
Gamma & LUT Preference settings:
1) Enable Gamma/Lut Correction.
2) Gamma 2.2
3) Affect Material Editor
4) Input Gamma: 2.2
5) Output Gamma: 1.0
Vray Settings:
1) Max FB: off
2) Vray FB: on
3) Dark Multiplier: 1.0
4) Bright Multiplier: 1.0
5) Gamma: 2.2
6) Affect Background: checked
should i check the Linear workflow checkbox ?
when i check it, image look very dark.
lwf checkbox is meant to apply inverse 2.2 correction to the image. because in general images are 2.2 corrected. it means they have a 2.2 profile embedded. which is the case for most of the images which are in sRGB profile. so logically i should correct all my input images so the 2.2 profile is removed. this is done by the input images 2.2 parameters in the max gammaLUT panel.
if i check lwf checkbox in color mapping, it will apply another inverse gamma correction. so twice inverse corrected textures . not good. so it seems i don’t have to check it, right ?
same goes for the colors. it seems it is not necessary to check lwf at all finally.
but if vray VFB and vray in general ignores totally the gamma/LUT display panel settings, then i have to check this box.
that’s where i’m asking myself the question. does the vray vfb ignore the 3ds max gamma lut settings for textures input and output, or not ?
and since all the input images are gamma corrected to remove the 2.2 embedded profile, we can say they are “linear”. So why the hell should we put the gamma parameter of vray color mapping to 2.2 ? indeed, the 2.2 valus will correct all the colors during the computation, as if they were not already inverse corrected, right ? but we just invert corrected the textures with the max gammaLUT input value !!! so verything is linear ! why vray should be aware of any gamma thing since we put the right input values ?
Try out rendering an image with a lot of shadows and you will see the difference when using gamma 2.2 settings in VRay vs just using them in Max. The settings have to be adjusted in both places. Max for the textures, VRay Color Mapping for the lighting.
just to recapitulate and to get rid of my remaining doubts, the display and input max gammaLUT 2.2 value are not “redundant” with the vray color mapping gamma 2.2 value(right ?)
so the 2.2 display/input in max and the gamma 2.2 in vray color mapping seem to you to be a correct config, right ?
ok for the max DISPLAY gamma settings, which acts only on max frame buffer.
but the max input gamma setting affect the textures,with both vfb or max frame buffer. right ?
that is why it is un necessary to check lwf checkbox, because that would correct the textures twice if i understoode correctly.
as soon as i think i understood coorectly, another doubt comes to hit my head.
do
is the texture correction is made with the vray 2.2 color mapping gamma or with the max bitmap 2.2 gamma?
that would result in the same thing, but it’s a important difference. if the texture gamma correction is made with vray 2.2 color mapping gamma, there is no need to set the bitmaps input gamma to 2.2…
and if the texture correction when rendering in Vray frame buffer is not made by any of the two, then i have to check the damn lwf box…
technically, you should have in the colour mapping:
linear, darks=1, lights=1, gamma=2.2 and have “don’t affect colours adaptation only” enabled, you view the image in sRGB space within the Vray Frame Buffer, and save your files as .exr or .vrimg. If you do this, you have a linearly encoded file, which will be fine for doing re-compositing of the render elements. If you don’t do this, and you bake the 2.2 gamma into the image, any re-compositing of elements WONT give correct results, as the file isn’t linear data anymore; it’s got a gamma curve baked into it…
i tried your method, and with sRGB enabled, the results are quite the same in terms of colour/contrast…
what is the advantage of using this method when doing re-compositing of render elements ?? (or what problem can i meet when i’m not using it?) what do you mean when you say it WONT give correct results ?
It strictly depends on which level of accuracy are you looking for
I often bake gamma when I know from the begining that image is almost totally done straight from frame buffer; it’s faster
When I know that I’m going to make an amount of compositing, color correction or whatever post producion work, I always save WITHOUT baking gamma and apply an inverse gamma 2.2 curve layer in photoshop (top of the stack).
Trying to summarize the reasons I can say that it’s related with the following:
linear space: rgb 128 + rgb 128 = middle gray + middle gray = full black
non linear space: the previous equation is not totally correct
Probably it’s not a totally correct explanation but more or less it is
Sorry I might have mentioned that the Max gamma settings affect the display gamma in the VRay frame buffer, they don’t. They only affect the bitmap gamma in the VRay frame buffer. The input gamma settings for bitmap files only affect bitmaps in the VRay Frame buffer, here are some screenshots of this in action:
The sRGB button is the one that switches on and off the Gamma 2.2 in the VRay frame buffer. Although sRGB is not exactly 2.2 gamma, but it’s pretty close.
so, the idea is you just apply the colour transformation in the viewer only, and you keep all the data as linear, floating point. Vray calculates everything in floating point.
if your doing things like multiplying elements together in a post compositing program, you have to work with linear data, otherwise the resulting recompositing of elements, WONT be correct - i.e. won’t be the same as the Beauty RGB image in the VrayFrameBuffer. If you keep the data linear, you can do the correct math in post, otherwise your trying to put back together elements of an image that have the incorrect values (colours) which of course won’t work.
FYI - the images posted above - the second image I don’t think is correct… the input gamma of 2.2 removes the sRGB profile from the jpg texture. you then re-apply this by viewing the image in sRGB space…
