photo studio light calibration

I have found this great photo studio on blendswap.

and i tryed to calibrate the basic scene to blender filmic.

first i created a sphere object and applied a diffuse grey material to it.the grey has 18% value of white (0.18).
this sphere i placed on the tripod object.and created a plane with a colorchecker image on top of it.

i deactivated caustics, for faster and less noise rendering.activated denoising with basic settings.

then selected filmic ,in the colormanagment ,with high contrast look.after a testrender, with 500 samples,i saw artefacts from the denoising,so i increased it to 2000 ,there it looks clean,and go back to 1500 samples, there the artefacts starts to appear.

the tricky part to calibrate was,if the blender filmic look is set to say basic contrast,then the colorchart render appears to dull in the highlights,and if it sets to very high contrast the shadows are do low in the render.

so middle high contrast and high contrast seams good ,for this test ,to get the allmost same RGB values, from the reference color checker pic.

i leave the lightstrengths from the lightsources how it was.so i rendered with different exposure settings,in the film tab,to match the light intensity.

here are my results from the test,with filmic high contrast look,and exposure at 0.88

i used a colorpicker at the test to compare the values from the render and reference color checker pic.


a close up in the viewport


Nice experimenting

thank you lsscpp.here the render with filmic medium high contrast and exposure 0.87.i reduced the exposue because the overall colors apperence was a bit to bright.if you use a colorpicker,you can see the problem that occurs with lower contrast filmic look.
the neutral grey is near at 160 in all RGB channels ,but the 200 grey has a drop to ca 180 and the white 243 dosent come over 200.


here with very high contrast filmic look .as you can see, the highlights are brighter,but the shadows are lower vs the reference pic,and the colors are altered,more saturated and a overall different look vs the reference pic.


Remember that a colour chart would be a series of Albedo values based on paper. This means that the sRGB values you measure will never align with a virtual photograph that maps a broader dynamic range to the image than any reflected paper would hold.

Don’t expect 1:1, as it is like trying to compare apples to oranges. The best you can do is assert that the chromaticities of the various colour swatches match the sRGB triplets. Trying to get the arbitrary sRGB numbers that someone applied to a ColorChecker would be like trying to take a photograph of the chart and expecting a match.

Right. I’d be pleased already if H and S (in HSV color notation) get close

yes troy, everything is right what you say.this test was simply to see
if you texture a model,wich filmic look represent the reference texture colors the best.

Right. I’d be pleased already if H and S (in HSV color notation) get close

thats what i mean.

and its nothing wrong with Blender Filmic,its still one of the best things we have in Blender.

In my work, I have to match a set of RGB color swatches (.JPG output) as close as possible in the final rendered image, so a good amount of experimentation is required. It’s not an exact science, so experiments like this are helpful to me.

It is a bit easier than this.

Get the xy coordinates of the spot colour in question and make sure that your colours match in the reference space. So for example, currently REC.709 primaries are assumed by default in Blender, and Filmic follows this convention. If you deduce the xy coordinates for the spot colour via proper transformation from whatever reference sources you have, you can transform to the precise xy coordinate ratios in REC.709.

Intensity would need to be massaged as with any photograph, but the actual colour can be accurately transferred.