Linear Workflow washes images out

[axezine]

I have to say, probably the biggest problem is that a lot of people set their materials (colors, textures, what have you) to be WAY too bright. Then you light it in some fashion to get the intensity you want and bam, low contrast (because you're using GI, and if your material is too bright then secondary bounces become a huge source of illumination). For instance, if you look at photos of the moon (ie, stuff taken during the apollo missions) you'll see that they are incredibly contrasted - in fact, a lot of people use that as a reason as to why the landings "never happened". What is really going on is that average albedo for the Moon is just about neutral gray 0.12! That is quite dark...

Look up some real material information and see what their albedos are (this is not the best measure to set up materials because it's an overall value, but better than nothing) - you'll see that almost no common materials have an albedo even close to 1, 0.5 is probably way brighter than typical bright-white paint. It doesn't help of course that most textures are 8-bit images, which absolutely need to use the full 0-255 range to have decent color range, but most textures should be way way darker than they are...

Hope this helps...

Cheers!

Jorge

[soulburn3d]

Jorge, yup, at work, we follow the rule that human skin is about brightness 0.18, and everything is adjusted colorwise accordingly. It's not purely scientific, but it gives us generally good color results. Assuming of course you're going after realism.

- Neil

[soulburn3d]

Added a lesson on my website about the stuff we discussed here. Thanks for all the input!

http://www.neilblevins.com/cg_educat...difference.htm

- Neil

[Donald2B]

Great write up, Neil!

[Dariusz Makowski (Dadal)]

Im not going down on to why u guys should use proper linear workflow and so on. Ill simply say if you don't like 2.2 because of wash up. In past I found out that nearest nicest balance is 1.8 I don't advise it but its nice area to work in.

[Bruce Hart]

I also use a gamma of 1.8 to achieve a contrasty look. One of my biggest mistakes in the beginning with Vray was having the materials too bright (as mentioned above). I think that is why some people suggest reducing the secondary GI bounce, but it's probably better to leave at 1.0 and fix the colour mapping and materials IMO

[FSGFX]

Jorge, yup, at work, we follow the rule that human skin is about brightness 0.18, and everything is adjusted colorwise accordingly. It's not purely scientific, but it gives us generally good color results. Assuming of course you're going after realism.

- Neil

hi Neil, what do you mean by .18 brightness for skin..? as color this would be near black..?

[vlado]

hi Neil, what do you mean by .18 brightness for skin..? as color this would be near black..?

No, it's not... 0.18*255~46 or about RGB 46, 46, 46 which is far from black.

Best regards,
Vlado

[FSGFX]

thanks... i muddled something...

[soulburn3d]

What Vlado said. I was speaking in 0 to 1 scale instead of 255 scale.

Although to note, some more recent theory is that skin, depending on race, can actually be far more reflective than 18% (as high as 30-35%). But the overall idea holds true, things in the real world are generally far less reflective than we assume in traditional cg.

- Neil

[grantwarwick]

Things don't look washed out to me anymore and like others have mentioned, I introduce contrast with lighting, just like any photographer would. If we are doing a logo or something that needs contrast, the retouchers handle my output like they would if a photographer shot the logo in real life.

[jpmartin]

The skin tone value got me thinking my paint colors are way too bright...

So, say you have a specific set of paint colors you have to enter. I usually go to the website (mostly sherwin Williams) and use the material editors eye dropper to sample the color. The rgb values are often above 200. Just curious how everyone deals with that? Adjust all colors globally by a certain amount (say .8 maybe?)

[voltron7]

The skin tone value got me thinking my paint colors are way too bright...

So, say you have a specific set of paint colors you have to enter. I usually go to the website (mostly Sherwin Williams) and use the material editors eye dropper to sample the color. The rgb values are often above 200. Just curious how everyone deals with that? Adjust all colors globally by a certain amount (say .8 maybe?)

I do similar procedure. In PS, I have loaded some paint mfr color palettes from their websites, i.e. Sherwin Williams, Ben Moore, etc. I go into PS and make 32x32 bitmaps each time I need a new color. Now, I many saved, so when the same color is used, it's already in my library. I do this for whites, too. Clients do not remark about incorrect colors, so for me this it works and seems to eliminate guesswork on my end... I usually go for 1.0 and 1.0 for GI bounces and for max gamma 2.2 in and 1.0 out.
I used to pick the color from the website, but feel slightly better about using their freely provided palettes.

[jpmartin]

Thanks, Voltron.

I was considering a similar workflow, but still don't quite understand what to do with rgb values that are close to 255. For example, Sherwin Williams Navajo White is R-234 | G-223 | B-201 and has a LRV of 77. I've always tried to keep my whitest whites around 185, so these rgb values seem quite high. Do you lower the overall color value to get them within an acceptable range? (if so, I would imagine the other colors in the render would have to be lowered by the same percentage in order to keep the same intended relationship).

I'm wondering if the LRV is anything we should be paying attention to as well. I'm thinking that this is saying the value should be around 196, which is .77*255.

Just trying to get a gauge of what is working for everyone else...

Thanks!
-Jesse

[voltron7]

Mildly embarrassed to say I never really investigate any of the RGB values for those paint color palettes.