ACES filmic tone mapping

[another_commander]

It has already: https://github.com/OoliteProject/oolite ... tag/latest

For Windows, any change that goes into trunk becomes a nightly approximately 8 minutes after the respective commit.

[Cody]

Ah, okay - it hadn't made it into the nightlies section.
GitHub confuses me these days - as do many things!

[Nite Owl]

If the Nightly Release is not an option for me (reasons discussed elsewhere many moons ago) would copying and pasting the new Shader Files (as posted on GitHub) into the Shader Folder of the current Release Version (v1.90) work or is there other Core Coding in the Nightly Release that would prevent this from working? Have given this copying and pasting method a shot and all seems to be working fine but decided to check with the experts just to be sure.

[phkb]

would copying and pasting the new Shader Files (as posted on GitHub) into the Shader Folder of the current Release Version (v1.90) work

Yes.

[another_commander]

During experimentation with colored planet dark side illumination I came across this and thought I'd share just to make a point of what a great job ACES does for tone mapping. I hereby present to you comparison images of it vs our previous tone mapping algorithm.

What we are looking at: This is the USA on Earth's night side. For the purposes of the demonstration I modified the city lights' color to very intense HDR blue (rgb 0.0, 0.0, 10.0), then screen-shotted the same scene using our former tone mapping function, ACES without the highlights correction we recently applied and ACES as we have it currently in our shaders. I think the results speak for themselves:

1. Hejl / Burgess-Dawson filmic tone mapping: Appearance of electric blue city lights, it seems a bit "gamey". Looks more or less OK, but something feels off.



2. ACES filmic tone mapping w/o highlights correction: This looks just wrong. It very clearly demonstrates the problem ACES has with blue highlights. We are not even looking at blue here anymore, everything has turned magenta and purple. It does look more like city lights, though. Shame it's just plain wrong.




3. ACES filmic tone mapping w/ highlights correction: This is just awesome. Lights DO look like city lights and they appear bright and pleasant because the ACES algorithm causes colors to de-saturate as their brightness increases, which is what we normally observe in real life and on film.



Very happy with the current result.

[Cholmondely]

But what if you just changed the colour of the lights in the first shot to that of the third?

[another_commander]

But what if you just changed the colour of the lights in the first shot to that of the third?

See, this is the entire point: the color of the lights is exactly the same in all three shots.

[Cholmondely]

In which case, yes... the third system is clearly superior...

Maybe other species tend towards other colours for their city lights? Lobsters probably go for deep black...

On visiting eastern Africa some years ago, I was told that the locals tended to as low wattage light bulbs as possible for lighting - and that they were eagerly awaiting the development of the 0 watt bulb! (I was visiting a refugee home on the outskirts of town, which consisted of one room with a bed and a bulb, and outside was a small walled open-air space with a fireplace. That was it! I doubt they had much money to spend on electricity...).