The original one reminds me of the place I live - Sao Paulo Brazil - when it is in the same conditions. The other one reminds of the pictures I see on the web. View attachment 15951 edit: "original one" = as it is in RF2 currently, like in the screenshot I submitted
Not sure I understand; do you mean you find more realistic the one without the blue ambient or the opposite? Looks like you are applying two different sentence to the same picture...
Are both of those images using the same ambient color, or based on the sky hemisphere, or arbitrarily/best-guessed by yourself? Has the sun color been manipulated in any way, either? In other words, I'm assuming that the original direct sun color was set to approx. 6500K (neutral white in sRGB color space)--which would actually be the mixed color of the sun and sky/ambient light contributions. Since the new ambient term is supposed to represent the sky lighting contribution by itself, its color contribution needs to be removed from the direct lighting as well; otherwise, you get the net effect of having the sky color mixed twice.
I find the one from Mores without the blue ambient more realistic. The new Mores one with the blue ambient that you posted reminds me of the images I see on the web and from games like MGSGZ.
I might've been editing my last post when you responded....but doesn't that have the net effect of mixing the sky color twice, since, the original direct light color contribution was essentially sky and sun mixed. Wouldn't you have to remove the sky color contribution from the direct light? It's interesting, I was doing some tests this morning with my camera. I wasn't able to get it to take a picture that was both as warm as I saw it, with my own eyes, in direct sunlight, and as cool as I saw it in shade--with snow as my subject (arguably the most neutrally-balanced material).
About the one that you posted from Malaysia, it does not look realistic also. But while Mores that you posted was not horrible, this one look like a really bad calibrated monitor. And the current one has a really well lit tarmac. Its like if HDR could make you happy View attachment 15952 But of course: in MGSGZ, that have the so called light probes, you have the blueish shadows and the warmth from other places at the same time.
What do you mean with "remove the sky color from the direct light". They are independent, and that color we are talking about, it's affecting ambient .RGB only, not direct which still produce his own .RGB set of values in relationship with just the sun source as white to orange/reddish for twilight conditions. The difference it's just on ambient where you get a locked .RGB rotation during the day (white to black), when ambient it has been set to stay monochrome....while you get independent .RGBs values when unlocking the ambient to be paired to the sky dome (not the direct light). The result is direct is doing the same as before, while Ambient is producing colors instead monochrome variations, due the sky sampling (this should fake the missing indirect light). To make it simple, if you are asking if direct is now producing blue, the answer is no. View attachment 15953 View attachment 15954
So, that's actually an issue for me. Before, by making the ambient light a neutral color, you'd decoupled ambient light color (wavelength) from ambient light intensity (power/illumination). So, in order to get a neutral color in direct light (white), the direct light itself actually contained the mixed color of both the sunlight (which is, at its hottest, above the atmosphere (according to wikipedia) about 5900K) and the skylight to get a perfect white-balance (because something that is in direct sunlight is being lit both by the direct sun and the scattered light in the sky). However, it [the direct term] only has the intensity of the direct light, not the ambient light. By changing the ambient light to the skydome color, you've re-coupled the ambient light intensity and color, which means that that color contribution needs to be "un-mixed" (subtracted) from the direct lighting contribution. In other words, you need to now make the direct sun light warmer, because it no longer contains the sky color contribution. This should make it so the stuff in direct sunlight should have less of the bluish cast.
I do play a bit around with RGB LED's and what i have experienced is that the lightning source is the most important, in this case it would be the sun. How about to get the sun right before manipulating something else. The current sun has some unnatural colors in the mix, what drives me to ask if it is due to the individual balancing it or if it was by purpose to compensate something, or even if it is due to bad calibrated monitors.
We do not use static colors, they are the result of a computation due the nature of a dynamic environment. Said that; doing what you are suggesting in that quote, would make everything in direct sunlight looking greenish (aka acid) and also this is going to screw the physical scene/sky consistency, which is already taking into account the air medium per sun elevation (aka dusk/dawn) through a BIG field of physics equations. We use real math models for sky computation, nothing is static or canned. The complexity of making all this stuff consistent is because the nature of TMs playing between different sky/sun conditions around the planet. Something which needs another pass of math to make everything working properly in the same domain. Forcing the sun to generate a complementary element of the dynamic .ambient RGB will pollute the key aspects of that model and on top of that will not make any white balance due the wrong operation. Also both elements are NOT static and are NOT acting the same way at the same time (they are NOT paired into a linear domain)...this would mean getting a yellow dominance at the time the ambient is still white, or a greenish layer at the time the ambient is going more blue...etc.
Yes, sorry, I could've been clearer, and was actually about to clear it up a little bit when I saw you'd replied. I understood that the skydome and sunlight are dynamically calculated. I also understood that a neutral white object isn't always going to be D65, nor 'should' it always be (as I'd mentioned in my earlier post about the snow this morning), according to the time of day, etc. I was just using D65 as an example, since it evaluates to about a neutral white (for clarity's sake). I also appreciate that the direct light is dynamically calculated according to all those different properties as well. However, I'm now curious if that means that the direct lighting is calculated just based on a single path through the atmosphere to the point of illumination, or does it take into account the whole hemisphere? If it's the latter, can those calculations be decoupled (a direct path vs. a hemispherical path), and use those values directly for the ambient and direct terms? If it's the former...well, then....carry on!
The normalized sun light path it is calculated starting outside the scene, from the outer space and then basic light scatter components are computed at the time the light pass through the atmosphere layer to reach the scene. This direct light is then computed via VS to be scattered back to the sky dome, with albedo, ozone, water etc... all parts of the same model (also that light is computed in real kcd/m^2 before the TM). Ambient starts here, between the sky dome and the scene/enviro. But the problem is GI is not playing here, so while the sky model is VERY advanced and physically correct (theoretical due the nature of the applied TM moving everything inside a working range), the indirect light still remains something you can just fake with the cost of sucking back part of that realistic model. So, whatever you do inside these kind of game engines, you have to face that process of searching the photorealism (a term I hate because the misinformation about it) will be truncated at the time your system is not using any indirect light algo.
Interesting experiment! The ambient light is way too blue in Tuttle's shot's. Looks like something around rgb 155 210 255 (the color of the sky?) to me. It would be correct if the building were the only object in the scene, but it's not and there should be a lot of scattered light from other surfaces. Because we can't calculated all these lights we make a good guess and use instead rgb 205 227 255. Unmodified screen shot. Legacy hdr profile Ambient light for the shaded area of the building 155 210 255. (Photoshopped, multiplied a blue area with the shaded parts of the building) Ambient light 205 227 255. That's what I would like to see.
I was not understanding almost nothing until now, only a bit (on Tuttle pics). But the Tosch´s example nailed it for me. Interesting topic. Now I see with other "eyes".
This is a simple hack we already tried (rF1 docet), keeping the Ambient in the monochrome state (to avoid the sky dome sampling) and then hacking just the ambient RGB to be multiplied by constant factor (static) adding a bit of blue. It is not going to work apparently and still may introduce a bit of fight (inconsistency) with the dynamic sky/weather. Unlocking the ambient light for sky sampling can't be forced to work inside capped ranges, as it is part of the sky computation where the output is the result of math, not a static input.
And btw Tosch, meanwhile I would say I like your shots (for me the first still remains the more convincing as a compromise), they are all not correct if I've to take into accounts what I see IRL and what my reference shots are showing. I expect bluish just when there is a BIG sun occlusion, like a big building casting his entire shadow along the road and not just for a very small part like the screenshot. In that situation, as for the screenshot, the strong amount of direct sunlight will keep things alive, even on shadows. Looks at this ref shot (the bluish is WAY thinner than mine and your shot too): View attachment 15959 View attachment 15960
The photo gives a good idea how the ambient light "works" on surfaces facing in different directions. The underside of the roof receives only scattered light from the surfaces below and no blue ambient light from the sky. The red color is more saturated. The vertical surfaces interact with the blue ambient light which desaturates the red color. The vertical surfaces receive a mix of blue sky light and scattered light from the surfaces around. What would happen if the sun hides behind one of the clouds and the scattered light from the surfaces is reduced to scattered light from the blue sky? Light is an interesting topic. Here is a lecture with my all time favourite physicist and bongo player. Maybe it's a bit off topic but it helped me a lot to understand what light really is like and how it interacts with the stuff around us. I always watch this kind of videos before I nod off.
