日出日落場景制作——第二部分

第二種情況有點棘手。 讓我們從完成的場景開始，除去太陽以外的所有燈光。 如果我把太陽放在我想要的地方，并渲染，顏色都是錯誤的 - 冷，藍，，而不是戰(zhàn)爭的棕色和橙色。簡單來說，太陽太高了，沒有合適的暖色。

The second scenario is bit more tricky. Let's start with the scene I've finished, and remove all lights except the sun. If I set the sun where I want it, and render, the colors are all wrong - cold, blue, instead of war browns and oranges. Simply the sun is too high to have proper warm color.

如果我將VraySun換成標準方向燈，我可以完全控制它的顏色。 我用日落的HDR照片替換了VraySky（在Vray覆蓋標簽中進行照明）（說實話，即使沒有它，場景也可能工作，因為它的效果是微妙的，大部分的照明將是 但是，這是一個起點），碰到了主要反彈乘數(shù)，玩AO設置和Vray相機設置- 顏色開始看起來是正確的，但場景是太黑暗。

If I swap the VraySun for the standard Directional Light, I can have full control of its color. I replaced the VraySky (the one doing the lighting, in Vray override tab) with a HDR photo of a sunset (To be honest, the scene would probably work even without it, as its effect is subtle, and most of the lighting will be done by hand. Still, it's some starting point.), bumped up the Primary Bounces multiplier, played with AO settings, and Vray camera settings  - and the colors start to look right, but the scene is way too dark.

街道的墻壁處于陰涼處......他們可能應該這樣做，但是我希望他們能夠捕捉到一些光線，所以我在拱門下方放置了一個壓扁的球形Vray Light。 正確的墻壁有一個輕微的反光材料（作為蟲漆添加到基礎著色器），所以這里有一個很好的細節(jié)圖案。

The walls of the street are in shade... as they probably should, but I'd like them to catch some light, so I put a squashed, spherical Vray Light under the arch. The right wall has a slightly reflective material (added as a Shellac to the base shader), so there is a nice detailed pattern there.

另一盞燈放在屋頂上方，向右側(cè)墻上投射一些背光。

Another light was placed above the roof, to throw some back-light on the wall on the right.

而另一個相當大的放置在街道上方，模擬了來自天空的光線- 但還不夠，所以攝像機上方還有另一個更大的面向場景，提供了一些補光 向前的部分。 使用大面積燈具有一些優(yōu)點 - 您可以添加來自某個方向的光線，但是沒有會使圖像混亂的尖銳陰影，并且在CG，點/光點分布方面沒有特征，這并不總是需要的。 除此之外，它在現(xiàn)實世界中的作用就像一個巨大的柔光/反彈，無論是在攝影工作室，還是在電影中。 如果陰影的采樣不夠好的話，那么缺點是渲染時間，有時還會產(chǎn)生噪聲。

Yet another, quite big one, placed above the street, simulates the light coming from the sky - but it's not enough, so there's another, even bigger one above the camera, facing the scene, providing some fill light on the forward-facing parts. Using big area lights has some advantages - you can add light coming from certain direction, but without sharp shadows, which would clutter the image, and without a very characteristic in CG, point/spot light distribution, which is not always desired. Besides, it works like a big softbox/bounce in real world, either in photography studio, or on a movie set. The downside is rendertime, and sometimes noise, if the sampling of the shadows is not good enough.

在最終渲染之前，我調(diào)整了屋頂上方的背光，使其稍微飽和一點，再稍稍旋轉(zhuǎn)一點，真的沒什么大不了的。 這里還有另外一個問題 -角落里有輕微的出血。 為了解決這個問題，我可以使用更高質(zhì)量的輻照度地圖，或嘗試使用細節(jié)增強（以犧牲渲染時間為代價），但由于圖像將以高分辨率渲染，因此我現(xiàn)在的設置應該足夠了（以高分辨率， 即使低輻射圖設置提供足夠的信息，以獲得一個干凈的渲染）。 事實上，我已經(jīng)把它們放低了，但是需要做一些小小的修改。

現(xiàn)場缺乏的是一些氣氛 - 我把它留到了后期，進入后期制作階段。 我做了一個Zdepth pass，并在Photoshop中添加了一些小霧，以及其他一些簡單的調(diào)整。 - 最后的圖像看起來像。

Before final rendering, I tweaked the backlight above the roof a bit - less saturation, bit more intensity, and rotated it a bit - no big deal, really. There's another problem here - a light bleeding in the corner. To fix that, I could use higher quality Irradiance Maps, or try to use Detail Enhancement - at the expense of rendertime - but as the image will be rendered in high resolution, the settings I have now, should be enough (at high resolutions, even the low Irradiance Map settings provide enough information to get a clean rendering). Actually, I've lowered them even more, but that required some slight fixing to be done.

What the scene lacks is some atmosphere - I left that for the very end, for the postproduction stage. I rendered a Zdepth pass, and added slight fog in Photoshop, along with some other simple tweaks. - and the final image looks like.

看到這兩種方法，我們可以得出一些結(jié)論。 自動的太陽/天空系統(tǒng)是一個很好的起點，在某些情況下，它本身可能是足夠好的。 但是它的表現(xiàn)并不總是足夠靈活，一些場景將會從幾個放置良好的額外燈光中獲益匪淺 - 有些燈光將主要由手工照亮 - 一旦有了明確的方向，就不那么難了 你想達到什么目的。 這就是一些研究可以非常有用的地方。

Seeing those two approaches, we can draw some conclusions. The automatic Sun/Sky system is a great starting point, and in some cases, it's probably good enough by itself. But as good as it is, it is not always flexible enough, and some scenes will greatly benefit from few well placed additional lights - and some will have to be lit mostly by hand - which is not that hard, once you have a clear direction of what you want to achieve. And that's where some research can be very helpful.