[upd]
[upd]
[upd]
[upd]
Key lights: are the most bright lights in a scene. They define the dominant angles for shadows and illuminations.
Bounce Lights: They act as the reflection of any lightsource inside the scene. usually, they consist of wide spotlights, with very small brightness and a desaturated color, of the surface from what it is reflecting of.
Fill Lights: illuminate the other direction of the key light, they help lighting the "not lit by keylight"-surfaces. It is also reduced in intensitiy/brightness and you can try the complementary color of the keylight for your fill light. Or blue or grey for outdoorscenes. Softer not so dark shadows.
Rim Lights: define the outer shape of an object or a scene. Use a very bright light, perhaps cold colors and a light shadow
Final Image. To get a good feedback, of what your lights are doing, also always render them alone. Only then you can estimate the lighting effect.
Here the first lighting attempt after modeling. Lightsources, that you can physically see in a scene are so called: Practical Lights. As you see there are plenty of them. In this first attempt i created not only the keylights, but also their bounce lights and rims on the railing. And also some local fill lights. but the scene totally lacks of an ambient fill, which should represent the color of the sky in this case.
Here you see the ambient fill, which is blueish. The modeling was also somewhat cheesy, so i changed it.
Here we see another change in geometry, and the adding of volume fogs to the practical key lights.And last but not least, a whole scene fog, just to add some harbour feeling. But that was photoshopping, because i was too tired ;)
This is a Normalmap tutorial, i also published on my site. Perhaps it makes sense publishing it on the blog also.
Software: Photoshop, Nvidia Plug (Nvidia.com -> Developerzone - Photoshop Plugin)
Goal: Creation of a normalmap through a diffuse texture
Creation of normalmaps in Photoshop is not done by simply using the Nvidiaplugin only one time on a diffuse texture, but by applying the filter a couple of times at different blurring levels of the diffuse map.
Starting with the fine detail structures, you copy the source texture and blur it, to transform hard contrasts to gradients, to extract the more and more larger details.
The Nvidia plug wont convert hard contrasts without gradients to „deep“-structures.
The more you blur your ressource texture, the higher you set the „scale“-value in the Nvidiaplug
So lets start:
This is our ressource texture
fig1
fig2 (the y normal was inverted, because our game engine at Piranha-Bytes needs this, don't do this at home)
fig3
fig4
Workflow Part 1:
- duplicate your diffuse textur one time with „create new document from history" in history. (fig1)
- apply the nividia filter to one of the copys for getting a detail structure normalmap. Setting: Scale=4, Alternate Conversion=Max(RGB) (fig2) (the y normal was inverted, because our game engine needs this)
- put the normalmap aside for later compositing. (Tex1)
- get the other copy and desaturate (strg+shift+u)
- apply Gaussian Blur Filter with small Radius (2-3 pix) (fig3)
- duplicate your diffuse textur one time with „create new document from history" in history.
- apply the nividia filter, set „Scale " to 15
- put the normalmap also aside for later compositing. (Tex2)
- get the other copy
- apply Gaussian Blur Filter with higher radius (4-5pix)
- duplicate this document one time with „create new document from history" in history.
- apply the nividia filter, set „Scale " to 25
- apply Gaussian Blur Filter with small Radius (1-1,5 pix) to your normalmap in order to reduce moiré-effects, that came by using the gaussian-blur with high radius. (fig4)
- put the normalmap also aside for later compositing. (Tex3)
- get the other grey texture copy
- apply Gaussian Blur Filter with higher radius (6-7)
- apply the nividia filter, set „Scale " to 30-35
- apply Gaussian Blur Filter with small Radius (1-1,5 pix) to your normalmap in order to reduce moiré-effects, that came by using the gaussian-blur with high radius. (fig4)
- This normalmap is our base for the final normalmap. It defines the largest structures. (Tex4)
We are now going to compose the 4 different normalmap levels together by using the „overlay“-blending. One after another over the base-normalmap.
But after every single „overlay“-blending, you have to apply the nvidia filter once again in the „alternate-conversion“ setting: „normalize only“ in order to reduce wrong normalmap colors, that appear after every „overlay“-blending.
fig5
fig6
Workflow Part 2:
- Copy all content out of [Tex3] (ctrl+a, ctrl+c) as a layer into [Tex4] (ctrl+v)
- use blendingmode „Overlay" or „Soft light" (for less hard blending) on the new layer (fig5)
- flatten image to background layer
- use nvidiafilter with „Alternate Conversion"-Mode: „Normalize only" (fig6)
- Copy all content out of [Tex2] (ctrl+a, ctrl+c) as a layer into [Tex4] (ctrl+v)
- use blendingmode „Overlay" or „Soft light" (for less hard blending) on the new layer (fig5)
- flatten image to background layer
- use nvidiafilter with „Alternate Conversion"-Mode: „Normalize only" (fig6)
- Copy all content out of [Tex1] as a layer into [Tex4]
- use blendingmode „Overlay" or „Soft light" (for less hard blending) on the new layer
- flatten image to background layer
- use nvidiafilter with „Alternate Conversion"-Mode: „Normalize only"
If the Texture should be tilable, you now have to inspect your seams. And perhaps correct them.
Use the „offset“-filter in order to do this. Put exactly the half of your pixelsize of your document into the fields „horizontal pixels“ and „vertical pixels“.
Then use the blur tool or copy stamp tool, to reduce your seams..after this use the offset filter again. And then also use the nvidia plug again in „normalize only“-mode to for color corretion.
Now your normalmap is final.“
I hope this was helpful to you! :) cya
sascha
