Go download TG2 http://www.planetside.co.uk/content/view/21/36/
Open TG2, and open the template file included below
http://www.mediafire.com/?gwgniou1ixn
On the right side, you’ll see the preview window, it’s probably all dark (well gray). It currently has the lighting and clouds turned off. Turn them on by clicking the buttons on the top left corner of the preview window area – the cloud looking button and the sun looking button.
Now you should be able to see a preview of the sky. It’ll look strange at first, because of the way TG2 renders it. As the percentage complete (upper right corner) increases, it’ll look more and more like a sky.
Ok lets start by decreasing render time. Click on atmosphere, then atmosphere 01 (upper left corner window). Go to the “quality” tab, and turn the quality way down. Do the same for altocumulus 01. We’ll dial these numbers back up for our final rendering. But for previews, you don’t need them so high.
Let me start by explaining the final rendering process. If you click on “Cameras” at the top, you’ll see a long list of cameras – North, South, A1, A2, B1, B2, etc. The directional camera’s (east, west, etc) are useful for doing quick preview renders. They will show a 90* FOV of the sky. The camera’s you will use for your final renders are A1, B2, C1, etc. The only camera’s that are truly necessary are A1-8, B1-8, and C1 and C2. The other cameras may be useful in stitching together your final sky.
How do you use a camera? Well, you can use it in the preview window. In the lower left corner of the preview window there is a button “select different view cameras”. Pretty straight forward right? Just pick a different camera to see a different preview of your scene.
You can also click on the “Renderers” tab. You will see 3 renderers already set up for you. I use “Full Preview”, along with a directional camera, for a quick preview rather then using the preview window. To do this, click on the “Full Preview” renderer, select your camera, and hit render image.
Final rendering is done with the” Final Render” renderer. What we are going to do is render a minimum (20) 800x600 images. For each render, you will have to manually change the camera from A1 to A2 etc etc until you’ve rendered all of A1-8, B1-8, and C1 and C2. (you should be able to skip the BC camera’s too). I have had the best luck saving these final renders as .bmp files. They each take 15 minutes to an hour (or more) depending on your computer. So yes, it will take a long while.
Make sure you increase the quality level of your clouds and atmosphere before you do final rendering.
But Garasaki, why in the world do we need these 20 images??? Well TG2 restricts the free version to a maximum image size of 800 wide by 600 tall. So if you just did east, west, north, south, and up, you’d end up with 600x600 pixel images. Its my opinion that the game requires images of 1024x1024 to make a decent looking sky. So in order to accomplish this, we are going to render small snapshots of the sky and stitch them together. Each of the images these camera’s render is a small shot which actually duplicates some of the information of the other shots – this helps stitch them together later.
It’s useful to understand how the images are arranged. All the A images are next to each other, with some overlap, all the B images are next to each other with some overlap, the B images sit on top of the A images with some overlap, and the C images are pictures looking straight up, with some MIRRORED overlap in the middle. These should also overlap the B images. For instance, if you put A1 (on the left) next to A2 (on the right), you’ll notice the images duplicate a vertical column of information in the middle. If you put B1 on top of A1, you’ll notice the images share a horizontal column of information in the middle.
So basically, the A images are a 360 strip of your sky, at the bottom, the B images are a 360 degree strip, closer to the “top” of your sky, and the C images fill in the blanks at the very top.
Making changes to the cameras and renderers I have setup have the potential to ruin the entire thing - mess with these at your own peril.
We will stitch these 20 images together using “hugin”. Get it here http://hugin.sourceforge.net/
I was never able to get the current version to work so I use a legacy version, 0.7, available here
http://sourceforge.net/projects/hugin/f ... e/download
Hugin, hypothetically, is simple. You just add your 20 or so images, tell hugin your camera’s FOV was 60, and then hit “align images”.
Best case scenario is that hugin is able to align your images perfectly, then you simply export your panorama. You will know you are this point when hugin pops up a panorama preview, you inspect it, and it looks 100% complete. Then close down the preview window, and click on the last tab (“stitcher”). Then click “calculate optimal size” for your Panorama Canvas size. Then choose the file formats you want and click “stitch now”. Name your file, let hugin do it’s thing, and you will have a completed skydome.
Worst case scenario with hugin is that, when you try to align your images, hugin will tell you it couldn’t align them automatically. This happens when your color variations are slight and subtle or you have large expanses of the same color. At this point you have 3 options:
- Try rendering the BC cameras and or the C3, C4 cameras, and adding those images to hugin to add some new control points. I’ve not had much luck with this.
Manually input your control points between different images/groups of images. This will make your head spin the first time you do it – but if you understand how the images are linked its not so hard.
Start over and change your scene to include more contrasty colors.
At this point, you can use JLV’s skydome batch file to create the skybox images the game uses.
http://mxsimulator.com/benchracing3/vie ... f=2&t=2868
You will also want to use some of the techniques DJ outlines here
http://www.mxsimulator.com/benchracing3 ... =11&t=1916
to create a lighting file that matches your new sky.