Creating A Star Wars Starfield



A completely digital starfield generated by my starfield program.
This starfield contains 2000 stars.

Starfield images are something I'll have a great need for once I get my Star Destroyer model competed. I have actually been trying to create a realistic starfield for 10 years now and I have tried many different techniques to generate them, with varying degrees of success. This webpage details all my past attempts to create starfields as well as my latest, and probably final, computer based solution.

Creating Star Wars (Non-Hyperspace) Starfields

Before proceeding, a little history about how I believe non-hyperspace starfields were dealt with in Star Wars is in order. It should be noted that I have no firm knowledge about what I'm going to say, I just vaguely remember reading about this being the way that starfields were dealt with in the original trilogy.

I believe that for general shots of starfields, both static and moving but non-hyperspace, they were actually created and photographed by means of placing a starfield on the inside surface of a large backlit vertical plastic cylinder that may have been many feet in diameter and height. A camera was then placed on a long vertical rod of some sort so that the nodal point of the camera was aligned with the center axis of the cylinder. The camera had two axes of movement including vertical up/down movement as well as horizontal pan. By placing the camera inside the cylinder at this position, focusing it on the interior starfield wall of the cylinder and combining various movements along theses two axes, the camera could record the starfield as ships would later move by the camera.

Past Non-Computer Based Attemps

My first attempt (1991) at creating a starfield involved me using a series of differently sized straight pins to literally punch hundreds of small holes in a large piece of heavy black construction paper. I then rear projected a light onto the back of the paper and photographed the front. The results where not very realistic as I tended to get a concentration of very bright stars at the center of the image and a trailing off of star intensity as you went out to the edges. This was due to the fact that my light source was essentially being projected as a cone of light with the brightest part at the center. It was at this point that I gave up on rear projection to light up stars.

After this first attempt, I switched to a front projection type of lighting. In this scheme, I placed the light source on the same side that I was photographing the stars. Now I was using the light source to bounce light off some material placed on an 11x17 inch piece of black lithographic plastic film. You can find this lithographic plastic at most image processing labs. Selection of a star reflection material was the next problem.

For many years I used aluminum dust sprinkled on the black lithographic plastic as my actual stars. This aluminum dust is the same material that I use as the twinkle effect seen during my transporter special effect. But using this material for starfields has its problems. First, for aluminum dust to produce bright enough stars to be viewed, the material requires that the light source be pointed almost straight down at the lithographic plastic which is lying on the ground. This introduces some reflection problems as the camera must also be pointed straight down on the plastic as well. Through the use of prudent light placement and black velvet and black construction paper placement around the camera and tripod, this problem can be largely mitigated.

I also found that aluminum dust results in stars that are to large as well as having an irregular shape. This problem still persists even when I screened the aluminum dust so that I was only using the smallest dust particles. In addition, I found that because the light source had to be placed at a rather steep angle of about 70 degrees to the horizontal ground and within about three feet to the plastic, I get those annoying variations in reflected light intensity.

In 2000 I decided that I could solve all of these problems by finding a star material that is semi-transparent. I looked at three very simple materials to help me solve my problems. I first looked at just sprinkling flour on the plastic. Flour has the problem that it tends to want to clump together when you spread it on the plastic. It also has the problem of being to fine grained as well. I then tried using ordinary table salt. I found that salt works much better because it is a semi-transparent material and it has a very consistent crystalline shape which makes nice star points. The only problem I found with salt is that it's crystal size creates stars that are a little to big.

My Current Non-Computer Based Solution

The last material I tried was regular table sugar. I found this to be just about an ideal material to sprinkle onto my plastic sheet. It's crystal size is just about right size and it is semi-transparent. Semi-transparency is a very important characteristic because using a material like this meant that the light source I used to illuminate the crystals didn't have to be pointed straight down at the crystals. In fact, when I photographed the sugar starfield, I actually placed the light source on the ground completely perpendicular to the camera's photographic axis. Sugar's light diffusing characteristics also resulted in very brightly lit crystals and this meant that I could place the light source about six feet away from the plastic and still get very bright stars. And being able to move the light source farther away from the starfield resulted in the elimination of bright spots on the starfield as well. And since the light source was throwing light that was parallel to the plane of the plastic sheet, that meant that there were no bad reflections coming up off the plastic itself. I did still have to use black velvet and construction paper to cover up portions of my camera's tripod, but these efforts were minimal. Below is the actual setup I used to record the above starfield.

(Image)

Non-Computer Based Results

All and all, I am very happy with the results that I'm getting with my sugar starfields. It is interesting to note that I used the above sugar starfield image as the input image into my hyperspace jump point generator special effect described elsewhere. Isn't it amazing that you can use such simple materials to create such a spectacular effect! Below is a list of the materials that I used to create this effect.

My Current Computer Based Solution

Relatively soon after I came to my sugar based solution to generating starfields, I decided that computer generated starfields would ultimately be the best way of generating starfields for several reasons. But since I didn't need any starfields immediately, it wouldn't be until January 2002 that I would turn my attention to creating a program for this purpose.

Advantages to Computer Generated Starfields

Digitally created starfields has several advantages to physically created starfields. First, with a starfield program I can specify as big a starfield as I want. Should I choose to shoot motion picture film with a model, I can now create a starfield that is large enough for me to actually pan and tilt across as the model moves past the camera.

Another advantage that digital starfields have over physical starfields is the sheer setup time required to produce the starfields. I could garentee myself that to shoot a physical starfield would require three or four hours. Creating a digital starfield takes less than a minute from start to finish.

Another advantage to shooting digital starfields is that I completely control the composition of the resulting stars in the image. That means that I can specify the look of the stars in the resulting grayscale image so that they will play nicely with my hyperspace generator program, should I want to generate a hyperspace jump from the starfield.

The Program

I have included the actual starfield generator program code here for you to study and use for your own personal use. Click here to view the source code. There are some notes below that you will want to understand.

Notes:
----------------------------------------
The resulting starfield images are single channel raw format images.
The starfield background is absolutely black.
The programmer can specify the starfield's width and height, the number
of stars to be in the image, the percentage of the total for each
star size and where the resulting starfield file is to be stored.

Last Updated: February 3, 2002
HTML URL: http://geocities.datacellar.net/~special_effect/starfield.html
E-Mail: special_effect.geo@yahoo.com or click here
1