TV static is really quite a clever optical trick. It's almost more mathematics than light.
The static appears at first glance to be random noise, but if you take any given frame and multiply the brightness of each pixel with that of the same pixel from one frame previous, then the noise begins to stabilize into something that could eventually resemble a coherent image.
The process of resolving these frames into a single image can't be done with one transformation. In reality, one must perform this same operation with every new frame over thousands of iterations before it begins to look like anything other than noise.
This can be done easily with modern computers and the employment of simple rendering algorithms, but the process is dependent on the pace at which the TV static moves from one "frame" to the next. It becomes tedious quickly, so our team automated it using a simple camera and tracking software.
Over the course of several hours we were able to resolve one relatively clear still image before it began to diffuse back into noise. The algorithm continued, and a few hours later the same image resolved once more. Or so we thought.
In the spirit of thorough investigation, we compared the two images using graphical analysis software and found several minute inconsistencies. One particularly bright member of our team hypothesized that we might not just be viewing images, but instead might be looking at frames of an animated recording; the inconsistencies could be explained by movement happening between the two frames.
Over the course of the next several weeks we continued to compare the images produced and confirmed that they did appear to be approximating movement from one to the next.
With no way to determine the length of the sequence we had begun to unearth, we had no choice but stretch our funding as thin as possible and prolong the study. We let the algorithm run for months without pause, working in the dark to reduce energy costs, our workspace lit only by the dim glow of the CRT television screen and the LCD readouts of our equipment.
It wasn't until approximately 14 months into the study that we reached something we were confident calling a "breakthrough." A point at which the images produced seemed similar to the first readings we took. The images become more and more similar to our first recorded frame until eventually, and we checked this over and over in our analytics software, we appeared to have looped back around to the beginning. The sequence had concluded and started again, and we had recorded it in its entirety.
The recording itself was eventually composited into a demonstration to be distributed to any scientific organization or government that might have an interest in our findings. The recording was not publicized until nearly six years later.
Today the footage is preserved here in its original quality.
