NTSC, the Amiga, and Some Little-Known Abilities

Methods to extract the most information from an NTSC signal, and some of the amazing effects that can result

By Bill Graham, Graphics Editor, nucmong@primenet.com

The Amiga has some special capabilities, as almost every user knows. All of us know about the Amiga's custom chips, its unique graphics and sound power, and more. But there are ways to use an Amiga that are not documented. For instance, most of us know that the Amiga is an NTSC- or PAL-compatible machine. And most of us know that means the Amiga is ideally suited to output either one of those video standards without much of the costly hardware necessary for a Mac or PC to accomplish the same thing.

I'm going to discuss the NTSC standard somewhat in reference to the Amiga, but most of this will apply to PAL users also. The current NTSC standard was agreed upon in the early 1950's to accomodate the then-new color television signal. What was possible even then and what we have now in regards to definition and picture quality are very far apart due to the government's insistence that the new signal be compatible with what was the then standard black and white television signal.

Every engineer I've known hates to have to design to older, inferior technology, especially when it is relatively easy to come up with something much better for the same amount of effort. And television engineers in the 1950's were no different. The result is that every color televison signal, whether broadcast, over cable, from live camera, or tape has significantly more information encoded into it than is readily visible. Much of this information was impossible for most people to access until the advent of affordable NTSC-compatible digital technology, namely, the Amiga.

This is due to the nature of the NTSC signal, which outputs 30 frames per second, with two fields per frame, for a total of 60 images per second. With two fields per frame, what is seen every 30th of a second is an interlaced image, with odd and even numbered horizontal scanlines. Since this happens 30 times every second, what we see is a continous motion.

The extra, invisible information encoded into the NTSC signal can be divided into three categories: Color, Detail, and Temporal. First, we'll discuss Color. The Amiga started out with a 12-bit color capability that was unmatched by any other machine. Over time, 24-bit color became the standard. The difference is the number of divisions, or shades, there are in the RGB (red, green, blue) color triplet. With the ECS color palette, there are 16 shades of each. 16x16x16=4096, for 12-bit color. With 24-bit color, there are 256 shades of each color, so that 256x256x256=16.7 million colors, give or take a few. For the Jay Miner historians out there, remember that the old 8-bit Ataris had 8 shades of each color, 8x8x8=256 colors total. The brightest pure red on an ECS machine, 15,0,0, is equivalent to the brightest pure red on a 24-bit machine, 255,0,0, or an old Atari 7,0,0 (remember the Setcolor command, anyone?).

These are RGB digital color values, and NTSC is an analog standard. Its color range, in comparison to the range of the RGB palette, is very limited. It can vary some, but most TV signals, when converted to RGB, fall into the 20,20,20 to 200,200,200 range in 24-bit color space. It can do a brighter white than this, but cannot handle the subtle gradations that RGB can. If you do the math, you'll see that the NTSC signal is roughly equivalent to 5.8 million colors.

In both cases, RGB digital and NTSC converted to digital, the number of colors far exceeds that which the human eye can differentiate. The range of color (note singular) with both is well within human perception, however. So how do we tease the extra color information out of an NTSC signal? By digitising the frames/fields, and "expanding" the limited NTSC converted colors into RGB colorspace. This is done with software like ImageFX and its Dynamic Range command. ADPro, Image Master, and other image processors have this ability also. By expanding the colorspace from NTSC to RGB, you are expanding the total color range. What can this do? Well, detail that was hidden in shadow can become visible. Areas that seemed to be a fairly uniform color now have subtle gradations. Also, having this expanded range allows you to "push around", or bias the middle color ranges via Gamma and Contrast adjustment, bringing out colors and patterns that were previously invisible. There are lots of possibilities with a good image processor, and the Amiga has the best.

The second type of hidden information in a digitised NTSC signal is Detail. Very soon after the Amiga was first introduced, its HAM color capability was recognized. It wasn't long before someone realized that digitised video was very expensive in processor overhead to animate, even on the Amiga. Fortunately, you can choose between an interlaced and non-interlaced display, thereby halving the amount of information the Amiga has to handle. And, an amazing thing was discovered in the bargain. A non-interlaced animation made from digitised video could actually look better on playback than the original video. Whether it did or not depended on a couple of things. Remember, each frame is actually two images superimposed. Either one of these images is sharper than the two combined. The difference would be the same with a film camera set to 1/30th of a second exposure. If the exposure was actually one image taken in 1/30th of a second, it would be sharper than if it was two images taken 1/60th of a second apart and merged. If the frame was a scaled frame-grab, i.e., a 364x452 HAM image scaled to 364x226, there was no difference in quality, unless the color expansion discussed above was done. But if the original frame was Deinterlaced, and the resulting animation was made from all Odd or Even fields, then the resulting playback was noticeably sharper than the original video. With the advent of HAM8 and the possibility of high resolution-high color animation, the difference was even more pronounced. Combining these two concepts, expanding color space and sharpening the animation by removing half the image soon revealed the way to extract the third type of hidden information from the NTSC signal.

Temporal resolution can take some time to comprehend. In reality time is analog. People apply arbitrary measurements to it, but time is analog just the same. But for movies, video, and computer animation, there are many different measurements. Movies are shot and projected at 24 frames per second. Even though they are shown on a huge screen, the motion is smooth and the individual frames cannot be seen. NTSC television is actually 60 images per second, interlaced as described above. Compared to movies, 60 images per second seems like overkill. Well, overkill equals greater temporal resolution. Using your Amiga, a frame grabber, and ImageFX or some other batch processor, you can grab a sequence of video, deinterlace it and save the odd fields numbered sequentially with odd numbers, then processing the frames again, this time saving the even fields numbered with even numbers. If you have a field capable video board like the VLab Motion or a PAR, this process is even easier, as they will save the fields in sequence. Taking video that was recorded at 60 images per second and playing it back as animation at 24 images per second is an incredible experience. It is not a simple matter of things occurring two and a half times slower. There is a fluidity and smoothness of motion that is both amazing and hypnotic. The normal blurring of hair blowing in the wind is greatly reduced or eliminated entirely. Eye movement is much more discernible. All the things we have come to know as a fast blur like a flag waving or a tree moving with the wind become a razor sharp, easily followed event.

By combining all three of these techniques using a capable batch image processing program, the Amiga user can extract all the information from a given NTSC signal. With enough RAM or a fast enough hard drive, an AGA HAM8 animation can be compiled that is much more detailed, much more colorful, and much more fluid than the original video and is a wonder to behold, even if it's just some digitised wedding video.

But, other than artistic merit, what is the purpose of all this? Well, obvious security or surveillance purposes come to mind. Sociologists and psychologists could possibly gain some insight with this new perspective. City traffic managers could perhaps find these techniques helpful in dealing with traffic problems. It's hard to put a finger on exactly what a new perspective can do. But with an Amiga and some extra hardware and the right software, the power is there to experiment with. And to top it all off, it is truly a lot of fun.