Digital systems operate on bits, the smallest unit of information that any computer can understand. A bit can only have two values: 1 (on) or 0 (off). It's how the system works binary.
Combining bits in a different order, number and with different values, as is done composing any digital information. The more bits, more info.
Thus, the color in digital images, is built by bits. For example an image of a bit, will only target values (1) black and (2), so that its color depth or bit depth will be limited if not minimal.
As we increase the bit color of an image, we will have more values that represent it.
distintos, tenemos suficiente para representar el color de una imagen y que resulte real para el ojo humano. Usually 8 bits, equivalent to 256 different values, we have enough to represent the color of an image and make it real to the human eye.
These 8 bits are actually bits per color channel (RGB), ie when talking about an image of 8 bits, we refer to 256 levels of red x 256 shades of green x 256 shades of blue = 16,777,216 colors.
(8×3), y la podemos por tanto ver definida de las dos maneras. Caution should be exercised with this nomenclature, and must specify what they mean bits, as for example an image of 8 bits per channel, is the same as an image of 24 bits per pixel (8 × 3), and we can only much more defined two ways.
- RGB 8 bits per channel (bpc) = 24 bits per pixel (bpp)
- 8 bits per channel RGB + Alpha channel = 32 bits per pixel (bpp)
- RGB 16 bits per channel (bpc) = 48 bits per pixel (bpp)
DEPTH OF COLOR IN THE DIGITAL VIDEO
While we have said that with 8 bits per channel can get an image with a color definition sufficient for the human eye in this video is not always so.
Most digital video formats generate images of 8 bits, others are capable of reaching the 10 bits, providing more dynamic range image, in particular 1024 values per channel, compared to 256 provided by the 8-bit .
- Bit depth ---------- Video Formats
- 8 bits per component -> Betacam SP, DV, DVC-Pro, DVC-ProHD, HD Cam, HDV
- 10 bit per component -> HDCAM SR, Digital Betacam, D5, D6
But what gives us greater depth of color video if an image of 8 bits is sufficient for the human eye?
Greater color depth gives a broader range of performance on the image and can adjust much more accurate color touch-ups, etc ... Chromas overall color depth is beneficial to any post-production process on the material we work with .
que necesitaremos, haciendo el proceso de trabajo mucho más lento. But the price to pay when working with higher bit-depth, is very high in terms of processing speed of our machine and disk space you need, making the process work much slower. It is therefore essential to know when and how to use a deeper color to the needs of our project.
WORKING WITH VIDEO A 10BITS OA 8bits
As we have seen, most video formats work at 8 bits, which provides values from 0 to 255, where the absence of value (0) would be black and the maximum value (255) would be white.
But in reality, 8-bit digital video, white is at the value 235, while black is raised to the value 16. The values located between 236-255 and between 0-16 are reserved for the super-white (headroom) and the super-black (footroom).
Here you can find more information on the subject, since we shall not dwell much on this point: Black and white Levels
8-bit images are usually sufficient to simple editing jobs, with little post-production, for as soon as we begin to "tighten" the capabilities of 8-bit-based filters and advanced retouching, we encotraremos with the first and most serious problem : the posterization or color banding.
The banding, occurs mainly in color gradients where the 256 levels are not sufficient to represent the full range of colors. This is something inherent in the 8-bit images, what happens to any color retouching and if it emphasizes what we are not careful with the flow of work and at some point in the recompressed image editing with a codec, the result may much worse.
Banding can be prevented, or at least alleviate it, using various techniques.
If we assume 10-bit material in the possibilities for 1024 are minimal banding color levels will be sufficient to prevent it.
In case you can not have 10-bit recorded material can do the following:
- Snap 10bits codecs. For example the recent Apple Pro Res 422 codec 10 bits allows quite manageable file sizes while maintaining a color depth of 10 bits. Another highly recommended option is to acquire a codec Cineform which also permits HD timelines to 10-bit processor barely suffering because they are highly optimized to take little and offer a high quality.
In case of not having any of these codecs, we can serve any other 10-bit, the downside will be the huge files that we will be forced to work and the consequent slowdown of work, which is not highly recommended.
- Capture in the native format to 8 bits, but set the project to a bi t more depth. Thus, any additional retouching, digital gradient effect or we add will be generated in the new color space for our project. After Effect For example there is the option of setting the timeline to 8-bit, 16 bits or 32 bits. While the work process slows down dramatically, we will gain in quality, especially color gradients and blurs.
Eye, for work in a higher bit depth can slow down the work up to 4 times, sometimes making the process unworkable if we have a good machine.
One solution would be to convert the bit depth right at the end of the project, before rendering, making the new software calculates all the filters, effects and more, only at the end.
Still, the images from 8-bit sources, can still show a slight banding, because even if they are converted to 10 bits, originally have only 256 levels and that can not be varied.
In this case we can resort to old tricks that still work like a blur to add gradients and blur to add a little noise filter
POINT FLOAT 32 BIT
32 bit or float point (floating) is the maximum depth of color that we can get in a digital system. Here there is no possibility of banding artifacts because there is no color levels as the 256 of the 1024 8-bit or 10 bits. Exists only in float point values from 0 (black) to 1 (maximum brightness) with intermediate values which may be infinite. For example if a system of 8-bit gray value is represented by 128, in a float point The gray can be 0.5892345 ... with all possible decimal ... so that the color range is virtually infinite.
The images generated by computer (CGI) and the images from the 3D software are created in this color space, but when saving image formats of 8 or 10 bits are automatically resample.
A project set to 32 bits will allow maximum quality in all we add digital elements such as gradients, in the implementation of filters such as blur and allow the color corrections do not cause damage in excess of 8-bit images with which we are working , but, again, can slow down our work up to 400%. It is likely that a project set to 16 bits we have enough.
, aumentaremos enormemente nuestra capacidad postproduccir imágenes de calidad. More and editing equipment are more powerful and can take advantage of greater color depth, if we add to this work, Possibility of codecs such as ProRes or Cineform codecs will greatly increase our ability postproduccir quality images.
