So I've been revisiting what it would take to make a video overlay system. My preferred route woudl have been FPGAs, but the hardware, development tools, and IP cores run the cost into the rate of $10K so that's obviously off.
My understanding of the structure of the video signal is somewhat lacking, but not non-existent. It'd be nice if someone knew of any links clearly describing (visually) what the structure of the composite and RGB video signals were.
It would seem that I am still limited to a microcontrollers...probably an ARM Cortex which is lacking for something like video. I considered that it might be possible to simply override the the analog video signal values at the correct times by using pull resistors/transistors to overlay the data without needing to digitize and pass all the data through the MCU which would be lots of processing power. I wasn't sure if this idea was valid so I went searching for similar circuits and yay I found one that seems to work this way.
I came by some RGB ADCs and DACs though that do digitize the data. THey seem "straightforward" to implement. I was thinking that perhaps a similar method could be used as I just described, except it would be used on the digitized RGB values rather than the analog values to provide more colour than usual on the overlay while still keeping processing power to a minimum. See my attached block diagram for what I mean. Let's assume we'll have the required composite to RGB and vice versa converters on the inputs and outputs of the system so that it works with most video sources and display devices.
As you can see in the attached diagram, a half bridge is applied to each of the digital RGB lines so that it could be overriden to be a 0 or a 1 or left floating so that the original digital value remains unchanged as it is passed to the DAC. In this case, because of limited MCU processing power all half bridges on each colour bus are tied each other so that when an override occurs the colour channel gets set to 0% or 100% (ie. no adjustments of colours by independently overriding each RGB colour bit). THen the possibly override colours would be the most intense hues of red, green, blue, black, and white (and I guess orange, yellow, and purple too). Basically all colours where all RGB components are either 100% or 0%.
THe timing information that the ADC derives has been left out just to make things simpler. But obviously the DAC and the MCU would have to be reading it so the MCU would know when to apply the proper pull-down/up overrides.
Is this concept sound? DOes anyone see any problems? It's jsut that graphical overlay hardware is just so damned expensive and in almost all cases either far too decadent for my purposes or far too lacking.
My understanding of the structure of the video signal is somewhat lacking, but not non-existent. It'd be nice if someone knew of any links clearly describing (visually) what the structure of the composite and RGB video signals were.
It would seem that I am still limited to a microcontrollers...probably an ARM Cortex which is lacking for something like video. I considered that it might be possible to simply override the the analog video signal values at the correct times by using pull resistors/transistors to overlay the data without needing to digitize and pass all the data through the MCU which would be lots of processing power. I wasn't sure if this idea was valid so I went searching for similar circuits and yay I found one that seems to work this way.
I came by some RGB ADCs and DACs though that do digitize the data. THey seem "straightforward" to implement. I was thinking that perhaps a similar method could be used as I just described, except it would be used on the digitized RGB values rather than the analog values to provide more colour than usual on the overlay while still keeping processing power to a minimum. See my attached block diagram for what I mean. Let's assume we'll have the required composite to RGB and vice versa converters on the inputs and outputs of the system so that it works with most video sources and display devices.
As you can see in the attached diagram, a half bridge is applied to each of the digital RGB lines so that it could be overriden to be a 0 or a 1 or left floating so that the original digital value remains unchanged as it is passed to the DAC. In this case, because of limited MCU processing power all half bridges on each colour bus are tied each other so that when an override occurs the colour channel gets set to 0% or 100% (ie. no adjustments of colours by independently overriding each RGB colour bit). THen the possibly override colours would be the most intense hues of red, green, blue, black, and white (and I guess orange, yellow, and purple too). Basically all colours where all RGB components are either 100% or 0%.
THe timing information that the ADC derives has been left out just to make things simpler. But obviously the DAC and the MCU would have to be reading it so the MCU would know when to apply the proper pull-down/up overrides.
Is this concept sound? DOes anyone see any problems? It's jsut that graphical overlay hardware is just so damned expensive and in almost all cases either far too decadent for my purposes or far too lacking.
Last edited:
