Theorizing how to replace a VIC-II by hardware.

Well... I'm starting this as a discussion about what would be needed to 
replace the VIC-II with hardware that would emulate it and 
expand/enhance it.

For a start, I'm a reader. I read, ideas comes to my mind and then I 
ask, so I can learn from people that actually know.

So...

AFAIK (and I'm telling this for people to know what is what I do know, 
accepting corrections of course) , the VIC-II do cares about

- Display text and graphics.
- DRAM refresh.
- System clock generation.

--For the first, it reads from the CHAR ROM for text, or from RAM when 
instructed to do so, mostly by KERNAL but also by custom routines. For 
graphics, it reads from RAM, and is the task of the CPU how and what to 
put in RAM for the VIC-II to render it as graphics (actually the text 
does also works this way).

- For the second, the VIC do read data from the RAM chips just for them 
to force their contents not to be lost.

- For the third, the VIC do send a clock pulse from one of its pins to 
one of the pins of the CPU, that then outputs this signal to the rest of 
the system.

So its inputs are power, of course, the clocks generated from the 8701 
or equivalent circuit, and the address and data bus.

And its outputs are the video signals, the DRAM refresh lines, the IRQ 
lines, the BA line to control the CPU and stop it when needed.

Then, there is the read/write pin and the chip select pin as control pins.

So: what would be needed to emulate first of all for a C64 to believe 
that the VIC-II is still there?

For me:

- The clock output. The CPU, and the rest of the system, won't work 
without it.
- The DRAM refresh lines. DRAM contents won't last without it.
- The raster line register. There are things the CPU does when the 
raster line is in certain locations, like triggering IRQs.
- The missing thing I am not aware of and that you are telling me.

With an FPGA or a PIC fast enough, maybe there would be a way to 
implement this **as a first measure** to make a C64 *think* there is at 
a logical level a VIC-II there, because at an analog level, the system 
doesn't care about if there is a video signal coming from it or not, or 
if the graphics are actually rendered: the CPU just read or write 
registers, gets controlled by the BA line or IRQ, and gets ticked in by 
the clock pin.

The next step would be reading from the proper places (ROM, RAM or COLOR 
RAM), and render the bitmap image taking care of the effects supposedly 
to be applied by the raster line tinkering that the demos do (of course 
sprites also have to be rendered).

And finally, get that bitmap converted to an output signal that would be 
in a format that could be converted to *real* HDMI (digital) or 
component / svideo (analog) video. It could be even encoded inside 
network packets to have a VIC that could be remotely viewed through a 
network connection. This final stage could get rid, maybe, of the 
PAL/NTSC (and SECAM) limitations, because digitally it could be upscaled 
and up-frame-rated to modern standards, applying filtering - or just 
having a 4K version of the classic VIC graphics with the pixels upscaled 
to the size of actual bricks. ;)

NOW is your turn (that I would actually appreciate as I love to learn)

- What do you think?
- Is it doable?
- Does it worth the work to do it?

Thanks all.

--

Saludos y cerveza.