Okay... can do... drew up a nice schematic:
R16..R19 are on the Oric mainboard (no changes there), R15 is replaced by this mod. RV1 is removed, the other components are added. RCA socket is the one from the RF can.
About the weak composite signal going into the RF modulator: looking at the Oric's schematic, one can see that R16..R19 in the 'D/A converter' are much higher values than the resistor divider formed by R15 (originally 62 Ohm) and RV1 (220 Ohm). This means that the digital outputs from the PROM (IC23) have a relatively small effect on the voltage on R15/RV1 point. Hence the trouble people have had trying to use this as a video signal directly. Transistor-buffered or otherwise.
BUT: since IC23 has TTL-level outputs, voltage swing on those outputs is much larger (probably from around 0.4V to 3.5V or so).
Which made me realize a simple solution: mess a bit with the 'load' on the right side of R16..19 network. I did this by changing R15 to a much higher value, removing RV1, and putting result into an emitter-follower to amplify the output current (not the voltage!).
Components:
R15 isn't critical. I started with 1.8 kOhm, but found that colors were a
bit more saturated with a lower value. Much lower & the voltage swing will be reduced. Much higher and DC bias for the transistor's base may become a problem when the PROM's outputs pull it towards 0V.
Transistor itself is an ordinary small-signal NPN type (just be sure to check it has a decent gain/bandwidth). I planned to use a BC550 but found I was out. Thus picked a BC546B instead. There are many types that will do the job.
R1 is a bit of a compromise. A lower value increases the drive power for down-going parts of the output signal, but increases power consumption of the circuit. 150 Ohm is a reasonable (and often used) value for this purpose.
R2 serves to match output impedance of the circuit to the coax cable. A bit less than 75 Ohms is needed. I re-used R15 (62 Ohm), not
quite optimal but does the job.
C1:
within reason, basically the bigger the better (I used 470 uF actually). Check polarity: + goes to transistor side, - goes to the RCA center pin. There's only some 2V over it, so eg. a 10V rated capacitor is fine. Strictly speaking it can be left out, but that puts a DC bias on the outgoing signal, which isn't how it
should be.
C2 is optional. It serves to smooth the edges of the signal before it goes into the cable, which usually improves image quality
slightly. Too high a value, and characters become a blur. Instead of C2 one could also put a much smaller capacitor (around 22..47 pF or so) between GND and the transistor's base. While drawing above schematic I found my mod actually has R2 and C1 swapped. This shouldn't make much difference I think (if at all).
Be sure to check where signals go on the board, and what happens around the RF can! I isolated both 5V and video signal going in, and inside disconnected whatever was on the RCA center pin. Then routed that pin to the mod circuit as shown in schematic. This also takes RF modulator's power consumption out of the picture. If you're feeling adventurous, you may try & obtain both composite
and RF output.
Where you put the extra components is up to you. Possibly they can fit in the RF can. I thought it was
too tight a squeeze (and didn't feel like ripping the RF can's contents out), so placed the parts where RV1 used to sit.
P.S. Image quality is useable but far from perfect! Screen is stable, characters readable, colors are there, and that's about it (I tried with 3 TV's so far, all produced a stable image). Colors look washed out compared to RGB output. And depending on color combinations, plenty of dot crawl to go around...
But most high-contrast combinations look decent imho.