The way of generating .wfn or .molden file using ORCA has been described at beginning of Chapter 4 of Multiwfn manual. Boot up Multiwfn, and then load wfn/molden file, then input following commands:
6  //Modify wavefunction
34  //Set occupation number of inner orbitals to zero
-1  //Return
5  //Calculate grid data
1  //Electron density
3  //High quality grid (Multiwfn provides rich and flexible options to set grid, thus you can also manually specify grid spacing, see Section 3.6 of Multiwfn manual for detail)
2  //Export the grid data to cube file
Now you have density.cub in current folder, which should be what you need.

I use Gaussian to produce valence electron density cubes for use by Gavezzotti's PIXEL program.

A typical GJF is below for G16W is below.

#######################################.

%nprocshared=  4
#MP2/6-31G** guess=core nosym density=MP2 pop=esp cube=(cards,frozencore)

#hps_2xxx 'P b c a'  fragm   1

0 1
S    0.095673   -1.850555   -0.192184
O   -0.578335   -3.026974    0.268055
O    1.419211   -1.991221   -0.716047
O    0.060886    2.160599    4.074644
N   -0.842085   -1.238984   -1.382957
C    0.128758    1.247642    3.072692
C   -0.841222    0.257627    3.065011
C   -0.838533   -0.701946    2.080861
C    0.141172   -0.679465    1.103616
C    1.100700    0.312314    1.103329
C    1.101477    1.272443    2.095758
C   -0.500670    0.037660   -1.946283
C   -1.357032    1.101201   -1.780318
C   -1.048737    2.322556   -2.348088
C    0.113646    2.484040   -3.051236
C    0.957435    1.423452   -3.216253
C    0.651559    0.181502   -2.680460
H    0.642395    2.828166    3.912882
H   -1.704276   -1.343272   -1.146732
H   -1.613443    0.240017    3.845856
H   -1.609888   -1.482480    2.071284
H    1.863642    0.338147    0.314538
H    1.874005    2.053319    2.107295
H   -2.281745    0.979033   -1.201508
H   -1.739754    3.168322   -2.232117
H    0.366876    3.462343   -3.479903
H    1.890432    1.556801   -3.779655
H    1.320499   -0.675004   -2.841322

c:\clp\doc\dens\hps_2xxxa.den
    0   -4.543032   -5.870974   -6.402253
  110    0.080000    0.000000    0.000000
  144    0.000000    0.080000    0.000000
  166    0.000000    0.000000    0.080000

########################################

I can get Orca to produce total electron density cubes in a format which the PIXEL program can read.
The problem I have is that the results are just not good enough even though the surfaces generated by the Gaussian and Orca cubes appear identical when examined in GaussView6.
The Gaussian cubes are not sensitive to the step size used to generate the cubes or the condensation level used within the PIXEL program. In contrast the Orca cubes give PIXEL results which are dependent on step size and condensation level. It is almost as though the Orca cubes were just not accurate.

My question really is can some wavefunction output from Orca be used to allow Multiwfn to compute better cubes?

I am aware that your manual reccommends that Molden style output from Orca be used with Multiwfn but I amjust not sure if what I am trying to do will work.

Angelo Gavezzotti no longer supports PIXEL it is now hosted on Simon Parsons web site and I asked Simon for help but apart from saying that it would be great if Orca could be made to work and trying the total/valence density switch in PIXEL Simon had no other suggestions.

Sorry to bother you with all this but if Orca could be used with PIXEL it would help those who do not have a Gaussian licence.