Hey Maya,
Thanks so much for the interesting reply! I absolutely did not realize CA was the specific cause of these bad bokehs, but more importantly I never realized that front bokeh and back bokeh circles would be so complimentary.
In the case of the tests Marianne Oelund has realised, it's rather spherical aberration that's at play, as she only tested for one color (green). And it's only valid in the centre of the frame, off-centre other aberrations pile up I believe to make things even way more complex.
Fine-tuning spherical aberration to enhance background blur comes at the cost of foreground blur. Besides in the case of the Nikon 58mm the aberration is so extreme that focus shift is significant and resolution not quite excellent.
Marianne has also made tests with a white point light source (sorry I couldn't find the thread she posted these photos in) :
I think that these full spectrum tests are less representative of a lens' spherical aberrations - perhaps because it's difficult with common means to test lenses with very small and precise point light sources - but more so of its CA in the transition zone.
A more concrete illustration of what the blobs above mean :
It's from Lenstip's AF / LOCA test, and it's the Nikon 105mm 1.4. That lens is specifically designed to preserve some degree of spherical aberration to enhance background blur smoothness and progressive transitions : you can still read the "lensalign" text much further beyond the focal plane and it progressively blends in the overall background blur in a gaussian way (that corresponds to a blob with a brighter centre than the edge, like an apodised lens). The front bokeh however has been made significantly harsher / busy (that corresponds to a blob with a brighter edge and a darker centre). When you close the aperture, it's the equivalent of cutting the extreme edge of the blobs above. So a lens like the one below may show traces of edgy bokeh at f1.4 but will look completely smooth at f2 :
This is the reason why I always prefer to shoot my Canon 50mm STM at f2.5 at a maximum, provided there's enough light. Wider than that and bokeh starts to get edgy in the centre.
An apodised lens can do what the Nikon 105mm 1.4 does for both the front and the background blur (it's not quite the same but kinda). To do so in an equal fashion between front and rear, it's better for it to be very "neutral" in terms of spherical aberration, hence why I'm a little bit worried abut the 85mm RF non-DS version.
These might be gimmicky but you could also have filters such as: 1) clear center spot surrounded by ND. Things would be be sharp front to back due to a pinhole effect, but also surrounded by blur. Almost like a double-exposure wide open and stopped down. 2) shapes such as stars. 3) random scattering of pinholes, giving the most jarring bokeh...
Bring on the gimmicks
. I'd love that. Although it's not at the aperture stop, has anyone experimented with the EF to RF drop in filter adapter to see if interesting effects can be obtained bokeh wise ?