Full test image, MFT sensor, Sigma 135mm f/1.8 coupled with reversed Sigma 70mm f?2.8 with paper aperture of 27mm between them.
Several people have used the Samyang/Rokinon 135mm lens as a tube lens for objectives, and at least one person has used it also as the base for mounting reversed lenses ("coupled lenses" or "stacked lenses"). Everyone were underwhelmed by the results. Mike (mawyatt) reported that the cheap Vivitar/Komine 135mm that I and other use was better. But we do not have reports yet about the new Sigma 135mm f/1.8 lens which is one of the sharpest lenses currently in production.
I tried it with various reversed lenses mounted in front of it, and found it was better than the Vivitar/Komine, but had lots of longitudinal CA when used as a tube lens. While there was variation depending on the reversed lens, all combos had some amount of loCA, more or less of the same colors. However, stacking (especially in PMax) could in most cases still produce an almost perfectly clean image since loCA is restricted to out-of-focus regions, if the reversed lens does not cause too much additional loCA.
For the reversed lens mounted on the 135 lens, I tested some highly-regarded but somewhat unusual lenses, all on the Oly PEN F MFT sensor used at normal resolution. Lighting and subject-camera angles were kept constant throughout tests. I used continuous light (Tronde light) and exposures around 1/30 s.
I tried the Oly OM 90mm f/2.0 macro, an old film-era lens with a large cult following. I liked this lens because it is a fast macro, capable of being stopped down a stop or two without entering diffraction territory. It also has a long flange-film distance and is fully manual. It does not expect a glass filter between it and the film, so nothing needs to be added to it's rear element when used in reverse.
I also tried the Oly 50mm f/2.0 macro for Four Thirds (not MFT). This is another highly regarded lens with a long flange-to-sensor distance. It expects a thick filter stack between its rear element and the sensor, so when used in reverse at wide apertures it needs a 4mm optical window between the rear element and the subject. I had a 3mm 1/10 wavelength window and made an attachment to put this on the rear of the lens for my tests.
I also tried the Sigma 70mm f/2.8 macro, regarded by some as the sharpest macro lens ever. It is made with exotic glasses that are no longer available. It has a long flange-sensor distance and expects a 2mm filter stack between rear element and sensor. I attached an ordinary filter to the rear of the lens for this test.
I also tried the 50mm f/1.4 Sigma ART lens, with filter attached to rear element.
As tube lenses, besides the Sigma 135mm and the Vivitar/Komine 135mm, I tried the exotic Oly Four-Thirds 150mm f/2.0, which also contains special glasses and has been tested as one of the sharpest production lenses at the time of its introduction. I also tried the 170mm Raynox DCR-5320.
For now I am short of time and will only post the winner and the most economical good option. In two weeks when I get back from the jungle I may post more results. These are results straight from the camera jpgs, no additional sharpening or exposure correction.
100% crops, Vivitar/Komine 135 on left, Sigma 135 f/1.8 on right, both with reversed 70mm Sigma macro lens and 27mm paper aperture. Is this difference worth $1300??
The best reversed lens was the Sigma 70mm, by far. This produced a 1.9x magnification when used reversed on the 135mm lenses. Even with both the Sigma 135 1.8 and the Sigma 70 2.8 wide open, it was sharp. That was very impressive. But I stopped it down a bit with a paper aperture of 27mm diameter, and did my final tests with that on all lenses (because most were not sharp wide open).
The Sigma 135 and the Oly 150 produced comparable results, but the Sigma was slightly cleaner and sharper. The Vivitar/Komine had less contrast and was not quite as sharp, but this was only just barely visible at 100%. The difference in price between $60 (Vivitar) and $1400 (Sigma) or $2000+ (Oly 150) is vastly disproportionate to the difference in quality, so unless you are printing your images large, you might as well stick with the Vivitar/Komine (serial numbers beginning with 28xxx...). I guess that on a FF sensor with its larger pixels, the differences would be even more minor. If I had used the High-resolution option of the PEN F to make 50Mp or 80Mp images, though, the difference would be more obvious. I will try this next time I have a free moment.
The reversed 90mm also did surprisingly well on the Sigma 135mm, though it had a warm color cast. It did better coupled than it did by itself, once the paper aperture was in place. But not as good as the Sigma 70mm.
The Oly 50mm macro did well but had more purple fringing than the other lenses. This could have been reduced in post-processing.
The Sigma 50mm ART lens was also quite good, but not as good as the Sigma 70mm.