Smearing towards the image corners and color cast mainly happen with symmetrical wide angle lenses and sensors of digital cameras with thick protection glass (basically all digital cameras and digital backs apart from digital Leica M which were and are intended to be as compatible as possible to their older wide angle lenses which were designed for film).
Medium and large format wide angle lenses for all kind of film cameras without mirror box (rangefinder cameras, technical cameras) mostly also are such symmetrical designs, so they won't work well very with digital sensors. Enclosed section views of the two Schneider 24 and 35 mm lenses.
Rodenstock designed 'partial asymmetrical' (= 'partial retrofocus') type of medium format wide angle lenses which are positioned already a bit further away from the sensor and therefore work a bit better with shift movements and digital sensors but still shift is limited. Enclosed section views of the two Rodenstock 23 and 32 mm lenses.
Most suitable are any kind of 'real asymmetric' (= 'real retrofocus') type of wide angle lenses, simply any wide angle lens that was designed for a mirror reflex camera. Such kind of lens provides a large distance between the last lens element and the sensor and withit the rays of light emitting from the back lens towards the sensor edges and corners (especially when shifted) are less much angled than with the two former lens types, therefore no smearing and barely any color cast issues occur. Enclosed section views of the two Canon 17 and 24 mm TSE lenses and the Contax 645 35 mm lens.
(sensor size shown is 40mm height of a 54x40mm sensor)
Currently I don't have all my three cameras (Sony A7RII, Fuji GFX, Alpa FPS IQ180) on hand to do and publish a direct comparison test with the mentioned retrofocus lenses, but from my practical usage I don't see any of the cameras behaving worse than another if one compares image sections of same field of view angles (means less shift movements for larger sensors).
Enclosed also a slightly revised version of my shift movement xls sheet, the first one had a few equation mistakes (sorry).
The sharp image circle of the TSE 17 is 58 mm, the image circle of the Fuji GFX 44x33 mm sensor is 55 mm, that means the lens covers that sensor and still allows slight shift movements keeping sharp corners. In my first sheet I showed a negative shift value which would tell that the last mm towards the image corners are already not perfectly sharp without shift, but that's not the case. It is the case using the TSE 17 with 54x40mm 60/ 80/ 100 MP sensors.
Enclosed also a link to an uncropped full resolution sample picture:
https://dl.dropboxusercontent.com/u/18437364/Guangzhou/Fuji-GFX+EF17f4TSE@f11+5mm-shift.jpg
Fuji GFX + Fotodiox Canon EF adapter + Canon EF 24f3.5 TSE (II), vertically with 5-6mm shift upwards.
The aperture was pre-set to f11 and I tried in lifeview to set the focus point in order to achieve sharpness from the bikes only a few meters in front of the camera until the skyscrapers at 'infinity' distance. According to my xls list, the TSE 24 should allow 5-6mm shift with still 'sharp' corners, if you check the top left corner, it's looks quite fine.
I mean we are talking here about a 20mm equivalent fov lens with shift on a 44x33mm medium format camera, I think it works cool;-)
A few words about the protection glass in front of the sensor of the Fuji GFX. Fuji has choosen a completely new solution, different to any other digital camera or digital back on the market: The protection glass is not glued directly on the sensor but it is positioned about 9mm in front of the sensor. This particular innovative postion has several advantages compared to a normal sensor-glass stack
1. Fuji was able to add ultrasonic cleaning by vibrating the glass instead the sensor. So Fuji is the only manufacturer to be able to offer that feature with this 44x33mm Sony sensor while all others (Phase One, Hasselblad, Pentax) aren't.
2. In case of scratching that glass by accident, only the (cheap) glass has to be replaced, not the (expensive) glass including the sensor it is normally glued to.
3. Dust on the protection glass is nearly invisible due to the distance to the sensor, the images by the Fuji GFX are much less prone to dust spots due to the distance and the ultrasonic cleaning.
Fuji of course incorporated the special position of the protection glass in the calculations af their newly designed GF lenses. It was unclear if that protection glass position has any negative effect on the image quality when using alternative lenses that were designed for film (no protection glass at all) or 'conventional' digital cameras.
Using the two super-wide-angle shift TSE lenses which were designed for a 'conventional' digital sensor-glass-stack and the Contax 645 35f3.5 and Hasselblad CF 50f4 FLE/ CF 100f3.5/ CF 180f4 lenses which were designed for film, I don't see any smearing effects towards the image corners, even fully shifted (apart from expected softness if shifted outside the usable image circle). That means, IMO the special protection glass position has no (or no worse) effect on image quality than a glass glued to the sensor.