Do these things work with an A7R and a wide angle lens like a Canon 24 TS-E? How wide would this go, 50? 30?
I found that DSLR sized sensors on view cameras demand a precision in zeroing that defeated a heavy studio monorail like the Cambo Ultima, so what hope for a cute little thing like these. Tilt where you don't want it is the enemy of deep focus.
I can see these are revolutionary with a Credo 50 on the back. I'll ask my guardian angel to magic up the money for me.
With an MFDB and live view these are probably the grail for a view camera lover.
With the Canon TS-E's you need a focal plane shutter and some electronic interface to control the aperture. Hartblei Hcam and Alpa FPS has this. Arca-Swiss got a focal plane shutter unit now, but I don't think they can control Canon lenses (yet). Correct me if I'm wrong.
The A7r sensor have crosstalk issues, and while the TS-Es won't bring it into crosstalk except in the very edge of the image circle (which is fuzzy anyway), tech wides will. Desaturation and demosaicing artifacts may follow. As it's subjective how much you can accept it's subjective how wide you can go. It also depends on how much you want to shift. Unfortunately I don't have any test files on A7r with tech wides, but have IQ250 files (equivalent to Credo 50) and from those I'd say if you keep within 25-30 degrees you're going to get fine image quality, and you can stretch it a bit in the corners and in horizontal orientation. The heavily retrofocus Canon TS-E delivers light at about 20 degrees at the image circle edge, and A7r show a little crosstalk there so I'd guess maybe that sensor is stable for 17 degrees or so.
Unfortunately there's no "max projection angle on the image plane" data in the lens data sheets. Since the wide lenses generally have quite large nodal point separation (meaning that the rear nodal point is considerable closer to the image plane than the center of the copal shutter, ie there's a steeper angle than one might expect) it's not so easy figured out from the available data sheets.
Schneider Digitar does have nodal point separation defined in their data sheets, but that does not really help that much as we need to know where the actual rear nodal point is. Anyway taking the flange focal distance, adding 5mm (to get to the center of the Copal) and then removing half nodal point separation seems to lead to an approximative position of the rear nodal point, which becomes equal to the focal length as these lenses are of symmetrical type. Angle is thus just the same as in the front, and thus the SK28XL delivers light at almost 60 degrees (!) at the image circle edge. Ie way past what the Sony CMOS sensors can do.
Rodenstock Digarons are retrofocus and unfortunately their data sheets don't have any nodal point separation specified so we can't do any approximation in the same way. Looking at a cross-section of one of these lenses it does seem like nodal point separation is quite big in these lenses, say 20mm or so. Then we can approximate the rear nodal point position as being flange focal distance + 5 - 20/2, ie flange focal distance - 5mm. Assuming this is right we get the following angles:
Digaron-S 23mm: arctan((70/2)/(44.8-5)) = 41 degrees at 70mm image circle edge
Digaron-S 28mm: 36 degrees
Digaron-S 35mm: 36 degrees
Digaron-S 60mm: 31 degrees
Digaron-W 32mm: 35 degrees at 90mm image circle
Digaron-W 40mm: 35 degrees
Digaron-W 50mm: 32 degrees
Using a laser pointer these angles could be verified, but unfortunately I don't have these lenses so I can't test (I have many of the symmetric Schneider Digitar lenses though). It does seem like Rodenstock have had some maximum angle of ~35 degrees as design target and adapted the degree of retrofocus to match that. The Rodenstock 32mm is thus more retrofocus (and more complex) than the 40mm, but will thanks to that not put the sensor more to the test than the 40. The 23mm seems to be an exception with a bit steeper angle. Nodal point separation can vary a lot between lenses though, so the actual value could be otherwise.
Anyway, with these angles it seems like you will with the Rodenstocks get some limitations with the Credo 50, and pretty large limitations with the A7r. Of course things get considerably worse with the Schneider Digitar lens line. If you would go to the 110 mm image circle edge of the 60XL you have 45 degrees angle there, ie more than the Digaron-S 28mm at its edge...
I have test files of IQ250 + 32 HR though. With the 32HR your trouble-free (very little desaturation and no demosaicing artifacts) zone is about 65-70mm image circle, which would put the IQ250 / Credo 50 limit at about 25 degrees. Or put it in other terms, you'll get 65-70mm image circle from the 32 and 40 and a little more with the 50 (but not full 90), and with the Digaron-S lenses 28 and 35 you get 50-55mm image circle (ie no shifting!). Depending on subject you can accept more crosstalk and expand this zone. With the A7r you get smaller circles, but on the other hand you have a smaller sensor and maybe you shift less then.
So the question "how wide would this go" is not so easily answered, unless you just say "it depends"...