agorabasta wrote:Pako, I believe you would be much better off using a remote flash in tele position ...
The flash being in tele position won't make any difference beyond brightness. In wide-angle position, a greater portion of the flash gun's output would go astray and be wasted, that's all. Provided the flash has a zoom reflector to begin with.
agorabasta wrote:... facing a deeply defocused long tele lens set at its medium aperture.
Huh!? Either you want collimation—then better focus accurately—, or you don't. And don't use any other aperture than the widest, or you risk vignetting. By the way, if you really want to use a telephoto lens (or any taking lens) as a collimator then the 'focusing' thereof won't get accomplished by the lens' focusing ring but by adjusting the lens' distance to the flash gun with the focusing ring firmly set to infinity (and the lens' rear end pointed towards the flash gun). But then, it would be a foolish idea to use a telephoto lens for this purpose anyway. Better use a close-up lens—as a collimator, that's the same thing as a telephoto lens optically but much easier to set up and to handle, and less likely to cause vignetting.
agorabasta wrote:The most important requirement for the lens properties is that it must be as much 'image side telecentric' as possible. That's why a long, bulky, preferably old and manual, tele lens has the best chance to safely collimate your backlight flux.
Except that a close-up lens has an even better chance. And no, image-side telecentricity is not required.
And what would 'unsafe collimation' be like?
agorabasta wrote:With a diffuse backlight, the lens on the camera has a very small chance to be the primary limiting factor in your setup.
Well ... if with direct light contrast becomes uncontrollable then diffuse lighting simply is better. After all, there's a reason why all slide copiers include such a white translucent diffusor screen. On the other hand, there also is a reason why these diffusor screens are removable. Pako, when you tried using direct light for copying before, was it with or without collimation?
And oh, by the way—in my previous post I said the collimator-to-slide distance didn't matter. I am afraid that's not entirely true. While this distance isn't critical, it still mustn't be too short or too long. A distance between approximately 1× to 2× (maybe 3×) the collimator's focal length should work fine ... provided you're willing to try direct collimated light in the first place ...