flowers said:
100 said:
Anyway, on a more serious note; I think the future is more about software than hardware.
I think you are right but I don't know if it's a good thing. Earlier lenses like 300/2, 200/2, 200/1.8, 135/2, 135/1.8, 135/1.5 were being designed and made. Now many lenses are just mk II III IV with no changes to lens design but only to coatings. Fastest lenses made today over 100mm (and sometimes even under) are f/2.8. The idea is to use a high ISO. "Nobody needs fast lenses" is the mentality. I think if this trend continues, more things will be delegated to software instead of hardware. Maybe tomorrow's fast lenses are compact f/4. Who knows.
Software can only go so far. You can do massive DNR, but at the end of the day, you're losing a lot of detail, too. We're to the point where in most typical lighting conditions, you can get away with f/4 lenses and IS, but only if you aren't looking at the pixels. You can usually get away with f/2.8 lenses and no IS, but again, only if you aren't looking at the pixels.
As pixel density increases, the light gathering decreases proportionally. That's one big reason why FF cameras have such a low-light advantage over crops. The pixel density is so much lower that you can get away with slow lenses in crappy light. Unfortunately, cameras are rapidly approaching the limit of what you can do in terms of sensor quantum efficiency (at best, it can improve by no more than about a factor of two—only one more stop), so most of that extra light gathering isn't going to come from better sensors beyond this point. Therefore, any future improvements in pixel density will require faster lenses just to break even. When we finally see a high-pixel-density full-frame camera, those f/1.2 and even f/1.0 lenses are going to start looking mighty tempting again, because the FF cameras are going to have the same terrible low-light performance as crop bodies unless you downscale at the end of your processing.
Now that's not to say that we won't eventually see hardware with electronic shutters that take thousands of images per second and use bats**t crazy advanced image processing to smart-merge the images and individually stabilize each part, compensating for motion, etc., but I'd imagine the compute power to do that in-camera is at least a decade out, and the storage requirements might push it even further out. So at least in the near term, fast lenses are going to continue to be crucial, IMO. And even in a decade, when we have such software capabilities, a fast lens will still be useful for obtaining shallow depth-of-field for artistic effect. Short of taking advantage of parallax in combination with light-field sensor tech, I don't see that being readily emulated in software.