rscheffler Offline Upload & Sell: Off
|
p.2 #16 · p.2 #16 · R5II + 100-300 = Ultimate field sports combo [image heavy] | |
docusync wrote:
Ron, I thought you were joking about the 200/1.4. No way this lens will ever materialize. Yes, it will be the size of a typical 400/2.8 just shorter. Same size entrance pupil though. Same $12k price or more since it would be a really niche lens.
The problem with the thumbnails - they are in focus, guaranteed, but now you have to find which one you like the best. There's not much difference between them (30 fps!).
This is much harder to do than to reject everything that's out of focus, and then you get three that are in focus, and you can definitely pick the best one because the other two are garbage.
I never saw girls hockey, but I think I can arrange that - my friend's daughter plays. That would be interesting to watch and take some pictures. I'm curious to see how face recognition works with helmets.
...Show more →
I was mostly joking. I can't envision any rational, realistic need for such a lens other than as a halo product to show off a manufacturer's capabilities. Modern cameras/sensors are very good now at high ISOs, to the point where one can easily get by with f/4 or f/2.8 in most situations. But if such a lens existed, for sure I'd want to use it from time to time through CPS.
If you're not seeing a lot of frame to frame difference at 30fps, maybe dial it back to 20? I make thumbnails fairly large in my culling app of choice so that I can better evaluate things like expressions without needing to view individual full-screen previews until the choice has been sufficiently narrowed down. I also find that if I need to be able to quickly find specific images among thousands from a game, it helps tremendously if I 'chimp' between plays and tag/lock the key images. At the least this lets me find those general plays quickly when on the computer where I can then decide whether or not I chose the best ones in-camera. I recently did this covering motorcycle racing, where I'd shoot maybe 2-3 thousand images per race but needed to turnaround 20-30 for a post-race gallery within a short time after the each race. I'd tag key moments like lead changes, crashes, victory celebrations, etc., isolate those in the culling software and pull the best for the gallery, ignoring the rest for later, more in-depth review (inevitably there will be good images that are initially missed this way, but if in a rush, good enough is usually good enough ).
At least with the R6II, subject recognition/tracking works pretty well at quickly zeroing in on heads/helmets during hockey games, but when there are multiple players in the scene from the same team wearing the same color helmets and crisscrossing in front or behind the player you want to follow, it can often switch to another person. Perhaps the R5II will be better in this regard because of newer algorithms and more powerful processors? As for actual recognition of a registered face, I'm not sure if or how well that will work with headgear such as helmets.
|