1
EOS Bodies - For Video / 5D3 video new vs. old firmware - might be different after all
« on: April 30, 2013, 03:11:42 PM »
A tweet from J.M. offers this sample of video taken using the internal recorder with the same settings, old vs. new firmware:
https://www.dropbox.com/s/vuwbn2582pv67i3/5d3121.png
So maybe I actually was correct in my initial take (a post I since deleted) where I said it looked like the video was sharper. I first I was sure that it was crisper, but then I decided I had tricked myself.
Internal vs external seems similar but new vs old firmware seems different.
Maybe it does grab more details and not mush details in shadows as much now.
(I think I have seen a few very weird artifacts with the new firmware though perhaps popping up in video. A couple nasty weird things, but not sure yet.)
(It doesn't look as crisp but artifact free as the 'raw' DNG stuff ML discovered though.)
https://www.dropbox.com/s/vuwbn2582pv67i3/5d3121.png
So maybe I actually was correct in my initial take (a post I since deleted) where I said it looked like the video was sharper. I first I was sure that it was crisper, but then I decided I had tricked myself.
Internal vs external seems similar but new vs old firmware seems different.
Maybe it does grab more details and not mush details in shadows as much now.
(I think I have seen a few very weird artifacts with the new firmware though perhaps popping up in video. A couple nasty weird things, but not sure yet.)
(It doesn't look as crisp but artifact free as the 'raw' DNG stuff ML discovered though.)


Nikon has about 5 or 6 recent releases, DX and FF, that just destroy the 5D3 for dynamic range. 5D3 even measured a full 0.4 stops worse dynamic range AND with worse banding than a 1Ds3 someone lent me some files from. The 5D3 actually has just about the worst read noise per photosite of all the Canon models. Worse than 40D,50D,60D,7D,1D4,5D2,1Ds3,etc. I had to go back to a 30D to find worse per photosite read noise. Not sure what they did with it in that regard. If they had it with 1D4 per photosite read noise it might have gained almost a half stop of DR compared to the 5D2 instead of losing 1/3 stop.