D800 dr

macgregor

macgregor

Veteran
I just made a quick DR test. I´m getting the following using the flaat10 profile:

+3.5 N -7 = 11.5 stops of DR

However, the darkest stops not only show some pattern noise but a lot of posterization, so I wonder how much of those values can be used in real world.
 
I rated Flaat_10 at 10 stops of DR, there's that extra bit of info you talk about but it's basically useless, it's stored at one or two shades for every stop of light
Flaat_11 is very useable too, and I'd rate it at 11 stops
the higher ones, Flaat_12 and Flaat_13 (12 and 12.7 stops), show a lot of that posterized noise you talk about; they can still be useful, but only with heavy noise reduction; I'm not sure you would like them
http://vimeo.com/groups/137445/videos/41935766
http://vimeo.com/groups/137445/videos/39941831
http://vimeo.com/39899742

and also:
if you're recording internally, my picture-control approach helps because the curve is applied to a better quality image (I think it's uncompressed 422 8-bit), whereas if ou apply it in post you're killing a lot of info and then applying the curve over H.264 8-bit 420
if you're recording externally, I haven't done any tests, but I'd think the picture-control approach wouldn't add a lot (i.e. you can get basically the same applying the curve in-camera with the picture control, or applying it in post)


plus: I see you studied here in Madrid!! are you still around, or did you fly away?
 
Last edited:
Stephenh said:
macgregor, can you post a jpg sample of both internal and external frame.
Click to expand...

From the DR tests? They look exaclty the same. Have in mind we´re talking about grey cards, so there's little detail and 0 motion, therefore compression is not an issue and the advantages of higher data rate don´t show.
 
I am of the view that as long as they keep line skipping to reduce sensor data to HD video specs, we are going to always see this issue. They need to come up with a new approach of down scaling for HD video.
 
Samuel H said:
I rated Flaat_10 at 10 stops of DR, there's that extra bit of info you talk about but it's basically useless, it's stored at one or two shades for every stop of light
Flaat_11 is very useable too, and I'd rate it at 11 stops
the higher ones, Flaat_12 and Flaat_13 (12 and 12.7 stops), show a lot of that posterized noise you talk about; they can still be useful, but only with heavy noise reduction; I'm not sure you would like them
http://vimeo.com/groups/137445/videos/41935766
http://vimeo.com/groups/137445/videos/39941831
http://vimeo.com/39899742

and also:
if you're recording internally, my picture-control approach helps because the curve is applied to a better quality image (I think it's uncompressed 422 8-bit), whereas if ou apply it in post you're killing a lot of info and then applying the curve over H.264 8-bit 420
if you're recording externally, I haven't done any tests, but I'd think the picture-control approach wouldn't add a lot (i.e. you can get basically the same applying the curve in-camera with the picture control, or applying it in post)


plus: I see you studied here in Madrid!! are you still around, or did you fly away?
Click to expand...


BTW after testing your 3 curves, I really like flaat10 (the others not so much). It's a bit cleaner than neutral and colors are warm and nice. Great job.

I just shot a real life test outdoors and I like it very much. The moment they come up with that mosaic AA filter this camera is going to be a very nice toy.
 
nice to hear - as I said, once you have an external recorders, the more aggressive curves don't make a lot of sense anyway - and also, current location, Lalaland, Madrid?
 
You must log in or register to reply here.
Back
Top