Problems in the sky

Started Sep 6, 2012 | Discussions thread
Shop cameras & lenses ▾
Andre Affleck
Senior MemberPosts: 2,194
Re: I see what you mean
In reply to Jack Hogan, Sep 10, 2012

Jack Hogan wrote:

Horshack wrote:

I've seen similar posterization-type effects in my D800 raws taken during the twilight/blue hour, at least how the image is rendered in CS5 using PhotoRGB on my 8-bit IPS display. I've never troubleshooted it to see if it's a display problem or if the problem is in the raw file.

Keep in mind that in any very clean 8 bit image, if we stare at it long enough we could be able to see a transition in tones in a very smooth, compressed gradient- like in the sky half way beltween the star and the right hand 'The Mirage' sign in your image: the green sRGB channel is around level 210 and in linear space the difference between level 209 and 210 is just above 1%, the Weber-Fechner criterion for visibility in relatively bright light. You definitely would not be able to see that same transition at 10 or more bits, so my guess is that the raw data is smooth.

I noticed the same thing with Horshack's sample. It is bit depth limited. The one thing that makes it a bit more apearant is that the transitions are clumped together. What looks like single pixel variations are actually 4 pixels of the same value clumped together, making the transitions coarser and more noticeable. If I do a Gaussian blur, they break up into single pixels and it becomes almost invisible. Not sure what is causing it, but it may be the demosaicing algorithm trying to get detail out of this area.

In any case, the OPs sample is far worse. I would say Horshack's sample is close to normal, but there is something definitely wrong with the OP's sample.

Reply   Reply with quote   Complain
Post (hide subjects)Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow