- Thread starter
- #21
Ryan Patch
Well-known member
- Joined
- Dec 4, 2007
- Messages
- 263
- Reaction score
- 0
- Points
- 0
- Age
- 40
- Location
- NY, NY
- Website
- www.bitterriver.com
Noirist -
We're hoping to test as follows:
Highlight recovery - have an area that's overexposed (maybe a couple of different overexposures) - this area needs to have both details and a flat area, so we can see how those look when we bring them down, what kind of noise is in them.
Shadow recovery - same thing. A dark area, (bracketed again), with both details (like a fine pattern or something), and a flat area, and maybe even someone's face. We want to bring it up to see what kind of noise and detail loss we're seeing. In a documentary you won't always get perfect exposure so this will be pretty important.
Now, ideally, in both of these tests, we want to have colors as well, to see how they stand up. Green/Red/Blue should be good. I'm not sure what the best way to engineer this shoot would be, but basically, we want to see what happens to color when we push it from the darks, or pull it from the whites. For example, I know in most compressions, and I think certainly in H264, the red channel gets subsampled. That means that if we are to push an image with a lot of red in it, there will be a lot of jagged edges, and all sorts of ugliness.
Perhaps one way to do it is just to get clothing with a lot of colors on them, arrange them in front of the camera, and control how much light falls on them, etc.
Then lastly, we want to look into detail - using either a focus chart or a detailed model or moulding.
If anyone has any ideas to add or comment on this, would love to hear them!
Ryan
We're hoping to test as follows:
Highlight recovery - have an area that's overexposed (maybe a couple of different overexposures) - this area needs to have both details and a flat area, so we can see how those look when we bring them down, what kind of noise is in them.
Shadow recovery - same thing. A dark area, (bracketed again), with both details (like a fine pattern or something), and a flat area, and maybe even someone's face. We want to bring it up to see what kind of noise and detail loss we're seeing. In a documentary you won't always get perfect exposure so this will be pretty important.
Now, ideally, in both of these tests, we want to have colors as well, to see how they stand up. Green/Red/Blue should be good. I'm not sure what the best way to engineer this shoot would be, but basically, we want to see what happens to color when we push it from the darks, or pull it from the whites. For example, I know in most compressions, and I think certainly in H264, the red channel gets subsampled. That means that if we are to push an image with a lot of red in it, there will be a lot of jagged edges, and all sorts of ugliness.
Perhaps one way to do it is just to get clothing with a lot of colors on them, arrange them in front of the camera, and control how much light falls on them, etc.
Then lastly, we want to look into detail - using either a focus chart or a detailed model or moulding.
If anyone has any ideas to add or comment on this, would love to hear them!
Ryan