Merge "Update gainmap documentation." into udc-dev
This commit is contained in:
committed by
Android (Google) Code Review
commit
19573fe42d
@@ -63,20 +63,20 @@ import libcore.util.NativeAllocationRegistry;
|
||||
* as follows:
|
||||
*
|
||||
* First, let W be a weight parameter determining how much the gainmap will be applied.
|
||||
* W = clamp((log(H) - log(displayRatioHdr)) /
|
||||
* (log(displayRatioHdr) - log(displayRatioSdr), 0, 1)
|
||||
* W = clamp((log(H) - log(minDisplayRatioForHdrTransition)) /
|
||||
* (log(displayRatioForFullHdr) - log(minDisplayRatioForHdrTransition), 0, 1)
|
||||
*
|
||||
* Next, let L be the gainmap value in log space. We compute this from the value G that was
|
||||
* sampled from the texture as follows:
|
||||
* L = mix(log(gainmapRatioMin), log(gainmapRatioMax), pow(G, gainmapGamma))
|
||||
* L = mix(log(ratioMin), log(ratioMax), pow(G, gamma))
|
||||
*
|
||||
* Finally, apply the gainmap to compute D, the displayed pixel. If the base image is SDR then
|
||||
* compute:
|
||||
* D = (B + epsilonSdr) * exp(L * W) - epsilonHdr
|
||||
* If the base image is HDR then compute:
|
||||
* D = (B + epsilonHdr) * exp(L * (W - 1)) - epsilonSdr
|
||||
*
|
||||
* In the above math, log() is a natural logarithm and exp() is natural exponentiation.
|
||||
* In the above math, log() is a natural logarithm and exp() is natural exponentiation. The base
|
||||
* for these functions cancels out and does not affect the result, so other bases may be used
|
||||
* if preferred.
|
||||
*/
|
||||
public final class Gainmap implements Parcelable {
|
||||
|
||||
|
||||
Reference in New Issue
Block a user