Changeset 28095 for trunk/ppStack/src/ppStackSources.c

Timestamp:

May 25, 2010, 2:56:44 PM (16 years ago)

Author:

Paul Price

Message:

Fixing sign error in magnitude correction that was producing magnitudes WAY too faint. The important thing is that the transmission (clouds) correction is negative. I think I must at some stage have flipped the sign on magCorr so as to get trans negative, without realising this threw the zero point off. This gives correct zero points compared to SDSS, while also applying the transmission correction properly. Thanks to Dave Young for finding this bug!

File:

• trunk/ppStack/src/ppStackSources.c (modified) (2 diffs)

trunk/ppStack/src/ppStackSources.c

@@ -77 +77 @@
     PS_ASSERT_VECTOR_SIZE(inputMask, sourceLists->n, false);
 
-#if defined(TESTING) && 0
+#if defined(TESTING) && 1
     {
         // Deliberately induce a major transparency difference
-        psArray *sources = sourceLists->data[1]; // Sources to correct
+        psArray *sources = sourceLists->data[0]; // Sources to correct
         for (int i = 0; i < sources->n; i++) {
             pmSource *source = sources->data[i]; // Source of interest
@@ -283 +283 @@
                 }
                 psArray *sources = sourceLists->data[i]; // Sources of interest
-                float magCorr = zp->data.F32[i] + trans->data.F32[i] - 2.5*log10(options->sumExposure);
+                float magCorr = zp->data.F32[i] - trans->data.F32[i] - 2.5*log10(options->sumExposure);
                 if (zpExpNum == numGoodImages) {
                     // Using measured zero points, so attempt to set target zero point
                     magCorr -= zpTarget;
                 }
-                options->norm->data.F32[i] = -magCorr;
+                options->norm->data.F32[i] = magCorr;
                 psLogMsg("ppStack", PS_LOG_INFO,
                          "Applying scale correction to image %d: %f mag (%f)\n",