rico Offline Upload & Sell: Off
|
chemprof wrote:
This is 100% crop from the upper right corner (shade, window frame, smooth wall). Note that this is NOT an anomaly. Now that I know what I'm looking for, I can find it in ALL of my images, including out of focus areas in any image. It's most noticable in mid-tone areas.
Gerald
Hi, John! I'll let this thread play out until I get a full JPEG sample. Plus, I can borrow Gerald's crop for a quick assessment. From it, I detect a statistical imbalance between alternating pairs of pixel columns to the amount of almost 3%. This will be increaingly noticeable as the contrast curve is stretched. That column-pairs are involved implies that alternating Bayer cells (2x2 sensor sites in the RGGB pattern) are out of balance (correctable in firmware).
The algorithm I deploy in the following, rebalanced crop of Gerald's does appear to improve matters greatly:
Canon DSLR I have tested (D30, D60, 20D) also exhibit a systematic imbalance, although different in its pattern: the two G measures within a given Bayer 2x2 is inconsistent by about 1%. The effect is seen as a grid or mesh pattern, but often masked by JPEG compression. Noone is perfect, eh?
Based on my quick results, I need to apply a secondary correction to fix some residual banding. Anyone with that full-sized D200 JPEG...?
|