Brian Beezley
Mu-43 Hall of Famer
StarZen Technologies, Inc; DataFlex tools and books
I like this bad-pixel utility, but I discovered that results are highly variable unless you carefully control the RAW processing parameters.
First of all, I use both Olympus Viewer 3 (OV3) and RawTherapee (RT) to process RAW files. They yield different results for bad pixels. For one thing, I discovered this morning that when RT saves a full image, it generates 4088 x 3076 pixels, not 4032 x 3024. Perhaps RT includes extra pixels at the edges of the frame the camera uses for IBIS. There are other differences between the two programs, including the demosaicing algorithm. I decided to standardize on OV3 for bad-pixel tests. I generate a 16-bit TIFF output file for the utility.
The first time you open a RAW file, OV3 initializes its processing parameters to the values the camera used when it captured the image. (On subsequent file openings, it uses the values you last set.) Many of these parameters affect the bad-pixel test. Some, like exposure compensation, you'd expect to have an effect. Others, like white balance and sharpness, you might not. I normally set my camera for automatic white balance, but there's no guarantee that the algorithm adjusts the balance the same way from one dark frame to another. So I decided to standardize on Daylight 5300K for bad-pixel tests. The Sharpness parameter can be deceiving. The camera defaults to 0, but this appears to sharpen the image. I normally use -2, which I believe does no image filtering. These two settings yield different bad-pixel results.
The other thing I discovered this morning is that the number in brackets next to an OV3 parameter setting isn't the current value as I'd always assumed. Instead, it is the "As Shot" camera value. This means that you must methodically open each drop-down parameter menu to determine the settings you're using.
If you standardize your settings, I think the bad-pixel utility can be useful for determining whether your sensor has developed new bad pixels. You might also use it to check image noise at various camera temperatures. I've used the X-Y coordinates the utility provides to examine questionable pixels with the OV3 X-Y cursor readout. But without standardized settings, I don't think reporting bad-pixel numbers is that useful.
Brian
I like this bad-pixel utility, but I discovered that results are highly variable unless you carefully control the RAW processing parameters.
First of all, I use both Olympus Viewer 3 (OV3) and RawTherapee (RT) to process RAW files. They yield different results for bad pixels. For one thing, I discovered this morning that when RT saves a full image, it generates 4088 x 3076 pixels, not 4032 x 3024. Perhaps RT includes extra pixels at the edges of the frame the camera uses for IBIS. There are other differences between the two programs, including the demosaicing algorithm. I decided to standardize on OV3 for bad-pixel tests. I generate a 16-bit TIFF output file for the utility.
The first time you open a RAW file, OV3 initializes its processing parameters to the values the camera used when it captured the image. (On subsequent file openings, it uses the values you last set.) Many of these parameters affect the bad-pixel test. Some, like exposure compensation, you'd expect to have an effect. Others, like white balance and sharpness, you might not. I normally set my camera for automatic white balance, but there's no guarantee that the algorithm adjusts the balance the same way from one dark frame to another. So I decided to standardize on Daylight 5300K for bad-pixel tests. The Sharpness parameter can be deceiving. The camera defaults to 0, but this appears to sharpen the image. I normally use -2, which I believe does no image filtering. These two settings yield different bad-pixel results.
The other thing I discovered this morning is that the number in brackets next to an OV3 parameter setting isn't the current value as I'd always assumed. Instead, it is the "As Shot" camera value. This means that you must methodically open each drop-down parameter menu to determine the settings you're using.
If you standardize your settings, I think the bad-pixel utility can be useful for determining whether your sensor has developed new bad pixels. You might also use it to check image noise at various camera temperatures. I've used the X-Y coordinates the utility provides to examine questionable pixels with the OV3 X-Y cursor readout. But without standardized settings, I don't think reporting bad-pixel numbers is that useful.
Brian
as the E-PM1 does in violet. I need a cheap'n'cheerful-looking body for some situations when I want to be considered just a holiday snapshot taker with a cheap point-and-shoot. Violet oh violet don't shrink on me, I want to find a good unit yet 
