New in Version 1.7
Summary
- The steps needed to make the mask for the sky should go more quickly now. Placement of the red dots in the sky is faster and better. Finding the mask is now faster and better. There are slightly different controls for placing red dots and finding the mask.
- The composition algorithms have been changed. There is an option to control the trade-off between star duplication and noise at the boundary between sky and ground.
- Alignment is faster, typically by about 20%. The alignment results are better when there are few visible stars or a lot of noise.
- Variable zoom levels are now supported.
- Other fixes and improvements.
Details
The search for possible stars that are used as an indication of where the sky is is now faster and better. The result is a more even spread of red dots with less likelihood of being too close to something in the foreground. Much less hand editing of the red dots is needed to get a good starting point for finding the sky. The menu of densities of red dots has been replaced by a push button to toggle between the automatically found dots and no dots.
The painting of red dots is now smoother. This helps in those cases where you have to paint dots to indicate where the sky is.
The generation of the mask for the sky is faster and more accurate. Two algorithms are supported. One will find disconnected parts of sky, for example patches of sky visible through trees. I call these patches of sky surrounded by ground “islands of sky”. The second algorithm will not find islands of sky. This second algorithm is useful in cases where parts of the foreground might be confused with sky. Most people, most of the time, will want to use the algorithm that finds islands of sky. There is a push button to toggle between these two sky masks.
There are many fewer small holes in the sky mask and fewer small gaps along the sides.
The composition algorithms have been changed. The algorithm “median” (which did traditional outlier rejection before computing the median) has been replaced by a new algorithm that integrates median with outlier rejection that is more appropriate for the noise typically seen in dark sky images captured with modern digital cameras. In testing this algorithm has always produced better results than any of the algorithms from version 1.6. There are two versions of this algorithm. “Min Horizon Noise” is designed to minimize noise at the edge of the mask for the sky. For most people, most of the time, this will produce the best results. Under some conditions, the algorithm “Min Horizon Noise” may trail or duplicate stars near the edge of the mask for the sky. In this case, the second algorithm “Min Horizon Star Dupe" may produce better results. The algorithm "mean" still does traditional outlier rejection. There are two version of mean to support minimum noise on the horizon vs. minimum star duplication on the horizon.
The test for consistency of exposure among all the images in a stack has been relaxed slightly so there should be less warnings about inconsistent exposures. (Some intervalometers and/or cameras seem to lack consistency in exposure duration).
Multiple zoom levels are now supported from “Fit” to 8x actual pixels. The ‘z’ key still toggles between “Fit” and “Actual Pixels.” Cmd- ‘-‘ and Cmd - ‘+’ now zoom out and zoom in respectively.
Fixed an error that caused alignment to fail if both the time between the exposures was too large (30 to 40 seconds was probably the limit) and the EXIF data did not have the sensor resolution (many cameras include this information in the EXIF data, but some do not).
A stronger warning is now issued if the input files have an unknown color profile. Under rare circumstances unusual profiles can cause a crash.
The metadata in the output tiff files is now cleaner and more complete. It had been distorted to work around a bug in Affinity Photo that has now been fixed. If you are unable to load file written by Starry Landscape Stacker into Affinity Photo, try upgrading to the latest version of Affinity Photo.
Fixed a problem that was preventing all the metadata from being copied from the input files to the output files. In particular, the XResolution and YResolution values which are often reported as PPI or DPI are now correctly copied.
Improved performance of scrolling. For Mac OS 10.13 and earlier, performance of scrolling and window resizing is very good. For Mac OS 10.14 after changing zoom level or resizing the window scrolling can be slow and jerky. After several seconds it usually becomes very smooth and fast. Window resizing tends to be jerky with significant hesitations. A future release should improve scrolling, zooming and window resizing for 10.14 and later.
Alignment is roughly 20% faster and is more accurate when there are few visible stars.
Clicking “Adjust Dots” (was “Adjust Stars”) no longer resets the zoom level.
Several small UI adjustments to improve clarity and consistency.