19 May 2020 - APP 1.080 will soon be released with full Fujifilm RAF support, so that will include SuperCCD & X-Trans camera's 🙂 !
2019 September: Astro Pixel Processor and iTelescope.net celebrate a new Partnership!
[Closed] Astro Pixel Processor 1.073
Astro Pixel Processor 1.073
- FIXED, Star Color Calibration, fixed a bug in the star color calibration tool. The bug would manifest itself when the sky background value was (very!) high in ADUs and it would cause star color calibration to fail.
- IMPROVED, Console Panel & Progress Monitors, scrolling or not scrolling text listens now to the mouse wheel as well. If the vertical scrollbar is set at the lowest position with a mouse dragging the scrollbar or the mousewheel, scrolling continues. If the scrollbar position is not at the lowest position, automatic scrolling of the text stops.
- IMPROVED, IMAGE VIEWER OpenGL and Non-OpenGL, it's now possible to zoom in/out with the mouse wheel if you have a mousewheel on your mouse. This enables quick zooming especially with the OpenGL image viewer.
- IMPROVED, IMAGE VIEWER OpenGL and Non-OpenGL, if you click on one of the mouse buttons and drag the mouse in any direction, the image will automatically scroll, so this enables mouse dragged scrolling in the image viewer. If a tool/feature is active in which you need to draw area selectboxes, you can draw these with the left mouse button, and mouse dragged scrolling is then still possible with any other button on your mouse, except the left one.
- FIXED, Data Calibration, when masters are assigned to the light frames, the iso/gain/exposure value matching of masters could give a warning (for instance, multiple masterdarks were loaded with different iso/gain/exposure values) when the warning really was not needed, due to floating point imprecision of the values. This will no longer happen. See the screenshot of the problem: the light frame has ISO/gain of 1.30999999..... = 1.31 and exposure time of 1800 seconds. The Masterdark with ISO/gain 1.31 and exposure of 1800 seconds, should have been matched without asking.
- FIXED, IMPROVED, Data Calibration, APP is now compatible with 32bits masters created by CCDStack.
- UPGRADE, IMPROVED, Local Normalization Outlier Rejection Filters, in 6) Integrate, you can choose from 4 different Local Normalization Outlier rejection filters. All 4 filters have been upgraded. Local Normalization Outlier Rejection is an innovative way to correct pixels values relative to each other before outlier rejection filtration is performed. It improves the robustness of outlier rejection and therefore is better at preserving real signal and thus better Signal to Noise ratio when compared to other outlier rejection filters like Linear Fit Clipping which is used in other applications as a way to correct pixel values relative to each other before outlier rejection filtration. The current upgrade includes local interpolation at pixel level between the local sky background values which were found when the integration layers were analysed before the actual integration starts. The old Local Normalization filters could give artefacts on bright stars, especially when data from different camera's and telescopes were used or if the star sizes and diffraction patterns change significantly between the frames to integrate. The following image shows such an artefact on H-alpha data combined from 5 different telescopes + camera's. On the top, you see the integrated data. On the bottom the Outlier Rejection Maps, which shows where pixels were rejected. The left side is from the old LN filter, the right side is from the new filter. Not always, is the artefact clearly visible in the integrated result, in some cases it is clearly visible. The outlier rejection map clearly shows that with the old filter, the rejection is not done uniformly or consistenly between adjacent pixels. With the new filter, thanks to local interpolation on pixel level of local background values, this artefact has disappeared. Please note, diffraction protection was enabled in this example which causes the diffraction spikes to be protected.
To illustrate how well the new LN filters work relative to non-LN filters, I show the Outlier rejection maps of the new LN filter compared to it's non-LN filter. For illustrational purposes, the data was integrated without Multi-Band Blending and Local Normalization Correction. This clearly shows the need for better outlier rejection filters that can compensate for local background differences due to suboptimal data normalization to start with...:
- FIXED, Multi-Channel & Multi-Session integration with non-equal weights and Multi-Band Blending enabled, an important bug was found by Mike Stutters @mestutters and he reported it in the following topic: https://www.astropixelprocessor.com/community/main-forum/odd-result-when-integrating-with-per-channel-and-all-option/ The bug showed a problem that seemed to be caused by the Multi-Band Blending (MBB) algorithm. If MBB was disabled, all worked fine. It turned out that the bug manifested itself due to an improper calculation of integration weights in the Multi-Channel and/or Multi-Session modes. So this means, that integration weights per frame were miscalculated previously as well in the Multi-Channel and Multi-Session modes. This is an important bug fix ! All weights will now be correct when non-equal weights are chosen in integration combined with enabling Multi-Band Blending. The error would show an integration like this:
- IMPROVED, MULTI-BAND-BLENDING (MBB), the MBB algorithm has been upgraded to a multi-threaded version. The algorithm was already quite fast on 1 cpu thread, but now, especially on big images, the MBB calculation will be significantly faster, resulting in faster integration time. Normally, on a 36 MegaPixel color image, the algorithm would take 2-3 seconds. This will now be less than 1 second depending on the amount of CPU threads. So using MBB on 100 large images, it would normally add 200-300 seconds to the total integration time, this will now be less than 100 seconds if you have mutliple cpu threads available.
- IMPROVED, DRIZZLE/MBB INTEGRATION OUTPUT WEIGHT MAP, the weight maps, are now properly scaled in 8bits depth. They used to be scaled in values of 0-100, now they are from 0-255. However, actual application of weights in integration is done in 16bits, because integration weights, drizzle weights and MBB weights are all combined to give a pixel in integration a certain weight.
- FIXED, DRIZZLE/BAYER DRIZZLE INTEGRATION using SONY ARW files, as reported by 2 different users ( @dav78 & @eshy76 ) drizzle and bayer drizzle integration did not work with the newly supported Sony ARW files. This has been fixed. Drizzle and Bayer Drizzle integration will now work fully with Sony ARW files.
- FIXED, IMPROVED, enabling Bayer Drizzle on Super Pixel data, If you have set the debayer algorithm on Super Pixel and the Integration mode to Bayer Drizzle, APP will now warn you that this is not possible. The Super Pixel debayer algorithm will not give correct registration parameters for Bayer Drizzle integration and even more importantly, the CFA holes in the images as a result of not-debayering or not there anymore due to the Super Pixel mode, so Bayer Drizzle integration on Super Pixel data is not possible.
- FIXED, IMPROVED, Switching Debayer Algorithm while processing, if you switch the debayer algortihm after 5) Normalize, you will be asked if you want to restart data normalization. If you change the debayer algorithm form or to Super Pixel, you will be asked if you want to restart all data processing. A complete data processing restart is needed in this case since image dimensions will change, and thus registration parameters, star locations and normalization details will change as well.
- FIXED, SUPER PIXEL DEBAYERING, super pixel debayering was not working when 32bits calibration masters were used. This is now fixed. The error that popped up was a java.lang.IllegalArgumentException: Unsupported data type 4:
- IMPROVED, NOISE INTEGRATION WEIGHTS, the integration weights calculatedm based on noise of the frames is now
noise_integration_weight = noise_of_frame_with_the minimum_noise / noise_of_frame_i
it used to be
noise_integration_weight = Square-root of (noise_of_frame_with_the minimum_noise / noise_of_frame_i)
So a frame with twice the noise of the frame with the least noise, will only have half the weight.
- FIXED, INTEGRATION WEIGHTS QUALITY, the actual calculation of the quality integration weights contained a small error. In some cases, the values were not the same as reported by APP after data normalization. This is now correct.
- IMPROVED, INTEGRATION WEIGHTS, APP will now report the actual used integration weights in both the console panel and the integration progress monitor.
- IMPROVED, INTEGRATION ANALYTICAL DETAILS, APP will now report the analytical details like FWHM, noise, SNR, star shape and Total exposure time in the console panel and the integration progress monitor.
- IMPROVED, MASTER CALIBRATION FRAME MATCHER, when processing binned and unbinned frames at the same time, you will have Masters of different sizes from the same camera. Previously, APP did not filter directly on the required image dimensions of the frame to calibrate. This is now added to the Master Calibration Frame Matcher module. So masters of different sizes will now be directly matched with the correct light frames, without APP popping up the question, which master needs to be assigned to which light.