|
Image!
Ok as promised I have a screenshot comparing APP and PI using same data as quick and dirty as a newbie would do. This does not involve tweaking of setting to get the best from either program. In PI I used the BPP script as is, and in APP I had to run the numeric tabs in order manually touching as little as possible. I tried to make it as fair as possible and not the best that either program should be capable of.
Annoying for me it contradicts my earlier quick tests and i still stand by those comments and can't account for the deviation I got for this test where i was trying to give each program a fair chance.
Shots were Nikon D800E with f2.8 200mm lens on tripod. Sets of lights (200 frames), darks(35), flats(77) and bias(38) were taken. The flats were over exposed so not ideal. In body light pollution filter in place as always.
PI I loaded everything into the BPP script set reference frame (first light) and save directory, hit the go button and went to bed. In the morning I had my finished integration frame with rejection maps attached. It chugged through it all, no crashes or user interaction needed.
APP however took days to go through it all and I had to step each tab in its process myself as it required some user input during it. So I would set steps off before bed or work as needed to go through each step. My quick laptop tests didnt require as much interaction.
My feeling is it took much longer on APP by default than PI just accounting for actual processing time. The screenshot attached shows the integration frame result from both programs left is PI, right is APP.
The flatter windows underneath sho closeups of the center of each integration frame.
Observations:
1 - application of calibration data has improved overall vignetting of the right side, the APP image. PI clearly exhibits diagonal lines in the vignetting and less smoothness overall.
2 - worryingly the APP stars all show a vertical streak artifact I have yet to see in any of my testing. PI is much better and shows my lens was out of focus, no hint of those streaks (they are not in my lights either). I think this is an integrationartifact of the default settings of APP, again this is the first time I've seen this so it doesnt confirm my earlier observations of better strs than PI.
3- more worrying APP if you look close produced an integration frame that is larger than the registration target in both width and height. You can see the ugly blank portion in the top right corner where lights didnt cover that
4- the colour cast I can only attribute to the differing default settings. APP takes you through a calibration step and a normalise one. Again with newbies in mind this gets you closer to a nice final image to show off with as little tinkering as possible.
Absolutely NO cosmetic processing has been done, both files loaded into PI were black with a few point stars visible, I used STF to adjust the levels for display, dragged this to the workbench and from ther dragged to each image so they both have the same visual adjustment applied for this comparison.
In recent months I've seen a number of people asking about moving beyond DSS and photoshop and I've tried to post with them in mind, experts in any package have nothing to fear from APP today. I still think it will improve and stabilise fast and it shows some promise to me, unfortunately my first level field test doesnt show that to you all. Until this I was exclusively doing full mosaic integrations which are of worth to me rather than to a registration target. But I cant account for the difference in integration sizes, both programs appear to have used the full raw dimensions but APP did a litlle more, perhaps its a stretch of the images, that may also account for the star deformation too.
I'll leave it there, need sleep and got python code to work on this weekend but I might point the author of APP to this and he can investigate if he wants.
|