20 January 2021: Soon to be released APP 1.083-beta2 : improved comet registration, updated tools, new Star Reducer Tool and more...
16 November 2020 - Wouter van Reeven has officially joined the Astro Pixel Processor Team as a moderator on our forum, welcome Wouter 🙂 !
The Batch/Resize tool only has a slider in integer % increments, and it's not quite accurate. I first ran into this issue with mixed sets of images binned 1x1 and 2x2. The 1x1 were 4096 pixels wide and the 2x2 were 2048 pixels wide. To use the RGB Combine tool the images must be the same resolution and bit depth, so I used Batch Rotate/Resize to scale the larger images by 50%. The results are 2049 pixels wide, not 2048 so I am looking for a method to scale to a specific resolution. Even if Batch Rotate/Resize did correctly scale to 2048, that would help in this specific case but it's quite possible to need to scale an image by a non-integer %.
Is there a way to precisely resize images? Or am I going about this the wrong way somehow?
I have successfully integrated 1x1 and 2x2 binned subs from the same camera but have never resized them beforehand. Just load all the frames you have together with calibration frames then run through the normal APP steps 4) Register to 6) Integrate and APP will automatically handle the resizing. I have always used a 1x1 frame as my reference and APP interpolates and upscales the 2x2 frames to match. I guess the opposite would apply if a 2x2 frames was selected as the reference frame but I've never tried it that way round.
best of luck
I'm not trying to integrate subs of different resolutions, I'm trying to RGB combine FITS images of different resolutions that are already integrated. My first thought was to resize in APP but don't see that as possible. I'm not sure why a 50% resize of 4096 pixels results in 2049 pixels.
I've never tried to accomplish what you are attempting so not sure quite what would be the best approach. I think if I was in your position I would still reload the separate RGB integrations back into APP as lights then repeat the APP processing steps starting from Star Analysis. This may not be the sole approach but I think you will definitely get you a result. In effect you are capitalising on APP's ability to handle images captured with different camearas and optical systems to correctly align and integrate subs - only in this instance there are only single subs to be 'integrated'. With only three subs the reprocessing will be done in moments.
I don't know how you created your separate RGB integrations but if they were not all created using the same reference frame then I suspect your RGB subs may not be properly aligned even if you sort the dimension issue.
If that fails I would think of doing the RGB combine in eg Photoshop once the size issue is sorted.
I'm sure Mabula or Vincent will step in if there are better solutions to your issue.
These aren't files I created, they are stacked files somebody else created. Each FITS file is a stack for a single LRGBHa set. Some are 1x1 binned and others are 2x2.
I'm trying to RGB combine stacked results from somebody else who didn't use APP. I don't have their lights.
I suppose I can convert to tif and resize in PS, but I was hoping there was something I was overlooking in APP, since I didn't expect the resize tool to be unable to do it.
OK, I'm getting to understand the circumstances. However I still think you should be able to load the stacked FITS files that you have into APP as 'lights' (Step 1 - Load). You should then be able quickly to process these through APP's normal workflow and at end have new 'stacked' LRGBHa files that have been registered and normalised by APP. These should will load into RGB Combine.
I'm not sure I follow... do you mean process them as multichannel with just the one file for each channel? I can try that after tonight's imaging session. I didn't consider that but I think I see what you are getting at. It would basically be a project where each channel just has a single sub. Thanks for the idea.
Yes, that it is what I am suggesting you try. Not sure if the FITS files will load as Lights or Other Processed but as long as they load I think they will process thru and thus all aligned and normalised at the finish.
Not sure if your original issue with Resize is a fault or a feature!
Hope it goes well
Imaging was cut short by clouds tonight so I gave this a try. It seems to have worked created 5 FITS files with width 2052. No idea how it came up with that number, and I kind of wish it would interpolate the bin 2x2 up to the resolution of bin 1x1 but for this purpose it is good enough.
But, lets say I want to do a LRGB project where I collect 2/3 time on Luminosity binned 1x1, and the other 1/3 time on RGB binned 2x2? I want to keep the full detail of the 1x1 Lum channel and just use the 2x2 RGB channels to colorize it. How do I do that, instead of having the Lum downsized to half the resolution? Some use of image scale?
I am pleased that the suggestion worked for you. Regarding the final image size of 2052, this may have something to do with the Composition mode that you selected at 6) Integration. For example if you chose mode = Full and if APP had felt your stacked image files were not perfectly aligned it would have extended the frame size to cover the extended image size. If you want the final 'stacked' image size to exactly match the size of an input frame then select mode = Reference.
Regarding your second question about stacking 2x2 binned RGB subs with 1x1 Luminance, then provided you select one of your 1x1 binned L frames as the reference frame then I'm sure you'll find that the resulting stacked image files for the RGB channels will be at the same resolution as the L frame (assuming of course that you don't use any of the drizzle parameters at 6!).
I'm not clear which method APP uses to handle the upscaling of the binned image sizes but I am sure it is robust - after all the normal case is that frames for integration at the pixel level will be somewhat rotated and misaligned / dithered so the contribution of each sub-frame pixel to the final image matrix must be carefully weighted. The problem is extended further if frames captured from with imaging set-ups / image scales are to be stacked.
To recap, in my experience APP always outputs integrated image files that are at the same image scale as the selected reference frame but the final image size will also depend on the composition mde chosen (Full, Reference or Crop).
I have only used DSS for stacking prior to using APP but I do think it is very good. Hope you enjoy it to.
James Paul, I agree with Mike that you need to ensure that a Luminance bin1 frame is chosen as the reference frame. I do a lot of LRGB using bin2 on the color channels. Beginning with APP 1.075 I noticed that the Analyze Stars process would pick any of the multitude of frames as the reference, regardless if it was bin1 or bin2. I now remind myself to scrutinize its choice and override when necessary.
Thanks for all the help! Using the bin 1x1 for the reference image makes sense, I expect everything else including drizzle probably works as expected. I'll experiment some more when I have the time. I also used DSS prior to APP and think APP seems far better in both features and results. Memory management and performance is also superior, I had difficulties even attempting 3x drizzle on my 32GB Windows10 system with DSS as it would often crash. APP has much more flexible drizzle parameters (not being limited to just 2x or 3x image scale alone is very nice) and performance is much faster and has never crashed for me, even when constraining it to 12 or 16GB RAM. In fact, even if I don't give it enough RAM it still doesn't crash and just impacts processing speed. Compared to DSS, the multichannel and multisession workflows are vastly easier to understand and manage, and APP's ability to process multipanel subs and mosaics makes what most people expect to be a complex process practically effortless.
I look forward to 1.076. A feature I'd like to see someday would be to save and load settings, so I don't have to remember to set the options on each tab every time for my own "defaults" that I use most often. I mitigate that a bit by often leaving APP open for days and just "clearing" from tab 1, so all my options are still set.
Thanks for all the advice in here, Mike here is correct as well, you can load any data of the same object (being it from various different telescopes or not) into APP and perform the registration etc on them so APP can nicely integrate them. I didn't practice with bin2 myself, but I'm assuming bin 1 will already be the reference frame to stack against as it has "better", sharper stars and such. APP 1.076 will come out, most likely this week (fingers crossed, but he is working on completing it right now).