windjammer

Quite marvellous! Is there a time lapse animation of the globe that you could do to go with it ? (that probably doesn't have to be annotated... ) Simon

Your first one of the three - the reddy, murky, mysterious one is fabulous.

An interesting choice, I would have PC+bits in the warm room and USB extenders to the scope. PC does double duty as the office PC ....

Ah, but Hale's pic clearly has canals !

Ooops

When you put it into manual guide mode does the scope move when you toggle E/W and N/S ? If it does then check the pulse durations in manual mode are the same'ish as the pulse durations in calibrate mode. if manual mode doesn't move the scope after twiddling pulse widths then clearly something is not communicating: driver installations etc. If you use ST4 then a scope on the control lines will show pulses - or not. If its fancy ASCOM then beyond my pay grade, unfortunately. Simon

Have you had a look at the Lindy active USB3 cables and hubs ? Expensive but work well - and capable of serious distance https://www.amazon.co.uk/LINDY-10m-Active-Extension-Port/dp/B00DYWRRES/ref=sr_1_2?crid=16RZPMSS34JAO&keywords=usb3+extension+lindy+hub&qid=1672879838&sprefix=usb3+extension+lindy+hub%2Caps%2C97&sr=8-2

I use an Atik 460EX mono on a Startravel 150 F5. The 6Mpxl Atik sensor is about 12mm by 10mm, fits nicely inside a 11/4inch filter window. I have used it with a 0.8x focal reducer at F4 and currently with a 2.5x barlow at F12.5 for galaxy hunting. It works well with mono RGB filters and narrowband - I have not noticed pincushion even at F4 with Atik sensor (which admittedly is quite small by CMOS standards). Regarding chromatic aberration, Blue is not as sharp as it could be but as someone has already mentioned, it all comes out in the processing wash. I have an Astronomic L3 blue cut filter in the image train that truncates the extreme blue that helps a bit. As someone else mentioned, mount fettling becomes the more important issue. So if you do go ahead, my 2 bits is get a cooled mono camera with a smaller sensor and you should get years of fun out of your 120. Good enough to learn the trade anyway! Every so often (like now!) I think about upgrading - but I cannot claim to have exhausted the potential of the current set up, not by a long way. I am no great shakes, but here is a link to some of my pics if you want to have a look Simon

What's wrong with spaghetti ??

I use one of these, an A4 tracing light box. https://www.amazon.co.uk/dp/B073F97NS3/ref=pe_3187911_185740111_TE_item Here it is in action: They come in A3 sizes and 5V and 12V versions Simon

That Mars occultation is a cracker

Yes, using onboard graphics (Dell motherboard), no 3rd party graphics card. They always give up the ghost sooner or later! Simon

Ah, just got it. BXT is Blur Xterminator ?

>>BXT has enabled me to find a processing workflow Are there any details you can spell out ? I need something to hunt down lurking details Simon

>> "Exploiting the Equalise Function" tutorial Do you have a link for that BTW ? Simon

Thanks for the comment Martin, and yes, a chronic problem of mine I was trying to converge on this Adam Block image from the 0.8m Mount Lemmon Schulman Telescope https://commons.wikimedia.org/wiki/File:M66_Galaxy_from_the_Mount_Lemmon_SkyCenter_Schulman_Telescope_courtesy_Adam_Block.jpg It seems to go with the territory using a mono cam. And irrecoverable errors get baked in early on, a bit like beginning lumps in gravy. Still, I was pleasantly surprised at the details that were lurking and came out: 0.15m vs 0.8m. I will have to reprocess and see if it takes a different path... Simon

Hi Here is M66, NGC 3627, part of the Leo triplet of galaxies. Mono RGB and Ha composite. The exposures were done almost 3 years ago(!) but until recently had taken too many stupid pills to get anything out them. Cloudy nights and freezing weather I wasn't stupid enough to go out in means they came to the top of the to-do list. I might have another go this Spring - telescope has been fed and fettled and looking for a challenge. Simon Details: M66 is the largest and brightest member of the Leo triplet of galaxies, a grouping of three gravitationally interacting galaxies in the constellation of Leo. Photograph taken in Astronomik RGB broadband filters, Ha narrow band filter. Total exposure time 7.5 hrs. R 2x2 bin - 15x300s = 1.3hrs, 22 March 2020 G 2x2 bin - 11x420s = 1.3 hrs, 23 March 2020 B 2x2 bin - 18x420s = 2.1 hrs, 24 March 2020 Ha 2x2 bin - 11x900s = 2.8 hrs, 26 March 2020 Rig: Imaging scope: SW Startravel 150mm F5 Refractor, 2.5x Celetron Luminos 2inch imaging barlow, Atik 460EX mono Guide scope: SW Evostar 90mm F10, with guiding XY stage, ZWO 120MM camera Guiding: 2 stage PHD: high frequency guide scope (mount tracking) and low frequency OAG image train guiding (guidescope flex) Mount: Home made German Equatorial pillow block mount, permanently rooftop mounted. Spring loaded DEC axis gearing. Other gadgets: ST4 based anti vibration shutter, ST4 based PEC Processing: PixInsight: Lights, Darks, Flats, Biases, Align Calibration, StarNet2 star removal/star layer GradXpert: Gradient removal Topaz DeNoise AI: Noise removal Affinity Photo: 32 bit image processing (curves, high pass masking, selective colour)

>>V2 on Windows has had a huge OpenCL stability upgrade, Fyi, I run Windows10. I had the OpenCL hardware acceleration turned off in V1 because of crashes. The download of V2 turned it on again. V2 had a spate of crashing, so it is now turned off, which got around the issue. So perhaps some more work to be done on that! Simon

Hi JR I have been using AP and your macros for the last few months - my processing has improved hugely. Great job. One thing I would like to see is the ability to export files as 32 bit FITS. I use external tools like graxpert and pixinsight, and using tiffs to move between AP and others doesn't seem quite right. I am sure bit depth gets mangled along the way! Simon

V - Still thinking abut this. I have some formalism but doesn't go very far beyond stating the obvious: Say G(x) is the transformed histogram H(x), G(x) = H(g(x)), g(x) the stretch function we want. The derivative G' = (dH/dg).(dg/dx) We specify an Xb, the bottom of the histogram below which we choose some linear function. We want G(Xb) = H(Xb) and G'(Xb) = H'(Xb) = (dH/dg).(dg/dx) Which doesn't offer many more clues on the g(x). I might play with adhoc methods to join up a linear bottom end with an interesting g(x). Perhaps just join up with something like quadratic interpolation through a couple of points - that would satisfy your G=H and G'=H' at Xb criterion. If it looks promising try and regularise it afterwards. An Xt can be defined as you suggest, the top of the histogram at which we want the pixel to be saturated, so that g(Xt) = 1. And all the fun is in between. Simon

Hello Ian Thats an interesting function - see pic. There is a possibility of a set of curves in there that deal with the foot of the histogram in a more natural way than just cutting it off at the knees. And also a set of curves that naturally move the histogram to the left after a stretch. It needs a play to narrow the parameter space a bit. Simon

Hi Martin That is a *very* fierce stretch - see first pic. After playing around with the parameters I went back to your basic f(x) = a * x / (x + b). A bit of a struggle with normalisation I ended up with f(x) =x* (1+b)/(x+b), b:(0.1). See 2nd pic. This looks very workable - looks as good as arcsinh or x^p, but very non linear in its variation with b. So I think a tool would come in 2 versions. Strong XXX, where b is scaled 0 to 0.15 and Auntie where b is scaled 0.2 to 1. I think I am still waiting for the magic function where the whole stretch bore is done and dusted in one click. Maybe AI is the answer... Simon

>>after that, I stretch manually but only above the background level. yes, the control described here has a black point set so that you can decide how much of the crud at the bottom you want to play with. >>f(x) = a * x / (x + b) Thats interesting, I will give it a try! Do you have a feeling for what a and b correspond to, or just take them as magic numbers ? Simon

I am amazed that horsehead is an unprocessed singe exposure! I love the image scale - no mucking about. There is a lot left even in that single shot to bring out. With a stack it would be remarkable.

So, as you point out, a colour preserving transform, which is (as I understand it) one of the plus points of the arcsinh transform, and that, we agree, is how it does it - and is of course entirely general for any constant across the channels. Colour gets lost in endless cycles of manually driven curves. (I'm not quite sure why that happens, but it does). The other advantage of arcsinh is (I'm told) that it gets to a stretched result in less iterations than curves cycles, and being algorithmic is (in principle - pesky sliders reappear) reproducible, which is my interest. Who really thinks that after grinding round a dozen curves iterations, shifting points around on the curve, you would get the same result if you started from scratch? Similar, yes, but not the same. But the point of the post was to find an alternative to arcsinh, which seems to be very popular for the above reasons, that was within the computational capabilities of Affinity Photo and would give similar results. So a simple function along the lines of x^p which has similar properties to arcsinh, and integrates easily into a non-destructive live layer with only 2 controls: black point and stretch. What's not to like ? Simon