rotatux

Interesting: so DSS does multiple iterations, while Regim does only one. Hence the very different "kappa" values. It might just as well fall back on my values range if using iterations=1. While I'm highly confident at the mathematical / statistical meaning of doing only one iteration, I'm more in unknown territory about doing many, especially as it will progressively shift the average value used as criteria. I still have to think about it as it seems to "just work" for many people. Is there some kind of synthetic report in DSS, where it would indicate the net proportion of kept pixels, so we can get an idea of the final selectivity of this algorithm ?

BTW this algorithm has a parameter, let's call it the "sigma factor". What values are you using ? /me depends on subs quality: poor => 0.5 or less, average => 1.0, good => 1.25 to 1.5, very good => 1.6 and above. I also noticed how the pixel keep rate seems related to the sigma factor, e.g. 0.5 gives around 25%, 1.0 => 50%, 1.5 => 75 to 90%. So that may also be a selection criteria if you want to keep a given pixel percentage of your subs.

For me it's another story as I am on holidays... I made a session for the whole night on 2017-05-24 from 10pm to 4:30am, had to stop after daylight had become too strong. Unfortunately some strong wind came after 11:30pm (black night starting around 11:00pm) and ruined most of my subs... the Nexstar SLT is so brittle that any wind will make my tubes shake , how are yours ? (especially the Synscan) And the show must go on, here's one which came out not so bad from what I could save. The Leo Triplet. Gear: Olympus E-PL6 with Skywatcher 130PDS 632mm/4.86 and SWCC at f/4.55 on Celestron Nexstar SLT Capture: 22 lights (/ 18% keep) x 30s x 3200iso, 23 darks Process: Regim 3.4, Fotoxx 12.01+

No, never astro-modded either the PM1 or the PL6. Currently I don't intend to, as IMO I'm unable to do it and would just detroy the sensor if trying. What would you think of the red sensitivity of the stock PL6, based on my image of America ? Intuitively I find it's much better than PM1, when I compare with e.g. that previous try.

Thanks Neil. I don't have Photoshop (no Windows, no Mac). If you know of a Gimp plugin then I could try something else... Actually if you look carefully there's both blue and red rings around bright edges (check that moon image where the red ring is obvious), but with different size and intensity. That's a type of CA (I think "color dispersion") which I have no plugin to handle; That's neither Directional CA nor Lateral CA, which I already have a plugin for. I'm ready to accept it as the price to pay for using the lens at full-aperture. Though I *can* reduce or get rid of it by using a DIY mask to reduce aperture to F/5 or F/5.6, but I'm just reluctant to use them because that fault is very uniform up to the edges (kind of artistic, one would say) and makes star colors more obvious.

Let me share two more processed images from 2017-05-16' ~night~ with the OM-Zuiko 200/4. First is Bode's galaxies, code M81 and M82. Two more NGC galaxies in the scene. Shows somewhat unexpected structure for that FL. Details in my gallery. Second is America Nebula (NGC7000) and Pelican Nebula. The 200 FL fits almost perfectly that targets. Details in my gallery. I tried some HDR to detach the brightest stars and give some 3D look, but that masked the faintest parts under the Pelican, so I stayed with the traditional version.

Hi Steve, thanks for comments. Absolutely no filter, as most of the time they remove too much wanted photons Setup details are in my gallery page: follow the rabbit link just under the first image ;-) . I was trying a new way to not clutter posts too much, would you prefer / should I rather copy the settings in each post ?

2 nights ago I managed to have another go at Markarian's chain, and tried different settings (mainly, higher ISO: 3200 instead of 2000). I find it's much better than previous try, maybe except blobby stars typical of refractors and strange star colors (though I see it also in SDSS images on astrometry.net). Was also much easier to process, as I didn't have to go at the limit of stretching to get decent details. Of course still more frames wanted, but I had other targets within that available 2 hours window. Details in my gallery. And annotated with astrometry.net job: Discovered Siamese Twins (left half-height) and strange NGC4299 (lower right border), going to read about them.

As requested, plate solve of my previous image through astrometry.net: (job)

About read noise: yes for my cam it is about constant (and low) when varying ISO, just like sensorgen says. I've also done my own tests and can confirm this, as those graphs show. I believe they would look the same on any modern CMOS sensor, apart from the data range (my sensor is 12-bits so 0-4095 only). What's more interesting for us is that the saturation capacity is not the same at all for each color (I don't speak of mono sensors of course), and it appears sensorsen only covered green capacity. BTW intermediate isos are real steps in the amplification stage, so worth investigating, contrary to what many sites say about staying to "round" values. You can also see that the E-PL6 has about max range at 800 iso in the reds where it's the weakest. So IMO there would be no point going to 400 or 200 iso, let apart to expose correctly for the chosen duration. Additionally I generally prefer to raise ISO because 1/ IMO half to quarter of the data range is acceptable (sacrifices only 1 or 2 bits that will be gained back with stacking), and 2/ I want higher numeric values ("ADU") to average out to higher precision numbers (this is debatable and depends highly on my 16-bit limited software processing chain). This is been debated here many times ;-) so I would prefer not to expand too much and come back with more actual results... i.e. DSO images

Yes exactly thanks for pointing to it Didn't exactly remember where I saw it, so cited from memory -- wrong of course.

I should explain more precisely: The background was limiting exposure because of the nearby moon and dust. I wanted to use 30s because that was the (theoretical) max in that zone to avoid FoV rotation, then had to severely limit ISO to get an acceptable bg level. But at such iso level I have quite high saturation capacity (I would say too much!), and the sensor could be used IMO more efficiently at higher ISO. Thinking back, higher ISO and shorter subs would have lead to more frames (and less noise) within given time.

nova.astrometry.net is down with 503 just now, but @Uranium235 already posted a even wider FoV of this region in a giant annotated image, in the "imaging with 130PDS" topic. Next try I shoot at this (and post a better version), I shall remember to add a plate solve version. In fact the result could have been even better if I pushed the ISO (which my new cam would easily allow) but the odds -- fov rotation, dust and nearby 68% moon -- were against me. Next try in a few days (weather and moon permitting).

After several months without getting outside because of bad weather, can't resist to post some results trying new gear (camera body and lens). Markarian's Chain and neighbours: Exposure's not right and result is missing depth, but as this is my first at this subject. Stunned by the number of catchable galaxies sitting there. Pleased with the result and the new lens, but not yet good enough to post to the 30sDSO challenge. See here for details.