rotatux

17 hours ago, Nigel G said:

I cashed in my scrap copper I have been saving ( 2 years worth ) and treated myself to a refractor  The Equinox 80ED Pro.

A big problem being I haven't seen any stars for over a week now, I'm itching to use it, first try will be on the Alt-AZ mount but how long will I have to wait.

30 minutes ago, Tangoringo said:

the other day Ive managed to grab an HEQ5 Pro with a William Optics Megrev 90 for 600GBP. Can't wait now to get my hands on it to have a play. Needless to say, expect a load of help requests over the coming months.

With all that new gear for many people, no wonder why it's been rainy or cloudy for several weeks ;-)

Congratulations both, all nice buys.

On 10/2/2017 at 11:08, The Admiral said:

I'm not sure exactly what's going on here Fabien. Here's an image of the Rosette I made when I first got into astrophotography, and which was posted ages ago on this thread.

[...]

Now this was taken with a 102mm f7 refractor, so from an imaging point of view very similar to your set up. I used 130 x 10s lights, so about half the exposure you used. Admitedly, I'm not in the suburbs of a huge city, and I'm not using a pollution filter, and Orion's belt is clearly visible to me (at least, when this never ending cloud shifts out of the way!). I've found my camera to have a good red response, even though it is unmodified. May be your sky background is just too large. What effect the filter will have I don't know, as I've not used one, but one might hope it would give a significant improvement. It would be interesting to see what the deep red response of your camera is; this site might give you a clue https://kolarivision.com/articles/internal-cut-filter-transmission/, though it doesn't have your actual camera you may be able to identify others with the same sensor. Also, as a previous Olympus user (for general photography) I am aware that the MFT sensors are quite noisy, even the latest ones suffer a bit with long exposures. I fear you may be fighting against the odds here.

As for stars blowing out, I think that is to be expected with exposures used to capture the tenuous nebulosity, particularly with a sensor not specifically designed for well depth. Personally, I don't worry too much about it, and I think if you were to try HDR to capture them then the exposure differences needed would be quite large, much more than a 30s to 20s change would accommodate.

I guess the other alternative in your viewing location would be narrow band imaging, but that would require a dedicated camera such as the AS1600 4/3.

Ian

PS. You might be interested in this review which discusses noise levels in the EM5 and EM1 http://www.wrotniak.net/photo/m43/em1-em5-dark.html

Yes your setup with APS-size sensor and no flattener/corrector has nearly exactly the same FoV as me with M4/3 and ComaCorr. The Rosette framing of both images indeed look the same. Nice picture Ian BTW ;-)

Looked at kolarivision site. I found a cam with same sensor as me, the E-P1. However it's only the characteristic of the filter, not the sensor. I know from sensorgen.info that my sensor has 38% quantum efficiency (though I don't know how to mesaure and check it myself), but I suspect that value also varies with wavelength, so I can't get any conclusion (the net light sensitivity should account for both the filter and the sensor).

About noise: The first generation sensor (12.3mpx) is from Olympus and indeed quite noisy, though manageable. But the 16mpx sensor found in E-PM2/PL5/6/7, E-M5/10 is from Sony and much better (3x less noisy, and 1.6x more sensitive at 60% QE) -- that's what I'm willing to get, maybe second hand. Your 2nd link conforts me in this opinion.

19 hours ago, Filroden said:

I wonder if you've lost some of the data when removing/minimising the light pollution during processing? I found the Rosette, Soul and Orion Nebula all very difficult to process because they filled my field of view, making it difficult to calibrate any background gradient. Pacman was much easier as it occupied much less of the frame.

Here's two of my images of the Rosette - these are both from the same data (just one is rotated 180deg)! The biggest difference between them is that I improved my background model and preserved more of the nebula which existed in the unprocessed image.

Interesting... makes sense given my technique, I need to double check my "NG darks": as the target was rapidly climbing in the sky, I may have captured the wrong gradient and mismatched a bit with the lights background. Oh damn, I have yet to write a blog entry on that "NG dark" technique of mine :-P

18 hours ago, Nigel G said:

I wonder if you add bias frames if it would help with noise levels.

You have captured a lot of data, it seems over powered by noise or light pollution. ( your loosing a lot during processing gradients ) maybe a lower ISO setting ?

Getting better though  

Thank you Nige. I don't think separate bias would help as I already use traditionnal (not scaled) darks dedicated to each session or even subject.

I hope it's LP and it can be sorted out, the filter was here to help. I would like lower ISO if I could achieve longer subs, but the mount is so erratic I'm afraid I wouldn't capture any signal as lower ISO + 20s. Will retry on occasion though.

On 31/1/2017 at 16:57, The Admiral said:

I presume that you've read Craig Stark's article on "The Effect of Stacking on Bit-Depth"?  (http://www.stark-labs.com/craig/resources/Articles-&-Reviews/BitDepthStacking.pdf)

I found he brilliantly explains how stacking can get you additional bit depth (though there seems to be a off-by-one error in bit gains for each of his stacks).

(reading links from your last post, will follow up later ;-) )

Two new pics finished (actually one has to stop somewhere), one half-success and another failed.

First is my #2 try at nebulas around Alnitak. Much better than my try #1, still progress to be done. I guess I can't catch more unless going to a darker site (was taken from Paris suburbs).

Capture: 101 good of 123 lights x 25s x 2500iso, 30 NG darks, Olympus E-PM1 with Skywatcher 130PDS on Celestron Nexstar SLT, Skywatcher ComaCorr and TS-UHC filter. Processing: Regim, Fotoxx.

Second is Rosette on which I failed to capture enough SNR (sorry for your eyes ;-)). I had to stretch so much I had to process the noise, even then the result isn't satisfying. Surely I was too ambitious when reducing the sub length to 20s, I suspect the UHC filter dims the overall image and requires at least 25-30s; Stars were burnt in my previous 30s try, so I decided to reduce it, but I might need HDR on it (or stop trying to image from Paris suburbs from which I can only barely see Orion's belt).

Capture: 50 good + 42 average lights (of 122) x 20s x 2500iso, 56 NG darks, Olympus E-PM1 with Skywatcher 130PDS on Celestron SLT mount, Skywatcher ComaCorr and TS UHC filter. Processing: Regim, Fotoxx.

A few days ago Ken gave me the idea to try to calibrate the color levels coming on the sensor through my filters, so I'm in the process of making some kind of flat in different filter configurations. Hoping to understand my filters characteristics and those bad star colors.

1 hour ago, Filroden said:

Olly has a good graphic about the f-ratio myth that is much clearer.

Maybe you mean that one ?

2 hours ago, happy-kat said:

Nice... I don't know that (Android) app, using Stellarium to plan my framing, what is it ?

20 hours ago, Nigel G said:

Cassegrain Alt-AZ scopes seem to be more commonly used with guiding, for some reason.

As Ken reminded us, the longer focal length requires more precision in tracking. I may just as well retry my MAK on small objects, as mine is somewhat shorter than Ken's.

19 hours ago, happy-kat said:

Looks like my ST80 would capture pretty much the same as my 135mm if it's as above 80√ /400 but then you seem to include the aperture in the calculation?

Yes aperture and/or focal ratio (also based on aperture). 80/400 is f/5 so you would have to close your 135 past 5 (e.g. f/5.6 and smaller) for your scope to bring you more photons on the sensor than the lens. Which you probably won't do, as most 135 are good enough from f/3.5 or f/4 (I dream I could find a f/2 one :-P).

19 hours ago, Nigel G said:

I have already got my Orion planetary imaging/guide cam and ordered the Orion mini 50mm guider scope, so will be able to guide as well.

BTW, is there such a thing as Alt-Az guiding ? Field rotation put apart, it could be a way to achieve longer subs. Or said otherwise, achieve subs as long as field rotation allows, without being annoyed with balancing or tracking errors. Thinking about it, that would severely impact transportability and ease of setup though (add laptop, guide scope+cam, cables, etc), but I wonder whether anyone has done it.

14 hours ago, Nigel G said:

Here's what I mean, these images are cropped from 135mm lens image to compare against 150p images.

They are all similar exposure times, the 135mm lens appears to gather more.

Pretty obvious indeed. IMO this helps to bring an answer to a long asked question about the equivalence between various apertures.

In all cases it's agreed that the bigger the (true) aperture, the more photons you get and the less exposure time you need. And the longer the focal length, the more magnifying and less photons (*) you get and the more exposure time you need. True aperture is focal / focal ratio, so 150mm for your scope and 48mm (2.8) or 39mm (3.5) for your lens.

(*) actually it's less photons per pixel, since the same quantity of photons flowing through the same aperture are just spread over a wider surface.

Question is by how much those two factors combine and compensate each other, and which one wins over the other. I've read many articles about it. Some say / demonstrate the captured light level per pixel is proportional to D²/F (or F/R², R being the focal ratio F/D), others say / demonstrate it's proportional to only D/F (= 1/R).

If the former was right, your scope (150²/750 = 30) should capture more light at the same sub length than your lens (135/2.8² = 17.2). Your shots seem to prove the latter is right, i.e. 1/5 < 1/3.5 or 1/2.8.

Of course post-processing vary, you should really compare out-of-sensors RAWs, or JPEGs processed strictly equally. But it also correlates with my own experience as a wide-field imager.

18 hours ago, Nigel G said:

Now my wife has agreed to let me build a summerhouse (observatory) at the bottom of our garden 100ft from the bungalow, which gives me 360 degree imaging field and be able to leave the mount set up.

Project for the coming months.

Good luck. A fixed observatory certainly alleviates the burden of EQ setup, when it's mostly done once. Hoping your observatory will get enough clear skies to enjoy.

14 hours ago, Stub Mandrel said:

That gives me a focal length of 588mm (using my formula) which is so close to 585 millimetres, well it's basically 0.5%.

So yours is fully conformant. With the SWCC mine gives 565 or 590 depending on the spacing (shortest or longest, about 18mm between the two). Pretty cool actually, as it gives more FoV on demand to my smaller sensor

14 hours ago, Stub Mandrel said:

I wonder what's up with Irfan view?

Maybe it's actually loading the JPEG  "thumbnail" embedded in the RAW, rather than the RAW itself ? It's a feature I have in my "ORF" RAW format, so other RAW formats may have it too.

3 hours ago, Stub Mandrel said:

In one of my RAWs Tejat and Propus are 1971 pixels apart and my camera is 5.19um per pixel, so that's 10.229mm distance on the sensor

I'm afraid you must not have measured the right stars :-)

In my Stellarium Propus and Tejat are 1.865° apart, so the approximate corresponding sensor distance should be on the order of tan(1.865°)x650mm = 21.17mm.

That's way off your measure, so either you missed the right stars in your image or you are measuring a resized image (maybe onscreen display rather than true-image pixels ?).

Apart from that I agree on the formula. The /2 division on both sides isn't really necessary because of small angles.

Hi all, my first post in this long-time read thread. 1 year ago it finished convincing me and I bought a 130PDS, and don't regret it. These days I essentially use it for Alt-Az imaging though I also have an Eq (but don't use it often).

Incidentally I used some of my past months images to check and calibrate the pixel resolution with different combinations of filters, CC position, oculars (I also do projection imaging sometimes). It's pretty easy: knowing the sensors dimensions and number of pixels, and reference angular distances between stars (thanks to measure tool of Stellarium and catalogs), and just apply some basic trigonometry.

For reference I also calculated the resolution with no additional optics (the cam barely adapted to the OTA), and I found 632.5mm ± 1.3mm.  This is also confirmed indirectly by other optical combinations.

I know for normal that actual focal varies a bit from theorical / sold specification, mine is away by ~2.7%. Is it a normal value and what's your own values ?

You are on a good way, maybe not aggressive enough in your stretching.

Look at what I get with only an 8-bit stretch of your JPEG, this was with only levels black point and mid point. So you must be able to get far more from your 16-bit image.

The main idea is 1/ strech with levels (gamma / mid point), or brightness/contrast, or curves, then 2/ trim the backgroung with levels black point.

Ok, now unless I'm wrong you will see you some apparent (strong?) vignetting, so you will need flats. But keep that for after you are comfortable with processing

4 hours ago, The Admiral said:

The 'fullness' of the wells when the DAC saturates (which will also affect DR) will be governed by the gain between the sensor and the DAC, i.e. the ISO value.

Exactly, that was the "saturation-limit" part of my reasonning. Though a physicist will argue the sensor's pixels keep the same "full-well" capacity whatever the ISO, only the maximum representable electronic level drops when ISO raises.

4 hours ago, The Admiral said:

I presume that you've read Craig Stark's article on "The Effect of Stacking on Bit-Depth"?  (http://www.stark-labs.com/craig/resources/Articles-&-Reviews/BitDepthStacking.pdf)

I've read many similar, but maybe not this one. I learned many of those articles have to be taken with care as there are often gotchas in the reasonning, sometimes small sometimes big, so you need to keep awake. Always interesting, gonna read it, thanks Ian.

My go too

Not as stretched as others and I tried to calibrate the color and get back some core (failed at the latter). I saw core is not totally saturated but didn't find the right tool to get it back. Most stars are saturated too so had to resort to background stars for B-V calibration.

Was funny, thanks again for letting us play

And Nige... WOW you got the core back and nice nebula color changes.

6 hours ago, Filroden said:

Here's my latest rendition of the Rosette using Ha as Luminance and some very poor RGB data to add colour (there is now no Ha in the colour data).

Not so poor ! That RGB did much good to your image, with nice star colors (even if it lacks a bit of blue) and delightful and delicate color change between red, pink and orange in the nebula. Eventually it's not all red ;-)

4 hours ago, Herzy said:

I took this on Saturday with the clubs SE mount and an AT72ED from a dark site. I can't get the core to look right. The nebula was so bright even in 45s subs that it appeared all white. I layered some old pictures I had over it to make it more visible, but it still is blown out. I also don't like the blurred feeling to the image. Seems like the detail is all blurred.

Thanks for that superb image. I knew all this region is dust and feature rich, but seing it all revealed like this blows me off. Don't mind blurring too much, some images (including this one) are better appreciated as a whole rather going into the details.

1 hour ago, The Admiral said:

The other thing, as I understand it, is that it is not strictly necessary to have sufficient gain to get from 0 to 1 out of the DAC. First, sky background will take you above the 'floor', and averaging the signal from all the subs will, provided you are not doing integer arithmetic, allow interpolation between DAC units. The reason is that the photon and other statistics will sometimes give a zero, and other times 1, and perhaps occasionally 2, and the average will be somewhere between. So may be it might be possible to use a lower ISO to advantage.

Thanks for pointing that Ian. It was already discussed in the thread (which I have read once entirely) so I was aware of it, and is actually one of the triggers which made me come back at Alt-az imaging

But it's also a matter of precision: if you get one bit (literally) of signal out of each sub, it would take 256 subs to gain 8 bits of precision (not taking noise into account) and have something to stretch -- maybe less bits would do, I take 8 as a known reference. Also, depending on sub exposure, you could also get less than a full bit of signal (lol! I like quantum physics ), needing even more subs. There's also the matter of image depth through imaging software chain: if I get subs with 1-bit of signal each it means they are at position 12, so to get 8 bits of precision requires software to process at least up to 20 bits.

That's where I may be hitting Regim's 16-bit depth limit (of images). On M42 all was fine but on Rosette and Horsehead I feel I'm at the stretching limit. With the number of subs I take, the remaining noise should be near level 0 or 1. I stop stretching steps when noise becomes significant, but at the same time I understand there's no more signal to come because I've used all the bits within available depth. Feels like a dead end -- though I'm in the process of trying to "pre-scale" the images, which *will* saturate more stars but I hope allow stacking to expose more bits of signal.

3 hours ago, Filroden said:

UHC filters remove most of the green/yellow wavelengths (see the image below for an example). This probably affects calibration, particularly of the yellow stars. You may first have to manually rebalance the image and then run a more automated calibration.

You must be right, though I could not find the specific response curve for my filter. When I look through it during day, I see all seems red and blue, something like a mix between 2 eyes of a red-blue 3D glasses. At first I thought it would remove essentially yellow and a bit of red, but it's narrower actually and removes green as well (confirmed by looking at trees folliage and grass, which turn brown or dark red).

So Regim's calibration must be trying to compensate for the lack of green in stars, and the image ends with too much green. Sounds correct.

Now, what's the best and more practical way to get back a correct green channel ? If I play with balance to raise green level, I may end with the same green cast as auto-calibration since there's not enough green from the start (but will try). Capture green from another session, like you do with L+R+G+B(+Ha) ? Apart from the inconveniences of multi-session and number of subs, I would have to do it with a narrow green filter to be efficient -- if it exists   and that would somewhere negate the use of a UHC filter. Or synthetize a fictive green channel from the two others ? Would be wrong as the red+green/blue ratio is precisely what changes from star and star and enters the definition of the B-V index. Back to home to think and try :-P

On 27/1/2017 at 22:22, The Admiral said:

That's pretty reasonable for only 7 minutes exposure Fabien. Looking forward to your 120 x 20s set. The transition from red to blue is, I think too, unreal, and your B-V calibration is probably giving you nearer to the truth. A lot of images one sees with red/blue are of course NB images with colour mapping.

Thanks. I hoped some shots I've seen had real colors rather than mapped, but maybe not after all. As Ken reminds us, the Rosette is very bright in red. Just wonder then why the sensor captured 2 different colors.

On 27/1/2017 at 22:22, The Admiral said:

I wonder also whether you really need ISO2500, as you are reducing the DR significantly I feel. I used 1600 on my Rosette, and have used 400 on M42 and M45 where I wanted to retain as much information in the bright areas/stars.

That's a matter of camera. Actually I've tested 800, 1000, 1250, and 1600. With only 12-bits depth and 30% quantum efficiency (taken from sensorgen), most if not all deep sky (not clusters) subjects' data just happens to be below the minimum first value above zero with short exposures. Said otherwise, lower ISOs just don't give me enough data to stretch within my exposure limits. So I need to amplify more than all of you 14-bits sensors owners, and 2500 is my camera's maximum of analog amplification (3200 and above is digital). Would be different if my mount accepted to do 40-50s subs

Good news is my read noise is about the same at every iso, so I prefer to always use the same ISO level and vary the length and number of subs. Bad news is my noise is quite high so I need about 80-100 subs to tame it.

BTW don't be fooled by the DR such as indicated at sensorgen.info, it's for a single shot: If you consider stacking removes the noise, you get full DR for any analog-scaled ISO from a sensor-only point of view; What matters then is to avoid saturation of your subject wanted parts (you may saturate some subs for HDR to get faint parts), and hence match the subject brightness range to your camera by adapting the exposure. It's good if you have the choice to adapt the exposure, to lower ISO in addition to varying sub length, I don't think I have that choice.

Just to not let unfollowed my horrible Alnitak shot proviously shown, this is my 2nd try at Alnitak and its nebula friends. Much better I hope, though still not on the par.

Capture: 101 good of 123 lights x 25s x 2500iso, 30 NG darks, Olympus E-PM1 with Skywatcher 130PDS on Celestron Nexstar SLT, TS-UHC filter.

Thanks to much more subs, there's much more detail.

However I had difficulty with color calibration: I had to revert to the semi-manual version, the full automatic gave green stars rather than blue. Again I suspect the UHC filter to be the cause, just as if it was ripping out some information essential to the calibration But I've no idea what yet. On my calibrated displays this results in white/grey rather than blue stars, including Alnitak itself (and some nebula too), but I'm happy I eventually managed to avoid getting green stars.

Gives me a headache: On one hand the filter would allow me up to at least 30s subs (if not limited by mount accuracy) and catch many targets, while without it I am limited to 8-10s subs by light pollution and unable to catch Rosette or Horse Head. On the other hand the filter seriously reduces Regim's ability to auto-calibrate colors. And I love good (correctly) colored stars...

13 hours ago, mxgallagher said:

Not edited much, just had a quick play with the levels in gimp

I feel there's a lot more detail than what appears, so it's a good shot ! And the same time the core is quite saturated... Typical of my own historic stretches with Gimp levels You could try using Color->Curves rather than ->Levels so that you can avoid to stretch the brighest parts and only affect low brightness parts.

If you feel like it, you could also try 16+-bit aware imaging software, starting with Gimp 2.9 since you are a Gimp user -- though I have yet to try it myself. This will help recovering faint details.

A good surprise, I already got back my Internet at home, so here the image I talked about.

It's 7 good + 7 average lights x 30s x 2500iso and 30 NG darks (from another subject in the same session) usual Alt-Az train + my new TS UHC filter. Ignore the noise...

This is without any color calibration, direct colormap from the sensor. Usually I get really weird star colors and a B-V calibration is required. But with the new filter I think the colors are already very good. I had a doubt when seeing those very red stars but I checked their B-V index and the color is plain justified (there's one at 2.30 !).

Just the stars of the cluster are cyan when they should be pale blue, and the brighest of the cluster should be orange rather than white (or whatever). Means there's too much green ? I can't believe it, where would it come from and why only those stars.

I also perceive a noticeable color transition from red around to blue or grey in the center, which I think is nice, if not physically real. A thing I rarely notice in most other shots.

However I find the stars are too saturated at 30s x 2500iso, that's why I only did 20s for the real session. I just tried a quick BV but not good, most stars are good but the nebula turns all red, loosing the center blue/grey. I hope the big stack will turn better.

16 hours ago, Nigel G said:

I can't balance my scope and camera, its always top heavy, the dove tail is to short, to add counter weight to the bottom of the scope will add to much weight.

If you can't add weight at the back, you could try removing some at the top: attach a few (helium) balloons to the finder, for example.

Er, I admit it could look like a joke but it might work...