Live LRGB captures

[Martin Meredith]

I had a chance to try out live LRGB captures in Jocular last night for the first time. Its a fairly random collection of objects...

M92.

 

M5

 

Abell galaxy cluster 2151

 

NGC 6709, an OC in Aquila. This was quite low but there are very few OCs in the south until after midnight at present. 

 

M57

Definitely a blue colour cast on this last one that needs to be checked (there is automatic gradient removal in each filter channel but it doesn't look to be doing the right thing here).

There are still a few tweaks to make especially to stack combination, and there is no manual colour balancing yet.  I wasn't measuring total exposures but they are 15s subs except for Abell 2151 where I used 30s. The total sub count is at the top, but in most cases there are excluded subs so these are upper bounds.

Generally I collected 4 or 5 subs in each of RGB first then added the rest as L. I'm finding that relatively small amounts of colour info is needed along with LAB colour saturation to preserve colour even when a lot of L is included. 

Mostly,  I'm using LAB colour space (it is possible to switch to HSV, RGB or a fast hybrid LRGB combination, live, but LAB works best). The only colour controls are saturation, fractional colour binning, and colour stretch (CS in the screenshots). 

All of these used StarlightLive as the capture engine and for controlling the filter wheel. I still seem to have a fair amount of tilt or something in the optical train that needs to be sorted out too.

cheers

Martin

 

[RobertI]

Great results Martin, I particularly like the globs. I can’t judge colour accuracy (apart from M57!) especially on my small phone screen, but the colours have come through very well. It’s good news that you can use plenty of luminance without washing out the colour, definitely seems to be a problem with SLL. 

[RobertI]

Btw, I have failed to get Jocular working, it may just be my 8 year old laptop which has been loaded with various junk over the years, I may need to try with a clean new laptop! 

[Bill S]

This is looking very promising indeed, Martin. Not convinced I can shoehorn a powered filter wheel in between my Lodestar and focuser, at least with a focal reducer in the train. Might have to move the primary mirror up a bit...

You say you ran RGB then L. Does the order of  RGB make a difference to the colour cast? (Not sure if you can run in a different order or not with SLL.)

Rob - at what stage do things not work with Jocular? Have you got Python installed and apparently running OK? The 32 bit laptop I use for imaging must be a similar vintage as yours.

[Martin Meredith]

Thanks Rob. Yes, it was the difficulty of getting a good combination of L and RGB in SLL that set me on the path of using LAB space. To be fair, I don't think SLL was designed with LRGB in mind. It does require  specific processing to ensure that the L doesn't overwhelm colour information. These results were obtained with the saturation slider quite low, actually. It is possible to apply very large amounts of saturation too, but the results are not necessarily very pleasing! BTW I hope you can get Jocular going. I'm not much help as I don't use Windows these days.

Cheers Bill. When I started off last night I collected L first and had problems aligning subsequent RGB data. It might just have been issues with the mount -- I can't be sure -- but when I switched to collecting RGB first I had no alignment issues. I think my alignment algorithm may have a problem going from a large number of stars in one frame to a much smaller number in the next, but not the other way round. If so, it can be fixed and made more robust. It will still be a while before I release the colour version.

My filter wheel (from SX) doesn't add much to the optical path (15mm or so?) because the camera sensor can be push down to within a whisker of the filters within.

All this live LRGB will be much easier when it is scripted from Nebulosity or similar. At present it involves a lot of clicks in SLL and name changes (Jocular identifies the filter from the occurrence of red, green or blue in the FITs name for the moment -- a temporary solution). I guess there are maybe 50+ keystrokes for a complete LRGB capture, but this could in theory be reduced to one!

Martin

[Martin Meredith]

Actually Bill, I noticed I didn't answer your question re colour cast. I think the two things are not related, because I'm creating an independent stack for each filter and applying gradient removal to these stacks independently too, so it won't affect alignment (since stars are extracted prior to each sub joining its stack). What I'm not doing (and ought to be) is realigning each of the N stacks to each other. That is the next step. In general I want to be able to realign from an arbitrary sub (perhaps automatically) as soon as an alignment issue arises, since in my experience a great many failures to align can be fixed by choosing a different sub for the key stars.

Yes, one can run in any order, including interleaved, in SLL. Its just a case of getting it to move the filter wheel in between each one, then having to type red green or blue in the saved .FITs name (which is tedious if one interleaves).

Also, the number of subs of each type (RGB) shouldn't affect the colour cast, again because I compute the stacked RGBs independently. Having more Rs for instance will just lower the noise in that channel. 

Martin

[Bill S]

Martin - thanks for the explanation. It all sounds rather complicated so it's a good job you are figuring it out and producing the software.

A few measurements later and a look at the Starlight Xpress USB filter wheel details and I'm thinking things should fit and focus using the type of adapters you have described in other posts. (T thread on both ends of the filter wheel and T to 1.25 inch and/or 2 inch fittings for the Lodestar and the focuser.) I believe you have used Baader filters for LRGB. I see ZWO do some cheaper ones (probably not so good, but...). However, I think the ZWO ones are thicker and won't fit in the filter wheel. I think a phone call to FLO may be wise. I'm getting tempted by multi-spectral.

[umasscrew39]

On 01/06/2019 at 03:13, Martin Meredith said:

I had a chance to try out live LRGB captures in Jocular last night for the first time. Its a fairly random collection of objects...

M92.

 

M5

 

Abell galaxy cluster 2151

 

NGC 6709, an OC in Aquila. This was quite low but there are very few OCs in the south until after midnight at present. 

 

M57

Definitely a blue colour cast on this last one that needs to be checked (there is automatic gradient removal in each filter channel but it doesn't look to be doing the right thing here).

There are still a few tweaks to make especially to stack combination, and there is no manual colour balancing yet.  I wasn't measuring total exposures but they are 15s subs except for Abell 2151 where I used 30s. The total sub count is at the top, but in most cases there are excluded subs so these are upper bounds.

Generally I collected 4 or 5 subs in each of RGB first then added the rest as L. I'm finding that relatively small amounts of colour info is needed along with LAB colour saturation to preserve colour even when a lot of L is included. 

Mostly,  I'm using LAB colour space (it is possible to switch to HSV, RGB or a fast hybrid LRGB combination, live, but LAB works best). The only colour controls are saturation, fractional colour binning, and colour stretch (CS in the screenshots). 

All of these used StarlightLive as the capture engine and for controlling the filter wheel. I still seem to have a fair amount of tilt or something in the optical train that needs to be sorted out too.

cheers

Martin

 

Excellent views and great progress, Martin.  Congrats to you for sticking with this effort.  It is a shame that SLL is non longer supported as I am sure Paul could have done amazing upgrades to what many of us think was the best EAA software ever made available.  Unfortunately,  because of the lack of support and my night skies seemingly worse when I moved from California to Florida, I've converted to CMOS cams and SharpCap.  The CMOS design makes a dramatic difference in my views with  much less noise but SharpCap is a chore to learn.   Best of luck as you continue to develop this model.

Bruce

[JamesF]

I'd not heard that SLL was no longer supported.  That's a shame.  I know from working on oacapture that it can be a huge commitment of time though, and sometimes that time just isn't available.

James

[Martin Meredith]

Thanks Bruce. I hope you get some decent skies soon and post some CMOS images. At some point I'm sure I'll acquire a CMOS camera myself but I'm still enjoying myself too much with the Lodestar!

Martin

[roelb]

Martin, is a Windows executable of Jocular possible?

The Jocular installation seems to involve a lot of specialized steps.

Thanks in advance.

Roel

[Martin Meredith]

Hi Roel

Yes, it is possible, but I don't have/use Windows so it would be a case of someone giving it a go. There are recipes to do it although it isn't straightforward I suspect (I tried a build for the Mac and spent some hours on it without resolving all the issues, so went for a code distribution instead). Jocular installation isn't necessarily difficult once Python and the required libraries are installed. And then the benefit is that future releases don't need these steps and it is just a case of dropping the new release in the right place.

cheers

Martin

[roelb]

Thanks Martin

On 06/06/2019 at 09:12, Martin Meredith said:

Hi Roel

Yes, it is possible, but I don't have/use Windows so it would be a case of someone giving it a go. There are recipes to do it although it isn't straightforward I suspect (I tried a build for the Mac and spent some hours on it without resolving all the issues, so went for a code distribution instead). Jocular installation isn't necessarily difficult once Python and the required libraries are installed. And then the benefit is that future releases don't need these steps and it is just a case of dropping the new release in the right place.

cheers

Martin

Thanks Martin. I'll try the Python installation.