Tulip widefield from RASA 8.

[ollypenrice]

Our first 'proper' image from the RASA/ASI2600/Avalon. Two panel mosaic, 36x3 minutes per panel, so just 3.6 hours for the lot!

Notes:

This is an Ha-dominated target so a broadband capture can only take you so far. If I lean on the Ha signal in the data any harder it does give more but at a cost of noise in the red channel. We aim to try a dual or tri-and filter later.

No darks or flats, yet colour gradients and vignetting were minimal, so no DBE or other gradient removal. The two panels were stretched, eyeballed for colour balance and blackpoint, and Registar just joined them seamlessly. Remarkable.

Paul manned the plethora of software packages from the UK while I bolted the bits together (which is the easy bit...)

Olly

Edit: scroll down for a version made from data software binned 2x2 in 'superpixel' mode.

 

Edited September 6, 2021 by ollypenrice

[vlaiv]

Just a quick question if you don't mind.

ASI2600 is 6284x4176 pixel camera, and yet this image is two panel composite that is only ~1300x1000 px in size.

Any particular reason you've chosen to present it reduced in size, and how did you and at what stage perform reduction?

[gorann]

Nice one Olly and you seem pleased with what the rig can accomplish, which is reassuring. However, I agree with Vlaiv that it would be interesting to see it in higher resolution to peep at your corner stars. If they are perfect or nearly so it may convince me to make that sensor-tilt-adjustemnt set-up you just described in another thread.

It is great that we both went for the ASI2600MC that has a built-in tilt plate. I have been following a thread about the cheaper competitor (Touptek sold under various names) with the same sensor but without tilt plate, and with that camera some have experienced tilt and can only adjust it by adding a tilt plate and there would be no space for that on a RASA8.

Cheers

Göran

Edited September 6, 2021 by gorann

[ollypenrice]

3 hours ago, vlaiv said:

Just a quick question if you don't mind.

ASI2600 is 6284x4176 pixel camera, and yet this image is two panel composite that is only ~1300x1000 px in size.

Any particular reason you've chosen to present it reduced in size, and how did you and at what stage perform reduction?

The full sized image is far too big to upload easily with my internet connection so this is just resized in Photoshop. I also pushed the Ha signal pretty hard with a view to a final presentation below full size. Since this is a widefield image without anything crying out for a high resolution view I would normally have stacked it in a different mode giving a binned output, but I was curious to see how the data would perform at full size. I'll try processing a binned version to see if it will let me push the Ha signal harder. I suppose it will.

I haven't yet read up on what AstroArt is offering to do with the raw OSC data. This has all changed since I last stacked any OSC several years ago. In RGB debayer it offers me

- binning 2x2

- bilinear 1x1

- 2 gradients 1x1

- 4 gradients 1x1

If I knew what the last three meant I'd be a wiser man than I am!   Homework time...

I'm not a pixel peeper and post plenty of images at full size when appropriate, mostly galaxy crops from the high res rig, but I'll put in some crops from this at full size to give folks an idea of what to expect. I'll also re-do the image from a binned stacking and post that later.

Noise floor in Ha regions 

Corner crop, top left.

Corner crop, top right...

 

If the idea were to present at full size I would want more than an hour and a half per panel and, while the corners aren't perfect, I'm happy with them. Getting this far at F2 wasn't easy and we are outside the official limits of the image circle according to Celestron. Anyone wanting perfect stars should, I think, choose a refractor and take six times longer than the RASA. 

I'll chuck in one more: both the mosaic overlaps lie in the crop below, and this was on unflattened panels:

Olly

 

 

 

 

[ollypenrice]

1 hour ago, gorann said:

 

It is great that we both went for the ASI2600MC that has a built-in tilt plate. I have been following a thread about the cheaper competitor (Touptek sold under various names) with the same sensor but without tilt plate, and with that camera some have experienced tilt and can only adjust it by adding a tilt plate and there would be no space for that on a RASA8.

Cheers

Göran

I just followed your choices because they worked, Goran!  I would not consider this cheaper camera for the RASA. I think tilt adjustment is essential for F2. I'm not at all sure that my tilt tuning was correct. All I know is that, in one orientation, it gave acceptable corners (acceptable to me) for the first time. The fact that it didn't do so well at 90 degrees suggests something is still out. Possibly one tilt is correcting another as things stand. For now we just want to get on with it. 

Olly

[gorann]

Just now, ollypenrice said:

I just followed your choices because they worked, Goran!  I would not consider this cheaper camera for the RASA. I think tilt adjustment is essential for F2. I'm not at all sure that my tilt tuning was correct. All I know is that, in one orientation, it gave acceptable corners (acceptable to me) for the first time. The fact that it didn't do so well at 90 degrees suggests something is still out. Possibly one tilt is correcting another as things stand. For now we just want to get on with it. 

Olly

Olly, I think your stars look as good as could be expected from such a fast instrument with a wider than recommended sensor so I would be happy and do like you say: just get on with it. I aslo usually downsample the images I put out, both because it makes the files smaller and more managible (still usually around 8 Mb as jpg). Most often I found 4000 x 3000 pixels or even a bit less to be sufficient to present all the details that really matters in the image.

[vlaiv]

22 minutes ago, ollypenrice said:

- binning 2x2

- bilinear 1x1

- 2 gradients 1x1

- 4 gradients 1x1

These are just debayering modes I presume.

Binning 2x2 is actually Super pixel mode - it creates one RGB pixel out of RGGB quad by using R and B as is and averaging two G pixels. This reduces resolution (pixel count) of the image by x2 in each direction.

Bilinear is just simple average of pixels of corresponding color to produce missing values. It takes R from one RGGB quad and R from adjacent RGGB quad and creates average of those two to be placed between two R values. Other components are handled similarly.

Gradients approach (whether 2, 4 or VNG - variable number of gradients), analyze image for gradients obviously and based on that interpolate missing values. It is a bit higher order method of interpolation than simple bilinear.

In any case - options 2, 3 and 4 give "full" resolution in term of pixels by interpolating (fancy word for educated guess ) missing colors while first one is closer to actual sampling rate of color sensor.

By looking at images - I'd say that 3.76µm pixel size is too small for RASA8 and that you'd be better by binning your image x2 in the end if you use any of 2, 3 or 4 debayering technique. I think that spot diagram for RASA8 goes in line with this as it is quoted as being 4.6µm RMS or less over imaging circle (which is equivalent of 10.8µm FWHM with sampling being best with 6.77µm pixel size).

[ollypenrice]

21 minutes ago, vlaiv said:

These are just debayering modes I presume.

Binning 2x2 is actually Super pixel mode - it creates one RGB pixel out of RGGB quad by using R and B as is and averaging two G pixels. This reduces resolution (pixel count) of the image by x2 in each direction.

Bilinear is just simple average of pixels of corresponding color to produce missing values. It takes R from one RGGB quad and R from adjacent RGGB quad and creates average of those two to be placed between two R values. Other components are handled similarly.

Gradients approach (whether 2, 4 or VNG - variable number of gradients), analyze image for gradients obviously and based on that interpolate missing values. It is a bit higher order method of interpolation than simple bilinear.

In any case - options 2, 3 and 4 give "full" resolution in term of pixels by interpolating (fancy word for educated guess ) missing colors while first one is closer to actual sampling rate of color sensor.

By looking at images - I'd say that 3.76µm pixel size is too small for RASA8 and that you'd be better by binning your image x2 in the end if you use any of 2, 3 or 4 debayering technique. I think that spot diagram for RASA8 goes in line with this as it is quoted as being 4.6µm RMS or less over imaging circle (which is equivalent of 10.8µm FWHM with sampling being best with 6.77µm pixel size).

Thanks. I just processed a super pixel stack and think it's considerably better, certainly on this target and with just 1.5 hours per panel. I'll need to do it again because I made a bad decision early on which has created colour artifacts in stars but the reduced noise lets me get a lot more out of it. Lots to learn with a new rig.

Olly

[ollypenrice]

OK here's a version done from the 2x2 'superpixel' stacking algorithm. Obviously you've more depth to play with. Stars weren't so good though, tending to have a blue edge and a yellowish edge on the other side. It would be a lot easier if I weren't trying to extract NB signal from a broadband image. Using the dual or triband filter on targets like this would remove this pressure. This is still reduced to 5mb but will take an age to upload!

Olly

 

 

[gorann]

Interesting Olly, I thought of it but never tried superpixel binning my RASA images. Wonder why your stars suddenly got a blue rim from that. Should give one of my data sets a try. But tonight I will be busy collecting data.

[Magnum]

Nice first light Olly, You might be interested in my latest 2 dual narrowband images with my 533 the  little brother to the 2600, now ive developed a decent work flow im very happy with the results, ive changed a couple of steps since the data I sent you on Orion.

Using the either the  Altair Triband or Optolong L Enhance In comparison to real narrowband imaging with mono, I find OSC with dual band filters give a totally different look, its more like broadband image with some ha data added, as the stars don't get suppressed hardly at all, I did try a L Extreme, but I didnt like that as I found I then lost the range of colours in the targets ( ie the reds were all 1 shade ) as well as loosing the yellow stars. I actually really like the effect of having a fairy dust multitude of stars at same time as the Ha, and for me its not going to replace real narrowband, I have a triple rig now so im producing full colour with the 533 triband setup, then if I want to isolate the ha and suppress stars I shoot pure HA mono images with either of the other 2 scopes as I still love mono images.

Only thing id watch with the dual narrowband filters is halos, you need to get the filter as close to the sensor as possible when using fast optics, especially with my samyang f2 lens but only on some red stars.

Lee

 

 

Edited September 7, 2021 by Magnum

[Magnum]

Hi Olly, I presume you have already Seen this on the ZWO site on what they recommend. https://astronomy-imaging-camera.com/tutorials/which-debayer-algorithm-is-the-best-for-asi-cameras.html

I download the trial of APP to try it, but found I could already achieve the same in MaximDL5 just by playing with the combination of high quality & Under sampled check boxes in the debayer window, Maxim still impresses me even though its ancient now.

Lee

Edited September 9, 2021 by Magnum

[ollypenrice]

6 hours ago, Magnum said:

Hi Olly, I presume you have already Sean this on the ZWO sight on what they recomend. https://astronomy-imaging-camera.com/tutorials/which-debayer-algorithm-is-the-best-for-asi-cameras.html

I download the trial of APP to try it, but found I could already achieve the same in MaximDL5 just by playing with the combination of high quality & Under sampled check boxes in the debayer window, Maxim still impresses me even though its ancient now.

Lee

No, I hadn't seen that, so thanks.

Olly

[peter shah]

That's superb Olly, star colour to die for, stupendous FOV filled with colour.....just wonderful!.....I have an 11in RASA I'm setting up....really looking forward to feeling some of that pain 😉