A7R3 pixel shift nightscape

[Astrophotographer 10]

I took a series of nightscape images at my dark site a few weeks ago under very good conditions.

I used Pixel Shift to emulate what I do with mono CCD colour images which combine 4 images taken with clear,red, green and blue filters. Pixel shift does much the same thing.

The camera was tracked.

30 seconds ISO6400 Sigma Art 14mm F1,8.




The Orion Arm of the Milky Way and the Magellanic Clouds A7r3 pixel shift 3 image mosaic

Greg.

 

[sharkmelley]

Beautiful image! Pixel shift is an interesting and hopefully useful technique to use for astroimaging.

The green sky - is that caused by airglow?

Mark

 

[Astrophotographer 10]

Beautiful image! Pixel shift is an interesting and hopefully useful technique to use for astroimaging.

The green sky - is that caused by airglow?

Mark

--
Takahashi Epsilon 180ED
H-alpha modified Sony A7S
http://www.markshelley.co.uk/Astronomy/

Thanks Mark. Yes that is airglow and so are the magenta streaks in the sky. I often get a lot of airglow there. Perhaps because its at 770 metres altitude.

I also did a series with an A7R2 with the same lens, same scene but with star eater turned off using the Jim Kasson trick of setting it to Continuous Lo drive mode. I'll post that soon as a comparison.

Greg.

 

[Nikolai Vassiliev]

Oh, great.

How do you combine 4 ARW?

With Imaging edge or manually ps/some-astro-soft?

P.S.

And what is a jim kasson's trick?

 

[Tristimulus]

Very interesting image - on its own (terrific skyglow) - and technically (using the pixel shift feature and tracking).

Pixel shift is exposing 1/4 red pixels, 2/4 green pixels and 1/4 blue pixels per run, so the filtered monochromatic camera should still be more effective? Following your work and very curious about upcoming results and comments. Thumbs up!

 

[Astrophotographer 10]

Very interesting image - on its own (terrific skyglow) - and technically (using the pixel shift feature and tracking).

Pixel shift is exposing 1/4 red pixels, 2/4 green pixels and 1/4 blue pixels per run, so the filtered monochromatic camera should still be more effective? Following your work and very curious about upcoming results and comments. Thumbs up!

Mono CCDs also use a clear filter for the detail and that makes a strong image. Pixel shift does not have a clear filter. There was talk of a clear RGB colour filter matrix at some point. Perhaps its hard to implement as it would throw off their normal colour adjustments.

Kodak made one about 12 years ago no. They call it Truesense I think. It increased sensitivity of a colour sensor by a large amount. From memory it doubles the sensitivity.

I wonder if Sony will use that technology in say an A7S model at some point? Its probably one of their last remaining options to increase sensitivity in sensors.

In mono CCD world a BSI sensor is way more sensitive than one that is front side illuminated. By about a 1/3rd. QE of an A7r3 is right up there with a mono CCD sensor which are usually around 59% which is what I believe the A7r2/3 is. It seems about right as I use both types of cameras.

Greg.

 

[Astrophotographer 10]

Oh, great.

How do you combine 4 ARW?

With Imaging edge or manually ps/some-astro-soft?

P.S.

And what is a jim kasson's trick?

--
http://flickr.com/photos/qwz

Yes with Imaging Edge to process the ARQ files and then export them out of there as a TIFF file into PT Gui for stitching.

Greg.

 

[Pogo65]

What I find difficult to gras- is that to take full advantage of this there cannot be any shift of the pixels during tracking in order for this to work correctly. Certainly if you expose for 30 seconds per color. You would need sub arcsecond precision to do that, which in my opinion , even with a perfectly aligned astrophysics mount is still hard to do. Of course it is a wide angle shot, and I would have to do the math to see how many arcseconds a single pixel would cover.

Besides all that, excellent shot, but I wonder how much better this is compared to just 4 30 second shots averaged. Did you try?

regards,

Peter

 

[Tristimulus]

What I find difficult to gras- is that to take full advantage of this there cannot be any shift of the pixels during tracking in order for this to work correctly. Certainly if you expose for 30 seconds per color. You would need sub arcsecond precision to do that, which in my opinion , even with a perfectly aligned astrophysics mount is still hard to do. Of course it is a wide angle shot, and I would have to do the math to see how many arcseconds a single pixel would cover.

Besides all that, excellent shot, but I wonder how much better this is compared to just 4 30 second shots averaged. Did you try?

regards,

Peter

The pixel size of the A7RIII is close to 4.5 micron.

The plate scale of the 14 mm lens is 66 arc seconds per pixel, so plenty of leeway here.

Use a 135 mm lens and the plate scale is 6.9 arc seconds per pixel.

So pixel shifted images should be possible with a good tracker and a 135 mm if the sub exposures are not too long (something like 15 to 30 sec tracked pixel shifted images should be possible with this focal lenght - even more leeway if using shorter lenses).

To get a plate scale corresponding to 1 arc sec per pixel we need 930 mm focal lenght (and if wanting this resolution along the diagonal 1300 mm focal lenght is needed).

So sub arc seconds resolution is a bit over the top if using ordinary camera lenses...

 

[blue_skies]

Impressive capture, Greg

 

[The Davinator]

Amazing how deep these shots can go. I remember hypering Kodak PPF400 and Tech Pan 2415 and running for 45 to 90 minutes.

Nice shot.

 

[aSevenArr]

I took a series of nightscape images at my dark site a few weeks ago under very good conditions.

I used Pixel Shift to emulate what I do with mono CCD colour images which combine 4 images taken with clear,red, green and blue filters. Pixel shift does much the same thing.

The camera was tracked.

30 seconds ISO6400 Sigma Art 14mm F1,8.


The Orion Arm of the Milky Way and the Magellanic Clouds A7r3 pixel shift 3 image mosaic

Greg.

It's a beautiful image, well done.

However as far as pixel shifting is concerned we can't even get close to native camera resolutions on our computer screens. All your fine details are simply being thrown away by compression algorithms whenever you share an image (which inevitably means as a JPEG).

 

[Astrophotographer 10]

What I find difficult to gras- is that to take full advantage of this there cannot be any shift of the pixels during tracking in order for this to work correctly. Certainly if you expose for 30 seconds per color. You would need sub arcsecond precision to do that, which in my opinion , even with a perfectly aligned astrophysics mount is still hard to do. Of course it is a wide angle shot, and I would have to do the math to see how many arcseconds a single pixel would cover.

Besides all that, excellent shot, but I wonder how much better this is compared to just 4 30 second shots averaged. Did you try?

regards,

Peter

I have got 4 30 second stacked data. Just have to finish processing it.

The Imaging Edge software seems to take some of the need for exact accuracy out as you can see it remove the coloured edges of the slightly blurred landscape out of the image.

Also this is not really more demanding than mono filtered CCD imaging with a telescope.

I am more interested in the deeper colour rather than a sharpening increase. I had sharpening set to zero in Imaging Edge.

Greg.

 

[Astrophotographer 10]

I took a series of nightscape images at my dark site a few weeks ago under very good conditions.

I used Pixel Shift to emulate what I do with mono CCD colour images which combine 4 images taken with clear,red, green and blue filters. Pixel shift does much the same thing.

The camera was tracked.

30 seconds ISO6400 Sigma Art 14mm F1,8.


The Orion Arm of the Milky Way and the Magellanic Clouds A7r3 pixel shift 3 image mosaic

Greg.

It's a beautiful image, well done.

However as far as pixel shifting is concerned we can't even get close to native camera resolutions on our computer screens. All your fine details are simply being thrown away by compression algorithms whenever you share an image (which inevitably means as a JPEG).

Yes I agree. I am more interested in the improved colour depth.

I have regular images ready to stack for a comparison. I'll process 3 types. Pixel shift, 4 x 30 second stack A7r3 and 4 x 30 second stack A7R2 with star eater turned off shortly.

Greg.

 

[Astrophotographer 10]

Impressive capture, Greg

 

[Astrophotographer 10]

Amazing how deep these shots can go. I remember hypering Kodak PPF400 and Tech Pan 2415 and running for 45 to 90 minutes.

Nice shot.

Thanks Davinator. I agree, the depth modern cameras can capture is a reminder of just how sophisticated these cameras are. For example backside illuminated CCD sensors typically sell for more like US$30K. That is a for a 4mp sensor too!

Greg.

 

[Pogo65]

What I find difficult to gras- is that to take full advantage of this there cannot be any shift of the pixels during tracking in order for this to work correctly. Certainly if you expose for 30 seconds per color. You would need sub arcsecond precision to do that, which in my opinion , even with a perfectly aligned astrophysics mount is still hard to do. Of course it is a wide angle shot, and I would have to do the math to see how many arcseconds a single pixel would cover.

Besides all that, excellent shot, but I wonder how much better this is compared to just 4 30 second shots averaged. Did you try?

regards,

Peter

The pixel size of the A7RIII is close to 4.5 micron.

The plate scale of the 14 mm lens is 66 arc seconds per pixel, so plenty of leeway here.

Use a 135 mm lens and the plate scale is 6.9 arc seconds per pixel.

So pixel shifted images should be possible with a good tracker and a 135 mm if the sub exposures are not too long (something like 15 to 30 sec tracked pixel shifted images should be possible with this focal lenght - even more leeway if using shorter lenses).

To get a plate scale corresponding to 1 arc sec per pixel we need 930 mm focal lenght (and if wanting this resolution along the diagonal 1300 mm focal lenght is needed).

So sub arc seconds resolution is a bit over the top if using ordinary camera lenses...

Got you, that makes absolute sense! I will give this a shot myself on my trip to Tenerife next month.

 

[aSevenArr]

I took a series of nightscape images at my dark site a few weeks ago under very good conditions.

I used Pixel Shift to emulate what I do with mono CCD colour images which combine 4 images taken with clear,red, green and blue filters. Pixel shift does much the same thing.

The camera was tracked.

30 seconds ISO6400 Sigma Art 14mm F1,8.


The Orion Arm of the Milky Way and the Magellanic Clouds A7r3 pixel shift 3 image mosaic

Greg.

It's a beautiful image, well done.

However as far as pixel shifting is concerned we can't even get close to native camera resolutions on our computer screens. All your fine details are simply being thrown away by compression algorithms whenever you share an image (which inevitably means as a JPEG).

Yes I agree. I am more interested in the improved colour depth.

I have regular images ready to stack for a comparison. I'll process 3 types. Pixel shift, 4 x 30 second stack A7r3 and 4 x 30 second stack A7R2 with star eater turned off shortly.

Greg.

Indeed but sadly I don’t think compression respects colors any more than it respects detail. And worse than that every single screen out there is showing your image with different colors.