Bubble Nebula

[Rodd]

I need some advise.  My Bubble Nebula image was selected for a gallery display.  It was taken with the TOA 130 and ASI 1600.  Since my best version was a re-worked JPEG, a reprocess was necessary to ensure the highest quality print.  To my surprise the printing company (Bay Photo in CA) prefers high quality JPEGs.  So I reprocessed taking care to save to 300 dpi.  I worked with a master in XISF format (Pixinsight's version of FITs), and each change I saved to JPEG--always maintaining my XISF master-so there was no JPEG modification.   But I have serious concerns.  Fist of all, the 16mp image is quite small (the Bubble Nebula is small at 1,000 mm focal length, necessitating a crop).  PI is telling me that the 300dpi dimensions will be 9x8"--not a very impressive size for a professional gallery.  The following questions are at issue

1) If the viewing distance will be 5 feet--or even 10 feet, can I bump up the size to 20x16"--or even larger?  The display will not be meant for pixel peeping at 6 inches.  The image will be printed on metal, so the visible detail will pretty much be preserved.  And fine scale blemishes, such as noise, or sharpening artifacts (not many of those) will be hidden.

2)Now--CC on the image.  Is it too saturated?  Should I leave it a bit over saturated for printing?  On my good 4K monitor it looks quite nice.  On my iphone it looks like a cartoon.  In the morning I can't believe I created such drivel.  By the evening I am feeling proud.  This hobby will be the death of me.

3) I realize I might be nit picking, but I get one shot at this--there will be an element of permanence to it; no alterations possible after its printed.  Every time I look at the image, two issues predominate: saturation and brightness.  But that is for screen viewing--maybe I should make it twice as bright and twice as saturated for printing.

Thoughts?  I guess the question really is....should I leave it alone and go with it, or reprocess the image.

 

[MarkAR]

I would have thought that the print company would want the highest resolution possible. That would mean anything better than a jpeg, tiff at the least.

Can you save as higher resolution jpeg ?

On my screen that image looks superb, well done.

 

[Kinch]

Hi Rood....it looks fine on my monitor as it is.

I think you may have Photoshop....if so....you could try resizing it there.  Just a thought!

....and remember....if it is for public display....most won't have a clue about what we astro-photographers see when pixel peeping.....so I would not worry about that aspect of it.....just make it stand out 🙂

[Rodd]

3 minutes ago, MarkAR said:

I would have thought that the print company would want the highest resolution possible. That would mean anything better than a jpeg, tiff at the least.

Can you save as higher resolution jpeg ?

On my screen that image looks superb, well done.

 

Me too.  They say I can submit a TIFF or PNG--but a high quality JPEG is preferred.  I once sent an image in 3 formats and believe it or not, the JPEG one was better for printing.   Anyway--probably doesn't make much difference--thanks for you input.

[Rodd]

4 minutes ago, Kinch said:

Hi Rood....it looks fine on my monitor as it is.

I think you may have Photoshop....if so....you could try resizing it there.  Just a thought!

....and remember....if it is for public display....most won't have a clue about what we astro-photographers see when pixel peeping.....so I would not worry about that aspect of it.....just make it stand out 🙂

Thanks Brendan--I do have a copy of PS but I have never used it.  The resampling tool in PI allows for resizing, but the unknown is the difference between print and screen (and viewing distance).  

[Filroden]

11 minutes ago, Rodd said:

Thanks Brendan--I do have a copy of PS but I have never used it.  The resampling tool in PI allows for resizing, but the unknown is the difference between print and screen (and viewing distance).  

I've just thrown the image into Affinity Photo and used a simple bilinear resample to make the image 48" x 40" at 300 dpi and it still looks good (to my eye). The file size on a 95% JPG quality setting is ~55Mb, but there are a LOT of pixels in there  Photoshop allows similar simple resizing. I probably wouldn't use PI as its not really print production software.

As for preferring JPG, I'm not surprised. Even at 300dpi, the compression in JPG at very high quality settings should still be unnoticeable to the eye at a proper viewing distance. 

[Rodd]

1 minute ago, Filroden said:

I've just thrown the image into Affinity Photo and used a simple bilinear resample to make the image 48" x 40" at 300 dpi and it still looks good (to my eye). The file size on a 95% JPG quality setting is ~55Mb, but there are a LOT of pixels in there  Photoshop allows similar simple resizing. I probably wouldn't use PI as its not really print production software.

As for preferring JPG, I'm not surprised. Even at 300dpi, the compression in JPG at very high quality settings should still be unnoticeable to the eye at a proper viewing distance. 

Wow--thanks Ken.  Do you really think I could print at 48x40?  If I do--do I need to resize, or can I just tell them to print that large?

[Filroden]

1 hour ago, Rodd said:

Wow--thanks Ken.  Do you really think I could print at 48x40?  If I do--do I need to resize, or can I just tell them to print that large?

You'd need to ask the printers whether they are happy to do the resizing or whether you would need to do it.

I don't know if I'd push it as high as 48" by 40". Below is a slice from the image I created at that scale (300dpi). I think it holds up well. And don't forget the viewer will not be as close to the image as you are to the screen. Stand back 4 or 5 feet and tell me it's not an impressive image!

I've printed my own images (ZWO 1600ASI images) at close to a cropped 40"x40" and they look awesome. My printer will do fairly cheap "tester" prints where they print maybe 1 inch of the full width of the image to show you how it would look (I get to chose which slice they print). It's worth talking through the options with them.

Also, there are many algorithms to rescale the image. I only tried the default so better results might be found from the others.

 

[Rodd]

7 minutes ago, Filroden said:

Stand back 4 or 5 feet and tell me it's not an impressive image!

I think defects are starting to show--but that is huge.  I am thinking 16 x 20.  But all of this confuses me.  What is the difference between resizing and just zooming in to make the image bigger on the screen.  I understand the concept of the more pixels you have the larger you can print--beyond that....I have have no understanding of it

[Filroden]

5 minutes ago, Rodd said:

What is the difference between resizing and just zooming in to make the image bigger on the screen.  I understand the concept of the more pixels you have the larger you can print--beyond that....I have have no understanding of it

Zooming beyond the point where 1 pixel of image fills 1 pixel of screen will start to pixelate, i.e. go blocky. By rescaling the image you're effectively adding many many more smaller pixels by using fancy maths to try and estimate how they should be filled (you can't create more detail but you can fake it!). That allows you to zoom further. The better the maths, the more real it might appear. However, resizing does like images that are smoother in graduation (easier to calculate the missing bits of data). Astro images can be quite detailed and because its stretched, can lose that smooth graduated feel of a daylight image making it harder to rescale.

I've never found a good way to test what something would look like printed by viewing the screen. Screen pixels aren't the same size as printer pixels so viewing 100% on screen isn't what you'd see in print. I end up creating the right sized image using the DPI and desired print size then I zoom the resulting image until it broadly looks to be the right scale across, i.e. 20" of intended print area takes up roughly 20" of screen size (which might mean zooms lower than 100%).

[Rodd]

1 hour ago, Filroden said:

Zooming beyond the point where 1 pixel of image fills 1 pixel of screen will start to pixelate, i.e. go blocky. By rescaling the image you're effectively adding many many more smaller pixels by using fancy maths to try and estimate how they should be filled (you can't create more detail but you can fake it!). That allows you to zoom further. The better the maths, the more real it might appear. However, resizing does like images that are smoother in graduation (easier to calculate the missing bits of data). Astro images can be quite detailed and because its stretched, can lose that smooth graduated feel of a daylight image making it harder to rescale.

I've never found a good way to test what something would look like printed by viewing the screen. Screen pixels aren't the same size as printer pixels so viewing 100% on screen isn't what you'd see in print. I end up creating the right sized image using the DPI and desired print size then I zoom the resulting image until it broadly looks to be the right scale across, i.e. 20" of intended print area takes up roughly 20" of screen size (which might mean zooms lower than 100%).

thanks Ken.  I think I will go with 16x20--that's half of what you did.  My monitor is about 9x14, and at something less than fullscale it extends way beyond the screen boundaries.  That seems to me to be similar to what you say, but the way I see it I can either choose 8x10, which is too small, or something in between 8-10 and 16x20.  At a viewing distance of 6 feet--I think it will be nice.  i guess there is only one way to find out!

[Filroden]

Take a photo of it once it's displayed!

[Rodd]

7 minutes ago, Filroden said:

Take a photo of it once it's displayed!

OK--unless it looks totally horrible.  I want to get an ASI6200 camera.  It has 60 megapixels.  Making 40x30 images with that would be easy.  The trouble is it tales 2" filters--which would cost almost as much as the camera.  Oh well

[tooth_dr]

17 minutes ago, Rodd said:

OK--unless it looks totally horrible.  I want to get an ASI6200 camera.  It has 60 megapixels.  Making 40x30 images with that would be easy.  The trouble is it tales 2" filters--which would cost almost as much as the camera.  Oh well

That looks real nice Rodd, I'm sure it will print well.  The 6200 has almost twice the sensor size so stands to sense it would print out larger more easily.

[Rodd]

9 minutes ago, tooth_dr said:

That looks real nice Rodd, I'm sure it will print well.  The 6200 has almost twice the sensor size so stands to sense it would print out larger more easily.

Thanks Doc....yeah, I would love a full frame

[Rodd]

I set it off to be printed--3 ways (I only have 2 weeks so I don'

t have time to "see how it comes out" and try again.  The above image wax 1 version, and these next two images have been improved--IMO.  The top of the bubble looks more HSTish.  These two are very close, its just the first one was pushed a bit too hard--maybe, so I backed off on the second.  Like a fool I was not careful enough, and the core of the bright star to the right of the Bubble is wack...maybe no-one will notice?

[Rodd]

Yay!  I fixed the stars (big deal, I know....but little things like that drive me crazy).  This is the first image from above with fixed stars--the one that may be pushed to hard...not sure.  Do you think I should get a fourth one printed--or will the star not be noticed in the gallery?

 

[ramdom]

There are two things I can recommend, one I learnt early one and one recently. The first is about size. I prefer to size up a large image in PixInsight using drizzle if I am going to print it on a poster or blow it up somewhere.  Based on Ken's test, you may not need to do this but it is something to keep in mind for now and future. Drizzle in PI basically sizes up my image  to x times as I want it but I have to start early in your processing to get this going.  I believe with drizzle I can blow up a image to any size and not lose any resolution (and in fact, gain resolution if I'm undersampled). 

The second is about the stars. There is a way to minimise artifacts including the ringing and that's something I've just started to do and had some success with.  That is by doing most of your processing on a starless version of the image. You can create a starless version in PinInsight (there's a tutorial posted in the forum there) or via the starnet++ program...

Hope that helps... but TBH I think your last version looks pretty good as is. 

 

--Ram

 

 

 

[Rodd]

11 minutes ago, ramdom said:

Hope that helps... but TBH I think your last version looks pretty good as is

Thanks RAM.  I considered drizzle, but That would have required me to realign the subs into the masters and I do not have the subs any longer--only the masters.  Your right about thge stars--when undersampled--they turn round.  This data was shot quite oversampled though (.78 arcsec/pix).  So it probably would not have helped much.  4x the size would have been nice though!

[kirkster501]

I've always found this a very tough object to process in HOO.  I'm happy with the OIII.  Equally happy with the Ha.  But when I HOO them, it all goes pear-shaped.  I need to have another go, hopefully to get something as good as you

[Rodd]

57 minutes ago, kirkster501 said:

I've always found this a very tough object to process in HOO.  I'm happy with the OIII.  Equally happy with the Ha.  But when I HOO them, it all goes pear-shaped.  I need to have another go, hopefully to get something as good as you

Bicolored images always give me a lot of trouble too....always.  In fact except for the Veil, Thor's helmet and...well, that might be it, actually, I have never been able to pull one off effectively.