How to get more detail/sharpness is images?

[Z3roCool]

 

Hi - I have just started taking photos and pretty happy with my first attempts.

I have been wondering how to get more detail/sharpness in the nebula detail?

I guess it is a mix of a few things:

• Getting more data - This photo used a number of 5 sec, 20 sec and 120sec subs (20 of each) Also added some darks (Not for the 120sec though). Would I benefit from shooting longer subs (5mins+) over more shorter subs?
• Guiding - Currently not guiding other than NEQ6. I have a 50mm guide scope and Altair camera. Will look at guiding next shoot. With a decent PA (Use SharpCap and get 'Good' PA) what kind of subs should I expect before images start to trail? Do subs over say 5 min+ give a lot better data over more shorter subs?
• Getting a telescope - Currently, have the Canon 400 5.6L (Canon 7D) lens which is very sharp with normal photography. Would a telescope improve (APO Triplet) over my Canon lens?
• Focus - I focussed this using Backyard EOS and pointing at a star and then moving the focus until the lowest number appeared (Cannot remember the name of the method). Is this pretty accurate, over say a Bahtinov Mask?
• Stacking - I am using DSS to stack. I am pretty much using the defaults at present. IS it worth looking into this more and trying different ways of stacking? Any other methods other than DSS?
• Processing - I am testing out processing methods at the moment. Using PS and LR. Learning how to stretch an image with curves and levels. Did not try on this photo but learning how to mask and get the core back from being blown out. Do I just need to get better? Are there any good plug-ins that help with nebula detail that anyone uses?

Really just looking at if I need to get a telescope or just carry on with what I have and get more data and see where that takes me...

Thanks for looking,

Tom.

 

[Louis D]

You can increase perceived sharpness by using unsharp masking, just don't overdo it or you'll get glaring artifacts.

[vlaiv]

1. Yes, more data is good thing - provided that you don't mix in bad data which can "spoil" the good data. Shooting longer subs is always better than using short subs, but depending on equipment and target there is a point of diminishing returns. So with one setup you can get from 75% to 98% by switching from 30s to 5 minute subs (those percentages should be interpreted as quality compared to single total exposure), while other setup can give you increase in quality from 96% to 97% by using same change in sub duration. Note that by using longer subs and having less than perfect gear and conditions (and no one has perfect either of those, but people do come close), you increase your chances of having bad data. For example if there is sudden strong gust of wind that throws off your guiding - with 30s subs you will be wasting only 30s of data, while with 5 minute sub you end up wasting 5 minute of data. So it is really a balance - selecting sub duration that will provide the most usable data - but that is, like I've said, different to each set of gear and shooting conditions. One way to go about it is to experiment, but I do encourage people to understand the whole picture (different types of noise, shooting conditions, guiding precision and its impact, ....) when choosing what to experiment with, because there are so many variables and so little time to test out each combination.

2. Trailing in subs happens for two reasons. One is imperfect PA - and you can work on that, develop routine that will get you better PA (like drift align). Other reason is not something that you can easily fix - it is PE (periodic error) of the mount. You can fiddle with your mount in mechanic way (like adjust tensions, change bearings, properly apply grease, mod stuff), and you can do PEC - periodic error correction, and both of these will lessen PE of your mount, but it can't be completely eliminated. So you can always expect some sort of elongation if you use long subs and don't guide. This of course depends on resolution that you are working on, and some other factors like seeing blur, general mount precision and stiffness, etc. This is why some people have round (ish) stars even if not guiding - but if you are after detail/sharpness in your data - then you don't want large round stars - you want tiny round stars

3. Yes, in general switching from lens to scope will have positive effect on detail/sharpness provided that scope is optically good. Two things have say in this - first, scope (in general) has larger aperture than lens, and second, scope tends to be better corrected (optically at infinity, with fewer elements) than (most) lenses. There is however drawback. Scopes have longer focal lengths, and that implies that you will be working on higher resolution (in terms of arc seconds per pixel). Higher resolution means that all of this will have tighter tolerances (you will need better guiding, better PA, PE will be more pronounced, etc ...). You also need to select telescope that has good optical characteristics across the whole field - APO + field flattener is going to be good choice.

4. I have bahtinov mask, but I also use same approach as you - I inspect star sizes in short "focusing" sub - I also look at change in star sizes across the whole field - there is some field curvature in my setups so I aim to minimize that. One can learn to hit good focus using this technique, and it minimizes chance of forgetting bahtinov mask on scope and wasting your subs . Also if using scope, you might want to check your focus (I do it by just looking at the stars in subs) if there is temperature change - there might be a need to refocus if scope cools or gets a bit warmer.

5. There are couple of different algorithms for stacking and it would be a good thing to get to know in principle how they work. They can help eliminate some sort of artifacts and noise. You may also find that different software solutions for stacking give you slightly different levels of sharpness of final image, but that has to do with other things and not stacking algorithms themselves. It has to do with how well software can determine centers of stars, and find optimum mathematical transform to align different subs (rotate, shift, scale, what ever is needed). Differences tend to be small, but some software does produce tighter stars in final stack (by small amount).

6. Processing does indeed have an impact on final result, but I see it as separate thing. It is skill that can be improved upon in time, so you can get back to your old data and process it differently. This is why I mostly concentrate on improving acquisition of data - since you can do that "only once" (not really true - heavens will be there for a long time, so you can always get more data on particular target, but I find that imaging time is rather precious - you need good conditions and clear night and those don't really come by that often).

As for detail/sharpness I'll throw in some more tips:

- learn about seeing, how to judge it and where to get forecast for particular night

- learn about transparency, and same as above, how to judge it and get forecast when planning a session.

Two above have significant impact on your results.

HTH

[Z3roCool]

4 hours ago, vlaiv said:

1. Yes, more data is good thing - provided that you don't mix in bad data which can "spoil" the good data. Shooting longer subs is always better than using short subs, but depending on equipment and target there is a point of diminishing returns. So with one setup you can get from 75% to 98% by switching from 30s to 5 minute subs (those percentages should be interpreted as quality compared to single total exposure), while other setup can give you increase in quality from 96% to 97% by using same change in sub duration. Note that by using longer subs and having less than perfect gear and conditions (and no one has perfect either of those, but people do come close), you increase your chances of having bad data. For example if there is sudden strong gust of wind that throws off your guiding - with 30s subs you will be wasting only 30s of data, while with 5 minute sub you end up wasting 5 minute of data. So it is really a balance - selecting sub duration that will provide the most usable data - but that is, like I've said, different to each set of gear and shooting conditions. One way to go about it is to experiment, but I do encourage people to understand the whole picture (different types of noise, shooting conditions, guiding precision and its impact, ....) when choosing what to experiment with, because there are so many variables and so little time to test out each combination.

2. Trailing in subs happens for two reasons. One is imperfect PA - and you can work on that, develop routine that will get you better PA (like drift align). Other reason is not something that you can easily fix - it is PE (periodic error) of the mount. You can fiddle with your mount in mechanic way (like adjust tensions, change bearings, properly apply grease, mod stuff), and you can do PEC - periodic error correction, and both of these will lessen PE of your mount, but it can't be completely eliminated. So you can always expect some sort of elongation if you use long subs and don't guide. This of course depends on resolution that you are working on, and some other factors like seeing blur, general mount precision and stiffness, etc. This is why some people have round (ish) stars even if not guiding - but if you are after detail/sharpness in your data - then you don't want large round stars - you want tiny round stars

3. Yes, in general switching from lens to scope will have positive effect on detail/sharpness provided that scope is optically good. Two things have say in this - first, scope (in general) has larger aperture than lens, and second, scope tends to be better corrected (optically at infinity, with fewer elements) than (most) lenses. There is however drawback. Scopes have longer focal lengths, and that implies that you will be working on higher resolution (in terms of arc seconds per pixel). Higher resolution means that all of this will have tighter tolerances (you will need better guiding, better PA, PE will be more pronounced, etc ...). You also need to select telescope that has good optical characteristics across the whole field - APO + field flattener is going to be good choice.

4. I have bahtinov mask, but I also use same approach as you - I inspect star sizes in short "focusing" sub - I also look at change in star sizes across the whole field - there is some field curvature in my setups so I aim to minimize that. One can learn to hit good focus using this technique, and it minimizes chance of forgetting bahtinov mask on scope and wasting your subs . Also if using scope, you might want to check your focus (I do it by just looking at the stars in subs) if there is temperature change - there might be a need to refocus if scope cools or gets a bit warmer.

5. There are couple of different algorithms for stacking and it would be a good thing to get to know in principle how they work. They can help eliminate some sort of artifacts and noise. You may also find that different software solutions for stacking give you slightly different levels of sharpness of final image, but that has to do with other things and not stacking algorithms themselves. It has to do with how well software can determine centers of stars, and find optimum mathematical transform to align different subs (rotate, shift, scale, what ever is needed). Differences tend to be small, but some software does produce tighter stars in final stack (by small amount).

6. Processing does indeed have an impact on final result, but I see it as separate thing. It is skill that can be improved upon in time, so you can get back to your old data and process it differently. This is why I mostly concentrate on improving acquisition of data - since you can do that "only once" (not really true - heavens will be there for a long time, so you can always get more data on particular target, but I find that imaging time is rather precious - you need good conditions and clear night and those don't really come by that often).

As for detail/sharpness I'll throw in some more tips:

- learn about seeing, how to judge it and where to get forecast for particular night

- learn about transparency, and same as above, how to judge it and get forecast when planning a session.

Two above have significant impact on your results.

HTH

Wow! What a great reply. Thanks for this, great information. Top stuff ??

[ollypenrice]

In my opinion the way to combine long and short exposures seamlessly is this one: http://www.astropix.com/html/j_digit/laymask.html

In general there is rarely a need to shoot different exposure lengths but M42 is different.

Be very careful never to clip your black point. I think you have probably done so here to a small extent.

Olly

[Davey-T]

9 hours ago, Z3roCool said:

Are there any good plug-ins that help with nebula detail that anyone uses?

Not for nebula particularly but Noels Actions automate a lot of things.

Dave

[Z3roCool]

9 hours ago, ollypenrice said:

In my opinion the way to combine long and short exposures seamlessly is this one: http://www.astropix.com/html/j_digit/laymask.html

In general there is rarely a need to shoot different exposure lengths but M42 is different.

Be very careful never to clip your black point. I think you have probably done so here to a small extent.

Olly

Thanks Olly, think you are right. Will take a look at the link ?

[Z3roCool]

9 hours ago, Davey-T said:

Not for nebula particularly but Noels Actions automate a lot of things.

Dave

Cheers Dave, will take a look ?

[wimvb]

Besides the mechanics (pec, polar alignment, sturdy setup, no differential flexure, guiding working, larger aperture, minimising diffraction in optics), and conditions (seeing/transparency), you also get better sharpness with better pixelscale (fewer "/pixel), up to a point.

By applying concepts from "lucky imaging", you can select frames with tighter stars over those with fuzzier stars. In PixInsight, I use a weightfactor based on signal to noise ratio, star fwhm, and star eccentricity when stacking my subs. While I haven't done a proper analysis, I hope this will give me a better image quality in general.

During post processing, I use deconvolution, hdr transformation and sometimes unsharp mask to improve perceived sharpness.

The mechanical part is always the more important one. You can't get a masterpiece from fuzzy data. NASA learned that the hard way. They had to send up astronauts in space to give Hubble glasses.

[Z3roCool]

3 hours ago, wimvb said:

Besides the mechanics (pec, polar alignment, sturdy setup, no differential flexure, guiding working, larger aperture, minimising diffraction in optics), and conditions (seeing/transparency), you also get better sharpness with better pixelscale (fewer "/pixel), up to a point.

By applying concepts from "lucky imaging", you can select frames with tighter stars over those with fuzzier stars. In PixInsight, I use a weightfactor based on signal to noise ratio, star fwhm, and star eccentricity when stacking my subs. While I haven't done a proper analysis, I hope this will give me a better image quality in general.

During post processing, I use deconvolution, hdr transformation and sometimes unsharp mask to improve perceived sharpness.

The mechanical part is always the more important one. You can't get a masterpiece from fuzzy data. NASA learned that the hard way. They had to send up astronauts in space to give Hubble glasses.

Cheers wimvb

I think maybe when stacking I am adding all the frames and not being strict enough (Precious Data) I use DSS to stack and often tell it to use the best 90% of subs, which is normally around 2 it ditches. Thanks for your input.

[wimvb]

2 hours ago, Z3roCool said:

Cheers wimvb

I think maybe when stacking I am adding all the frames and not being strict enough (Precious Data) I use DSS to stack and often tell it to use the best 90% of subs, which is normally around 2 it ditches. Thanks for your input.

I don't use dss (anymore), but I've always wondered how dss calculates which subs to use. If you always set it to use 90%, you may one time throw away good data, and another time keep poor data, since the measure is relative to "the best" sub in the stack, and not an absolute measure. In PixInsight, I can tell it to use only stars with a fwhm smaller than an absolute value. Maybe dss can also do this, I don't know.

[Z3roCool]

On 15/03/2018 at 15:35, wimvb said:

I don't use dss (anymore), but I've always wondered how dss calculates which subs to use. If you always set it to use 90%, you may one time throw away good data, and another time keep poor data, since the measure is relative to "the best" sub in the stack, and not an absolute measure. In PixInsight, I can tell it to use only stars with a fwhm smaller than an absolute value. Maybe dss can also do this, I don't know.

Do not think it does the same. Suppose if I manually check each frame and chiose the best and then do 90% in DSS it should tighten thungs up. Need to take a look at PixInsight.