Harnessing Dither, Drizzle, and Deconvolution for Greater Detail and Resolution

#...

· 2 likes

Nick Grundy:
Steeve, I love this approach. It's definitely worth the time and thanks for sharing it with us.

I do remember Russ mentioned at NEAIC that BXT does best on undersampled data. I assume that's pretty much what you accomplish with the 2x drizzle, no? (I could be confusing this aspect)

2x drizzle over-samples the image.

Dithering typically adds so little additional information that I suspect that you'll get the exact same results if you simply use an interpolated image sampled at twice the rate of the original, which is computationally far simpler than dithering.

John

10.80 Steeve Body #... · view views
John Hayes: Nick Grundy: Steeve, I love this approach. It's definitely worth the time and thanks for sharing it with us. I do remember Russ mentioned at NEAIC that BXT does best on undersampled data. I assume that's pretty much what you accomplish with the 2x drizzle, no? (I could be confusing this aspect) 2x drizzle over-samples the image. Dithering typically adds so little additional information that I suspect that you'll get the exact same results if you simply use an interpolated image sampled at twice the rate of the original, which is computationally far simpler than dithering. John Hi John. Thanks for chiming in on this. i invite you to watch Russell talking about BXt and drizzle combination https://www.youtube.com/live/5_Dg7bYu-E8?feature=share. Check at around 1h13 min. Drizzle + BXT definitely seem to work very nicely together. I’ll check interpolation at 2x and report back Like

10.80

#...

John Hayes:
Nick Grundy:
Steeve, I love this approach. It's definitely worth the time and thanks for sharing it with us.

I do remember Russ mentioned at NEAIC that BXT does best on undersampled data. I assume that's pretty much what you accomplish with the 2x drizzle, no? (I could be confusing this aspect)

2x drizzle over-samples the image.

Dithering typically adds so little additional information that I suspect that you'll get the exact same results if you simply use an interpolated image sampled at twice the rate of the original, which is computationally far simpler than dithering.

John

Hi John.

Thanks for chiming in on this.

i invite you to watch Russell talking about BXt and drizzle combination https://www.youtube.com/live/5_Dg7bYu-E8?feature=share. Check at around 1h13 min. Drizzle + BXT definitely seem to work very nicely together. I’ll check interpolation at 2x and report back

11.57 Wei-Hao Wang #... · view views · 3 likes
One thing that you forgot to test is to first upsample the subs by 2x and then do a non-drizzle stack, and then also a BXT. Drizzle works the best on under-sampled subs that come from an old-school CCD with small photosites (like those on the old Hubble, which is why drizzle was invented) without microlenses. It also works on under-sampled subs taken with modern CMOS (with nearly 100% microlens covering fraction), but the effect is much less dramatic. I think this difference (photosite size) is one of the reasons why recently people start to complain that drizzle is not very effective: the cameras are different now. On well-sampled subs, drizzle has very little real effects. It may appear to still give you a higher resolution than well sampled, un-drizzled stacks. But probably 90% of the effect is just an illusion that can be achieved by simply up-sampling the subs and then stacking. Edited ... Like

11.57

#...

· 3 likes

One thing that you forgot to test is to first upsample the subs by 2x and then do a non-drizzle stack, and then also a BXT.

Drizzle works the best on under-sampled subs that come from an old-school CCD with small photosites (like those on the old Hubble, which is why drizzle was invented) without microlenses. It also works on under-sampled subs taken with modern CMOS (with nearly 100% microlens covering fraction), but the effect is much less dramatic. I think this difference (photosite size) is one of the reasons why recently people start to complain that drizzle is not very effective: the cameras are different now.

On well-sampled subs, drizzle has very little real effects. It may appear to still give you a higher resolution than well sampled, un-drizzled stacks. But probably 90% of the effect is just an illusion that can be achieved by simply up-sampling the subs and then stacking.

Edited ...

22.40 John Hayes #... · view views · 1 like
Yeah, I hear what Russ is saying, but as I said, under most circumstance (like under all but the VERY best atmospheric conditions) any additional information that comes from drizzling falls somewhere very close to zero. If it does make a significant difference, that would concern me. That would mean that almost ANYTHING could cause noticeable swings in the output. I'll bet that simply increasing the sampling by a factor of two might change things as much as drizzling simply based on the fact that there is virtually no difference in the information content in either case. Remember that drizzling was developed for Hubble data where you actually can extract addition information by drizzling dithered data simply due to the fact that it's a diffraction limited system operating in a vacuum. I'll look forward to hearing what you find... John Like

22.40

#...

· 1 like

Yeah, I hear what Russ is saying, but as I said, under most circumstance (like under all but the VERY best atmospheric conditions) any additional information that comes from drizzling falls somewhere very close to zero. If it does make a significant difference, that would concern me. That would mean that almost ANYTHING could cause noticeable swings in the output. I'll bet that simply increasing the sampling by a factor of two might change things as much as drizzling simply based on the fact that there is virtually no difference in the information content in either case. Remember that drizzling was developed for Hubble data where you actually can extract addition information by drizzling dithered data simply due to the fact that it's a diffraction limited system operating in a vacuum.

I'll look forward to hearing what you find...

John

10.80 Steeve Body #... · view views
Wei-Hao Wang: One thing that you forgot to test is to first upsample the subs by 2x and then do a non-drizzle stack, and then also a BXT. Drizzle works the best on under-sampled subs that come from an old-school CCD with small photosites (like those on the old Hubble, which is why drizzle was invented) without microlenses. It also works on under-sampled subs taken with modern CMOS (with nearly 100% microlens covering fraction), but the effect is much less dramatic. On well-sampled subs, it has very little real effects. It may appear to still give you a higher resolution than well sampled, un-drizzled stacks. But probably 90% of the effect is just an illusion that can be achieved by simply up-sampling the subs and then stacking. I do have drizzle 2x, 1x and no drizzle as separate stacks without bxt and with bxt. This are the first two images I posted. Is that what you mean…? you may be correct about the upsampling effect however I also recommend you watch the video I posted in the post above and check Russell showcasing the effect. Edited ... Like

10.80

#...

Wei-Hao Wang:
One thing that you forgot to test is to first upsample the subs by 2x and then do a non-drizzle stack, and then also a BXT.

Drizzle works the best on under-sampled subs that come from an old-school CCD with small photosites (like those on the old Hubble, which is why drizzle was invented) without microlenses. It also works on under-sampled subs taken with modern CMOS (with nearly 100% microlens covering fraction), but the effect is much less dramatic. On well-sampled subs, it has very little real effects. It may appear to still give you a higher resolution than well sampled, un-drizzled stacks. But probably 90% of the effect is just an illusion that can be achieved by simply up-sampling the subs and then stacking.

I do have drizzle 2x, 1x and no drizzle as separate stacks without bxt and with bxt. This are the first two images I posted. Is that what you mean…?

you may be correct about the upsampling effect however I also recommend you watch the video I posted in the post above and check Russell showcasing the effect.

Edited ...

11.57 Wei-Hao Wang #... · view views · 1 like
Steeve Body: I do have drizzle 2x, 1x and no drizzle as separate stacks without bxt and with bxt. This are the first two images I posted. Is that what you mean…? No. Not what I meant. I meant first 2x up-sample the subs, and then do a regular stack of those up-sampled subs without any drizzling involved. Stack and then a 2x upsample will be less effective. I did watch Russ' video, and nothing contradicts with what I said. Like

11.57

#...

· 1 like

Steeve Body:
I do have drizzle 2x, 1x and no drizzle as separate stacks without bxt and with bxt. This are the first two images I posted. Is that what you mean…?

No. Not what I meant. I meant first 2x up-sample the subs, and then do a regular stack of those up-sampled subs without any drizzling involved. Stack and then a 2x upsample will be less effective.

I did watch Russ' video, and nothing contradicts with what I said.

10.80 Steeve Body #... · view views
Wei-Hao Wang: Steeve Body: I do have drizzle 2x, 1x and no drizzle as separate stacks without bxt and with bxt. This are the first two images I posted. Is that what you mean…? No. Not what I meant. I meant first 2x up-sample the subs, and then do a regular stack of those up-sampled subs without any drizzling involved. Stack and then a 2x upsample will be less effective. I did watch Russ' video, and nothing contradicts with what I said. Got you. Is there a module in Pixinsight that allows me to upsample all subs easily? I have always done that with photoshop… Like

10.80

#...

Wei-Hao Wang:
Steeve Body:
I do have drizzle 2x, 1x and no drizzle as separate stacks without bxt and with bxt. This are the first two images I posted. Is that what you mean…?

No. Not what I meant. I meant first 2x up-sample the subs, and then do a regular stack of those up-sampled subs without any drizzling involved. Stack and then a 2x upsample will be less effective.

I did watch Russ' video, and nothing contradicts with what I said.

Got you. Is there a module in Pixinsight that allows me to upsample all subs easily? I have always done that with photoshop…

10.80 Steeve Body #... · view views
John Hayes: Yeah, I hear what Russ is saying, but as I said, under most circumstance (like under all but the VERY best atmospheric conditions) any additional information that comes from drizzling falls somewhere very close to zero. If it does make a significant difference, that would concern me. That would mean that almost ANYTHING could cause noticeable swings in the output. I'll bet that simply increasing the sampling by a factor of two might change things as much as drizzling simply based on the fact that there is virtually no difference in the information content in either case. Remember that drizzling was developed for Hubble data where you actually can extract addition information by drizzling dithered data simply due to the fact that it's a diffraction limited system operating in a vacuum. I'll look forward to hearing what you find... John Thanks John. That’s exactly why I’m posting this here so I could get peeps like you who really understand this stuff to explain to me what I’m seeing… me I’m a tinkerer who likes to experiment. Could it be because my seeing is not that great (between 2.5 and 3.5 at best) that this is making more of a difference? I do like what I’m seeing though.. even though if it’s only due to upsampling effect. I’ll test upsampling and stacking and check the difference with drizzling. As an audio mixer I use upsampling a lot on audio material. Especially for equalization as it makes a pretty substantial difference in term of quality of the processing. Like

10.80

#...

John Hayes:
Yeah, I hear what Russ is saying, but as I said, under most circumstance (like under all but the VERY best atmospheric conditions) any additional information that comes from drizzling falls somewhere very close to zero. If it does make a significant difference, that would concern me. That would mean that almost ANYTHING could cause noticeable swings in the output. I'll bet that simply increasing the sampling by a factor of two might change things as much as drizzling simply based on the fact that there is virtually no difference in the information content in either case. Remember that drizzling was developed for Hubble data where you actually can extract addition information by drizzling dithered data simply due to the fact that it's a diffraction limited system operating in a vacuum.

I'll look forward to hearing what you find...

John

Thanks John. That’s exactly why I’m posting this here so I could get peeps like you who really understand this stuff to explain to me what I’m seeing… me I’m a tinkerer who likes to experiment. Could it be because my seeing is not that great (between 2.5 and 3.5 at best) that this is making more of a difference? I do like what I’m seeing though.. even though if it’s only due to upsampling effect. I’ll test upsampling and stacking and check the difference with drizzling. As an audio mixer I use upsampling a lot on audio material. Especially for equalization as it makes a pretty substantial difference in term of quality of the processing.

11.57 Wei-Hao Wang #... · view views · 1 like
Steeve Body: Wei-Hao Wang: Steeve Body: I do have drizzle 2x, 1x and no drizzle as separate stacks without bxt and with bxt. This are the first two images I posted. Is that what you mean…? No. Not what I meant. I meant first 2x up-sample the subs, and then do a regular stack of those up-sampled subs without any drizzling involved. Stack and then a 2x upsample will be less effective. I did watch Russ' video, and nothing contradicts with what I said. Got you. Is there a module in Pixinsight that allows me to upsample all subs easily? I have always done that with photoshop… Ha! I am also a Photoshop person. I think there is an "Integer Resample" function that can do 2x upsampling very fast. But if you want to quickly do it on many images, you will need to run it on an image container. I am not the best person to tell you how to do it, because, like you, I am a PS guy. Like

11.57

#...

· 1 like

Steeve Body:
Wei-Hao Wang:
Steeve Body:
I do have drizzle 2x, 1x and no drizzle as separate stacks without bxt and with bxt. This are the first two images I posted. Is that what you mean…?

No. Not what I meant. I meant first 2x up-sample the subs, and then do a regular stack of those up-sampled subs without any drizzling involved. Stack and then a 2x upsample will be less effective.

I did watch Russ' video, and nothing contradicts with what I said.

Got you. Is there a module in Pixinsight that allows me to upsample all subs easily? I have always done that with photoshop…

Ha! I am also a Photoshop person. I think there is an "Integer Resample" function that can do 2x upsampling very fast. But if you want to quickly do it on many images, you will need to run it on an image container. I am not the best person to tell you how to do it, because, like you, I am a PS guy.

10.80 Steeve Body #... · view views · 1 like
Wei-Hao Wang: Steeve Body: Wei-Hao Wang: Steeve Body: I do have drizzle 2x, 1x and no drizzle as separate stacks without bxt and with bxt. This are the first two images I posted. Is that what you mean…? No. Not what I meant. I meant first 2x up-sample the subs, and then do a regular stack of those up-sampled subs without any drizzling involved. Stack and then a 2x upsample will be less effective. I did watch Russ' video, and nothing contradicts with what I said. Got you. Is there a module in Pixinsight that allows me to upsample all subs easily? I have always done that with photoshop… Ha! I am also a Photoshop person. I think there is an "Integer Resample" function that can do 2x upsampling very fast. But if you want to quickly do it on many images, you will need to run it on an image container. I am not the best person to tell you how to do it, because, like you, I am a PS guy. Thanks! I only use pixinsight for pre processing… but I know about image containers in PX so I’ll investigate… I’m sure google will come to the rescue…! I’ll post back a comparison once I’m done and see if this effect was mostly upsampling…. Edited ... Like

10.80

#...

· 1 like

Wei-Hao Wang:
Steeve Body:
Wei-Hao Wang:
Steeve Body:
I do have drizzle 2x, 1x and no drizzle as separate stacks without bxt and with bxt. This are the first two images I posted. Is that what you mean…?

No. Not what I meant. I meant first 2x up-sample the subs, and then do a regular stack of those up-sampled subs without any drizzling involved. Stack and then a 2x upsample will be less effective.

I did watch Russ' video, and nothing contradicts with what I said.

Got you. Is there a module in Pixinsight that allows me to upsample all subs easily? I have always done that with photoshop…

Ha! I am also a Photoshop person. I think there is an "Integer Resample" function that can do 2x upsampling very fast. But if you want to quickly do it on many images, you will need to run it on an image container. I am not the best person to tell you how to do it, because, like you, I am a PS guy.

Thanks! I only use pixinsight for pre processing… but I know about image containers in PX so I’ll investigate… I’m sure google will come to the rescue…! I’ll post back a comparison once I’m done and see if this effect was mostly upsampling….

Edited ...

10.80 Steeve Body #... · view views
@John Hayes @Wei-Hao Wang So I did another test were I resampled all the lights using integer resample in Pixinsight at 2x and re-stacked them to compare... and indeed the result are very similar like you both predicted. I still do give a slight edge to the drizzle version as it has smoother gradient and better level of contrast on the real faint detail stuff by really that is debatable at this point. An advantage of the drizzle version would be time saving in the stacking, It took considerably longer to upsampling the lights and then stack them than just drizzle 2x in WBPP From left to right: drizzle 2x with BXT, upsampled 2x lights with BXT, drizzle 1x with BXT Like

10.80

#...

@John Hayes @Wei-Hao Wang So I did another test were I resampled all the lights using integer resample in Pixinsight at 2x and re-stacked them to compare... and indeed the result are very similar like you both predicted. I still do give a slight edge to the drizzle version as it has smoother gradient and better level of contrast on the real faint detail stuff by really that is debatable at this point. An advantage of the drizzle version would be time saving in the stacking, It took considerably longer to upsampling the lights and then stack them than just drizzle 2x in WBPP

From left to right: drizzle 2x with BXT, upsampled 2x lights with BXT, drizzle 1x with BXT

11.57 Wei-Hao Wang #... · view views · 1 like
Thanks for the test. Very close to what I expected. As for speed, my experience is that drizzle is much slower if not including the time to setup the image container. Before I run drizzle, I always first do a normal integration (with drizzle file included) to reject outlier pixels, and then a drizzle integration. Anyway, there are many details that can be changed and can affect the total execution time. Like

11.57

#...

· 1 like

Thanks for the test. Very close to what I expected.

As for speed, my experience is that drizzle is much slower if not including the time to setup the image container. Before I run drizzle, I always first do a normal integration (with drizzle file included) to reject outlier pixels, and then a drizzle integration. Anyway, there are many details that can be changed and can affect the total execution time.

10.80 Steeve Body #... · view views
Wei-Hao Wang: Thanks for the test. Very close to what I expected. As for speed, my experience is that drizzle is much slower if not including the time to setup the image container. Before I run drizzle, I always first do a normal integration (with drizzle file included) to reject outlier pixels, and then a drizzle integration. Anyway, there are many details that can be changed and can affect the total execution time. Yeah weird… Me when I use WBPP I set it to max “no compromise”. It does both a drizzle and regular integration at the same time. It took 50min to stack and drizzle all SHO and RGB filters and here it took 25min just to do Ha with no drizzle… Like

10.80

#...

Wei-Hao Wang:
Thanks for the test. Very close to what I expected.

As for speed, my experience is that drizzle is much slower if not including the time to setup the image container. Before I run drizzle, I always first do a normal integration (with drizzle file included) to reject outlier pixels, and then a drizzle integration. Anyway, there are many details that can be changed and can affect the total execution time.

Yeah weird… Me when I use WBPP I set it to max “no compromise”. It does both a drizzle and regular integration at the same time. It took 50min to stack and drizzle all SHO and RGB filters and here it took 25min just to do Ha with no drizzle…

22.40 John Hayes #... · view views · 1 like
Steeve Body: @John Hayes @Wei-Hao Wang So I did another test were I resampled all the lights using integer resample in Pixinsight at 2x and re-stacked them to compare... and indeed the result are very similar like you both predicted. I still do give a slight edge to the drizzle version as it has smoother gradient and better level of contrast on the real faint detail stuff by really that is debatable at this point. An advantage of the drizzle version would be time saving in the stacking, It took considerably longer to upsampling the lights and then stack them than just drizzle 2x in WBPP From left to right: drizzle 2x with BXT, upsampled 2x lights with BXT, drizzle 1x with BXT Steeve, I think that you might not have fully appreciated what I was trying to say—probably because I didn’t make it very clear. Here’s what I should have said: Drizzling beyond the seeing limit is functionally equivalent to simply interpolating the data to a higher sampling rate. When you up-sample using IntegerResample, that’s not what you are doing. You have to interpolate the image data to a higher sampling rate. There are a lot of good ways to interpolate the data (using things like spline fitting) but in this case, I think that a simple linear fit will work well to make the point. So, to do the right comparison, you need to use the Resample tool, which allows different options for interpolating the data. As Wei-Hao said, an image container is the easiest way to do that on a batch of images. Image containers are super simple to set up so this is a great opportunity to learn how to use them (if you haven’t already). I’m not trying to send you back to the drawing board, but I’m pretty certain that the only reason that you see a small difference between the drizzled data and the up-sampled data is because the data wasn’t properly up-sampled. John Edited ... Like

22.40

#...

· 1 like

Steeve Body:
@John Hayes @Wei-Hao Wang So I did another test were I resampled all the lights using integer resample in Pixinsight at 2x and re-stacked them to compare... and indeed the result are very similar like you both predicted. I still do give a slight edge to the drizzle version as it has smoother gradient and better level of contrast on the real faint detail stuff by really that is debatable at this point. An advantage of the drizzle version would be time saving in the stacking, It took considerably longer to upsampling the lights and then stack them than just drizzle 2x in WBPP

From left to right: drizzle 2x with BXT, upsampled 2x lights with BXT, drizzle 1x with BXT

Steeve,
I think that you might not have fully appreciated what I was trying to say—probably because I didn’t make it very clear. Here’s what I should have said: Drizzling beyond the seeing limit is functionally equivalent to simply interpolating the data to a higher sampling rate. When you up-sample using IntegerResample, that’s not what you are doing. You have to interpolate the image data to a higher sampling rate. There are a lot of good ways to interpolate the data (using things like spline fitting) but in this case, I think that a simple linear fit will work well to make the point. So, to do the right comparison, you need to use the Resample tool, which allows different options for interpolating the data. As Wei-Hao said, an image container is the easiest way to do that on a batch of images. Image containers are super simple to set up so this is a great opportunity to learn how to use them (if you haven’t already). I’m not trying to send you back to the drawing board, but I’m pretty certain that the only reason that you see a small difference between the drizzled data and the up-sampled data is because the data wasn’t properly up-sampled.

John

Edited ...

11.57 Wei-Hao Wang #... · view views · 1 like
Hi John, Please see if you agree with this. I think the best way to achieve a 2x resampling is to do it simultaneously with the image registration. For example, one may use an image that's already 2x up-sampled as the registration reference. Then register everything to it. This way, each image would have only undergone one resampling. If one first 2x up-samples and then registers, or first registers and then 2x up-samples, each image would have undergone two resamplings, which would be less ideal. Personally I think the difference between Steeve's 2x drizzled and 2x upsampled cases would be even closer if the later is done optimally (not suggesting Steve to do it again though). However, I expect that the drizzled case would still be better, very very slightly, perhaps so slight that only a very careful analysis can reveal the difference. In my work flow, drizzle takes so much longer. I no longer think it is worthwhile for the small improvement it brings (unless the original image is under-sampled, of course). Like

11.57

#...

· 1 like

Hi John,

Please see if you agree with this. I think the best way to achieve a 2x resampling is to do it simultaneously with the image registration. For example, one may use an image that's already 2x up-sampled as the registration reference. Then register everything to it. This way, each image would have only undergone one resampling. If one first 2x up-samples and then registers, or first registers and then 2x up-samples, each image would have undergone two resamplings, which would be less ideal.

Personally I think the difference between Steeve's 2x drizzled and 2x upsampled cases would be even closer if the later is done optimally (not suggesting Steve to do it again though). However, I expect that the drizzled case would still be better, very very slightly, perhaps so slight that only a very careful analysis can reveal the difference. In my work flow, drizzle takes so much longer. I no longer think it is worthwhile for the small improvement it brings (unless the original image is under-sampled, of course).

10.80 Steeve Body #... · view views
John Hayes: Steeve Body: @John Hayes @Wei-Hao Wang So I did another test were I resampled all the lights using integer resample in Pixinsight at 2x and re-stacked them to compare... and indeed the result are very similar like you both predicted. I still do give a slight edge to the drizzle version as it has smoother gradient and better level of contrast on the real faint detail stuff by really that is debatable at this point. An advantage of the drizzle version would be time saving in the stacking, It took considerably longer to upsampling the lights and then stack them than just drizzle 2x in WBPP From left to right: drizzle 2x with BXT, upsampled 2x lights with BXT, drizzle 1x with BXT Steeve, I think that you might not have fully appreciated what I was trying to say—probably because I didn’t make it very clear. Here’s what I should have said: Drizzling beyond the seeing limit is functionally equivalent to simply interpolating the data to a higher sampling rate. When you up-sample using IntegerResample, that’s not what you are doing. You have to interpolate the image data to a higher sampling rate. There are a lot of good ways to interpolate the data (using things like spline fitting) but in this case, I think that a simple linear fit will work well to make the point. So, to do the right comparison, you need to use the Resample tool, which allows different options for interpolating the data. As Wei-Hao said, an image container is the easiest way to do that on a batch of images. Image containers are super simple to set up so this is a great opportunity to learn how to use them (if you haven’t already). I’m not trying to send you back to the drawing board, but I’m pretty certain that the only reason that you see a small difference between the drizzled data and the up-sampled data is because the data wasn’t properly up-sampled. John John first of all thank you for taking the time to educate me on the matter. I truly appreciate it. I absolutely don't mind going back to the drawing board and re doing these things as I'm learning a lot through doing and it is something I quite enjoy doing anyway. Now I have done what you have suggested and applied the resample tool with the following settings to all the frames using an image container. FYI I'm applying the resample to all the light frames as well as the flats and darks... which I'm assuming is the right thing to do. this is the result I got. I still favour the drizzled 2x personally as this new looks just a tad soft compared to the drizzle 2x... but now the interpolation algo because a variable and I'm sure I'll be getting slightly different results depending on the settings... From left to right: drizzle 2x with BXT, upsampled interpolated using bicubic spline 2x lights with BXT, drizzle 1x with BXT and for reference the previous one doing interger resample From left to right: drizzle 2x with BXT, upsampled 2x lights with BXT, drizzle 1x with BXT In saying all of this though you are right in saying that it upscaling effect of drizzling 2x is what seems to be providing the biggest effect in combination with BXT, and that swapping drizzle for upscaling with interpolation provide the same benefits. So my logic would lead me to conclude that by upsampling the data, BXT has more pixel to work with to increase local contrast and detail in the nebulosity and therefore seem to reveal more details. Now I've spent a fair amount of time comparing the detail I got in this image to much higher resolution instrument and to my eyes at least the process didn't create details that were not there in first place... so from an ethical point of view I felt ok about using it in my workflow since I saw noticeable improvements. In you opinion do you think this process has merit or am I just fooling myself with snake oil here...? Like

10.80

#...

John Hayes:
Steeve Body:
@John Hayes @Wei-Hao Wang So I did another test were I resampled all the lights using integer resample in Pixinsight at 2x and re-stacked them to compare... and indeed the result are very similar like you both predicted. I still do give a slight edge to the drizzle version as it has smoother gradient and better level of contrast on the real faint detail stuff by really that is debatable at this point. An advantage of the drizzle version would be time saving in the stacking, It took considerably longer to upsampling the lights and then stack them than just drizzle 2x in WBPP

From left to right: drizzle 2x with BXT, upsampled 2x lights with BXT, drizzle 1x with BXT

Steeve,
I think that you might not have fully appreciated what I was trying to say—probably because I didn’t make it very clear. Here’s what I should have said: Drizzling beyond the seeing limit is functionally equivalent to simply interpolating the data to a higher sampling rate. When you up-sample using IntegerResample, that’s not what you are doing. You have to interpolate the image data to a higher sampling rate. There are a lot of good ways to interpolate the data (using things like spline fitting) but in this case, I think that a simple linear fit will work well to make the point. So, to do the right comparison, you need to use the Resample tool, which allows different options for interpolating the data. As Wei-Hao said, an image container is the easiest way to do that on a batch of images. Image containers are super simple to set up so this is a great opportunity to learn how to use them (if you haven’t already). I’m not trying to send you back to the drawing board, but I’m pretty certain that the only reason that you see a small difference between the drizzled data and the up-sampled data is because the data wasn’t properly up-sampled.

John

John first of all thank you for taking the time to educate me on the matter. I truly appreciate it. I absolutely don't mind going back to the drawing board and re doing these things as I'm learning a lot through doing and it is something I quite enjoy doing anyway.

Now I have done what you have suggested and applied the resample tool with the following settings to all the frames using an image container.

FYI I'm applying the resample to all the light frames as well as the flats and darks... which I'm assuming is the right thing to do.

this is the result I got.

I still favour the drizzled 2x personally as this new looks just a tad soft compared to the drizzle 2x... but now the interpolation algo because a variable and I'm sure I'll be getting slightly different results depending on the settings...

From left to right: drizzle 2x with BXT, upsampled interpolated using bicubic spline 2x lights with BXT, drizzle 1x with BXT

and for reference the previous one doing interger resample
From left to right: drizzle 2x with BXT, upsampled 2x lights with BXT, drizzle 1x with BXT

In saying all of this though you are right in saying that it upscaling effect of drizzling 2x is what seems to be providing the biggest effect in combination with BXT, and that swapping drizzle for upscaling with interpolation provide the same benefits. So my logic would lead me to conclude that by upsampling the data, BXT has more pixel to work with to increase local contrast and detail in the nebulosity and therefore seem to reveal more details.

Now I've spent a fair amount of time comparing the detail I got in this image to much higher resolution instrument and to my eyes at least the process didn't create details that were not there in first place... so from an ethical point of view I felt ok about using it in my workflow since I saw noticeable improvements. In you opinion do you think this process has merit or am I just fooling myself with snake oil here...?

22.40 John Hayes #... · view views
Wei-Hao Wang: Hi John, Please see if you agree with this. I think the best way to achieve a 2x resampling is to do it simultaneously with the image registration. For example, one may use an image that's already 2x up-sampled as the registration reference. Then register everything to it. This way, each image would have only undergone one resampling. If one first 2x up-samples and then registers, or first registers and then 2x up-samples, each image would have undergone two resamplings, which would be less ideal. Personally I think the difference between Steeve's 2x drizzled and 2x upsampled cases would be even closer if the later is done optimally (not suggesting Steve to do it again though). However, I expect that the drizzled case would still be better, very very slightly, perhaps so slight that only a very careful analysis can reveal the difference. In my work flow, drizzle takes so much longer. I no longer think it is worthwhile for the small improvement it brings (unless the original image is under-sampled, of course). Wei-Hao, Yes, I completely agree. Your approach sounds really good! John Like

22.40

#...

Wei-Hao Wang:
Hi John,

Please see if you agree with this. I think the best way to achieve a 2x resampling is to do it simultaneously with the image registration. For example, one may use an image that's already 2x up-sampled as the registration reference. Then register everything to it. This way, each image would have only undergone one resampling. If one first 2x up-samples and then registers, or first registers and then 2x up-samples, each image would have undergone two resamplings, which would be less ideal.

Personally I think the difference between Steeve's 2x drizzled and 2x upsampled cases would be even closer if the later is done optimally (not suggesting Steve to do it again though). However, I expect that the drizzled case would still be better, very very slightly, perhaps so slight that only a very careful analysis can reveal the difference. In my work flow, drizzle takes so much longer. I no longer think it is worthwhile for the small improvement it brings (unless the original image is under-sampled, of course).

Wei-Hao,
Yes, I completely agree. Your approach sounds really good!

John

10.80 Steeve Body #... · view views
Wei-Hao Wang: Hi John, Please see if you agree with this. I think the best way to achieve a 2x resampling is to do it simultaneously with the image registration. For example, one may use an image that's already 2x up-sampled as the registration reference. Then register everything to it. This way, each image would have only undergone one resampling. If one first 2x up-samples and then registers, or first registers and then 2x up-samples, each image would have undergone two resamplings, which would be less ideal. Personally I think the difference between Steeve's 2x drizzled and 2x upsampled cases would be even closer if the later is done optimally (not suggesting Steve to do it again though). However, I expect that the drizzled case would still be better, very very slightly, perhaps so slight that only a very careful analysis can reveal the difference. In my work flow, drizzle takes so much longer. I no longer think it is worthwhile for the small improvement it brings (unless the original image is under-sampled, of course). Are you using the WBPP script for stacking or you are doing everything manually? For me as it’s stands using the script and applying 2x drizzle works well for me without too much extra overhead on my MacBook m2 max so I’m probably going to stick to that for now. Also would you mind clarifying that statement for me so I make sure I got this right “one may use an image that's already 2x up-sampled as the registration reference. Then register everything to it. This way, each image would have only undergone one resampling” Do you mean by that pick a reference frame, upsample it 2x, register all the frames that have not been upsampled yet to it, then upsample all frames after they have been registered and stack that yeah? Edited ... Like

10.80

#...

Wei-Hao Wang:
Hi John,

Please see if you agree with this. I think the best way to achieve a 2x resampling is to do it simultaneously with the image registration. For example, one may use an image that's already 2x up-sampled as the registration reference. Then register everything to it. This way, each image would have only undergone one resampling. If one first 2x up-samples and then registers, or first registers and then 2x up-samples, each image would have undergone two resamplings, which would be less ideal.

Personally I think the difference between Steeve's 2x drizzled and 2x upsampled cases would be even closer if the later is done optimally (not suggesting Steve to do it again though). However, I expect that the drizzled case would still be better, very very slightly, perhaps so slight that only a very careful analysis can reveal the difference. In my work flow, drizzle takes so much longer. I no longer think it is worthwhile for the small improvement it brings (unless the original image is under-sampled, of course).

Are you using the WBPP script for stacking or you are doing everything manually? For me as it’s stands using the script and applying 2x drizzle works well for me without too much extra overhead on my MacBook m2 max so I’m probably going to stick to that for now.

Also would you mind clarifying that statement for me so I make sure I got this right “one may use an image that's already 2x up-sampled as the registration reference. Then register everything to it. This way, each image would have only undergone one resampling”

Do you mean by that pick a reference frame, upsample it 2x, register all the frames that have not been upsampled yet to it, then upsample all frames after they have been registered and stack that yeah?

Edited ...

1.20 Joseph Biscoe IV #... · view views · 1 like
Steeve Body: Wei-Hao Wang: Hi John, Please see if you agree with this. I think the best way to achieve a 2x resampling is to do it simultaneously with the image registration. For example, one may use an image that's already 2x up-sampled as the registration reference. Then register everything to it. This way, each image would have only undergone one resampling. If one first 2x up-samples and then registers, or first registers and then 2x up-samples, each image would have undergone two resamplings, which would be less ideal. Personally I think the difference between Steeve's 2x drizzled and 2x upsampled cases would be even closer if the later is done optimally (not suggesting Steve to do it again though). However, I expect that the drizzled case would still be better, very very slightly, perhaps so slight that only a very careful analysis can reveal the difference. In my work flow, drizzle takes so much longer. I no longer think it is worthwhile for the small improvement it brings (unless the original image is under-sampled, of course). Are you using the WBPP script for stacking or you are doing everything manually? For me as it’s stands using the script and applying 2x drizzle works well for me without too much extra overhead on my MacBook m2 max so I’m probably going to stick to that for now. Also would you mind clarifying that statement for me so I make sure I got this right “one may use an image that's already 2x up-sampled as the registration reference. Then register everything to it. This way, each image would have only undergone one resampling” Do you mean by that pick a reference frame, upsample it 2x, register all the frames that have not been upsampled yet to it, then upsample all frames after they have been registered and stack that yeah? This is an interesting topic. I'm curious if you got it figured out. But yes, choose a reference frame manually. Upsample it. And then use it for aligning. Aligning can be done with either the star alignment tool, or you can select a manual reference file in WBPP. Like

1.20

Joseph Biscoe IV

#...

· 1 like

Steeve Body:
Wei-Hao Wang:
Hi John,

Please see if you agree with this. I think the best way to achieve a 2x resampling is to do it simultaneously with the image registration. For example, one may use an image that's already 2x up-sampled as the registration reference. Then register everything to it. This way, each image would have only undergone one resampling. If one first 2x up-samples and then registers, or first registers and then 2x up-samples, each image would have undergone two resamplings, which would be less ideal.

Personally I think the difference between Steeve's 2x drizzled and 2x upsampled cases would be even closer if the later is done optimally (not suggesting Steve to do it again though). However, I expect that the drizzled case would still be better, very very slightly, perhaps so slight that only a very careful analysis can reveal the difference. In my work flow, drizzle takes so much longer. I no longer think it is worthwhile for the small improvement it brings (unless the original image is under-sampled, of course).

Are you using the WBPP script for stacking or you are doing everything manually? For me as it’s stands using the script and applying 2x drizzle works well for me without too much extra overhead on my MacBook m2 max so I’m probably going to stick to that for now.

Also would you mind clarifying that statement for me so I make sure I got this right “one may use an image that's already 2x up-sampled as the registration reference. Then register everything to it. This way, each image would have only undergone one resampling”

Do you mean by that pick a reference frame, upsample it 2x, register all the frames that have not been upsampled yet to it, then upsample all frames after they have been registered and stack that yeah?

This is an interesting topic. I'm curious if you got it figured out. But yes, choose a reference frame manually. Upsample it. And then use it for aligning. Aligning can be done with either the star alignment tool, or you can select a manual reference file in WBPP.

10.80 Steeve Body #... · view views · 2 likes
Joseph Biscoe IV: Steeve Body: Wei-Hao Wang: Hi John, Please see if you agree with this. I think the best way to achieve a 2x resampling is to do it simultaneously with the image registration. For example, one may use an image that's already 2x up-sampled as the registration reference. Then register everything to it. This way, each image would have only undergone one resampling. If one first 2x up-samples and then registers, or first registers and then 2x up-samples, each image would have undergone two resamplings, which would be less ideal. Personally I think the difference between Steeve's 2x drizzled and 2x upsampled cases would be even closer if the later is done optimally (not suggesting Steve to do it again though). However, I expect that the drizzled case would still be better, very very slightly, perhaps so slight that only a very careful analysis can reveal the difference. In my work flow, drizzle takes so much longer. I no longer think it is worthwhile for the small improvement it brings (unless the original image is under-sampled, of course). Are you using the WBPP script for stacking or you are doing everything manually? For me as it’s stands using the script and applying 2x drizzle works well for me without too much extra overhead on my MacBook m2 max so I’m probably going to stick to that for now. Also would you mind clarifying that statement for me so I make sure I got this right “one may use an image that's already 2x up-sampled as the registration reference. Then register everything to it. This way, each image would have only undergone one resampling” Do you mean by that pick a reference frame, upsample it 2x, register all the frames that have not been upsampled yet to it, then upsample all frames after they have been registered and stack that yeah? This is an interesting topic. I'm curious if you got it figured out. But yes, choose a reference frame manually. Upsample it. And then use it for aligning. Aligning can be done with either the star alignment tool, or you can select a manual reference file in WBPP. Yes I have done the upsampling test and it is working with pretty much the same results… but that being said I still just find it easier to hit the drizzle 2x button in WBPP and let it run Like

10.80

#...

· 2 likes

Joseph Biscoe IV:
Steeve Body:
Wei-Hao Wang:
Hi John,

Please see if you agree with this. I think the best way to achieve a 2x resampling is to do it simultaneously with the image registration. For example, one may use an image that's already 2x up-sampled as the registration reference. Then register everything to it. This way, each image would have only undergone one resampling. If one first 2x up-samples and then registers, or first registers and then 2x up-samples, each image would have undergone two resamplings, which would be less ideal.

Personally I think the difference between Steeve's 2x drizzled and 2x upsampled cases would be even closer if the later is done optimally (not suggesting Steve to do it again though). However, I expect that the drizzled case would still be better, very very slightly, perhaps so slight that only a very careful analysis can reveal the difference. In my work flow, drizzle takes so much longer. I no longer think it is worthwhile for the small improvement it brings (unless the original image is under-sampled, of course).

Are you using the WBPP script for stacking or you are doing everything manually? For me as it’s stands using the script and applying 2x drizzle works well for me without too much extra overhead on my MacBook m2 max so I’m probably going to stick to that for now.

Also would you mind clarifying that statement for me so I make sure I got this right “one may use an image that's already 2x up-sampled as the registration reference. Then register everything to it. This way, each image would have only undergone one resampling”

Do you mean by that pick a reference frame, upsample it 2x, register all the frames that have not been upsampled yet to it, then upsample all frames after they have been registered and stack that yeah?

This is an interesting topic. I'm curious if you got it figured out. But yes, choose a reference frame manually. Upsample it. And then use it for aligning. Aligning can be done with either the star alignment tool, or you can select a manual reference file in WBPP.

Yes I have done the upsampling test and it is working with pretty much the same results… but that being said I still just find it easier to hit the drizzle 2x button in WBPP and let it run

1.20 Joseph Biscoe IV #... · view views
Steeve Body: Joseph Biscoe IV: Steeve Body: Wei-Hao Wang: Hi John, Please see if you agree with this. I think the best way to achieve a 2x resampling is to do it simultaneously with the image registration. For example, one may use an image that's already 2x up-sampled as the registration reference. Then register everything to it. This way, each image would have only undergone one resampling. If one first 2x up-samples and then registers, or first registers and then 2x up-samples, each image would have undergone two resamplings, which would be less ideal. Personally I think the difference between Steeve's 2x drizzled and 2x upsampled cases would be even closer if the later is done optimally (not suggesting Steve to do it again though). However, I expect that the drizzled case would still be better, very very slightly, perhaps so slight that only a very careful analysis can reveal the difference. In my work flow, drizzle takes so much longer. I no longer think it is worthwhile for the small improvement it brings (unless the original image is under-sampled, of course). Are you using the WBPP script for stacking or you are doing everything manually? For me as it’s stands using the script and applying 2x drizzle works well for me without too much extra overhead on my MacBook m2 max so I’m probably going to stick to that for now. Also would you mind clarifying that statement for me so I make sure I got this right “one may use an image that's already 2x up-sampled as the registration reference. Then register everything to it. This way, each image would have only undergone one resampling” Do you mean by that pick a reference frame, upsample it 2x, register all the frames that have not been upsampled yet to it, then upsample all frames after they have been registered and stack that yeah? This is an interesting topic. I'm curious if you got it figured out. But yes, choose a reference frame manually. Upsample it. And then use it for aligning. Aligning can be done with either the star alignment tool, or you can select a manual reference file in WBPP. Yes I have done the upsampling test and it is working with pretty much the same results… but that being said I still just find it easier to hit the drizzle 2x button in WBPP and let it run Ok, well good to know. Thanks for testing. Learned a good bit reading through. Like

1.20

Joseph Biscoe IV

#...

Steeve Body:
Joseph Biscoe IV:
Steeve Body:
Wei-Hao Wang:
Hi John,

Please see if you agree with this. I think the best way to achieve a 2x resampling is to do it simultaneously with the image registration. For example, one may use an image that's already 2x up-sampled as the registration reference. Then register everything to it. This way, each image would have only undergone one resampling. If one first 2x up-samples and then registers, or first registers and then 2x up-samples, each image would have undergone two resamplings, which would be less ideal.

Personally I think the difference between Steeve's 2x drizzled and 2x upsampled cases would be even closer if the later is done optimally (not suggesting Steve to do it again though). However, I expect that the drizzled case would still be better, very very slightly, perhaps so slight that only a very careful analysis can reveal the difference. In my work flow, drizzle takes so much longer. I no longer think it is worthwhile for the small improvement it brings (unless the original image is under-sampled, of course).

Are you using the WBPP script for stacking or you are doing everything manually? For me as it’s stands using the script and applying 2x drizzle works well for me without too much extra overhead on my MacBook m2 max so I’m probably going to stick to that for now.

Also would you mind clarifying that statement for me so I make sure I got this right “one may use an image that's already 2x up-sampled as the registration reference. Then register everything to it. This way, each image would have only undergone one resampling”

Do you mean by that pick a reference frame, upsample it 2x, register all the frames that have not been upsampled yet to it, then upsample all frames after they have been registered and stack that yeah?

This is an interesting topic. I'm curious if you got it figured out. But yes, choose a reference frame manually. Upsample it. And then use it for aligning. Aligning can be done with either the star alignment tool, or you can select a manual reference file in WBPP.

Yes I have done the upsampling test and it is working with pretty much the same results… but that being said I still just find it easier to hit the drizzle 2x button in WBPP and let it run

Ok, well good to know. Thanks for testing. Learned a good bit reading through.

4.21 Miguel T. #... · view views · 1 like
Hello, I find this interresting. I have myself invested many hours playing around with DrizzleIntegration and combining it with BXT to see the outcomes. One thing I would've loved to see in your experiments is the usage of different Kernel functions in the DrizzleIntegration process. Some of them are very computationally intensive and take very long to process so this is an area that I have yet to fully explore. "Gaussian and variable shape drizzle kernels (...) can be used to improve resolution of the integrated image. However, these functions tend to require much more and much better dithered data than the standard square and circular kernels to achieve optimal results." The VarShape/Gaussian kernels tend to allow me to push the resolution of small galaxies a bit further, but looked too artificial with deconvolution when I tried them. Maybe due to the too small number of subframes or maybe that's just how it is... Edited ... Like

4.21

Miguel T.

#...

· 1 like

Hello, I find this interresting. I have myself invested many hours playing around with DrizzleIntegration and combining it with BXT to see the outcomes.

One thing I would've loved to see in your experiments is the usage of different Kernel functions in the DrizzleIntegration process. Some of them are very computationally intensive and take very long to process so this is an area that I have yet to fully explore.

"Gaussian and variable shape drizzle kernels (...) can be used to improve resolution of the integrated image. However, these functions tend to require much more and much better dithered data than the standard square and circular kernels to achieve optimal results."

The VarShape/Gaussian kernels tend to allow me to push the resolution of small galaxies a bit further, but looked too artificial with deconvolution when I tried them. Maybe due to the too small number of subframes or maybe that's just how it is...

Edited ...

10.80 Steeve Body #... · view views · 1 like
Miguel T.: Hello, I find this interresting. I have myself invested many hours playing around with DrizzleIntegration and combining it with BXT to see the outcomes. One thing I would've loved to see in your experiments is the usage of different Kernel functions in the DrizzleIntegration process. Some of them are very computationally intensive and take very long to process so this is an area that I have yet to fully explore. "Gaussian and variable shape drizzle kernels (...) can be used to improve resolution of the integrated image. However, these functions tend to require much more and much better dithered data than the standard square and circular kernels to achieve optimal results." The VarShape/Gaussian kernels tend to allow me to push the resolution of small galaxies a bit further, but looked too artificial with deconvolution when I tried them. Maybe due to the too small number of subframes or maybe that's just how it is... Really good question. I have not done the testing myself but I’m part of an astrophotographers group with lots of very clever folks who have tested this and so far these are the recommended settings based on testing: Drop shrink 0.35 Variable shape 1.5 if you have at least 200+ frames Square kernel if under 200 the default drop shrink is 0.9 which really should be the default setting at all… Edited ... Like

10.80