Rodd Posted February 17, 2018 Share Posted February 17, 2018 This image consists of about 3 hours each of RGB in 10 min subs using the TOA 130 with .7x reducer and STT-8300. I create a luminance with 51 RGB subs for a synthetic LRGB image. I have 1 clear night (hopefully) to collect either Ha, ora proper lum. As soon as a necessary adapter arrives I am scheduled to send in my camera for diagnostics due to the filter wheel tendency to jam. Then I will switch to the ASI 1600. meanwhile, tomorrow is supposed to be clear--then cloudy for a week or so. I have to decide if I go for Ha or lum--or a bit of each, but that may not be worthwhile. The Ha regions ARE rather pale. Any suggestions? This is a drizzle integrated image--so its 4x the size of a regular image. Sort of an experiment to see if drizzle makes sense for a smaller target like a galaxy. The image has not been down sampled so full resolution mode may be unsuitable--I won't know until I post. I edited and reduced the background a bit--may be a bit topo dark (not by much), but it was rather noisy at full resolution. Hopefully this is better. Link to comment Share on other sites More sharing options...
Filroden Posted February 17, 2018 Share Posted February 17, 2018 Another lovely image. I think lum data might add more than Ha and define more detail. Link to comment Share on other sites More sharing options...
Demonperformer Posted February 17, 2018 Share Posted February 17, 2018 Nice start on this one, Road. Link to comment Share on other sites More sharing options...
Rodd Posted February 17, 2018 Author Share Posted February 17, 2018 26 minutes ago, Filroden said: Another lovely image. I think lum data might add more than Ha and define more detail. Thanks Ken. I am not sure--I binned 1x1 so not sure lum will do much--unless I get ALLOT of it. Rodd Link to comment Share on other sites More sharing options...
Rodd Posted February 17, 2018 Author Share Posted February 17, 2018 7 minutes ago, Demonperformer said: Nice start on this one, Road. Thanks DP--Not sure about the drizzle. Its bigger, but not very appealing at that scale. So whats the point. Maybe with allot more data. Rodd Link to comment Share on other sites More sharing options...
Stub Mandrel Posted February 17, 2018 Share Posted February 17, 2018 Very nice, do you have good dark skies? I struggle to get M101 to stand out from the murk. Link to comment Share on other sites More sharing options...
Rodd Posted February 17, 2018 Author Share Posted February 17, 2018 1 minute ago, Stub Mandrel said: Very nice, do you have good dark skies? I struggle to get M101 to stand out from the murk. Thanks--no my skies are cruddy usually. they were OK for this, but not great. Rodd Link to comment Share on other sites More sharing options...
cfpendock Posted February 17, 2018 Share Posted February 17, 2018 That's nice Rodd. I especially like the soft detail of the core. But there seem to be some funny artefacts on the stars at full res. Is this the effect of drizzle? No, I don't think so. The blue overlaps at the top left and top right are in different directions????? But a very nice image. Chris Link to comment Share on other sites More sharing options...
Rodd Posted February 17, 2018 Author Share Posted February 17, 2018 49 minutes ago, cfpendock said: That's nice Rodd. I especially like the soft detail of the core. But there seem to be some funny artefacts on the stars at full res. Is this the effect of drizzle? No, I don't think so. The blue overlaps at the top left and top right are in different directions????? But a very nice image. Chris No...my processing. That's why I said full resolution is no good. Heck....its no good. Over processed. I did not protect the stars properly. Not enough data for drizzle. Too much noise. I am working on the non drizzle now. I guess I should have just posted that one. Well see how it comes out Rodd Link to comment Share on other sites More sharing options...
MarsG76 Posted February 17, 2018 Share Posted February 17, 2018 Looks great.. Ha data will definitely emphasis the nebulous regions. Link to comment Share on other sites More sharing options...
ollypenrice Posted February 17, 2018 Share Posted February 17, 2018 It's good, but lots and lots of real luminance will pull out faint stuff that you won't know is there till you try it. Really. Olly Link to comment Share on other sites More sharing options...
Rodd Posted February 17, 2018 Author Share Posted February 17, 2018 3 hours ago, MarsG76 said: Looks great.. Ha data will definitely emphasis the nebulous regions. Thanks--but I think the non drizzle approach is the way to go--at least with the data I have. See below 4 hours ago, cfpendock said: That's nice Rodd. I especially like the soft detail of the core. But there seem to be some funny artefacts on the stars at full res. Is this the effect of drizzle? No, I don't think so. The blue overlaps at the top left and top right are in different directions????? But a very nice image. Chris Heres the non drizzle version. 5 hours ago, Filroden said: Another lovely image. I think lum data might add more than Ha and define more detail. Non drizzle version..... 2 hours ago, ollypenrice said: It's good, but lots and lots of real luminance will pull out faint stuff that you won't know is there till you try it. Really. Olly Thanks Olly--That's what I thought. I guess I need both really. But regardless, I think the regular integration is better, at least with the "limited" data. here's the non drizzled version. I think the drizzled version looks ok until full resolution is attempted. I usually process my images with full resolution in mind, but the size of the drizzled image makes it tough--its just so big. Link to comment Share on other sites More sharing options...
Demonperformer Posted February 18, 2018 Share Posted February 18, 2018 Well, I'm no processing expert, but having viewed the images side by side, the thing I most notice is the reduction of the yellow glow in the centre of the galaxy. And, on that basis, I would say that I prefer the second one. Whether that is a direct effect of losing the drizzle ... well, as I am not sure what drizzle (in this context!) is, I couldn't say. But I knows what I likes! Link to comment Share on other sites More sharing options...
Rodd Posted February 18, 2018 Author Share Posted February 18, 2018 1 hour ago, Demonperformer said: Well, I'm no processing expert, but having viewed the images side by side, the thing I most notice is the reduction of the yellow glow in the centre of the galaxy. And, on that basis, I would say that I prefer the second one. Whether that is a direct effect of losing the drizzle ... well, as I am not sure what drizzle (in this context!) is, I couldn't say. But I knows what I likes! Thanks DP.....I agree. For drizzle to work they say you need allot of subs and you have to dither. I do dither, but 18 subs is not really enough I don't think for the RGB channels. The synthetic lum was made out of 51 subs and that came out much better. Drizzle increases resolution and if undersampled can really round out your stars. The image is 4x the size. I am still trying to determine the difference between using drizzle and just zooming ion so the image is 4x bigger. I have used it successfully in the past and it is amazing how large the image gets and if clean how many details are visible. But it depends on the data...and in this case, it was a dead end. Rodd Link to comment Share on other sites More sharing options...
peter shah Posted February 18, 2018 Share Posted February 18, 2018 love the star colour... Link to comment Share on other sites More sharing options...
Rodd Posted February 18, 2018 Author Share Posted February 18, 2018 22 minutes ago, peter shah said: love the star colour... Thanks Peter. Rodd Link to comment Share on other sites More sharing options...
Rodd Posted February 18, 2018 Author Share Posted February 18, 2018 Gorann suggested I lift the faint outer regions around the galaxy a bit. Thanks, Gorann! A bit rough, though after I get the lum I expect it to smooth out a bit. Rodd Link to comment Share on other sites More sharing options...
wimvb Posted February 18, 2018 Share Posted February 18, 2018 That's a nice image. I like the first version of the regular image best. If you drizzle, you get an unevitable increase in noise. The data from four subs goes to make up one new image. So, in essence your 51 subframe image, has the same noise as a 12 sub image, per pixel. I've tried drizzle a couple of times, but was never impressed by the result. The final image size is just too large, and takes too much computing power. Btw, to lift the faint outer parts of the galaxy, (in PixInsight) try a very gentle exponential stretch with a mask that only reveals the part you wish to enhance. Use the range mask process to create the mask. Link to comment Share on other sites More sharing options...
Rodd Posted February 18, 2018 Author Share Posted February 18, 2018 1 hour ago, wimvb said: That's a nice image. I like the first version of the regular image best. If you drizzle, you get an unevitable increase in noise. The data from four subs goes to make up one new image. So, in essence your 51 subframe image, has the same noise as a 12 sub image, per pixel. I've tried drizzle a couple of times, but was never impressed by the result. The final image size is just too large, and takes too much computing power. Btw, to lift the faint outer parts of the galaxy, (in PixInsight) try a very gentle exponential stretch with a mask that only reveals the part you wish to enhance. Use the range mask process to create the mask. Thanks win. I forgot about exponential transformation. Nothing will compare to 5-6 hours lum. I am wondering if I should go 15 min instead of 10. Less frames though Rodd Link to comment Share on other sites More sharing options...
Stub Mandrel Posted February 18, 2018 Share Posted February 18, 2018 Drizzle comes into its own for planets, when s/n ratios are very high with thousands of frames. Link to comment Share on other sites More sharing options...
Rodd Posted February 18, 2018 Author Share Posted February 18, 2018 3 hours ago, wimvb said: That's a nice image. I like the first version of the regular image best. If you drizzle, you get an unevitable increase in noise. The data from four subs goes to make up one new image. So, in essence your 51 subframe image, has the same noise as a 12 sub image, per pixel. I've tried drizzle a couple of times, but was never impressed by the result. The final image size is just too large, and takes too much computing power. Btw, to lift the faint outer parts of the galaxy, (in PixInsight) try a very gentle exponential stretch with a mask that only reveals the part you wish to enhance. Use the range mask process to create the mask. Het Wim--great idea, thanks. I redid the lifting bit using your idea. Its not as dramatic but I think its cleaner. Still not the same as a bundle of lum (and then a lifting perhaps), but for now, I think its an improvement. Just working with the jpeg for now. Thanks again, Rodd Link to comment Share on other sites More sharing options...
kirkster501 Posted February 19, 2018 Share Posted February 19, 2018 Looks really nice. I have also been through the "what subs should I get" quandary for three years and have vast incomplete data sets. I tried getting RGB all at 1x1 without luminance. But what I have found through trial and error (and with no formal research at all)- and I don't actually know why - is that you still need 1x1 luminance channel for galaxies and nebula, even if you have the RGB in 1x1 and its fine detail. Why?????? Star clusters (open and globular) not so important to get luminance with the exception of the Pleiades and the gas cloud they are immersed in that benefits from 1x1 luminance. So, I have now converged on Luminance 1x1 and RGB 2x2 for my data collection on galaxies and nebulae. If you get a Ha channel to blend with the RGB then I do that 1x1 as well and if the Ha is to act as the "Luminance" channel for the project it must be 1x1. This is what I have found. Of course, this is what everyone told me in the first place. But being the engineer and ever the doubting Thomas, I thought I knew better. And I didn't.... Also, I cannot detect much difference in an image with a 1x1 luminance and RGB 2x2 from an image that has all channels 1x1. I don't actually understand why it is so important to grab as much 1x1 luminance as possible if the other channels are 1x1. But it most assuredly is. Link to comment Share on other sites More sharing options...
Rodd Posted February 19, 2018 Author Share Posted February 19, 2018 4 hours ago, kirkster501 said: Looks really nice. I have also been through the "what subs should I get" quandary for three years and have vast incomplete data sets. I tried getting RGB all at 1x1 without luminance. But what I have found through trial and error (and with no formal research at all)- and I don't actually know why - is that you still need 1x1 luminance channel for galaxies and nebula, even if you have the RGB in 1x1 and its fine detail. Why?????? Star clusters (open and globular) not so important to get luminance with the exception of the Pleiades and the gas cloud they are immersed in that benefits from 1x1 luminance. So, I have now converged on Luminance 1x1 and RGB 2x2 for my data collection on galaxies and nebulae. If you get a Ha channel to blend with the RGB then I do that 1x1 as well and if the Ha is to act as the "Luminance" channel for the project it must be 1x1. This is what I have found. Of course, this is what everyone told me in the first place. But being the engineer and ever the doubting Thomas, I thought I knew better. And I didn't.... Also, I cannot detect much difference in an image with a 1x1 luminance and RGB 2x2 from an image that has all channels 1x1. I don't actually understand why it is so important to grab as much 1x1 luminance as possible if the other channels are 1x1. But it most assuredly is. Thanks Kirkster. To tell the truth, I am usually pressed to see much of a difference between an LRGB and an RGB image. Maybe its because I don't collect enough Lum. For narrowband its different--I always see an improvement when I use Ha as a lum in a SHO palette image (but not in bicolor). Ha as lum in a broadband image screws with the color palette-especially the red. I have not experimented much with binning. The only time I binned 2x2 was with the C11 Edge one time. I did shoot a few subs 2x2 by mistake and was very surprised when the 10 min lum came out blown out. I could not figure out why...somehow a setting got changed. I collected 32 10 min lum subs last night (27 usuable). I was going to go for 15 min, but my sky could not handle it last night (it rarely can). We'll see if it helps--should pull the noise down at least. Thanks for the perspective on binning. I'll post the LRGB when completed (if there is a noticeable improvement) Rodd Link to comment Share on other sites More sharing options...
bluedandelion Posted February 19, 2018 Share Posted February 19, 2018 Love both versions at the scale I am seeing them on my monitor. A promising start! I use PI and in my experience about 20 subs (more is better) and careful tweaking of drizzle parameters can give excellent results. The default settings are fairly good. Once the drizzled image is scaled back to the original resolution, for me it almost always wins against regular integration. The background is more filled in, has more contrast with the main subject, and there is a subtle sharpening of details in the bright wispy areas. Also I find the LRGB workflow more rewarding in the end though it takes longer. I'd take a lot of lum, as Olly advised, with dither. Drizzle 2x and then rescale to the original. No need to drizzle the RGB. If you use PI, I also recommend deconvolution to bring out more detail. I recently posted an M65 and that's exactly the workflow I used. All the best! Ajay Link to comment Share on other sites More sharing options...
Rodd Posted February 19, 2018 Author Share Posted February 19, 2018 1 minute ago, bluedandelion said: Love both versions at the scale I am seeing them on my monitor. A promising start! I use PI and in my experience about 20 subs (more is better) and careful tweaking of drizzle parameters can give excellent results. The default settings are fairly good. Once the drizzled image is scaled back to the original resolution, for me it almost always wins against regular integration. The background is more filled in with more contrast with thye main sublject like and there is a subtle sharpening of details. Also I find the LRGB workflow more rewarding in the end. I'd take a lot of lum as Olly advised with dither. Drizzle 2x and then rescale to the original. No need to drizzle the RGB. If you use PI, I also recommend deconvolution to bring out more detail. I recently posted an M654 and that's exactly the workflow I used. All the best! Ajay Thanks Ajay. I never scaled the dither image back--maybe that is a critical step. I'll have to give that a try. Usually for me its a battle between drizzle and Mure Denoise--as they are not compatible. I like Mure Denoise. The other thing about drizzle is I am under the impression that it is best when undersampled. At 1.59 arcsec/pix with poor to fair seeing I don't think I am hugely undersampled here. Hard to say. I did not know you could mix drizzle and non drizzle. that's interesting. I do use deconvolution to sharpen. If I am not careful, I tend to over sharpen, so I am very mindful of not goinmg too far. Rodd Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.