Hi everybody, first of all many thanks for your replies and help, all very important! I would have liked to reply you sooner, but parents visit (very appreciated) filled the weekend.
And was it maybe the first new moon and clear sky weekend after 2 months of clouds...?! Yes of course, but don't think about it...
I want to:
1) put my comments to your posts
2) publish underlying data, subs and integrated images
3) sum up the next steps based on yout suggestions, both for me and for those who are at my stage
I had to choose whether to reply to single posts, but fragmenting my reply, or put all my comments together in a more homogeneus post; I chose the second. Hope everying will be clear enough.
- Refocusing frequency -> The trade-off is of course how much time you want to loose in refocusing, so I settled on 2h and no temperature trigger using 5s exposures at Unity gain (111 on ASI183) and f5.6 on my RC8 carbon. I don't see evident changes in FWHM before and after refocusing in this way, as you can see in "FWHM" attachement. That's why I think it should be sufficient. The Elf manual focusing also could suggests refocusing may not be that crucial, given its great results.
Anyway, there's actually no reason not to trigger on temperature if one have an electronic focuser, so I will put it back in my refocusing procedure.
- Decovolution -> M51 is deconvolved with SPF calculated on actual image stars. I stop encreasing deconvolution strenght when I see stars start to "melt" with each other, creating sort of filaments between them. It's the same for you?
All of you have underlined that deconvolution IS mandatory on RC, so no more doubts on keeping this processing step.
- Denoising -> I found Jon Rista method to be the best I tried so far, especially in terms of repeatibility and effectiveness. I see softening coming in expecially when denoising after stretching, but I think careful (patient...) tuning of TGV and MMT parameters can avoid it, so I'll try to improve on this.
- Sharpening -> Weak on this point... I see visible effects only when using Unsharp Mask (Pixinsight), but it degrades overall image quality quite soon. So I don't have clever methods on this point actually.
- Seeing, FWHM and Eccentricity -> as Bokemon said, high frequency imaging could help to assess it. I don't actually know other ways to do it, other than beeing confident that all my set-up is perfectly configured and therefore seeing is measured by FWHM.
Actually I can't understand if my FWHT is seeing limited or not, but I guess not. See FWHM Attachment: avarage is 4.2", but my points are very scattered, far too much. Why, in your opinion? These are 2 minutes subs at unity gain (111 on the ASI183 native driver) from 1 night out of 2 of LUM on the M51.
Nothing to do with The Elf FWHM plot, much more regular and lower valued.
Eccentricity suffers the same problem, see the graph attached. Again, avarage quite reasonable on 0.45, but scattering all around. I guess my signal is too weak and I need to encrease exposure and/or gain. Maybe this will stabilize these values.
The Integrated LUM (attached, see point 2) of M51 has 4.2" FWHM as measured by Subframe Selector and Ecc 0.22 (so low?!), quite in accordance with FWHMEccentricity tool which gives 4.39" and 0.21. I often get very different results from this two tools, but that's another story and not the case here.
For completeness I attach also SNR and Stars from the same set of subs.
- Collimation -> I understood from many of you that perfect collimation is much more important than I hoped. I didn't choose a Newtonian (even though I saw the best pictures at 8" from these scopes) to avoid recollimating it every time I mount/dismount/travel, but this point come back to be crucial also with RC. I found that atmospheric instability (seeing) degrade defocused star patterns so much that evaluations based on these moving and vibrant patterns becomes (to me) very very unprecise; extending exp times of course stabilizes the pattern, but rings become less visible and the whole processo gets much longer... But I'm probably wrong, given you all do it. Does an artificial star maybe solve this problem completely?
As for FWHM and Eccentricity Contour plots, and referring to The Elf plots, I struggle to get them with the RC8 since my subs have too few stars, as PixInsight tells me... I probably need to abandon 60s and 120s exposures and go for 180s or 240s, maybe increasing gain also from Unity gain which I currently use.
Last, a friend on Facebook made me understand CCDInspector is absolutely necessary. There are other free tools around, but from my trials none of them is actually as straight forward as it is. Since we are saying that collimation must be as perfect as possibile it could be that this tools is actually mandatory...
- Focal Reducer -> As Dan reminds me, I'll abandon 0.67x reduction factor, going for 70mm backspacing (from flange) and 0.72x.
When I tried to image without Focal Reducer I found the same softness in my image, so the FR itself doesn't not play a big role in the overall softness.
- Image scale -> both M51 and Needle are shot at 0.45"/px. Do you think such strong oversampling contribute a lot to the final image softness? The answer is probably yes, but I'm still hesitant to accept it, since oversampling should add (useless but not harmful) information... And enlarging pixel size is apparetly no that easy with commercial CMOS sensors, coming all of them with pretty small pixels. Anyway, future 2x2 binning will go to 4.8um pixel size and 0.9"/px (with focal reducer), which is about 1/3 of avarage 3" seeing, so should be fine.
- OAG -> Large majority of you agree on OAG instead of guide scope. I'm currently guiding at 0.55"/px on my small OAG (8x8mm), so 5.5 times an avarage seeing of 3" and near the 0.45"/px of the main camera. All these image scales are too low. I was temped by a 70/400 guide scope, which should do the job if rigidly secured to OTA and run at 1,5"/px with the ASI290. It would have some advange, but consensus goes for the ASI174 and Celestron 12x12mm OAG (or similar), which is another investment by the way...
- Reference Point -> The Elf, your images set the best practices to follow, at least in this thread but I think in general. I guess you're quite "spot on", showing the improvements to do both on RC8 and RC10. It looks like your color layer is much softer than the LUM, because of those colored halos around stars and probabily all over the image. Don't you think? Sharpening color layer would probably boost perceived details even more.
Subs are here, together with master lights: https://drive.google..._ZM?usp=sharing. I put also darks and reference frames I used, just in case.
Upload is in progress, it will finish in about 4 hours.
Many thanks in advance to all of you willing to give them a look! (In case you use darks, "_A_" are for LUM and "_B_" are for COL subs).
Coming in the next post...
ps: how to paste images between lines of text, here on Cloudynights?
Edited by davide.cattani, 14 June 2021 - 09:55 AM.