Ritchey–Chrétien enjoyers thread

[digital nexus]

No I have those

Do those go first

?

[tall summit]

Yep

[digital nexus]

I need both right?

[tall summit]

Def 2inch one

[digital nexus]

I’m not gonna bother with the tilt plate until I see how bad the tilt is

[tall summit]

Honestly, I got it, but ended up not needing it

[digital nexus]

What’s all these golden knobs do

[oblique sun]

i assume one is to lock the focuser in place and the second idk

[digital nexus]

uhhh I don’t fit

[oblique sun]

its m54 i think

[tight lodge]

For a sec I thought you broke it💀

[digital nexus]

It’s m63

I using m54

I don’t have an m63 thing

[oblique sun]

time to adapt

[tall summit]

My focuser looks to be different from yours. I have two knobs, one is to fix it in place once focused, one controls how tight is the "grip"

[digital nexus]

Is this normal

[tall summit]

Nope

Mine has zero play anywhere in the train

[digital nexus]

Bruh why is it doing that then

@harsh matrix @frosty shard please hhelp

[harsh matrix]

I dont have time rn

Sorry

[digital nexus]

Its the mirror moving

Or the mirror is moving too

Is there a way to tighten the mirror?

[tall summit]

Maybe counter screws are loose?

[digital nexus]

The lock screws were loose

[tall summit]

Push or pull

[digital nexus]

I tightened them

[digital nexus]

What are counter screws

Does tightening the locking screws mess with Collimation?

[tall summit]

Which ones

On a side of ota?

[digital nexus]

Big one

[tall summit]

These ARE the primary collimation screws

[digital nexus]

not according to this tho?

[tall summit]

The are called "lock screws", sorry

Its a push-pull kind of thing

[digital nexus]

So Collimation is definitely messed up

[tall summit]

You loosen the big one as you need, then tighten the small one so it doesnt move back past that point

[tall summit]

Right now probably yes

[digital nexus]

Well it was just shipped from Texas to Pennsylvania. I was expecting that anyways

[tall summit]

What tools do you have?

Cheshire?

[digital nexus]

none

I thought you didn’t need any for the DSI method

[tall summit]

Well, you want to get as close as possible before going under stars and doing dsi

[digital nexus]

I have this though

[digital nexus]

I have 1.25” collimation cap….

[tall summit]

Thats a good

[digital nexus]

I never used anything else for my dob

[tall summit]

At least make sure all circles are as concentric and as centered as possible

Then do dsi

[digital nexus]

I can borrow a Collimation Hotech if I need to

[tall summit]

You might want to invest 30-40$ into making that tool I mentioned, going forward.

[digital nexus]

Making a Cheshire?

Would rather buy one

[tall summit]

Sure, if you dont mind dropping some $$ I would highly recomend OCAL

I have a diy clone, makes it very convinient to allign and center things

[digital nexus]

I do mind. How much

[tall summit]

Clone will be about 30$ in amazon parts, original I believe is around 200$

[digital nexus]

What is it

[tall summit]

https://youtu.be/cVVXd7KFH1w?si=oi9NGW7KRv_l-xlQ

Its basically an electronic cheshire

[digital nexus]

You made one?

[tall summit]

Yeah, want a link?

See how you do fist with what hou have, collimation cap also helps a lot with mechanical alignment

[digital nexus]

On how to make one? Sure

[tall summit]

https://youtu.be/iCMgB6f-6ls?si=UUkDK0EUmEF_1CH4

[harsh matrix]

That's wrong lol

[digital nexus]

well when i tightened the big screw it stopped bouncing the focusder

[harsh matrix]

Probably because the mirror carrier bottomed out on the small screws.

You essentially pulled the mirror carrer into the locking screws

[digital nexus]

I dont understand

[digital nexus]

So did i fixed it the wrong way?

[harsh matrix]

Yeah

[digital nexus]

What should i have done

[harsh matrix]

Since the big screws will have messed up collimation.

Tighten the small screws

[digital nexus]

collimation was already probably messed up

[harsh matrix]

Yeah I mean

I suspect that too

[digital nexus]

so i expected to have to do it anyways

so nothing i need to freak out over

so

Locking Screw - Small
Collimation Screw - Large

[harsh matrix]

It seems like this guy "collimated it" once by loosening the collimation screws to the point where the mirror could move on its own.

[alpine flame]

pretty sure thats the airy disk there

thats a sign of high quality optics

[high aspen]

@digital nexus are you using the rc to imag yer?

[digital nexus]

Can’t

[tight lodge]

That's why I prefer SkyWatcher over GSO

[tight lodge]

yay or nay?

[frosty shard]

Alright due to a change of plans we're not oneshotting the Andromeda Galaxy

We're oneshotting the Pleiades

The first of 12 panels tonight

[tight lodge]

Uuuuhm

[frosty shard]

honse (how many min)

[tight lodge]

3.3h

[frosty shard]

update: I don't think I can get to all the panels tonight

[quartz meadow]

other one is umm rotation??

[frosty shard]

Pleiades nuts

[frosty shard]

@candid flame you may like this

However, you might notice all my frames are marred with tilt :(

[alpine flame]

Brother doesnt skywatcher use GSO mirrors😭

Wait only secondaries i think

[high aspen]

why :(

[digital nexus]

Missing adapter

[high aspen]

buh

[candid flame]

Goofy Ahh mosaic

Also no signal = no bitches

[high aspen]

tilt gang

[digital nexus]

I need M63 to M48. I have M64 to M48

[high aspen]

oh 😭

[tight lodge]

Synta😅

[digital nexus]

I was suggested to buy a gate handle from Lowe’s and screw it into my RC6 lol

[high aspen]

someone eated it

[tight lodge]

@harsh matrix Your fish head nebula was quite high rez. Did you also got this

[fringe delta]

is that 3 stars or what

[high aspen]

truangley

[tall summit]

Hmm, the end of your focuser is M63? My adapters are all m56

[high aspen]

shiny

[tight lodge]

I think.

[tight lodge]

They almost look like gravitational lensing

[fringe delta]

yeah it does look like that a bit

[silver ibex]

The new rig is getting closer with every day

I found a guy interested in my mount and another interested in the SCT reducer. The CEM70 guy also finally replied to my offer and we've decided on 2200€ which is nice

Still waiting for the 10" RC dude to reply, there's also another CEM70 which might be better but the guy has yet to answer to me

[tall summit]

My 6" didn't resolve this

[high aspen]

barlow up >:3

[tight lodge]

What was your guiding RMS?

[tall summit]

All over between 0.5 to 1

[tight lodge]

😅

[tall summit]

I am contemplating taking the reducer off and embacing full FL at F9. When I started I had unmodded dlsr, so the difference between f9 and f6 was great

[tall summit]

The best I had with my CG4 was 0.38-39. But it takes an extra clear and crispy nights to get to those numbers

[high aspen]

i wanna see how that turns out ngl

bc i plan on doing that

[tall summit]

Waiting for my 60mm F4 guide scope to arrive

[tight lodge]

My guiding usually setters between 0.34 and 0.4". I can't get below 0.34" because that's half the diffraction limit of the telescope. So already is pushing the limits

[tight lodge]

OAG

No discussion

[tall summit]

Might come down to that, I'll see how that scope works out first. I kind of need it as a counterweight too, to compensate for the heavy back. My limit I think is 0.8x, without the reducer, so I am usually below it

hah, I do see it in my subs:))

[high aspen]

a oki

[high aspen]

make a ff for the guidescope and have it as a widefield imager

[tall summit]

@high aspen what is FF

@tight lodge I feel redeemed now lol

[tight lodge]

Maybe your star stretch was too aggressive and you bloated the stars

[high aspen]

feild flatner

[tall summit]

That, and I definitely didn't cull my subs aggressively enough to preserve this detail I didn't know about

[tall summit]

nah, It would make a shitty imaging scope I am sure:) I do have my old modded Canon laying around, kinda want to get a lens for it by summer to try some milky way shots

[digital nexus]

yea the focuser is m63. I wish it was m64 because my Askar adapter is

[tall summit]

Is it original, or aftermarket? I mean focuser

[digital nexus]

After market

[tall summit]

Oh, right, non carbon versions come with Crawford style usually

[digital nexus]

https://www.teleskop-express.de/en/telescope-accessories-5/focusers-adapters-motorfocus-296/focusers-and-micro-transmission-92/ts-optics-2-5-rack-and-pinion-focuser-with-m90x1-connection-and-360-rotation-6970

[tall summit]

feels robust

[digital nexus]

It’s heavy and solid that’s for sure

[tall summit]

Remind me, what ota are you upgrading from?

[digital nexus]

Not really upgrading, I’m keeping my other scope

It’s an Askar 71f

[tall summit]

thats a nice FL spectrum you can cover now

[digital nexus]

I can l do 200mm if I wanted to with my DSLR as well. Although it’s a vintage lens and causes some crazy stars

[tall summit]

490, 1370 and 920 with the reducer

very sweet

[high aspen]

[high aspen]

im gonna get a carbonfiber one :3

[dense quiver]

https://tenor.com/view/star-gif-16789607377123159596

[tight lodge]

Indeed I need more. But I have to split the night.
I also got this beauty

[dense quiver]

oo nice

[high aspen]

purple

[harsh matrix]

I don't believe it is in my fov

It's definitely not present in the final image

[tight lodge]

Did you post it somwhere?

[harsh matrix]

Yes

[tight lodge]

oh i found it, the object is just slightly out of frame 😭

[frosty shard]

NINA live stacking 🅱️ like

[tall summit]

Holy gradient

[frosty shard]

The flat corrects fine in SIRIL, it's the plugin's fault I think

[harsh matrix]

is it calibrating the lights with a master bias?

[frosty shard]

Ope

[harsh matrix]

that explains that I guess

[frosty shard]

Thanks for the quick sanity check

[tight lodge]

@harsh matrix , Is this comparable with what your 6" got last year? #1326833769495269376 message

[harsh matrix]

not really

that's got a long ways to go before I can get a better idea

[tight lodge]

Oh, I'm not talking about SNR, I'm just referring to the sharpness

[frosty shard]

If his dataset is available on #1021002743080558642 then you can estimate the PSF and account for their pixel scale

[harsh matrix]

it's not

[tight lodge]

De data is not available just yet. I need mooore

[silver ibex]

Damn I made an ad asking for RCs and this guy apparently had one with adaptive optics???

Also 3" focuser esatto and Astrophysics corrector, I'm afraid this costs 8 or 9 kidneys

[high aspen]

bwaa

[tight lodge]

How big the aperture?

[high aspen]

=1mm

[tight lodge]

Guess who got his flat pannel to open the full 270°😂

[silver ibex]

10"

[tight lodge]

Oooh. I know what AO he has. It's a Starlight Xpress. Something like this most likely

[silver ibex]

Yes, it's obviously not a deformable mirror

[tight lodge]

at 10" there's no reason to have one

[silver ibex]

At this point it's not really for combating seeing, more like an extra layer of guiding

But yeah normal guiding is absolutely fine

I'll just ask for the tube

[harsh matrix]

bruh

[harsh matrix]

I'm not talking about SNR either

Its hard to judge the sharpness of the object without sufficient integration imo

[harsh matrix]

If you want in on a little secret, I can tell by eye that his stars are bigger than those from my 8

Not by a whole lot, but by enough that I can see it

[tight lodge]

did i mention that the stretch was without star mask or starX? i just stretched everything at once out of curiosity

i just did a linear fit, channel combination, stretch and played a bit with the curves

[high aspen]

i think even without nb theres not much stars by the horsehead

[tight lodge]

no, but stretching them at the same time as the nebula can bloat the stars

[high aspen]

ohh

yea can

and blows them easily

[frosty shard]

I got locked out of my apartment after taking down my scope

Waited 2.5 hours for someone to open the door

[frosty shard]

I think I might have to do a more traditional SHO process for this region

that, and get more data

[silver ibex]

Ignore the terrible gradient, not my data as it's clear from the spikes, but my process

[silver ibex]

When you do get more data I HAVE to try it

[digital nexus]

ok RC6 update

1. I tried my new mount last night, got around 0.9-0.6”. I need to tune the setting a bit, hopefully can get it consistent at least. This is good for my Askar 71f, but clearly not the RC6.

2. Im working on getting an OAG from a club member. Unsure when but that’ll solve my guiding issue.

3. My RC6 focuser ends in M63. I have M64. So the previous owner of the RC6 is sending me a spare M63-M48 adapter to me. That’ll let me actually use it.

4. My rig uses too much power, too much amps for my sv241. So I need to buy another power supply to finish powering my rig.

So RC6 is unusable until I get an OAG and that M63 adapter. Great

[crisp flower]

Carried by noisex BUT i really Like it so far

[harsh matrix]

It would have had to have been some sort of miracle if it was all perfect on the first night

Nothing in this hobby is easy

Also that is essentially a second rig being built from scratch

That makes the road even more bumpy than it already was to begin with.

That's the frustrating thing about the hobby

But once you've stuck through it all, ironed out the problems with both rigs, and you nail down the small stuff, it will be incredibly rewarding.

Not just rewarding, but fulfilling as well.

[digital nexus]

well I havent had any issue with the RC6 yet. I havent done anything lol

Last night was just my tried and true fractor

new mount was the issue

I bought a 50mm guide scope. so if I dont get that OAG, il hopefully have the Ragdoll settings down where I can get at most 0.5" Then the RC6 can finally be used

[tight lodge]

Good subs for a 130PDS?

[frosty shard]

I'm trying to finish my Pleiades mosaic tonight at a dark site (well, in the sense that I get all panels, not necessarily getting all integration) but I can post the data I currently have and I'll definitely try to get more narrowband from home

[high aspen]

yipee

[vapid patio]

@digital nexus btw i have been trained by veloren for about a year so feel free to ping me im always able to help

i know a decent bit

[tall summit]

What refresh rate did uou have for guide cam?

[digital nexus]

I don’t know

My guide cam sucks though. It has micro disconnections

[tall summit]

Meant exposure, sorry. I think harmonic mounts want 1sec or even less

[digital nexus]

1.5 right now i think

proxisky doesnt recommend 0.5 for ragdolls

1 wasnt working too well

[tight lodge]

This has to be one of the sharpest images I took

[frosty shard]

All panels of the Pleiades have been acquired

I cannot wait to assemble them despite the low integration

[tight lodge]

Cursed Honse (2.5h)

[harsh matrix]

[frosty shard]

We got clouded out

[tight lodge]

Don't laugh at a honse next time😂

[high aspen]

what the

[frosty shard]

I'm about to process the remaining 7 panels of the Pleiades

[tight lodge]

Rate my honse

[frosty shard]

Extremely solid honse/10

Unfortunately I already picked @harsh matrix's image to serve as my browser background on one of my PCs

[tight lodge]

Foe 3.3h is not that bad. Imma grind it for the rest of the winter

[frosty shard]

I am tired of dim objects (IC 10) so I'm going to not grind and shoot easy objects lol

I can use your honse as the new tab background on my phone actually

[tight lodge]

I got absolutely sick of M31. One night and I got 2 gorgeous images rather than grinding on a target that I hate by default

[frosty shard]

Well I guess the Pleiades is kind of a grind as a 12 panel mosaic

[tight lodge]

Perfect fit😂

[frosty shard]

Absolutely

Excellent phone wallpaper material

[tight lodge]

Unintentionally I created a really nice phone wallpaper. Thank you Alnitak

[frosty shard]

Oh my god

I apologize for sending that as a JPEG, but it's 9999x7777 pixels

I think my sensor dewed up because Alcyone, Atlas, and Pleione have bright halos

[tight lodge]

Moooore

[frosty shard]

I am uploading the data if you want to have a crack at it

[tight lodge]

Later tonight tho

[frosty shard]

Warning: it is 1 GB of FITS data

in one file

[tight lodge]

Did you run MSG on it?

[frosty shard]

I don't think that functionality exists in SIRIL

[tight lodge]

Ah. I forgot about that. Imm a have a crack at it tonight and see how badly I ruin it

[quartz meadow]

honse but has been put under the hydraulic press

[tight lodge]

https://tenor.com/view/suboptimal-gif-2694570377890855135

[high aspen]

fish

[stiff mason]

How is it still sharp

[tight lodge]

Because the original was sharp

[stiff mason]

Stop making me regret buying a newtonian 😭

Well technically my mount is the bottle neck for my newt so it wouldn't matter anyway but like still

[tight lodge]

Why? That was shot with a newtonian

[high aspen]

your noot is just too cool clearly!

[tall summit]

I ordered a star diagonal and somewhat quality 4element barlow, will be actually looking through my RC6 for the first time soon

excited

am I the only one who never observed through RC visually?

[frosty shard]

I think most people only use their RCs for imaging

I am a big exception

[tall summit]

coming along nicely!

[tall summit]

I want to try to barlow it up and image jupiter using my phone

[frosty shard]

I was gonna say, you don't need to Barlow it for all practical visual observing purposes, but it will be very helpful for imaging

[sinful sapphire]

You can use your 2600

[tall summit]

not sure how i would mount it to digonal/barlow

As is, its too tiny for any practical imaging

[sinful sapphire]

Nosepiece

[tall summit]

hmmm

[sinful sapphire]

You screw it onto the camera and then attach it like you would an eyepiece

The imx571 is essentially just a bigger imx533 which is still a somewhat popular pick for planets

[tall summit]

damn, didnt occur to me at all

[sinful sapphire]

Will be better than any phone image

[tall summit]

is this what i need? https://www.amazon.com/Versatile-Adapters-Threaded-Standard-Accessories/dp/B0CYPLDFTL/ref=sr_1_4?

[sinful sapphire]

1.25'' to T2

[tall summit]

found this one

https://www.amazon.com/Acouto-Eyepiece-Adapter-Suitable-Telescope/dp/B07L8SRHH9/ref=sr_1_7_sspa?

@sinful sapphire this should do it, right?

[sinful sapphire]

Yeah

[tall summit]

oh maan, I am so happy I mentioned it, I was going to dust off my cell phone adapter

[sinful sapphire]

also note that this isnt good for DSO because it will cause vignetting on the corners

[tall summit]

yeaah

[sinful sapphire]

which is irrelevant for planetary but relevant for deep sky lol

[tall summit]

I did this with my cell phone, before getting in astro and knowing what I am doing

[frosty shard]

We interrupt your regularly scheduled programming to farm aura

[frosty shard]

@tight lodge return of the laser

[tall summit]

was proud of it as f*ck:))

[frosty shard]

I think you actually have the Galilean moons resolved as disks, which is crazy

[tall summit]

I always thought I just over-blew them. I did it with my 8" Dob and crappy 2x barlow

and shot with S25 ultra

[tall summit]

Order placed. Thanks a lot for putting me on to this, def opens up new possibilities

[stiff mason]

Then how good spikes

[tight lodge]

a newtonian can't have good spikes?

[stiff mason]

No but like did you upgrade them or sum

Cause mine look trash

Also you dont have that mirror glare which I'm assuming you added a mirror mask as well

[digital nexus]

How is it visually

I plan on using mine for visual a bit. Here and there

[frosty shard]

I haven't used the 6" but I actually prefer the 8" RC to an 8" SCT for deep sky

[digital nexus]

What about compared to a dob

[frosty shard]

It's not the best choice for planetary but that doesn't mean it won't deliver a sharp image

[frosty shard]

Dobs will be capable of a wider field, but an RC is actually the next best thing. I drop mine down to 39x for wide fields that are super consistent

[digital nexus]

I’m doing it for planetary and lunar only really

Visually

[frosty shard]

It's still pretty good for lunar and planetary

[digital nexus]

Just cause it can mount my star tracker better

[tall summit]

goto and tracking def go a long way making observation expirience better

[frosty shard]

The large CO means it won't be the ideal choice, but it will still gather plenty of light and deliver good resolution

It just won't have the high contrast of, say, an f/15 Mak or Dob

[high aspen]

f20 gregorian :>

gregorians on axis have like a 0.01um spot size

its crazy

and obv you got a smol co

you can make the co extemely smol with a greg

[tight lodge]

Oh yes. I do have both, mirror mask and CNC machined secondary holder.

Spider

[frosty shard]

But can you do it at f/8?

[high aspen]

i am too scared to try

[frosty shard]

Gregorians can get pretty long

There's a reason why Cassegrains are more popular: you don't need the secondary to be as far away from the primary as with a Cassegrain

Since you don't need to bring the light to focus before the secondary

[high aspen]

f 0.5 primary

[frosty shard]

That does work

Though whether it's practical to construct is a different story...

[high aspen]

and you dont have to use elliptical mirors

[frosty shard]

well the hyperbolic mirrors of classical Cassegrains and RCs are pretty hard to test

[high aspen]

sads

[frosty shard]

tl;dr: "Hubble moment"

Poor null corrector usage or manufacturing is not rare

[high aspen]

[frosty shard]

That's actually why ATMs would rather make an aplanantic Gregorian instead of an RC. The elliptical mirrors are easier to test (and IIRC they don't deviate much from hyperbolic)

I still need to figure out what the equations are for an aplanatic Gregorian

[high aspen]

oh

thats kinda cool actually

also, im now gunna try the quick greggy

i wonder if i could make it better somehow

[frosty shard]

Are you building a classical or aplanatic Gregorian?

[high aspen]

one with parabolic prim and elliptical secondary

which im guessing is classic

[frosty shard]

Yeah, with aplanatic you reduce the conic constants of both. I think this (see 8.2.3) should be able to get you the right values, but note the signs are different for Gregorians

[high aspen]

i managed to get it decent by eye

sub 0.04 is quite good

its only got a 300mm effl

[stiff mason]

I going to 3d print mine both the mask and the secondary

[high aspen]

hmmm

elliptical primary

hyperbolic secondsry

[frosty shard]

Since we're back on cloudy weather I need to figure out what unity gain on my camera actually is...
https://aifatelescope.github.io/datared/camera-charact.html

[candid flame]

Sinusoidal primary when

[high aspen]

whas

[candid flame]

You heard me

[high aspen]

https://tenor.com/view/cat-cat-screaming-machin3girl-gif-22623519

[candid flame]

Lemme get that Weierstrass mirror just to see what happens

[frosty shard]

@harsh matrix what mode do you use for your Touptek camera, and what are your gain/offset settings?

I just realized I am absolutely not using optimal settings for my camera

Which might explain why my data looks pretty noisy

[harsh matrix]

i use low noise readout mode at i use 100 gain (lowest i can set it) and I use an offset of 50

[high aspen]

50 offset is diabollical

[frosty shard]

Well it depends on what the offset setting corresponds to in terms of offset ADU

I ended up realizing the ATR533M, despite what I was initially led to believe, does not treat many of the settings the same as the ATR2600M

I should be using a gain around 316 or so

Gain 100 for me is like 2.1 electrons per ADU

[frosty shard]

It looks like unity gain is around 237 with my settings. I need to do more testing but oof, that is a lot of signal I've been leaving on the table

[tall summit]

New guide scope is in!

[digital nexus]

which is it

[tall summit]

@digital nexus Astromania 60mm 240mm fl

[digital nexus]

I just got a 50mm guide scope

hope it works well enough

[tall summit]

I was guiding with 30mm f4 scope before, so I am sure it will

[digital nexus]

my mount has some large spikes, and too high with my 30mm. proxisky recommended larger guide scope so im hoping this one will solve thjat

[tall summit]

Riight, mine is not harmonic

[digital nexus]

yea ik

[tall summit]

How is your mount expirience so far

[digital nexus]

meh

0.9-0.7 RMS"

Having high peaks of 4" or 3" though

not liking that

[tall summit]

Hopefully larger scope will improve things for you

I am really qurious if i'll see any change

No clear weather in sight

[tight lodge]

The honse is doing well on Astrobin

[frosty shard]

Link? Idk if I follow you

[tight lodge]

https://www.astrobin.com/n2kdx2/

[frosty shard]

@harsh matrix you may find this useful

[tight lodge]

This has to be the most detailed technical drawing of an RCT

[frosty shard]

It's the GSO RC8 design spec

[tight lodge]

Basically every RCT on the market

[frosty shard]

Well, this one doesn't have the baffle extension

[frosty shard]

1 hour 30 minutes on Orion from Sunday night

[high aspen]

onion!

[frosty shard]

Your flats weren't calibrating in high full well mode, right? Could that have been the reflections from the reducer?

[harsh matrix]

Correct and I dont think high fw mode works

[frosty shard]

What else happens when you use it? I have been using it this whole time

[harsh matrix]

Dunno

[harsh matrix]

need to paint the front of that reducer

im so nervous about it though 😭

[frosty shard]

Did you get the paint yet?

[harsh matrix]

yes

its how i go about painting it that makes me nervous

[frosty shard]

What's your strategy gonna be?

[harsh matrix]

im going to cover the lens somehow and use a thin brush to brush the paint on

i need to make sure i dont drip paint too

it's been years since ive painted anything so i dont have the slightest clue about how to go about this

[tight lodge]

I just realized. I have the ability to 3D print the whole thing

[frosty shard]

Bro can 3D print a mirror

[tight lodge]

I can not 3D print mirrors. But I can 3D print the whole tube

Aliexpress got my back to when it comes to mirrors

[stiff mason]

I need to do some astro it's been like 1 and a half months

But these stupid clouds wont let me

[harsh matrix]

guys

we're basically stealing a 585MM Pro

[tight lodge]

Do it

[harsh matrix]

it's already done

[tight lodge]

You gonna pair it with the RC8?

[harsh matrix]

no

i would be over sampled to hell and back

that would not be fun

it will pair fantastically with the quatt though

[versed roost]

where??

[frosty shard]

He got a big discount as compensation for the AM5N troubles

[harsh matrix]

looks like it may have paid off finally

[versed roost]

does the mount work now?

[harsh matrix]

yes

very well

[versed roost]

goated

[harsh matrix]

the new harmonic drive has very smooth periodic error

with a peak to peak of 10"

that beats all prior drives in both aspects

easily the best harmonic I have had in one of these mounts

it beats out a lot of other harmonics too

[stiff mason]

Yikes my friend wanted one so bad he couldn't find one at all

[tight lodge]

#1429897878897168405 message:AwkwardSmile:

[tight lodge]

@frosty shard do you want a challenge?

[digital nexus]

What reducer do you guys recommend for my rc6

I might get one eventually

[quartz meadow]

I mean the only thing you need to look out for, is that the reducer itself is not reflective

[silver ibex]

The Apex is interesting, @harsh matrix knows more about it

[tight lodge]

Costs more than the RC6💀

[digital nexus]

is there any that doesn’t cost twice what I paid for my scope….

[tight lodge]

Newtonian

[digital nexus]

I doubt I’d get an imaging newt for $400

[tight lodge]

Out of the box? No. With potential? Definitely

Brandon is still ignoring me

[digital nexus]

I spent $400 on my RC6, and it can basically be used for imaging off the bat. So an imaging newt would be more to get working properly?

[tight lodge]

Yes. And you will be better sampled

[digital nexus]

What imaging newt is the recommendation

With my 533 i thought i was sampled pretty good with my RC6

[tight lodge]

What's your sampling?

[digital nexus]

Nvm

[tight lodge]

😅

My 130PDS capped at 2.1"

1.9 was absolutely insane ngl

[digital nexus]

How bad is my sampling

I’m a refractor guy, Ive never nice worried about sampling

[tight lodge]

Do you have any data taken with it yet?

[digital nexus]

No I can’t even get my camera attached to it yet lol

I have an M64 adapter and need M63 smh

Getting that Friday. Then I can collimate it and then hope I can guide it properly

I might wait until I got my Ragdoll guiding good with my new guide scope.

Or il have to wait for an OAG

[tight lodge]

This is my sampling

[digital nexus]

Drizzle helps with over sampling right?

[tight lodge]

Undersampled

[digital nexus]

The RC6 is commonly used with the 533 so I’m not worried.

Oh

Well I’m not worried

[tight lodge]

Binning works with oversampling. However, CMOS doesn't do hardware binning, is more like downscaling

[digital nexus]

Yea il have to do some experimentation with binning.

[tight lodge]

So you can downscale your images in post-processing and get it fixed to some extent

This is what I got from binning my sensor

Very slight increase in SNR, but almost negligible

[candid flame]

the increase will always look negligible if you autostretch them separately, because it's using the noise characteristics to stretch the image

[tight lodge]

Then I should apply one STF to both?

That didn't strike me in the head💀

[candid flame]

You should just apply the same stretch at least

The binned one is stretched harder, making the noise look worse rn

[tall summit]

Apertura 0.67x reducer

[candid flame]

Or alternatively apply separate autostretches, but think of SNR as contrast instead of looking at the noise

[tight lodge]

The contrast is what I was looking for. In the bin 2x2, the dust appears ever so slightly darker

[candid flame]

Only slightly you think? If that doesn't impact sharpness too much I'm binning that any day

[tight lodge]

ok, i re did them and i applied the STF values from one to hte other so they get stretced to the same value

[candid flame]

are you sure those stretches are identical? the binned one looks much darker. Also, you won't be able to see the difference properly unless you zoom in to 1:1 or more

[tight lodge]

Here are the files

they are not sharp so keep that in mind. that night was pure junk that i decided to experiment with Bining rather than imaging

[stiff mason]

Well not anymore but i got my 6 inch for about that much

Now that same scope is over 600

[candid flame]

this is not from an RC is it?

[tight lodge]

nah

just a newt on a windy night

[candid flame]

i could tell because it's almost perfectly sampled 😂

[tight lodge]

oooooo they not. the ones i sent you are hideous

[candid flame]

how is it hideous?

[tight lodge]

that's not even half of what my OTA can do

[candid flame]

???

you fwhm should ideally be 4px and in this one it's 4.8px, so you're slightly oversampled

[high aspen]

wha

my fwhm is usually 3.8

so thats good????

[candid flame]

perfect 🔥

if it's in pixels and not arcsec

[high aspen]

yea

[tight lodge]

check this sub

[candid flame]

this is identical streched zoomed in enough for display problems not to occur

oh wow i think it's discord messing with it

[high aspen]

yea

[tight lodge]

The difference is so small that is hard to see it

[digital nexus]

You guys think this will be enough for my Rc6?

[high aspen]

left snr is good

[candid flame]

that's because it's identical strech, so now you should be looking at noise and not contrast

[high aspen]

ya

[candid flame]

bin2 is defo looking better imo

tho it's much easier to see it the other way where you look at contrast tbf

[candid flame]

turns out it's not discord, just a human brain moment... i swapped them around here

[candid flame]

SHEEESH

[tight lodge]

that has to be the sharpest i ever got

1.9"

[candid flame]

doesn't change the fact that sampling was good in the other image tho

it just had worse seeing

in fact, the sharp one is quite undersampled

[tight lodge]

this is where i am rn with hte sampling

[candid flame]

divide by 2 and 3 calculator 🔥 🔥 🔥

purely theoretically tho, your sampling should be 4x the seeing to not lose any information, tho the last 4x instead of 2-3x doesn't contain that much information anyways

basically you want to be slightly undersampled, because you get more signal per pixel, for a relatively small loss of sharpness

[tight lodge]

yeah, but at the same time, its good to be able to resolve very small details in the image

[candid flame]

ofc, but the difference in resolving power between 3x and 4x is almost negligible

that's why your divide by 2-3 calculator doesn't divide by 4 essentially

also that's why i'm saying your sharp image is undersampled, because an fwhm of 2px means your sampling is seeing/2, which is just on the border to undersampled, even by your calculator's standards

[tight lodge]

Only if I would have had a bigger aperture 😭

That's why I want the 10" quattro. Then reduce it to 670mm and have a similar sampling to the 130PDS buuut better resolving power and faster

[candid flame]

i have one i rarely use lol

[tight lodge]

[candid flame]

me looking at the RASA with 620mm when you say you wanna reduce the quattro to 670mm btw

[tight lodge]

ain't got the money for a RASA

nor that i like to manually swap filters

[candid flame]

worth it tho

the image circle on the quattro will be absolutely screwed if you put a .67x reducer on it btw

like i think even APS-C will be problematic

[tight lodge]

@harsh matrix

[candid flame]

right is much sharper damn

[tight lodge]

Left has better SNR

[candid flame]

are they denoised?

[tight lodge]

mine is, idk about Vel's one

[harsh matrix]

Mine is denoised quite heavily

I can see more resolved detail as well

Specifically at the top of that dark band in the bottom right

[tight lodge]

How much integration on yours?

[harsh matrix]

~66 hrs

[tight lodge]

[frosty shard]

this is proof star sizes don't directly correlate with resolved detail

[tight lodge]

66h?

[candid flame]

...because they're stretched differently

[harsh matrix]

that has no bearing on how much detail gets resolved

[harsh matrix]

yes but it's 66 hours at f/8

[candid flame]

star sizes don't??

[harsh matrix]

correct

[candid flame]

what

[harsh matrix]

well they dont determine resolved detail

[tight lodge]

and here i am with 4h and 10 min💀

[frosty shard]

I mean I'm also referring to central obstruction, which alters the PSF in a way that reduces contrast but not resolution

[harsh matrix]

star sizes arent really relevant in the context of a stretched image

[candid flame]

?????????

[candid flame]

yeah that's what i'm saying

in the linear image they absolutely are

[frosty shard]

I need to do a comparison now that I know what unity gain is for my camera. I have been so nerfed due to my bad camera settings...

[harsh matrix]

well here's the scoop alright

do to the way the math works with regards to sampling

the stars in Pickle's image will be bigger

automatically

[tight lodge]

also i'm at f5 sooo... a bit faster

[harsh matrix]

so technically he will always have bigger stars than me

[tight lodge]

yep. that's most likely because of the smaller apperture as well

[harsh matrix]

the aperture determines everything

[frosty shard]

When you have a larger central obstruction, you reduce the contrast (as seen on an MTF plot), but you can recover that contrast through deconvolution. What doesn't change (assuming your absolute aperture is the same) is the point where the contrast goes to zero. You can't recover any information beyond that point

https://www.telescope-optics.net/obstruction.htm

[harsh matrix]

it's likely contributing to a non insignificant amount of star bloat on your end

[frosty shard]

to be clear, this doesn't factor in sampling, it's just the optics here

[harsh matrix]

yeah Pickle's appears sharper because he has better contrast

his CO is way smaller than mine

although i dont know what the %/% breakdown is

[frosty shard]

I can't make any statements on how BlurXterminator works in terms of detail recovery, but at least with classical deconvolution methods, more iterations causes the image to become noisier. Of course, you can get around that with more data

[candid flame]

how do you measure contrast in a linear image if not just SNR?

[harsh matrix]

i can tell by eye

[candid flame]

go look at all the images then

[harsh matrix]

I don't have time to

[frosty shard]

47% for us, usually like 35% for an imaging Newtonian?

[harsh matrix]

oh also the Sii and Oiii never wanted to get less noisy after a certain point

i kept dumping more and more time into those channels and they reached a point where they werent getting better

[candid flame]

diminishing returns

[harsh matrix]

which contributed to me having to denoise them heavily

[frosty shard]

You use the modulation transfer function (which is the Fourier transform of the point spread function)

[harsh matrix]

that reduced some level of sharpness and destroyed some fine detail in those channels

[tight lodge]

i shot 2:1:2. might have been the right decision. i'd love to dump more hours on this one

[harsh matrix]

[tight lodge]

or the Horse

[harsh matrix]

i dumped a butt load of time into H-alpha

that has no noise

[candid flame]

that's a function, not a number...

[harsh matrix]

the H-alpha is carrying the detail lol

[frosty shard]

It is impossible to quantify contrast with a simple number like SNR

You can evaluate it at a particular frequency though

[harsh matrix]

even SNR in and of itself is a bit arbitrary

you still cant really quantify it

[candid flame]

ah, i understand now

[harsh matrix]

you can try but it's always some kind of guess

[candid flame]

what

SNR =μ/σ by definition

[high aspen]

https://tenor.com/view/cat-dragonfruit-eat-sauce-kitten-eating-gif-24483121

[frosty shard]

best illustration of different modulation transfer functions and how they relate to the point spread function

[harsh matrix]

the best that gets you is a guess though

it isn't going to be an objective analysis

there's way too many factors to accurately quantify it

[candid flame]

what do you mean a guess? it's a definition

[harsh matrix]

this is a definition

but the result is a guess

[candid flame]

what

[harsh matrix]

alright

[candid flame]

using the definition is not a guess.. it's a definition

[harsh matrix]

it's complicated

it's not as straight forward as you believe that it is

[candid flame]

it straight up is, your image just need to be absolutely calibrated, and ofc you can only do it on a uniform area to not catch the spread of the signal but only the noise

[frosty shard]

Perhaps the better term is estimate? You can still get a decent estimate of SNR, since you definitely know μ (just the signal of a linear image) and σ is approximately proportional to the square root of μ assuming the dominant contributor to the noise is shot (Poisson) noise. (You could even factor in read noise and thermal noise, then add them in quadrature, but for a decently exposed image these are small contributions)

But for a given image, the exact contribution can't be known just because the noise might be closer to or farther from the ground truth in a particular image.

[harsh matrix]

yeah estimate is a better term

[high aspen]

FISH NOISE?????

[candid flame]

or you can just measure σ

[harsh matrix]

the thing is, pixel size, aperture, and focal length all play a role in SNR

[candid flame]

yeah, because they change μ and σ

[high aspen]

o

[frosty shard]

I'm thinking in the case of a single image where you don't have individual subs to do statistics with, or do you have a way to do this for a single image?

[harsh matrix]

then you have to consider light pollution which doesn't exactly have a linear impact on your signal

even the properties of the light pollution change based on location

[high aspen]

guys, just get adaptive optics, simple as

[frosty shard]

Oh yeah, I forgot, normalization is going to alter your statistics

[candid flame]

i do, you just grab a uniform area in your image and calculate σ

[harsh matrix]

in theory you can do that

[candid flame]

...and in practice

[harsh matrix]

...except the result will only be a rough estimate and will be different for everybody in the context of contrast present in each image

it still is not an exact science

[candid flame]

how is it an estimate???? it's a literal definition

[candid flame]

SNR is not an exact science????????

boy do i have something to tell you lol

[frosty shard]

Because σ is an estimate (sample standard deviation)

[harsh matrix]

oops

[candid flame]

σ also has a strict definition. no estimation.

[high aspen]

proof

[frosty shard]

It does have a strict definition, but you are estimating it (unless you happen to know exactly what the ground truth)

those are not incompatible statements

[high aspen]

can you explain each part pleas

[candid flame]

what am i estimating??? i'm just plugging into the formula bro

[harsh matrix]

this is what I was trying to say but I couldn't quite put my finger on it

[harsh matrix]

you are estimating the histogram iirc

[candid flame]

what do you even mean by this??

[harsh matrix]

im trying to put it into words lmao

there's a graphic in my head i cant quite put into words atm

[candid flame]

it's all just definitions with formulas. there is no guessing. there is no estimation. it's just a calculation, the result of which we have defined to be called SNR

[harsh matrix]

anyway, by using the above equation, that is a derviative where you are estimating the area under a curve

the curve is the function that sigma defines

derivatives by definition are estimations

approximations

my bad

[candid flame]

what? there are no areas or curves. it's just a definition

[harsh matrix]

you are missing the entire point

have you ever done calculus?

[candid flame]

yes

[harsh matrix]

okay you should get this lol

[candid flame]

but this is not calculus. it's a simple definition

[frosty shard]

So the square of the standard deviation (σ) is the variance, and the variance is calculated by summing the squares of the differences between every individual data point (x_i) and the mean (μ), dividing it by the number of samples (N). Then you square root it to get the standard deviation

[harsh matrix]

this is calculus

the function is a derivative

[candid flame]

of course if you assume a posson distribution, you can do calculus, but this has nothing to do with a possion distribution

that's just a special case

[harsh matrix]

it is a function being used to approximate the area under the curve

what does N define here?

[frosty shard]

number of samples

[harsh matrix]

because that determines the accuracy of the approximation

okay

[candid flame]

the number of pixels

pretty discrete imo, idk about your sensor tho

[harsh matrix]

samples or pixels?

[high aspen]

woaw

[harsh matrix]

samples meaning pixels or subs?

[candid flame]

same thing in an image

pixels

there is only one image

we don't need multiple images

[harsh matrix]

that makes even more sense to me

[high aspen]

the humble onion:

[harsh matrix]

the approximation of Sigma becomes more accurate with more pixels

however, the caveat is that Sigma is still an approximation

[candid flame]

it's not an approximation of sigma. it is sigma

[harsh matrix]

not an exact number

okay you are right

since it isnt approximating Sigma

however, you are still approximating something

[frosty shard]

Actually, no, it's still an approximation.

[harsh matrix]

which is the point

[frosty shard]

Critical: https://en.wikipedia.org/wiki/Standard_error

[candid flame]

i'm not approximating anything

[high aspen]

i strech moar

[harsh matrix]

well cause the result is still an approximation correct?

[frosty shard]

Yes, though it's bounded, assuming the distribution is well-behaved

[harsh matrix]

what you say you are doing vs what the function is doing are 2 completely different things

i could say i am adding everything in that but that isnt what the function is doing

[candid flame]

we're not assuming any particular distribution. and i fact, it has to be a discrete distribution.

[harsh matrix]

i could say it is a sum which is a gross under representation of the purpose of the function

[frosty shard]

It can be continuous as well, the same logic holds. But some distributions lack a mean or standard deviation, like the Cauchy distribution

[harsh matrix]

you ultimately have to bound the distribution though

[candid flame]

i'm not saying anything about what i'm doing except "use the function, that's the output of your function"

[candid flame]

nope

you can always calculate σ even without knowing the distribution... you just plug into the formula

because we're not calculating the std of the distribution we think describes our data. we're calculating the std of the data itself

[frosty shard]

which necessarily only estimates σ, because we don't know the distribution our data is derived from

[harsh matrix]

that means you are taking a guess

i dont have time to argue in circles all day

this is a waste of my time

[candid flame]

yeah it's a good estimate of the std of the distrubution which we think describes our data. But that's not the std in the formula. the std in the formula for SNR is the std of the data, not the distribution which we think describes it

basically there's σ_dist and σ_data. σ_data is an approximation for σ_dist, but the σ in the formula for SNR is σ_data, NOT sigma σ_dist, which means SNR doesn't include any approximations

and by the law of large numbers, σ_dist = σ_data as N goes to infinity, and the approximation will be practically perfect for like 10k pixels, which is just a 100x100 pixel area

(and also long before 10k pixels for that matter)

[frosty shard]

This isn't true in every case, which might matter for astrophotographers

[candid flame]

what... law of large numbers is true for everyone lol

ofc it assumes you have a uniform area of arbitrary size, which can be hard to find

but still, it doesn't even matter because we're using σ_data and not σ_dist

[frosty shard]

The statement (σ_dist = σ_data as N goes to infinity) relies on the central limit theorem, which most distributions of interest to us follow, but some (like the Cauchy distribution) don't.

Try finding μ or σ by sampling a Cauchy distribution. It never converges no matter how many samples you take.

The standard error in those descriptors is unbounded

Though to be fair you can get around that by using a different estimator

That's the point of clipped averages/pixel rejection, since these rare events can be hard to account for

Though there are also issues that arise with using different estimators

[candid flame]

sure, but this std of the distribution is still irrelevant to the discussion

[candid flame]

and also no it doesn't rely on CLT

σ_dist will also go to infinity if they are cauchy distributed

[harsh matrix]

I think that very small difference does matter in the context of contrast in one set up vs another, especially when the CO is involved.

[frosty shard]

Well, I guess the point here is that if SNR uses σ_data, which is an estimate, then SNR itself is an estimate - which makes me interested in the bounds on the SNR estimate generally

I don't think those are particularly large though

[harsh matrix]

It sort of feels like tou are throwing the baby out with the bath water in this discussion which was the discussion about contrast in the first place

[candid flame]

still no, σ_data is not an estimation. it is a definition

[frosty shard]

It's both

[tight lodge]

What have i made

[harsh matrix]

Something which can be incredibly difficult to distinguish between different setups due to a number of factors.

[harsh matrix]

A mistake

[candid flame]

to be exact σ_data estimates σ_dist, but that's irrelevant since σ_data is what we have to use in the formula

[tight lodge]

i did not think you guys would go this deep into it 😂

[frosty shard]

Hey I'm just in this to sharpen my knowledge of statistics, something I've only learned through practice, and I'm literally a PhD student in the physical sciences

Chemists aren't required to learn statistics

[frosty shard]

I guess my question then becomes, are you interested in σ_dist at all?

[candid flame]

not for SNR

[frosty shard]

What about generally?

[candid flame]

in fact, i'd even rather say it's the other way around. σ_dist estimates σ_data

because the data is what i'm truly interested in in the end

[frosty shard]

I'd take the opposite approach personally

Instead I'd say σ_dist generates σ_data, and I want to know what the distribution looks like for each pixel in an image

(The rationale: I want to write a denoiser)

[candid flame]

what do you mean by distribution of a single pixel in a single image? there is only one value

[frosty shard]

Well, if you take one subexposure, sure, but what if you take multiple

[candid flame]

oh yeah so distribution of a single pixel between many images

[frosty shard]

One issue I see with (non-AI) denoisers is that they really would benefit from more input data

[candid flame]

like denoise while stacking basically?

with all subs as input

[frosty shard]

Yup, because you can't assume simple Poisson statistics when you're trying to denoise a normalized stack

Also, most denoisers assume additive white Gaussian noise

Which is not very relevant to astrophotographers

[candid flame]

i mean, CLT makes it pretty relevant but i see your point

[frosty shard]

Makes what relevant?

[candid flame]

Gaussian noise

[frosty shard]

That's relevant, it's the additive part that's the issue.

[candid flame]

What does additive mean in exactly in this context? No negative value?

[frosty shard]

It means that we're corrupting the image by taking the original data and adding a random value from a Gaussian distribution to each pixel

Shot noise is submultiplicative, by contrast

The noise is proportional to the square root of the signal

Though there is also an additive component from background glow

[candid flame]

yeah but that requires you to know the mean value of your signal in absolute units

[frosty shard]

either way, it implies a denoising method built for shot noise should alter dark and light pixels differently

[frosty shard]

To get the skyglow noise contribution?

[candid flame]

just in general, because N/sqrt(N) != 2N/sqrt(2N)

[frosty shard]

Oh, right, gain would also change this

[candid flame]

yeah exactly

also are you sure they use additive gaussian with the same distribution across the whole frame? when you want to simulate more noise with pixelmath, you usually multiply the gaussian by the pixels value (or sqrt maybe i don't remember)

[frosty shard]

The papers I've read on denoising tend to build algorithms around this assumption. How it's implemented in software is a different story

[candid flame]

hmmm, i mean if you have much more light pollution than signal i guess

which is probably true for most of us 😭

[silver ibex]

Nuh uh I'll define it like this:

[silver ibex]

Limits, derivatives and integrals are not approximations, they are exact values. It's a misconception that I see a lot, for some reason.

[candid flame]

i think he meant the sum that defines the std of N samples is an approximation of the std of the distribution that generated them

[digital nexus]

Will this be stable enough

[high aspen]

💀

pls push the weight further forwards

[digital nexus]

It don’t move at all

[digital nexus]

no

[high aspen]

[harsh matrix]

That's because that's the way it was taught to me by 2 different professors.

The definition of a derivative that I was given in my college was that a derivative is an approximation of the area under a curve.

So, from my perspective, I don't understand how what your saying is the case.

Especially since it wasn't just a single professor that made this "error"

[silver ibex]

You mean integral? Either way, it's not an approximation, it's literally the definition of "area" under a curve lol

Like, the limit of 1/x as x approaches +infinity isn't approaching zero, it is zero.

[harsh matrix]

I am going to be mighty embarrassed if my understanding has slipped so much in such short a time

And it could be that im stressed the hell out at work because of a major holiday tomorrow

Either way, I do not think it is appropriate for me to continue to discuss this because I am not thinking clearly at all

[silver ibex]

I mean, it's really just a conceptual difference I guess

[harsh matrix]

https://youtu.be/SxbmvSIO7rg?si=21ihTCIykbXYIZrd

This guy is my source, albeit, the information is warped in my mind after not looking at it for a while.

[candid flame]

kinda... basically like with almost all of math you can start many different places and derive the rest. You often start with the definition of area of a rectangle and compute the area under the curve as a riemann sum. If you're even more hardcore you go full Lebesgue on it and invent measure theory

[harsh matrix]

He draws some pretty important distingushments between the theory and the practice of SNR calculation in his series.

[harsh matrix]

The riemann sum is what I was referencing

That's what that equation looks like to my eye

[candid flame]

riemann sum is still not an approximation tho, as it's evaluated in the limit

[harsh matrix]

Idk my understanding is that it was an approximation

[silver ibex]

We do it like that too, still, we define the area under a curve as the integral, of course there's a very evident geometrical interpretation behind it

[candid flame]

for any finite N, it is an approximation, but the whole point is to evaluate it in the limit where it is exact

because the approximation gets arbitrarily good basically

for most functions*

[silver ibex]

The notation we use is cursed too 😭

I love the random star

The heart too

[candid flame]

italian 😭

[candid flame]

this is not saying SNR isn't rigorous, it's saying the calculators are bad lol

[harsh matrix]

I guess depending on the variables used to constrain the SNR, say mag/"^2 then that could be the case.

Which is more or less what I was trying to argue.

[frosty shard]

you go full Lebesgue on it and invent measure theory
Meanwhile

[harsh matrix]

Which is the more variables you consider in the calculation, the less objective the result becomes

[candid flame]

wdym more variables? there are only two (which can be objectively measured)

[high aspen]

they just love maths

[silver ibex]

I think he meant more error

[silver ibex]

Good, me too

[candid flame]

there is no error tho

[high aspen]

i think i failed the maths test i had yesterday

i got 2 more in the next few weeks to fail too

[harsh matrix]

That's what I mean

Hmmmmm hang on

Alright so I was getting ahead of myself

I know where the misunderstanding is coming from

[silver ibex]

Bruh

[harsh matrix]

Measuring SNR in your own data is fine

[silver ibex]

Just... study, I guess?

[harsh matrix]

But this conversation came up because of a difference in sharpness aka, edge contrast

If you want to compare edge contrast by using SNR as the guiding principle, the reason why this is not a good idea is because the differences between Pickle's and I's setups are influenced by far more than SNR alone

Things like CO and light pollution must be taken into account to accurately measure edge contrast by using SNR as the guiding principle.

That's why I don't think using it to gauge edge contrast is a good idea.

There's too many variables at play to draw an objective conclusion based on SNR alone.

[candid flame]

oh yeah i totally agree that SNR is not a perfect measure of contrast, that's why we were talking about the Fourier transform of the PSF

[harsh matrix]

The misunderstanding comes fromt he fact that you and I were on different subjects entirely

I got lost in the intricacies and got confused as hell

[harsh matrix]

Yaaayyy

In that case Pickle's image will definitely appear sharper because his sampling is far more coarse.

[candid flame]

how so?

[harsh matrix]

My image was shot at native focal length and my seeing wasn't all that impressive by the end of that project which degraded the overall sharpness.

[harsh matrix]

By the way of how the Fourier PSF is influenced by sharpness.

I can give you some great examples with my own data later

It couldn't be more obvious once you've become accustomed to several setups of varying sampling rates.

[candid flame]

trust me i've tried a myriad of setups lol. Surely you don't just mean the sampling tho? like having bigger pixels would make your sampling coarser, but that definitely doesn't make it sharper

and honestly the seeing should be the limiting factor by far, unless you're using some mini frac

but really what originally started it, is whether the sharpness is determined by star shapes alone, which i still believe it is

[harsh matrix]

Not necessarily sampling alone no

The seeing can mostly be disregarded in how disbursed the PSF is for smaller telescopes like Pickles since it takes a very unstable atmosphere to smear stars out on a shorter focal length telescope.

By nature of imaging at 1628 mm, my RC's stars are quite spread out

So the influence on the PSF is a softer star

[candid flame]

Ahhhhhhh, now I see the real misunderstanding

[harsh matrix]

Not something that appears small and restricted like with Pickle's rig.

[candid flame]

I see the seeing's influence as part of the star shape and PSF

Or well it's debatable whether it's part of the PSF, but it's definitely a part of the star shape

Because that way star shapes really do determine everything about the sharpness

[harsh matrix]

Yes I thought so.

Yeah I don't necessarily mean in regards to how the seeing has effected the data in specific subs.

I mean in regards to how much the seeing effects imaging with a longer vs shorter focal length system and how THAT influences edge contrast.

Even THEN that's still not the only factor, unfortunately.

CO still plays a role in edge contrast, especially for my RC.

[candid flame]

Yeah but that's also part of the star shape and PSF

[harsh matrix]

Which is why a frac is often touted as been preferred, it's because they lack a central obstruction which creates far better edge contrast that mine or pickle's scopes alone.

[harsh matrix]

In this situation, I'd argue that is the primary source of error in this comparison.

My seeing wasn't all that bad throughout the duration of that project and decon could undo it relatively easily, however the CO still heavily influences edge contrast on structures too.

[candid flame]

Of course, I'm not saying CO doesn't affect sharpness, I'm just saying it affects the stars in exactly the same way as the rest of the image, making the sharpness of the rest of the image be uniquely determined by the star shape

[harsh matrix]

Something that while seeing does play a role with, doesn't affect it as much as one may think in relatively good seeing conditions.

That entire project stack had a final FWHM of <2.7"

The first 2 nights was about 2.1"

Also that's for the entire image, the FWHM in the center was way better.

[harsh matrix]

I understand that but I wanted to stress that the massive CO on these RC's can't be overlooked unfortunately.

A small newt is relatively lucky in that they can kind of get around this with their much smaller CO's but I can't ignore it with my RC

[candid flame]

Yeah but my argument is still the exact same: anything that influences sharpness affects stars equally and vice versa, making star the star shape uniquely determine the sharpness of the rest of the image

There exists a bijection 🤓 ☝️

[candid flame]

Like this is all I'm arguing against lmao

Tho I guess he technically did say size and not shape

[frosty shard]

Yeah, this is the important bit, which is also influenced by how you stretch the image too

At least when producing a final image

[candid flame]

Oh yeah a lot

But does that mean we agree that star shapes determine sharpness completely?

*in the linear image

[frosty shard]

To be rigorous: what do you mean by sharpness?

[harsh matrix]

And now you understand why the sharpness discussion between 2 completely different setups is a pain in the hiney

That's why I don't look at sharpness as an indicator, I look for the fine details, which define the resolution of an image.

Folks try to conflate resolution and sharpness as one and the same but to my eye, and my mind, they are completely different topics.

[candid flame]

Great question actually lol

[frosty shard]

If you're talking about small-scale contrast, then yeah, the contrast distribution is entirely determined by the PSF, which can be measured by star shapes.

[candid flame]

Something along the lines of the kernel you would convolve a perfectly sharp image with in order to get your image, but that really just makes the statement trivially true

But really I think the statement is true for any reasonable definition of sharpness

[harsh matrix]

Yes so in that sense

Pickle's is sharper

Objectively

[frosty shard]

If you're talking about the minimum scale of resolvable features, then you're limited by where the modulation transfer function becomes zero.

[candid flame]

That indeed also sounds reasonable, and also seems to be determined by the star shape

[frosty shard]

Deconvolution essentially scales up certain parts of the MTF

Obviously you can't scale up something that's zero

That's part of the reason why deconvolution amplifies noise: if you push up the higher frequency components, that increases the local alternation in pixel values and the image becomes grainier

You can get around this by collecting more data though

[candid flame]

angular sharpness to be specific btw

[harsh matrix]

WOW?!?! REALLY????

[candid flame]

Nah watch me: 0*2=0

Just scaled up 0 to be twice as big

[harsh matrix]

You sure it's not 2?

[frosty shard]

0*2 = 2 for certain definitions of 0, 2, *, and =

[harsh matrix]

[runic violet]

r u guys ok

[harsh matrix]

No

[candid flame]

sharpness = SNR if you try hard enough?

[harsh matrix]

Unironically kind of, in dark skies.

[frosty shard]

Hmm...perhaps sharpness could be defined in terms of the SNR associated with each frequency component of the image?

I would have to think about how to define that

[candid flame]

Why only dark skies?

[harsh matrix]

Lemme see if you can catch my drift

[candid flame]

Can't you always trade sharpness for SNR with a convolution, and vice versa with deconvolution?

[runic violet]

what the hell are you guys talking about

[harsh matrix]

Specifically with regards to how light pollution, extreme light pollution like in my yard, affects the edge contrast of structures.

ESPECIALLY with regards to faint outskirts of objects like galaxies

From home, the cores of my galaxies look super sharp, but the faint outer arms are always mushy and look like crap

[candid flame]

But that just sounds like worse SNR tbh

[harsh matrix]

But when I shot M33 from bortle 3 this year, I noticed it was just about equal sharpness throughout the entire galaxy.

[harsh matrix]

It is! It's because light pollution acts like acid imo

It corrodes the sharpness of an image

In my opinion

That's how I feel it affects my data, ignoring the worse signal overall.

[frosty shard]

If a particular frequency component is poorly sampled, you wouldn't be able to scale it up in a deconvolution operation without introducing more noise to the image. You could define this across subs easily, but I guess you could get some notion of this in an image if you take its (possibly padded) Fourier transform

[candid flame]

I wouldn't necessarily say it affects sharpness, tho possibly contrast. Like if you gathered enough data from your backyard to get the same SNR as from bottle 3, wouldn't the images be practically identical?

[candid flame]

Yeah but that's the point isn't it? Deconvolution trades SNR for sharpness, exactly because it introduces noise

[harsh matrix]

That's what worries me about what I am bringing up

The answer is no

For the most part

[runic violet]

assuming LP is only poisson noise and has no spatial variability

[frosty shard]

Yup, I'm thinking about how you could transfer the concept of SNR to sharpness

[harsh matrix]

The fainter regions of galaxies ive imaged in the past remain mushy and gross despite integration time.

[harsh matrix]

I think I can see a relationship as well

It's hard to put into words though

I understand why I am thinking it is possible to have a relationship there and why that doesn't necessarily make sense by math alone.

[frosty shard]

Well perhaps I can. Let's assume (and this is really rough) that you have a sensor that contaminates the captured image with Gaussian noise

The modulus of the Fourier transform of the noise is 1 for every frequency component (it's complex-valued with random phases)

This corresponds with the fact that there's no correlation between the noise of individual pixels

But the modulation transfer function of the optical system drops off with frequency, unlike the noise