Yes but I think, this is not really linked to total exposure time , it is linked to number of images you process : with 100 images your SNR is 10 times better than with 1 image And I think, individual exposure time is very important : Peter Zelinka do no speak about individual exposure time ? You need to expose long enough to have sky light > noise on a picture. It happen's after +-120 seconds on a F5.6 telescope and an ASI2600 camera. On earth, the best dark sky in Atacama, send us 22 mag/arc.sec^2. So, I think, there's no way to receive light from an object fainter than that, no matter individual exposure time
@rafberu it is total exposure time. Not number of images. Each image adds noise, the read noise of the camera. So a perfect 10minute sub would have a better signal to noise ratio than ten 1 minute subs. A stack of 10 total minutes of exposure timen would have half the noise of a stack of 2.5minute totalexposure time (2.5x4 = 10, assuming the same number of sub frames in each stack). Where people get confused, is that there is a practical limit to sub length. So most of the time when you say "more subs" you are really saying "more total time". But you can't improve the SNR of any length of exposure time by just dividing it in to more subs to have more files to stack. Only more total signal, meaning more total time, raises SNR.
@@bbasiaga So why stacking ? Please, just Google "what is image stacking astrophotography" IMO this is the stacking, by averaging the value of each pixel wich increase SNR. I think SNR of 20 x 3mn exposures is more than 4 time better than 1 x 60 mn exposure.
@hunter133official square root of two is about 1.4, so 40% improvement in SNR for 2x the exposure time. Definitely noticeable. 1/1.4 = 0.71...so yeah pretty close to 75%.
Just a note on the mathematics of f-stops: These numbers are expressed as ratios between focal length and the diameter of the entrance pupil, thus the slash: f/4. Two lenses with the same focal ratio have the same effective illumination of the image, regardless of focal length or size of aperture. Also, since light gathering ability is a function of the area of the aperture, it is proportional to the square of the diameter. This is why, for a given focal length, f/2.8 and not f/2 is twice as bright as f/4.
That Andromeda photo you showed at the beginning is a great example of total exposure. For someone to discover a new nebula in one of the most photographed dso is astonishing to me. We are truly living in a great age of amateur astrophotography. My father shot in film 40 years ago. When i got serious in this hobby 15 years ago with a dslr, I was more concerned with getting a bunch of objects at 30-60 mins. Now i max out at around 4-8hours since clear moonless nights are few and far between. Good enough for me at this stage.
I know this question will come up in the comments, so here's the data behind the photos: *Crescent Nebula* - ASI 2600MC Duo, Askar V (80mm + Extender - f/7.5), Optolong L-Enhance, Bortle ~6 | 5 Minute Exposures, Gain 100 *Iris Nebula* - ASI 2600MC Duo, RedCat 51 (f/4.9), Bortle ~2 | 5 Minute Exposures, Gain 100 *Pleiades* - ASI 2600MC, RedCat 51 (f/4.9), Bortle ~2 | 5 Minute Exposures, Gain 100 *Orion* - ASI 2600MC Duo, RedCat 51 (f/4.9), Bortle ~2 | 3 Minute Exposures, Gain 100 *Elephants Trunk* - ASI 2600MC, Askar V (60mm + Reducer - f/4.5), Optolong L-Enhance, Bortle ~6 | 5 Minute Exposures, Gain 100 All images were stacked without Darks, Flats, or Bias. Just light frames.
From 10:00 to 10:05 as you go through the different exposure times, it seems like you've captured some sort of temporary emission in the nebula. It's just "southeast" of center frame, at a node of the gaseous structure. You can see it brighten, peaking in the 8-hour image, then seeming to fade away again in the next master once the 16-hour timespan averages it back down. Very interesting. I wonder if a time-lapse of the data, grainy as it must be, would show anything interesting!
Hey Peter, thanks for this! The comparison between monochrome and color isn't too hard, even without experimental results. There's a few primary factors to consider: 1) The bayer pattern. Color sensors are primarily differentiated from monochrome by adding the bayer grid which splits each grouping of 4 pixels into 2 green and 1 red and 1 blue pixel. So, you basically automatically get double the green data in every shot, and if you're using filters which block light that falls in one of those color spectrums, then that portion of the sensor becomes useless for that shot. 2) Quantum efficiency. The quantum efficiency of monochrome sensors is much higher, owing to no loss from the bayer filter (even the passband won't transmit perfectly). So even if you intend to shoot broadband (e.g. for galaxies), you'd still benefit from monochrome sensors (even using the exact same sensor in a mono vs OSC configuration), providing you use RGB filters which are at least as high a quality as the bayer filter would have been, since you can save time by not gathering double the green data that you need. In cases where you want to add narrowband filters, the benefit is even greater, since you don't get the loss due to doubling green or the general bayer pattern loss, AND you avoid the loss from double-stacking filters. The benefit of OSC is primarily just cost and simplicity of process.
I was actually just about to comment something similar. At 13:50, Peter states that if you imaged for 2 hours, "you're technically getting 2 hours of data on your red, green, and blue color channels". As you partially alluded to, this is not correct. Unfortunately, this creates a slight complication to explain as you create something similar to the "full frame equivalence" we see in photography. One huge benefit Peter has that cannot be understated is dark skies. 1 hour in a Bortle 0 or 1 is equal to 8 - 16 hours in a Bortle 4-5 sky. That truly benefits color sensors so much. Ignoring bayer matrix and QE for now, in 2 hour imaging session, each pixel on the sensor is still technically getting 2 hours of data. The problem is that each pixel has a color, so not all the data counts. For a color sensor, a Green photon falling on a Red or Blue photosite is ignored. Same is true for the other colors as well. This cuts down on the exposure times, as well as the technical resolution. There is also not very much green emission in space so much the sensors resolution is actually wasted. A way to explain it is that, assuming perfect efficiency, in a given 2 hour session on a color sensor, Green would capture the equivalence 1 hour of data, Red gets 30 minutes, and Blue gets 30 minutes, if compared to a monochrome sensor. This is just a conceptual way to understand it, the red pixels still get 2 hours of data but 75% of the captured photons are thrown away because they fell on a Green or Blue pixel. For monochrome, all photons of a given filter are accepted by all photosites and contribute to all of the resolution. So while on a color sensor, the red pixels got 2 hours of data, they only represent 1/4 of the image's resolution. Assuming an even RGB filter exposure time, on a monochrome sensor, Red will only receive 40 minutes of data but is contributing 100% of the image's resolution. Though we don't want to jump into the deep in too much, monochrome sensors also have a much higher Quantum Efficiency. In short, this is percent chance that a given photosite will convert a photon to an electron. This makes matters even worse for color sensors as we are already losing 50-75% of the photons captured to the bayer matrix, and using the 2600MC as an example (and these are averages as QE is a curve), we then lose another 22.5% for Red (600-700 nm is peak of 90% dropping to 65%), 10% for Green (500-600 nm), and 38% for Blue (400-500 nm peaking at 80% and dropping to 45%), due to QE. So now, we down are to some pretty heavy losses and in reality, technically we still have read noise and shot noise to account for. So when Peter states that a monochrome sensor would need 6 hours compared to color sensor, that is simply not true. For a 2 hour session, a monochrome sensor would be able to capture the image in less than half the amount of time. Using the 2600 MC and 2600MM as the comparison, in 2 hours: - the 2600MC effectively captures 30 minutes of Red, and after QE losses, is left with 24 minutes. Green caputes 1 hours and is left with 54 minutes (high green QE on the MC). Blue captures 30 minutes and is left with just 18.6 minutes. - Assuming an even split of the time for the 2600MM, Red, Green, and Blue each captures 40 minutes. Red is left with approx 28 min (QE for Red on the 2600 is poor), Green is left with 36 min, and Blue is left with 36 min. That is almost 1 stop light increase for Blue, which contributes heavily to DSO. Most people who switch from color to mono also increase their overall exposure time. But if someone is happy with their color images, they could switch to mono and get the same image (but with more resolution) even faster.
@@Cornerstone_Creative Yes, I didn’t touch on the resolution advantage of mono, but that’s a major factor too. A mono configuration of the same sensor will cut your average FWHM in half vs a OSC shot, since each pixel can be used individually instead of having to group 4 pixels into one “pixel”.
@@Cornerstone_Creative The other way to get better pictures with a color camera is to change the telescope to a faster one. Then you will beat every single mono camera that shoots with slower optics. The other option is if someone takes an 8 hour photo with a mono camera with the same telescope, and you take 16 hours with your color camera, then you will have the same result, if not better. It's all a matter of time and whether the result satisfies you. You save at least the three RGB filters, the expensive filter wheel and a bigger mount to take all that weight. Because the difference between mono and color for an hour's duration is that the mono sensor absorbs 400 photons during that time, and the one shot color camera 240 photons of light. It's not exactly double the difference. This explains why so many people continue to shoot in one shot color camera. It's not better than the mono camera, but if you spend enough time and that's easy because you won't be changing filters at night and refocusing them on each filter, you'll also save money, because I don't know an astrophotographer who got rich from this hobby. Then you will have good results even with the one shot color camera. And if you post your photos on social networks where most people look at them on their phones, then the sleepless mono nights for a slightly better detail which can be seen with a magnifying glass become meaningless.
EXCELLENT video Peter. YES indeed the brightness of the night sky makes a HUGE difference. It boils down to S/N. Each doubling of exposure adds 3 dB of S/N. Thus a darker night sky has the same effect has adding more exposure. One magnitude darker is about 2.512X. If your sky is say a limiting magnitude of 4, and you go to a sky with limiting magnitude of 7 then you get about 15.85X improvement of S/N, or about 4 stops!!! Another way to look at it, it would take 16 hours of exposure under a mag 4 sky as compared 1 hour under mag 7 sky to the same S/N!!!!!!!!!!
Great video Peter. One small point on the hyperstar. (You said, attach it to the back.) You attach it to the front. I've used the hyperstar for years on my sct 8". It's takes the focal length from about 2000mm down to about 396mm. What it does do for me is give me a wider field of view. Take the same scope, take the hyperstar off attach an OAG and you have a focal length of about 1700mm. Two scopes in one. Sorry I got side tracked.
when you swap out the secondary mirror with the Hyperstar and then go back - do you have any troubles with columniation or does the swap effectively leave the secondary mirror unchanged (since it is being screwed back in as it was before)? I have been thinking about a Edge-HD 8 to replace my old Meade 8" SCT just for Hyperstar.
It retains collimation well. I used one on my Edge HD 11 when I first got the telescope and the HyperStar works as advertised. I recently sold the HS for two reasons. Mainly, the time and effort required to change the setup from the HS to longer focal lengths. It is easier to leave the telescope in one configuration. The other drawback was using a manual filter drawer to change filters. It's not too bad because the HS is so fast but still requires manual attention to change filters as opposed to a filter wheel. I have seen some setups where a small filter wheel has been used with the HS with only a small light blockage. I have another refractor set up (complete mount/camera/etc.) at f3.9 for wide angle stuff and leave the Edge set up at 2800 FL for small DSO's.@@allenbaylus3378
The reason Peter said (at least once) that 'at 16 hours you see diminishing returns' is that you need to focus on Signal to Noise ratio. Each image brings in. signal (the light you want) and noise (the stray stuff you don't want). If you take a 1 hour image, it takes 4x more time to double the S/N (2 squared). So 1 hour to 4 hours to 16 hours to 64 to 256...). To see a double from 8 hours is 64 hours. When you compare 8 hours to 16 hours.... you are nowhere near making a dent at improving your image. Just as Peter said when he started the video (paraphrasing again), 'you can't just take 3-4 more hours and expect a good photo'.... well you need to go to 64 hours from 8 hours to see a decent (double) improvement of your image.
Thanks. Of course the question for many of us is “how to get a pleasing image” rather than “a perfect image”. So my question would be “what role does noise reduction software play?” Can you process each set with NR and THEN compare? I find even 30 min shots look “clean” after passing through Denoise or tweaking with Affinity and this saves me hours to look at other nice spots in our night sky! 😊
The basic principle underscoring this is "The Mathematical Theory of Communication." Basically, when you try to transmit a signal, you can change two variables: power and bandwidth. More power means that you're shouting more loudly and are easier to hear. More bandwidth means you can send data faster. A consequence of this, is that for a fixed transmission power, you can guarantee whatever minimum signal quality you want - given you're willing to slow down the data rate (this is how we can still talk to Voyager, which has the transmit power of a 100 W bulb). On the receiving end, this manifests as "integration time" and "antenna gain." The longer you observe a signal with a particular frequency, the more certain you can be that it exists even when washed over by interfering noise. The bigger your antenna, the more of that weak signal you can capture at once. Thus for telescopes, we are perpetually on the receiving end and have two modulation choices: aperture and integration time. A larger aperture will capture more light per instant, allowing us to gather what data there is more quickly. A longer integration time allows us to slowly pick out the faint signal in the noise. Mathematically, we're limited by the diminishing returns of the logarithmic nature of communication.
Whike the video itself is superb resource for someone like me who's just starting out what really sealed the deal for me is OG Deus Ex music in the background. You are a sir.
I need to recommend Craig Stark's "what do all great shots have in common?" Video. I would say it's the ultimate video in conceptualizing the basics of image noise and what to do to address it.
🎯 Key Takeaways for quick navigation: 00:58 📷 *To improve image quality, double your total exposure time; add more hours of data if your photos appear grainy.* 02:37 🌌 *The aperture of your telescope significantly impacts image quality; wider apertures like F2 gather more light and detail.* 05:38 🌟 *Telescopes with larger apertures gather more light, but focal length affects the apparent aperture size; accessories like the HyperStar can help.* 08:35 📸 *Increasing exposure time from 1 to 16 hours significantly enhances the quality of astrophotography images, especially in capturing faint details.* 11:45 💡 *Light pollution reduces image contrast, making it challenging to capture clean astrophotography; shooting in darker skies is preferable.* 13:34 ✨ *Difference in data collection between color and monochrome cameras: color cameras capture all color data simultaneously, while monochrome requires separate captures for each color.* 14:56 🪐 *The brightness of the target affects the required exposure time; brighter objects may need less exposure time compared to darker and dustier regions.* 17:40 🌃 *Light pollution demands longer exposure times to achieve comparable image quality to those taken in darker skies.* 18:36 📊 *Doubling the total exposure time is a rule of thumb to enhance image quality; beyond a certain point, diminishing returns occur in terms of data collection.*
Another big factor is the quality of the camera. Less noise = a better signal to noise ratio. A camera with very low noise can have the gain turned up further for a comparable image to a less expensive camera. And, more resolution is not necessarily helpful. More resolution means smaller pixels. Smaller pixels means less light captured in each pixel. Less light = less signal : noise. It is always interesting to figure the trade-offs.
To add to the complicated equation--I've found that a very fast telescope like a RASA or Hyperstar is awesome to a point.. you may end up collecting the light so fast that your individual images become mere seconds long, and you're dealing with thousands of them. That becomes unwieldy from a harddrive space and processing time standpoint.
Good content Peter. I am using a triple setup one at F4.9, F5.6, and F2 - all with nearly identical FOVs when paired with the proper camera, Your theory is sound logic and the images I am collecting have almost little noticeable noise. Longer integration time does help.
Another well thought out and produced video, Peter. Some thoughts while watching... As a long-time pro magazine and advertising photographer, I'm well acquainted with the basic science behind imaging. It seems to me that given a dark sky, fast sharp optics, and plenty of exposure, another variable is sensor ISO (ASA) sensitivity. Hopefully, in the future, astro cameras can operate at vastly higher levels without introducing grain/noise. I can only imagine flatfield f/2 optics (with the appropriate fl) in tandem tracking all night with a large sensor operating at like a million ISO! Another thought I've had is: What happens if you simply "copy" a good sub 100 times and stack those vs collecting individual subs over many hours??? I'm sure there's a good reason for the difference. Your thoughts, please. Clear Skies, Michael
The interesting concept, using two identical scope setups to capture light, has an added benefit… twice as many pixels. The unlikely chance that those pixels from each rig would be perfectly on top of one another means there would be added resolution.
Amazing video. Can you please do an HDR version of the Orion nebula? Merge together a let's say 7.5 min up to 8 hours worth of data? So, 7 photos of different exposures to one! Just to save the highlights!
Ok, but tell me WHY it's never occurred to me to run 2 rigs at the same time, on the same object, as a no-brainer hack for getting double the exposure time??! 🤦🏽♀️😅 1hr × 2 rigs = 2hrs of total data So simple & yet I didn't even consider it until you mentioned it in this video 😂 Thank you for that!
@@TJ-ki3gp As long as you like it! 😊 The 20, 30 secs are in the hidden menu, but only with 10 sec you can do lovely pix! Just make sure to stack it externally, like with SiriL.
In the area I reside, clear night skies are scarce, making it nearly impossible to attain the necessary exposure time. Thus, I heavily depend on noise reduction software, effectively enhancing the image quality equivalent to an additional 2 stops of exposure time. Naturally, it doesn't match the quality of actual exposure time.
Very interesting! I learnt a lot from this! You didn't touch on exposure time per sub? Does it make a lot of difference doing a 1 minute sub vs 5 for example?
Excellent video, thank you! And I’m glad to see that we have some of the same equipment (Askar V, ASI2600 Duo, along with an AM5 & ASIAIR+). It makes me feel like I’m on the right track. Now, just need to improve my processing skills.
Peter, not sure if you've covered this before, but might be a good future topic: magnitude. I'm a beginner and had some challenges with Horsehead and Soul Nebulas. I started doing some learning on magnitude, and compared magnitude values for these two targets, versus others where I've had success - that was it. In connection with this video, it seems like some deep sky objects need more time than others?
Hi Peter, just some thoughts here. If your having to deal with some light pollution (glow, wash out), extending the time is just going to further wash out the image. Or over expose. So in perfect\great sky conditions, yes you could extend the amount of exposure time (more light) or use a wider aperture. You need to take into account the possible over exposure. With all of that, are you speaking in terms of total exposure time (50 photos x 30 seconds) or the amount of time per image?
Wow, this video is genius, I'm on my fourth watch. I did some shooting a few nights ago in the GSL desert and had this very question - if I'm already getting good results at 30 minutes, why do I need more data? Thanks for making the connection between exposure time and aperture vs. noise reduction and clarity - bringing f-stops into the equation makes so much sense. Your layered examples of doubling light showed the principle perfectly. The question which remains for me is the third point on the exposure triangle: ISO/gain. How bright should my light frames be? I am shooting with a mirrorless Nikon full frame which is terrific at high ISOs. Even if I take somewhat noisy shots at ISO 6400, the grain seems to get sampled out in stacking. If my telescope is a slow f/7, how much of a compromise is boosting the ISO if I'm going to average the noise anyway? Thanks for the great video.
ISO/gain is just the same set of relations: Higher gain mimics longer exposure, greater aperture and wider f/ratios. The price paid is in grain & noise. Kodak Tri-X was fast film at 400ASA, but it was grainy.
❤ Lovely, thank you! You see; INTEGRATION TIME + POST PROCESSING. That's it. That's why I gave up with my desired and 💲 William Optics Zenithstar 61 + mount + guide scope + guide cam... even with my existing M4/3 Lumix GX80, the cost would be at least £2000...! No way. Just go for the $500 Seestar S50 ❤ Integration time + post processing. Period 🎉
So for something like the Orion nebula, what is the proper exposure target? Do you expose for the bright nebula or let that get blown out and try to expose for the dust? Or, do you expose for the bright part AND the dust and then combine in post? Or... Does the stacking just reduce the noise in the image so that you can see more of the detail when stretched? That sounds likely, but not really sure.
Peter,lovely review. Almost makes me want to get a RASA. This is also why I’ve been using my Samyang135 at f/2.0. Unfortunately, from Bortle 6, I’ll still need 16+ hours for the Spaghetti Nebula (SH2-240).
Hey @peter great video. What about attacking that imaging time from another vector - gain. In photography, the exposure triangle is aperture, shutter speed and ISO. But in AP, you rarely hear anyone talk about using gain to compensate for less imaging time or longer f/ratios. Being new to the refractor space, I’m finding the imaging time required for even f/5.6 is substantial. I just got an f7.5 system and I shudder to think the number of hours it will require to get a clean image.
Thanks Peter! Very well put together, and brought home focal length, aperture, f - ratio, light quality, integration time as a single relationship - with both logical and visual examples. I consider this as some of your best work. Kudos!
Thanks for this interesting comparison. Which influence does the exposure time of the individual photo have? I.e. 300 seconds of exposute time vs. 600? Best, Claus
thanks for this video. I want to use a 600mm lens with f6 on my camera. With a 2x teleconverter I get 1200 mm. My camera is quite insensitive to high ISO values. Does it make sense to use the teleconverter and increase the ISO? what exposure time and what ISO would you recommend?
Hello Peter. I'm a professional astronomer. I find some interesting concepts in your video, but unfortunately, you mix and confuse light collection with f-ratios. F-ratio tells light density in the focal plane not how much light is collected. Light collection from an object in the scene is proportional to aperture area times exposure time. It has nothing to do with focal length or f-ratio. F-ratio is not in the equation. For example, which collects more light from M51, a 50 mm focal length f/2.8 lens or a 200 mm focal length f/4 lens? A 50 mm f/2.8 lens has an aperture diameter of 50/2.8 = 17.86 mm. Area = 250.5 square mm. A 200 mm f/4 lens has an aperture diameter of 200/4 = 50 mm. Area = 1963.5 square mm The 200 mm f/4 lens collects 1963.5 / 250.5 = 7.8 times more light in the same exposure time for any object in the scene, whether a galaxy, a nebula, a star, a bird in a tree, or a persons face. Bin the 200 mm pixels 4x4 and the resolution in terms of pixels on the subject would be the same as in the 50 mm image, but the light in those pixels will be 7.8 times brighter in the binned 200 mm image. Try this with your 11-inch telescope. Choose a target like a galaxy that fits on your sensor in the f/10 configuration. Take one image at f/1.9. Take another at f/10 with the same exposure time. Bin the f/10 image by summing 5x5 pixels. You'll find the same amount of light per binned pixel and the same pixels on the object (within 5% because the f/1.9 is not a factor of 5 from f/10). You state in the video (at about 5:50) that change to hyperstar increased light collection by 25x. But the binning demonstration shows that the light is there, just distributed differently. Better to computer signal per square arc-second or arc-minute. By focusing on the subject, it will become clearer what the variables for light collection are. For example, redcat 51 (51 mm aperture) vs Celestron 11-inch (279 mm aperture) ratio = (279 / 51)^2 = 29.9 times more light from any object in the scene, e.g. a star, a galaxy, a square arc-second, a square arc-minute.. It has nothing to do with f-ratio. On the plus side, at the end of your video you talk about buying a larger telescope, but unfortunately you don't explain correctly why. I'll end with a comparison to Hubble, JWST, and other professional telescopes. Hubble and JWST are great deep sky telescopes. Hubble is an f/24 system, and the WFPC3 camera operates at f/31. JWST is f/20.2. I have done most of my professional work at terrestrial observatories with the NASA IRTF on Mauna Kea, Hawaii (f/38) and at the U Hawaii 88-inch (2.24 meter) f/10 telescope. By the flawed f-ratio ideas in this video, a redcat 51 (51 mm aperture diameter) with f/4.9, or your 11-inch hyperstar (f/1.9) would collect more light than these huge telescopes. NOT. Key is to computer the light per object area, like per square arc-minute. Quiz: assuming the same wavelength of light, how much light per pixel do the cameras on JWST, and Hubble collect per pixel compared to your redcat 51 with your camera (f/24 vs f/31 vs f/4.9, respectively), assuming the same sensor quantum efficiency? The LSST, Vera C. Rubin telescope is going to only take a pair of 15-second images per position (in each filter) and is expected to come online in January 2025. en.wikipedia.org/wiki/Vera_C._Rubin_Observatory It is an f/1.25 system. Do you really think that your 11-inch f/1.9 telescope with 1 hour or 16 hours of exposure time will collect more light from NGC 6888 in your video than the LSST in 30 seconds? Again, the key to light collection is aperture area times exposure time.
Why not start with actual aperture (size of entrance pupil) then the same size of the same light intake size will give the same signal regardless of F-stop or focal length. Focal lenth will just change the magnification of noise and object (worse F-number). Downsize the image of the longer focal length image will lower noise. Actual aperture size is the signal strength and affecting the noise PER DETAIL per the actual photographed object (nobula/planet). Not the focal ratio number (f-stop,). Larger aperture (entrance pupil) decreases actual noise per real object (nebula). Better (lower) f-stop (focal ration) just lower the noise per image. Not per same object photographed.
Yes! This is what I've been trying to figure out as well. Is it accurate to say that f-ratio gives snr per pixel, while aperture gives snr per arc second? Although the first produces pretty pictures, it could be at any focal length, and if it's super low, you won't be able to resolve much detail. So it's really aperture that matters rather than f-ratio at the end of the day (depending on your goals, but mine is to resolve detail)
4:01 man got Internet Historian flashback of his Fallout 76 review, lmaooo. But yeah, this is good stuff!! Have a SVX130T with a ZWO294 camera with a reducer. And having that reducer makes me feel a lot better lol. Still very new, but excited to learn more
Very interesting stuff! I’m still trying to decide on my first scope and assorted accessories. Just curious, what’s the longest exposure Astro photo you’ve ever taken?
Peter, its been many years since I studied the f stop math. You show it as doubles ie 2 4 8 16 etc. my understanding of f stops is the difference between f stops is not doubles its logarithmic so a F11 takes 84 times as long to get the same light a f2 gets in one hour?
The answer depends on the brightness of what. The 8” Celestron concentrates its light gathering power onto a smaller area. So if we talking about an evenly illuminated sky, they both collect the same total number of photons, but the Celestron is collecting those photons from a smaller area of sky at a faster rate per square degree of sky. So if you’re talking about a small target that fits in the field of view of both scopes, the Celestron will have much higher light gathering power for that target. It will capture more photons of that target per second. An analogy is a flashlight that has a lens that allows you to adjust it to a narrow beam or a wide beam, in both cases, the flashlight is the same brightness, but when you focus it with a narrow beam, all that brightness is focused on a small area and that small area is illuminated much more brightly. That’s like the Celestron, but in reverse.
How to use a Tracker and mirrorless camera for my astrophotography. Have you completely switched to telescopes from dslr/mirrorless cameras? If so I will find your videos less interesting but try to always watch them.
Wow, you've lost me immediately at "add a stop". Been astrophotographing quite a while now but have no idea what that is. Is that a DSLR thing? Gonna have to look that one up...
Just curious, when you say you have 2 hours of overall data what would you estimate is the avg time for the individual photos before they are stacked? For example i hour of data could be 60 60 second exposures or 4 15 minute exposures if I am thinking correctly.
How would I double my data that I get. Let's say I get 1 hour of imaging. Can I double that by running the photos through deepsky stacker twice to get 2 master images then stack those master images to get 2 hours of data. Or would that not work?
my camera can trace the stars for 5min (using ibis to shift the sensor to match Earth's rotation) and I thought that was a crazy long exposure time...my mind is blown
No, there's nothing magical about the exact hours. But 9-14 isn't as big a delta as 8-16 so it won't be as pronounced. However, 8hr vs 10hrs won't be as different as 2 vs 4 hours. 8 vs 10 isn't going to look a lot different.
Excellent video. Since ca,era sensor noise I’d somewhat random, would it be better to combine more images at a smaller shutter speed over the same amount of time or would that not matter. I am assuming the longer shutter speed is still short enough that it does not introduce motion blur.
@@samwarfelphotos of course. My question was whether it would reduce noise more to combine more shots or less shots, each totaling the same time. Intuitively, since noise is random, it would be better to have more shots in order to average out noise.
@@davidligon6088 But having a lower signal-to-noise ratio in each sub will also hurt your overall image... Most serious astrophotographers I know are pushing for longer and longer subs (10 mins, 15mins)... limited only by their sky brightness, since these are much better at bringing out very faint detail than 15 mins of 2 min subs would be.
I'm gonna ask a stupid question... :-) You're basically taking the same picture over and over again. Would it be possible to take just 1 picture and duplicate it 100 times and stack it to get the same result? Except for the noise beeing in the same place, you should end up with enough data to get a good image, or...?
To reduce the noise (graininess) you need to average different photos so that those random fluctuations cancel each other out, leaving the image you want which isn't changing. Like others, I do wonder what the role of ISO is and the noise reduction algorithms used in many DSLR cameras. When I look at images taken with different ISOs, I don't see a big difference in the background noise level (graininess) but the light pollution gets brighter along with the DSO so there seems to be limited benefit to using higher ISOs.
Peter, absolutely amazing Orion view. Please clarify how you did that widefield image. I put the ASI2600 and the redcat 51 in the Telescopius field of view and it was not nearly as wide as your amazing image.....Thanks!
This is interesting. But, coming from a world where I can count my clear, new-moon, not-committed nights without taking off my shoes and socks, I am maybe taking a different lesson. First, that I will rarely get a wonderfully clean image as I will never get multi-night imaging for a single object. Second, the fact that adding a bit more time will not buy me dramatically cleaner images. It would take multiples more time. So, maybe just reconcile myself to interesting, but somewhat grainy images, like Tri-X B&W film when I was a kid.
There is so many mistakes in this video, i don't know where to start. If the two scopes have the same aperture, then they both collect the same amount of light from a single target. The lower f ration scope will have a wider field of view, whereas the higher f ratio will have a narrower field of view. If you are imaging a single target, as shown at 3:37, then the scope with the lower f ratio will have less detail of that target than the scope with the higher f ratio-over the same imaging time, due to the lower f ratio compressing that image over less pixels of the camera sensor. There is so much incorrect information getting around when it comes to f ratios. There is no point in trying to compare f ratios using two far different size apertures.
While this is appreciated and well done, I'm thrown by the notion of data in terms of time. Data is not measures in terms of durration, its measured in terms of volume, such as gigabytes, megabytes,, etc. When using the terms of 4 vs. 8 hours, are you really meaning exposure time? I'm going to suggest that the amount of data is relatively the same, certainly no where near 2X per f/stop. Please clarify this. Furhter, are you suggesting photo stacking, or are you saying a single exposure for 8 hours? Thanks
In other words, or you need a lot of money to buy expensive equipment, or a lot of time to take zillions of subs, or… you go for planetary imaging - a couple of hours, and you’re done, a decent image to show to your friends 🥳 no flats, no guiding, no post-processing headaches! Though, I wish we’d more planets in our Solar System, a couple of dozen or so, better five dozen, just for variety… 😂
Another confusing video comparing f ratios without normalizing the aperture. It f ratios were important to capture light the James Web Telescope would not be f20, and would not have 6.5m meters of diameter.
Remember that noise reduction is a square root relationship. So you need 4x the exposure time to halve the noise. 1 hr to 4 hours, 4 to 16, etc.
Yes but I think, this is not really linked to total exposure time , it is linked to number of images you process :
with 100 images your SNR is 10 times better than with 1 image
And I think, individual exposure time is very important : Peter Zelinka do no speak about individual exposure time ? You need to expose long enough to have sky light > noise on a picture. It happen's after +-120 seconds on a F5.6 telescope and an ASI2600 camera. On earth, the best dark sky in Atacama, send us 22 mag/arc.sec^2. So, I think, there's no way to receive light from an object fainter than that, no matter individual exposure time
@rafberu it is total exposure time. Not number of images. Each image adds noise, the read noise of the camera. So a perfect 10minute sub would have a better signal to noise ratio than ten 1 minute subs.
A stack of 10 total minutes of exposure timen would have half the noise of a stack of 2.5minute totalexposure time (2.5x4 = 10, assuming the same number of sub frames in each stack).
Where people get confused, is that there is a practical limit to sub length. So most of the time when you say "more subs" you are really saying "more total time".
But you can't improve the SNR of any length of exposure time by just dividing it in to more subs to have more files to stack. Only more total signal, meaning more total time, raises SNR.
@@bbasiaga So why stacking ? Please, just Google "what is image stacking astrophotography"
IMO this is the stacking, by averaging the value of each pixel wich increase SNR. I think SNR of 20 x 3mn exposures is more than 4 time better than 1 x 60 mn exposure.
So twice the exposure would reduce the noise to 75% of the previous stop?
@hunter133official square root of two is about 1.4, so 40% improvement in SNR for 2x the exposure time. Definitely noticeable. 1/1.4 = 0.71...so yeah pretty close to 75%.
Just a note on the mathematics of f-stops: These numbers are expressed as ratios between focal length and the diameter of the entrance pupil, thus the slash: f/4. Two lenses with the same focal ratio have the same effective illumination of the image, regardless of focal length or size of aperture. Also, since light gathering ability is a function of the area of the aperture, it is proportional to the square of the diameter. This is why, for a given focal length, f/2.8 and not f/2 is twice as bright as f/4.
The amount of effort that went into collecting the data for this video is astonishing but incredibly useful. Thank you!
That Andromeda photo you showed at the beginning is a great example of total exposure. For someone to discover a new nebula in one of the most photographed dso is astonishing to me. We are truly living in a great age of amateur astrophotography. My father shot in film 40 years ago. When i got serious in this hobby 15 years ago with a dslr, I was more concerned with getting a bunch of objects at 30-60 mins. Now i max out at around 4-8hours since clear moonless nights are few and far between. Good enough for me at this stage.
I know this question will come up in the comments, so here's the data behind the photos:
*Crescent Nebula* - ASI 2600MC Duo, Askar V (80mm + Extender - f/7.5), Optolong L-Enhance, Bortle ~6 | 5 Minute Exposures, Gain 100
*Iris Nebula* - ASI 2600MC Duo, RedCat 51 (f/4.9), Bortle ~2 | 5 Minute Exposures, Gain 100
*Pleiades* - ASI 2600MC, RedCat 51 (f/4.9), Bortle ~2 | 5 Minute Exposures, Gain 100
*Orion* - ASI 2600MC Duo, RedCat 51 (f/4.9), Bortle ~2 | 3 Minute Exposures, Gain 100
*Elephants Trunk* - ASI 2600MC, Askar V (60mm + Reducer - f/4.5), Optolong L-Enhance, Bortle ~6 | 5 Minute Exposures, Gain 100
All images were stacked without Darks, Flats, or Bias. Just light frames.
Thank you. I was wondering what your exposure times were.
From 10:00 to 10:05 as you go through the different exposure times, it seems like you've captured some sort of temporary emission in the nebula. It's just "southeast" of center frame, at a node of the gaseous structure. You can see it brighten, peaking in the 8-hour image, then seeming to fade away again in the next master once the 16-hour timespan averages it back down. Very interesting. I wonder if a time-lapse of the data, grainy as it must be, would show anything interesting!
Hey Peter, thanks for this! The comparison between monochrome and color isn't too hard, even without experimental results. There's a few primary factors to consider:
1) The bayer pattern. Color sensors are primarily differentiated from monochrome by adding the bayer grid which splits each grouping of 4 pixels into 2 green and 1 red and 1 blue pixel. So, you basically automatically get double the green data in every shot, and if you're using filters which block light that falls in one of those color spectrums, then that portion of the sensor becomes useless for that shot.
2) Quantum efficiency. The quantum efficiency of monochrome sensors is much higher, owing to no loss from the bayer filter (even the passband won't transmit perfectly).
So even if you intend to shoot broadband (e.g. for galaxies), you'd still benefit from monochrome sensors (even using the exact same sensor in a mono vs OSC configuration), providing you use RGB filters which are at least as high a quality as the bayer filter would have been, since you can save time by not gathering double the green data that you need.
In cases where you want to add narrowband filters, the benefit is even greater, since you don't get the loss due to doubling green or the general bayer pattern loss, AND you avoid the loss from double-stacking filters.
The benefit of OSC is primarily just cost and simplicity of process.
I was actually just about to comment something similar.
At 13:50, Peter states that if you imaged for 2 hours, "you're technically getting 2 hours of data on your red, green, and blue color channels". As you partially alluded to, this is not correct. Unfortunately, this creates a slight complication to explain as you create something similar to the "full frame equivalence" we see in photography. One huge benefit Peter has that cannot be understated is dark skies. 1 hour in a Bortle 0 or 1 is equal to 8 - 16 hours in a Bortle 4-5 sky. That truly benefits color sensors so much.
Ignoring bayer matrix and QE for now, in 2 hour imaging session, each pixel on the sensor is still technically getting 2 hours of data. The problem is that each pixel has a color, so not all the data counts. For a color sensor, a Green photon falling on a Red or Blue photosite is ignored. Same is true for the other colors as well. This cuts down on the exposure times, as well as the technical resolution. There is also not very much green emission in space so much the sensors resolution is actually wasted.
A way to explain it is that, assuming perfect efficiency, in a given 2 hour session on a color sensor, Green would capture the equivalence 1 hour of data, Red gets 30 minutes, and Blue gets 30 minutes, if compared to a monochrome sensor. This is just a conceptual way to understand it, the red pixels still get 2 hours of data but 75% of the captured photons are thrown away because they fell on a Green or Blue pixel.
For monochrome, all photons of a given filter are accepted by all photosites and contribute to all of the resolution. So while on a color sensor, the red pixels got 2 hours of data, they only represent 1/4 of the image's resolution. Assuming an even RGB filter exposure time, on a monochrome sensor, Red will only receive 40 minutes of data but is contributing 100% of the image's resolution.
Though we don't want to jump into the deep in too much, monochrome sensors also have a much higher Quantum Efficiency. In short, this is percent chance that a given photosite will convert a photon to an electron.
This makes matters even worse for color sensors as we are already losing 50-75% of the photons captured to the bayer matrix, and using the 2600MC as an example (and these are averages as QE is a curve), we then lose another 22.5% for Red (600-700 nm is peak of 90% dropping to 65%), 10% for Green (500-600 nm), and 38% for Blue (400-500 nm peaking at 80% and dropping to 45%), due to QE. So now, we down are to some pretty heavy losses and in reality, technically we still have read noise and shot noise to account for.
So when Peter states that a monochrome sensor would need 6 hours compared to color sensor, that is simply not true. For a 2 hour session, a monochrome sensor would be able to capture the image in less than half the amount of time. Using the 2600 MC and 2600MM as the comparison, in 2 hours:
- the 2600MC effectively captures 30 minutes of Red, and after QE losses, is left with 24 minutes. Green caputes 1 hours and is left with 54 minutes (high green QE on the MC). Blue captures 30 minutes and is left with just 18.6 minutes.
- Assuming an even split of the time for the 2600MM, Red, Green, and Blue each captures 40 minutes. Red is left with approx 28 min (QE for Red on the 2600 is poor), Green is left with 36 min, and Blue is left with 36 min. That is almost 1 stop light increase for Blue, which contributes heavily to DSO.
Most people who switch from color to mono also increase their overall exposure time. But if someone is happy with their color images, they could switch to mono and get the same image (but with more resolution) even faster.
@@Cornerstone_Creative Yes, I didn’t touch on the resolution advantage of mono, but that’s a major factor too. A mono configuration of the same sensor will cut your average FWHM in half vs a OSC shot, since each pixel can be used individually instead of having to group 4 pixels into one “pixel”.
@@Cornerstone_Creative The other way to get better pictures with a color camera is to change the telescope to a faster one. Then you will beat every single mono camera that shoots with slower optics. The other option is if someone takes an 8 hour photo with a mono camera with the same telescope, and you take 16 hours with your color camera, then you will have the same result, if not better. It's all a matter of time and whether the result satisfies you. You save at least the three RGB filters, the expensive filter wheel and a bigger mount to take all that weight. Because the difference between mono and color for an hour's duration is that the mono sensor absorbs 400 photons during that time, and the one shot color camera 240 photons of light. It's not exactly double the difference. This explains why so many people continue to shoot in one shot color camera. It's not better than the mono camera, but if you spend enough time and that's easy because you won't be changing filters at night and refocusing them on each filter, you'll also save money, because I don't know an astrophotographer who got rich from this hobby. Then you will have good results even with the one shot color camera. And if you post your photos on social networks where most people look at them on their phones, then the sleepless mono nights for a slightly better detail which can be seen with a magnifying glass become meaningless.
EXCELLENT video Peter. YES indeed the brightness of the night sky makes a HUGE difference. It boils down to S/N. Each doubling of exposure adds 3 dB of S/N. Thus a darker night sky has the same effect has adding more exposure. One magnitude darker is about 2.512X. If your sky is say a limiting magnitude of 4, and you go to a sky with limiting magnitude of 7 then you get about 15.85X improvement of S/N, or about 4 stops!!! Another way to look at it, it would take 16 hours of exposure under a mag 4 sky as compared 1 hour under mag 7 sky to the same S/N!!!!!!!!!!
This is the best and possibly only complete and best explanation ever. Thanks Peter.
Great video Peter. One small point on the hyperstar. (You said, attach it to the back.) You attach it to the front. I've used the hyperstar for years on my sct 8". It's takes the focal length from about 2000mm down to about 396mm. What it does do for me is give me a wider field of view. Take the same scope, take the hyperstar off attach an OAG and you have a focal length of about 1700mm. Two scopes in one. Sorry I got side tracked.
when you swap out the secondary mirror with the Hyperstar and then go back - do you have any troubles with columniation or does the swap effectively leave the secondary mirror unchanged (since it is being screwed back in as it was before)?
I have been thinking about a Edge-HD 8 to replace my old Meade 8" SCT just for Hyperstar.
Allen, I have had problems a couple times over the years. But easily corrected. The Edge is a very nice scope.
It retains collimation well. I used one on my Edge HD 11 when I first got the telescope and the HyperStar works as advertised. I recently sold the HS for two reasons. Mainly, the time and effort required to change the setup from the HS to longer focal lengths. It is easier to leave the telescope in one configuration. The other drawback was using a manual filter drawer to change filters. It's not too bad because the HS is so fast but still requires manual attention to change filters as opposed to a filter wheel. I have seen some setups where a small filter wheel has been used with the HS with only a small light blockage. I have another refractor set up (complete mount/camera/etc.) at f3.9 for wide angle stuff and leave the Edge set up at 2800 FL for small DSO's.@@allenbaylus3378
The reason Peter said (at least once) that 'at 16 hours you see diminishing returns' is that you need to focus on Signal to Noise ratio. Each image brings in. signal (the light you want) and noise (the stray stuff you don't want). If you take a 1 hour image, it takes 4x more time to double the S/N (2 squared). So 1 hour to 4 hours to 16 hours to 64 to 256...). To see a double from 8 hours is 64 hours. When you compare 8 hours to 16 hours.... you are nowhere near making a dent at improving your image. Just as Peter said when he started the video (paraphrasing again), 'you can't just take 3-4 more hours and expect a good photo'.... well you need to go to 64 hours from 8 hours to see a decent (double) improvement of your image.
Thanks. Of course the question for many of us is “how to get a pleasing image” rather than “a perfect image”. So my question would be “what role does noise reduction software play?” Can you process each set with NR and THEN compare? I find even 30 min shots look “clean” after passing through Denoise or tweaking with Affinity and this saves me hours to look at other nice spots in our night sky! 😊
The basic principle underscoring this is "The Mathematical Theory of Communication." Basically, when you try to transmit a signal, you can change two variables: power and bandwidth. More power means that you're shouting more loudly and are easier to hear. More bandwidth means you can send data faster. A consequence of this, is that for a fixed transmission power, you can guarantee whatever minimum signal quality you want - given you're willing to slow down the data rate (this is how we can still talk to Voyager, which has the transmit power of a 100 W bulb). On the receiving end, this manifests as "integration time" and "antenna gain." The longer you observe a signal with a particular frequency, the more certain you can be that it exists even when washed over by interfering noise. The bigger your antenna, the more of that weak signal you can capture at once.
Thus for telescopes, we are perpetually on the receiving end and have two modulation choices: aperture and integration time. A larger aperture will capture more light per instant, allowing us to gather what data there is more quickly. A longer integration time allows us to slowly pick out the faint signal in the noise. Mathematically, we're limited by the diminishing returns of the logarithmic nature of communication.
Whike the video itself is superb resource for someone like me who's just starting out what really sealed the deal for me is OG Deus Ex music in the background. You are a sir.
I need to recommend Craig Stark's "what do all great shots have in common?" Video. I would say it's the ultimate video in conceptualizing the basics of image noise and what to do to address it.
This has been extremely helpful, I normally shoot for an hour. I’ve gotta up my game. Many thanks and clear skies forever and a day Peter
Best 20 minutes I've spent in quite a while. Thank you!
🎯 Key Takeaways for quick navigation:
00:58 📷 *To improve image quality, double your total exposure time; add more hours of data if your photos appear grainy.*
02:37 🌌 *The aperture of your telescope significantly impacts image quality; wider apertures like F2 gather more light and detail.*
05:38 🌟 *Telescopes with larger apertures gather more light, but focal length affects the apparent aperture size; accessories like the HyperStar can help.*
08:35 📸 *Increasing exposure time from 1 to 16 hours significantly enhances the quality of astrophotography images, especially in capturing faint details.*
11:45 💡 *Light pollution reduces image contrast, making it challenging to capture clean astrophotography; shooting in darker skies is preferable.*
13:34 ✨ *Difference in data collection between color and monochrome cameras: color cameras capture all color data simultaneously, while monochrome requires separate captures for each color.*
14:56 🪐 *The brightness of the target affects the required exposure time; brighter objects may need less exposure time compared to darker and dustier regions.*
17:40 🌃 *Light pollution demands longer exposure times to achieve comparable image quality to those taken in darker skies.*
18:36 📊 *Doubling the total exposure time is a rule of thumb to enhance image quality; beyond a certain point, diminishing returns occur in terms of data collection.*
❤ Good job, thnx 🎉
Another big factor is the quality of the camera. Less noise = a better signal to noise ratio. A camera with very low noise can have the gain turned up further for a comparable image to a less expensive camera.
And, more resolution is not necessarily helpful. More resolution means smaller pixels. Smaller pixels means less light captured in each pixel. Less light = less signal : noise.
It is always interesting to figure the trade-offs.
To add to the complicated equation--I've found that a very fast telescope like a RASA or Hyperstar is awesome to a point.. you may end up collecting the light so fast that your individual images become mere seconds long, and you're dealing with thousands of them. That becomes unwieldy from a harddrive space and processing time standpoint.
Good content Peter. I am using a triple setup one at F4.9, F5.6, and F2 - all with nearly identical FOVs when paired with the proper camera, Your theory is sound logic and the images I am collecting have almost little noticeable noise. Longer integration time does help.
Another well thought out and produced video, Peter. Some thoughts while watching... As a long-time pro magazine and advertising photographer, I'm well acquainted with the basic science behind imaging. It seems to me that given a dark sky, fast sharp optics, and plenty of exposure, another variable is sensor ISO (ASA) sensitivity. Hopefully, in the future, astro cameras can operate at vastly higher levels without introducing grain/noise. I can only imagine flatfield f/2 optics (with the appropriate fl) in tandem tracking all night with a large sensor operating at like a million ISO!
Another thought I've had is: What happens if you simply "copy" a good sub 100 times and stack those vs collecting individual subs over many hours??? I'm sure there's a good reason for the difference.
Your thoughts, please. Clear Skies, Michael
Loved the video. Great practical demonstration of theoretical concepts.
The interesting concept, using two identical scope setups to capture light, has an added benefit… twice as many pixels. The unlikely chance that those pixels from each rig would be perfectly on top of one another means there would be added resolution.
Super video. Food for thought. It certainly helps focus attention on the priorities for improvement. Thank you
Amazing video. Can you please do an HDR version of the Orion nebula? Merge together a let's say 7.5 min up to 8 hours worth of data? So, 7 photos of different exposures to one! Just to save the highlights!
Ok, but tell me WHY it's never occurred to me to run 2 rigs at the same time, on the same object, as a no-brainer hack for getting double the exposure time??! 🤦🏽♀️😅
1hr × 2 rigs = 2hrs of total data
So simple & yet I didn't even consider it until you mentioned it in this video 😂 Thank you for that!
Oh yes.
Like 2x $500 Seestar S50 working roboticaly for 2 hrs, that's 4 hrs integration time! 😍
For 1000 bucks 😂 ❤
@@PureAwareness76What exposure should we use on S50? Default is 10 sec but experimental gives us 30 sec and that seems to be limit of this mount.
@@TJ-ki3gp
As long as you like it! 😊
The 20, 30 secs are in the hidden menu, but only with 10 sec you can do lovely pix! Just make sure to stack it externally, like with SiriL.
In the area I reside, clear night skies are scarce, making it nearly impossible to attain the necessary exposure time. Thus, I heavily depend on noise reduction software, effectively enhancing the image quality equivalent to an additional 2 stops of exposure time. Naturally, it doesn't match the quality of actual exposure time.
A great analysis Peter……more data is my take home now!
Cheers
Si
Best explanation to date in this matter!
👊🏼
Now let’s combine it with the sub exposure argument😂
Seriously. Loved the video….
Very interesting! I learnt a lot from this! You didn't touch on exposure time per sub? Does it make a lot of difference doing a 1 minute sub vs 5 for example?
Excellent video, thank you! And I’m glad to see that we have some of the same equipment (Askar V, ASI2600 Duo, along with an AM5 & ASIAIR+). It makes me feel like I’m on the right track. Now, just need to improve my processing skills.
I meant to ask if you are an educator? You have excellent skills for presenting and making what you teaching, easy to understand.
Peter, not sure if you've covered this before, but might be a good future topic: magnitude. I'm a beginner and had some challenges with Horsehead and Soul Nebulas. I started doing some learning on magnitude, and compared magnitude values for these two targets, versus others where I've had success - that was it. In connection with this video, it seems like some deep sky objects need more time than others?
Hi Peter, just some thoughts here. If your having to deal with some light pollution (glow, wash out), extending the time is just going to further wash out the image. Or over expose. So in perfect\great sky conditions, yes you could extend the amount of exposure time (more light) or use a wider aperture. You need to take into account the possible over exposure. With all of that, are you speaking in terms of total exposure time (50 photos x 30 seconds) or the amount of time per image?
Wow, this video is genius, I'm on my fourth watch. I did some shooting a few nights ago in the GSL desert and had this very question - if I'm already getting good results at 30 minutes, why do I need more data? Thanks for making the connection between exposure time and aperture vs. noise reduction and clarity - bringing f-stops into the equation makes so much sense. Your layered examples of doubling light showed the principle perfectly. The question which remains for me is the third point on the exposure triangle: ISO/gain. How bright should my light frames be? I am shooting with a mirrorless Nikon full frame which is terrific at high ISOs. Even if I take somewhat noisy shots at ISO 6400, the grain seems to get sampled out in stacking. If my telescope is a slow f/7, how much of a compromise is boosting the ISO if I'm going to average the noise anyway? Thanks for the great video.
ISO/gain is just the same set of relations: Higher gain mimics longer exposure, greater aperture and wider f/ratios. The price paid is in grain & noise. Kodak Tri-X was fast film at 400ASA, but it was grainy.
Peter, were you broadcasting this video from the bridge of the Starship Enterprise?
Great video! Nice examples.
I really love my samyang 135mm f2. With a dedicated camera is a killer!
❤ Lovely, thank you!
You see; INTEGRATION TIME + POST PROCESSING. That's it.
That's why I gave up with my desired and 💲 William Optics Zenithstar 61 + mount + guide scope + guide cam... even with my existing M4/3 Lumix GX80, the cost would be at least £2000...! No way.
Just go for the $500 Seestar S50 ❤ Integration time + post processing. Period 🎉
Excellent information. Thank you!
So for something like the Orion nebula, what is the proper exposure target? Do you expose for the bright nebula or let that get blown out and try to expose for the dust? Or, do you expose for the bright part AND the dust and then combine in post? Or... Does the stacking just reduce the noise in the image so that you can see more of the detail when stretched? That sounds likely, but not really sure.
Peter,lovely review. Almost makes me want to get a RASA. This is also why I’ve been using my Samyang135 at f/2.0. Unfortunately, from Bortle 6, I’ll still need 16+ hours for the Spaghetti Nebula (SH2-240).
Great video! Really easy to follow and useful
Hey @peter great video. What about attacking that imaging time from another vector - gain. In photography, the exposure triangle is aperture, shutter speed and ISO. But in AP, you rarely hear anyone talk about using gain to compensate for less imaging time or longer f/ratios. Being new to the refractor space, I’m finding the imaging time required for even f/5.6 is substantial. I just got an f7.5 system and I shudder to think the number of hours it will require to get a clean image.
Great explanation sir, thank you.
Thanks Peter! Very well put together, and brought home focal length, aperture, f - ratio, light quality, integration time as a single relationship - with both logical and visual examples.
I consider this as some of your best work. Kudos!
Thanks for this interesting comparison. Which influence does the exposure time of the individual photo have? I.e. 300 seconds of exposute time vs. 600? Best, Claus
thanks for this video.
I want to use a 600mm lens with f6 on my camera. With a 2x teleconverter I get 1200 mm. My camera is quite insensitive to high ISO values. Does it make sense to use the teleconverter and increase the ISO?
what exposure time and what ISO would you recommend?
Hello Peter. I'm a professional astronomer. I find some interesting concepts in your video, but unfortunately, you mix and confuse light collection with f-ratios. F-ratio tells light density in the focal plane not how much light is collected.
Light collection from an object in the scene is proportional to aperture area times
exposure time. It has nothing to do with focal length or f-ratio. F-ratio is not in the equation.
For example, which collects more light from M51, a 50 mm focal length f/2.8 lens or
a 200 mm focal length f/4 lens?
A 50 mm f/2.8 lens has an aperture diameter of 50/2.8 = 17.86 mm. Area =
250.5 square mm. A 200 mm f/4 lens has an aperture diameter of 200/4 =
50 mm. Area = 1963.5 square mm
The 200 mm f/4 lens collects 1963.5 / 250.5 = 7.8 times more light in
the same exposure time for any object in the scene, whether a galaxy,
a nebula, a star, a bird in a tree, or a persons face. Bin the 200 mm
pixels 4x4 and the resolution in terms of pixels on the subject would be
the same as in the 50 mm image, but the light in those pixels will be
7.8 times brighter in the binned 200 mm image.
Try this with your 11-inch telescope. Choose a target like a galaxy that fits on your sensor in the f/10 configuration. Take one image at f/1.9. Take another at f/10 with the same exposure time. Bin the f/10 image by summing 5x5 pixels. You'll find the same amount of light per binned pixel and the same pixels on the object (within 5% because the f/1.9 is not a factor of 5 from f/10). You state in the video (at about 5:50) that change to hyperstar increased light collection by 25x. But the binning demonstration shows that the light is there, just distributed differently.
Better to computer signal per square arc-second or arc-minute. By focusing on the subject, it will become clearer what the variables for light collection are.
For example, redcat 51 (51 mm aperture) vs Celestron 11-inch (279 mm aperture) ratio = (279 / 51)^2 = 29.9 times more light from any object in the scene, e.g. a star, a galaxy, a square arc-second, a square arc-minute.. It has nothing to do with f-ratio.
On the plus side, at the end of your video you talk about buying a larger telescope, but unfortunately you don't explain correctly why.
I'll end with a comparison to Hubble, JWST, and other professional telescopes.
Hubble and JWST are great deep sky telescopes. Hubble is an f/24 system, and the WFPC3 camera operates at
f/31. JWST is f/20.2. I have done most of my professional work at terrestrial observatories with the NASA IRTF on Mauna Kea, Hawaii (f/38)
and at the U Hawaii 88-inch (2.24 meter) f/10 telescope. By the flawed f-ratio ideas in this video, a redcat 51 (51 mm aperture diameter) with f/4.9, or your 11-inch hyperstar (f/1.9) would collect more light than these huge telescopes. NOT. Key is to computer the light per object area, like per square arc-minute.
Quiz: assuming the same wavelength of light, how much light per pixel do the cameras on JWST, and Hubble collect per pixel compared to your redcat 51 with your camera (f/24 vs f/31 vs f/4.9, respectively), assuming the same sensor quantum efficiency?
The LSST, Vera C. Rubin telescope is going to only take a pair of 15-second images per position (in each filter) and is expected to come online in January 2025.
en.wikipedia.org/wiki/Vera_C._Rubin_Observatory
It is an f/1.25 system. Do you really think that your 11-inch f/1.9 telescope with 1 hour or 16 hours of exposure time will collect more light from NGC 6888 in your video than the LSST in 30 seconds?
Again, the key to light collection is aperture area times exposure time.
thanks for the video.
Does doubling the ISO correlate to doubling the "stop"?
Thx Peter ❤ it so helpful video.
Why not start with actual aperture (size of entrance pupil) then the same size of the same light intake size will give the same signal regardless of F-stop or focal length. Focal lenth will just change the magnification of noise and object (worse F-number). Downsize the image of the longer focal length image will lower noise. Actual aperture size is the signal strength and affecting the noise PER DETAIL per the actual photographed object (nobula/planet). Not the focal ratio number (f-stop,). Larger aperture (entrance pupil) decreases actual noise per real object (nebula). Better (lower) f-stop (focal ration) just lower the noise per image. Not per same object photographed.
Yes! This is what I've been trying to figure out as well. Is it accurate to say that f-ratio gives snr per pixel, while aperture gives snr per arc second? Although the first produces pretty pictures, it could be at any focal length, and if it's super low, you won't be able to resolve much detail. So it's really aperture that matters rather than f-ratio at the end of the day (depending on your goals, but mine is to resolve detail)
4:01 man got Internet Historian flashback of his Fallout 76 review, lmaooo.
But yeah, this is good stuff!! Have a SVX130T with a ZWO294 camera with a reducer. And having that reducer makes me feel a lot better lol. Still very new, but excited to learn more
Some people have 4 rasa setups. That will be crazy amount of light gathering capacity
Very interesting stuff! I’m still trying to decide on my first scope and assorted accessories. Just curious, what’s the longest exposure Astro photo you’ve ever taken?
I use the redcat 51 and like it a lot. It's a bit pricey for your first scope, but the size is manageable, and you will get good images.
Hey great video. How many minutes are you using to take the photos?
Peter, its been many years since I studied the f stop math. You show it as doubles ie 2 4 8 16 etc. my understanding of f stops is the difference between f stops is not doubles its logarithmic so a F11 takes 84 times as long to get the same light a f2 gets in one hour?
Also to reduce the noise by half you need to use squares 2 4 16 etc..
The Hyperstar is MAGIC!
The Hyperstar is just a reducer that REALLY reduces.
So…if I have an F/2 Rokinon 135 and an F/2 Celestron 8 would the brightness be similar?
Yes, same image brightness
The answer depends on the brightness of what. The 8” Celestron concentrates its light gathering power onto a smaller area. So if we talking about an evenly illuminated sky, they both collect the same total number of photons, but the Celestron is collecting those photons from a smaller area of sky at a faster rate per square degree of sky. So if you’re talking about a small target that fits in the field of view of both scopes, the Celestron will have much higher light gathering power for that target. It will capture more photons of that target per second. An analogy is a flashlight that has a lens that allows you to adjust it to a narrow beam or a wide beam, in both cases, the flashlight is the same brightness, but when you focus it with a narrow beam, all that brightness is focused on a small area and that small area is illuminated much more brightly. That’s like the Celestron, but in reverse.
How to use a Tracker and mirrorless camera for my astrophotography. Have you completely switched to telescopes from dslr/mirrorless cameras? If so I will find your videos less interesting but try to always watch them.
Great job. thank you very much
Very cool video, I have learned a lot. THX
Hi Peter, is the grain "strat" working in that way already considering the presence of darks, flats n biases or you exclude it?
Can you tell us what the individual sub exposure times were?
Great comparisson!
Wow, you've lost me immediately at "add a stop". Been astrophotographing quite a while now but have no idea what that is. Is that a DSLR thing? Gonna have to look that one up...
I was wondering if I do 3 hours for each LRGB filter, is that considered as 12 hours total exposure? Does different filters adds up?
what are your feelings on filters - like L-Extreme as a solution for mid bortle zones (bortle 5) over a long trip to a bortle 3 site?
Just curious, when you say you have 2 hours of overall data what would you estimate is the avg time for the individual photos before they are stacked? For example i hour of data could be 60 60 second exposures or 4 15 minute exposures if I am thinking correctly.
Gostei das dicas. Estando perto da cidade o celular já substitui o rádio e dá pra deixar no viva voz fora da foto😅
How would I double my data that I get. Let's say I get 1 hour of imaging. Can I double that by running the photos through deepsky stacker twice to get 2 master images then stack those master images to get 2 hours of data. Or would that not work?
my camera can trace the stars for 5min (using ibis to shift the sensor to match Earth's rotation) and I thought that was a crazy long exposure time...my mind is blown
What about camera gain or ISO?
Great thanks🎉
So does this mean there is almost no difference in results between 9hrs or 14hrs? It won't make a material difference until you get to 16hrs?
No, there's nothing magical about the exact hours. But 9-14 isn't as big a delta as 8-16 so it won't be as pronounced.
However, 8hr vs 10hrs won't be as different as 2 vs 4 hours.
8 vs 10 isn't going to look a lot different.
Very interesting and informative and totally makes sense😊
Excellent video. Since ca,era sensor noise I’d somewhat random, would it be better to combine more images at a smaller shutter speed over the same amount of time or would that not matter. I am assuming the longer shutter speed is still short enough that it does not introduce motion blur.
When tracking and guiding there should be no motion blur
@@samwarfelphotos of course. My question was whether it would reduce noise more to combine more shots or less shots, each totaling the same time. Intuitively, since noise is random, it would be better to have more shots in order to average out noise.
@@davidligon6088 But having a lower signal-to-noise ratio in each sub will also hurt your overall image... Most serious astrophotographers I know are pushing for longer and longer subs (10 mins, 15mins)... limited only by their sky brightness, since these are much better at bringing out very faint detail than 15 mins of 2 min subs would be.
@@samwarfelphotos thanks!
How long were your subs on Orion?
I'm gonna ask a stupid question... :-)
You're basically taking the same picture over and over again. Would it be possible to take just 1 picture and duplicate it 100 times and stack it to get the same result? Except for the noise beeing in the same place, you should end up with enough data to get a good image, or...?
To reduce the noise (graininess) you need to average different photos so that those random fluctuations cancel each other out, leaving the image you want which isn't changing. Like others, I do wonder what the role of ISO is and the noise reduction algorithms used in many DSLR cameras. When I look at images taken with different ISOs, I don't see a big difference in the background noise level (graininess) but the light pollution gets brighter along with the DSO so there seems to be limited benefit to using higher ISOs.
Peter, absolutely amazing Orion view. Please clarify how you did that widefield image. I put the ASI2600 and the redcat 51 in the Telescopius field of view and it was not nearly as wide as your amazing image.....Thanks!
This is interesting. But, coming from a world where I can count my clear, new-moon, not-committed nights without taking off my shoes and socks, I am maybe taking a different lesson. First, that I will rarely get a wonderfully clean image as I will never get multi-night imaging for a single object. Second, the fact that adding a bit more time will not buy me dramatically cleaner images. It would take multiples more time. So, maybe just reconcile myself to interesting, but somewhat grainy images, like Tri-X B&W film when I was a kid.
thanks you bro very nice
It could have been great if you measured actual SNR for each picture.
There is so many mistakes in this video, i don't know where to start. If the two scopes have the same aperture, then they both collect the same amount of light from a single target. The lower f ration scope will have a wider field of view, whereas the higher f ratio will have a narrower field of view. If you are imaging a single target, as shown at 3:37, then the scope with the lower f ratio will have less detail of that target than the scope with the higher f ratio-over the same imaging time, due to the lower f ratio compressing that image over less pixels of the camera sensor. There is so much incorrect information getting around when it comes to f ratios. There is no point in trying to compare f ratios using two far different size apertures.
Steve Morris's Rasa triplets.
While this is appreciated and well done, I'm thrown by the notion of data in terms of time. Data is not measures in terms of durration, its measured in terms of volume, such as gigabytes, megabytes,, etc. When using the terms of 4 vs. 8 hours, are you really meaning exposure time? I'm going to suggest that the amount of data is relatively the same, certainly no where near 2X per f/stop. Please clarify this. Furhter, are you suggesting photo stacking, or are you saying a single exposure for 8 hours? Thanks
Basically this is the inverse square law.
In other words, or you need a lot of money to buy expensive equipment, or a lot of time to take zillions of subs, or… you go for planetary imaging - a couple of hours, and you’re done, a decent image to show to your friends 🥳 no flats, no guiding, no post-processing headaches! Though, I wish we’d more planets in our Solar System, a couple of dozen or so, better five dozen, just for variety… 😂
Thanks for the video. However it doesn't actually answer the question, other than "just use trial and error"
Why'd you Todd Howard me? 😢
Another confusing video comparing f ratios without normalizing the aperture. It f ratios were important to capture light the James Web Telescope would not be f20, and would not have 6.5m meters of diameter.
If it's taking 8 or 16 hours of summed data to "smooth" out the image, then your subs are way too short.
Some folks say 6 frames, others say 20 frames, others a 100.
Why do people make these kinds of videos and then do not answer any of the questions being asked??? 😵💫