Becausr the full-res version links went down often yesterday, I'm hosting it on my IPFS node, here's the link (for those on IPFS, if you can afford the bandwidth please pin it to spread the load :) )
https://ipfs.io/ipfs/QmVLTMHtLRhnft3QspDx4qTJeXY6hiib1j77UfQ...
Alternatively, if you have a slower connection I made a lossless webp here which should result in the same output at about 1/10th the size: http://files.jjcm.org/mosaic.webp
As an amateur astronomer who has looked directly at the moon through a telescope hundreds of times, this photo looks strange and artificial, though I know it's not. I pondered why this was for a bit, and came to the conclusion that its otherworldly quality is actually due to being superior to what one can see directly. At any one moment, direct observation will have artifacts of the air moving between you and the target object. This picture quashes all of that at once, lending an "uncanny valley" feeling I think spurred by the fact that my brain just isn't used to that level of fidelity looking at a real object. This work is fantastic.
No no, it is strange and artificial. The stars are comped in from something else, and the moonglow looks like a gaussian blur. Still, it's pretty, and has hopefully inspired people to go stare at the moon.
To be fair, most astrophotography is "enhanced" one way or another - I can spend hours screwing around with a single DSO shot after stacking.
Here's a quick and dirty 30 second reconstruction using a shot of the moon from last summer, acquired in much the same way as OP - except using an EdgeHD 14 with an EOS 7D in video mode. It's pretty crap as last year was my first foray into planetary imaging - I usually do DSOs.
Amateur astronomer here as well that has also done a bit of lunar photography. (envious of your EdgeHD 14...)
I made a similar comment about the moonglow, but thinking a bit I think it's a combination of:
- Higher dynamic range than the human eye (combination of stars, moonglow, and high contrast lunar surface both in shadow and in sunlight)
- High resolution imaging, and the digital signal processing aspect (noise reduction, deconvolution, etc.)
They combine to make something that looks kinda like reality but is really a mix, and maybe triggers a little bit of the uncanny valley effect.
PSA: Update your Transmission. A good chunk of you are running 2.92, which is susceptible to CVE-2018-5702[0] (which allows basically anyone to write arbitrary files to your computer from a webpage).
Also, cool trackerless magnet link, worked a treat.
bittorrent is the smartest way to share any file that's being viewed by lots of people at the same time. I am continually amazed why it's not used more often.
Aside from the nontrivial security mechanisms that would be required to implement it safely, is there another reason why native browser support for torrenting doesn't exist yet?
This is great; as another pointer, http://clarkvision.com is my go-to source for beautiful astrophotography with technical explanations. Look for astro gallery and tutorials if interested
A quick word of caution. For anyone who might be interested in astrophotography I really wouldn't recommend any of the tutorials on the clark-vision site. This is what the co-author of an astronomical image processing program had to say about them: http://forum.startools.org/viewtopic.php?f=4&t=912
"The "articles" exhibit a litany of errors, nonsense and misunderstandings of how some basic algorithms and mathematical concepts work in image processing, including (but sadly not limited to) light pollution removal, deconvolution, the commutative property in mathematics, stacking, linearity vs non-linearity of data, basic physics and color theory.
After politely inquiring why he thought any of this was good practice or scientifically accurate and getting nowhere, even after demonstrating the mathematics and writing code to disprove some of his more fanciful claims and pointing out errors and untruths. The fact that I'm having to put this warning up, gives you an indication that that didn't go very well."
This can definitely be true and I personally find that the writing style of Clark's tutorials could use some polish -- his random switches between scientific and colloquial styles within a single tutorial can annoy both engineers and artists. I also heard that he does not like debate that he deems repetitive and is not be the friendliest opponent to his critics.
That said, the criticism above is very scarce on details. What exactly is the nonsense about mathematical concepts that he is objecting to? Roger's data describe his metrics, methodology, data and conclusions (not to the level required to publish in Nature, but better than most websites on the topic). One can argue with either or all of those, but such arguments should include the rigor at least to the level used on Clark's site.
That way we can understand what are some reasonable simplifications that are OK even if not strictly correct (e.g., when we tell middle school kids that equation x^2 = -1 has no solutions without into complex numbers or axioms of R) and which are genuine, major errors.
But the strongest argument going for Clark, IMO, is his stunning pictures. If his methods describe how he produced them, I am interested in reading more. My 2c, corrections welcome.
> That said, the criticism above is very scarce on details. What exactly is the nonsense about mathematical concepts that he is objecting to?
There are examples in the thread I linked to.
> But the strongest argument going for Clark, IMO, is his stunning pictures.
I've seen countless amateurs produce far better images with much cheaper equipment. He also misses the point of modding cameras, improving Ha sensitivity allows more structure to be captured. A case in point, this is my quick and cheap North America nebula with an old modded DSLR and Ha filter: https://c1.staticflickr.com/5/4490/37423232795_8a37a7ecbf_h....
I tried opening the image, but it made my entire laptop unresponsive[1]. So let's hope the graphics manufacturers catch up with the pixel trent before those 8K monitors become mainstream.
[1] On Chrome for MacOS. Granted my macbook is a couple years old, and the Apple graphics drivers are garbage.
Try firefox, it has an integrated decoder-downscaler, i.e. it never has to keep the full-size image in memory before downscaling to the target resolution. But that does not help if you zoom to 1:1.
My laptop is very much bog standard, but gimp had no problem loading it. Generally this may more come down to software implementation, rather than drivers
I don't have too great a desktop (i5 from more than 4 years ago (launch date was apparently 2011) and 16 gigs with more than half already used, no graphics card) and I could open the 291M file just fine in viewnior.
I do have an nvidia card, but since MacOS 10.14 the nvidia driver manager reports that my driver (387.10.10.10.40.105) is not compatible with my OS, and it also reports that there are no newer drivers available when using the update function. So I'm currently back at the default Apple driver.
I have ran the nvidia driver before MacOS 10.14, and it did feel slightly faster than the driver supplied by Apple. I doubt if it will improve opening an 8K*8K image in a browser though...
I mean would the display need to know - just GPU drivers should be able to send a 4k signal that has 4px squares each of one colour? Interesting though.
Well they're usually bigger. IMHO 1080p for monitors maxes out at about 22-24" before it starts to look blurry, after that you want 1440p, until about 28", which is when 4k starts to make sense.
By that logic Apple's Retina displays would be worse than non-Retina displays. Down-sampling decreases eye strain and makes the image smoother. It is only an issue when dealing with fine lines that our eyes watch very closely, such as in text.
If you turn your apple retina display to 1080p mode, then I would think it's not that good when there's a 1:1 correspondence between physical pixels and pixel data. However when the size stays the same, I wouldn't expect much difference.
A 32 inch 4k display downsampled to 1080p, well... :)
A down-sampled image has less noise than an image in its native resolution. You can experiment on your own monitors if you don't believe the theory, just remember to use font smoothing for non-native resolutions. You can also ask about it on the appropriate stack exchange, it is a frequent topic of confusion.
This is one of the advantages of shooting at a higher resolution than the intended target delivery resolution. When downscaling an image, the noise gets 'averaged' out of the image. For HD 1920x10280, a 2.7K resolution was a good choice before having to jump all the way to 4K. There's other technical reasons as well for shooting larger than intended use, but just re-enforcing the 'noise removal' process of downscaling an image.
Pretty much every download link is down, but the torrent is working: magnet:?xt=urn:btih:j6ixrzpitvpmkx5tgroy6utcpqr5hhhj&dn=updated%20mosaic.png&xl=304863698&fc=1
Assuming “lossless”, the main reason is you don’t want to “lose” information. For example, if someone wanted to actually use this picture to geo-reference (or would it be Lunar-reference =p) a particular reference a particular feature, you would want it to be accurate. Specific use case: you found some ice, and want to know exactly where it was, as well as composition. If you didn’t use a lossless format, you wouldn’t have confidence it wasn’t as artifact of the compression.
Edit: misstated Leo instead of lunar, Leo is low earth orbit
This is so beautiful, thanks! I'll cut some pieces to be used as desktop backgrounds. The southern part of the moon and surrounding space if turned upside down would make a really nice one.
BTW, I spent some time zooming around the stars and objects and some of them look rather interesting, at least to a completely astronomy illiterate like me, such as the closest objects to the moon at roughly 2 o' clock.
If anyone wants to look at this image properly then you need OpenSeadragon and VIPS.
OpenSeadragon enables you to view this image at full resolution without your device keeling over, VIPS enables you to make the image tiles required to get OpenSeadragon to work.
OpenSeadragon has only ever made it to the academic world, in this age of Instagram people want average resolution selfies that take three seconds to look at. OpenSeadragon enables a deeper look.
Openseadragon may make sense for huge images, not for merely large ones as this one, that fits comfortably in memory. Besides, not everyone is keen on using "web-based" programs just because.
The UI in that link looks really old. If you want the more modern, 3D view, I'd recommend going to maps.google.com, switching to satellite view, and zooming out.
Tangentially, I recently was using Google Earth, and zoomed all the way out so that the background stars, Milky Way, etc could be seen. Anyone know how accurate that scaling is of the background relative to Earth is? What caught my eye was the scale of Orion.
nomacs is a free, open source image viewer, which supports multiple platforms. You can use it for viewing all common image formats including RAW and psd images.
This makes me lose my shit every time I want to see a simple jpg image in Windows 10. WHY?? Even my 14 year old linux laptop can open 100+ MB images faster than W10.
Components of the OS are 20+ years old, it seems. The image handler is one of them. Simply right clicking on it takes 12 seconds for the menu to pop up.
GLiv [1]. Firefox and GLiv both take ~5 seconds to load the image for me, although whilst Firefox takes another ~5 seconds to re-render after zooming in to 100 %, GLiv is instant.
No, it's partially artistic, partially light scattering in the Earth's atmosphere from the moonlight and the exposure settings he used. The moon does have a very tenuous atmosphere but nothing thick enough to cause this effect.
Are you referring to Earth's atmosphere as an air sea? Is that an actual term being used? I couldn't find anything about this term in a quick ddg search.
It’s just a poetic expression I borrowed from somewhere (the original was in another language). The sibling commenter is right, ”ocean of air” is probably more commonly used in English.
Not even close. (81 megapixels is 9000 by 9000, and the moon only takes up about the middle 1/9th of the image, so it's about 3000 pixels across. So that's over a km per pixel at the very front of the moon, and even worse at the edges where the surface is at an oblique angle to our view)
Also, I can't remember the source (maybe it was a "What If?"), but I seem to recall that it's actually to create a lens that allows us to focus a point on the moon to allow us to make out current man-made items on the surface of the moon.
Also made it available in torrent form (with the IPFS URL as a web seed) http://www.urlhash.com/124410/mosaic.png.torrent