While working on the article about 4 galaxies and 4 clusters captured in one picture from the Pixel 8a, I got curious about the star trails when zooming in. Making some assumptions, and trusting Google’s calculations for the angular size a pixel covers, the results seem to make sense.

Here’s the details, starting with the unprocessed picture from the phone’s camera. For more about the picture itself, see the article mentioned above. These were saved in the phone as JPEGs and not the raw DNG format.

Here’s the same zoomed in view as shown in the other article.

In a 4 minute exposure the image shifts due to Earth’s rotation by about 1 degree, which is about half the width of the zoomed in section above. The software in the phone builds that image up by stacking 16 second subframes; the image shifts by about 6 arc-minutes in that time period.

The effective pixel size on the 8a (according to Gemini — it seems like it calculated this correctly) is 62 arc seconds (about 1 arc minute) as the phone bins 4 physical pixels (I’m skipping the details of the quad-Bayer mode that the sensor is operated in).

In 16 seconds (one sub-frame) for stars near the celestial equator, which runs from outside the top right to just inside the bottom right side of the first picture, we would expect to see trailing of around 5 or 6 pixels and oriented vertically with a slant to the right to match the imaginary line of the celestial equator.

That would seem to match what we see, given that even the dim stars appear spread out to cover a few pixels. As we move away from the celestial equator the actual motion in the sky will be less; a star at the celestial north pole would note move at all.

To help visualize that we can see in the 12 hour exposure below that stars close to the celestial north pole in the picture center trace out a very small circle compared to those at the edge.

In the first picture Cassiopeia is in the upper left and those stars are at 60° north, so we would expect they would have a very small — almost unnoticeable — trail.

Of course there’s also other ways to get trailing, including motion of the camera or its mount from wind, foot steps, or errors in the stacking alignment from the software being used.

We can check another photo, this one was taken in May of 2023 with a Pixel 7 when Ursa Major would have been almost directly overhead. The pixels on the 7 are larger than the 8a, the same calculations come up with 72 arc seconds (1.2 arc minutes) so we’ll expect a trail in the 4–5 pixel range along the celestial equator. The top center of the picture is about 20° N of the celestial equator and the bottom center is about 80° N of the celestial equator.

The results from this second photo also seem to agree with the predictions. Always a chance that this was all coincidental or explainable by some other mechanism.

What’s needed is the confirming experiment: On the TODO list has been mounting the phone on a barn-door type mount as that would remove the star trail problem across the short 4 minute exposure times. Alternate putting the phone on a scope on a tracking mount. If set up alt-az there will be a small amount of field rotation but that would be very small over the 16 second sub-frames.

Note: Google Gemini was used for the the Pixel 7 and 8a information. Any other mistakes are all mine.

Starting with a picture taken with a Pixel 8a on August 29th at approximately midnight from Cape Cod, Massachusetts, USA. If you can spot all eight of the objects right from the start then skip this article; everyone else read on for how to find objects in your own pictures. There’s also a 5th open cluster that’s findable in the image, but it’s small and non-obvious.

The unprocessed picture above was taken in Bortle 4.5 skies — looking East out over the Atlantic. This direction has considerably less light pollution than to the west at the observing location. Still enough light pollution to leave behind a grey fog that also shows the vignetting that’s typical of the limited optical systems in cell phones. Oddly it’s not symmetrical. There’s no provision in the phone’s built in photo software to fully correct for vignetting.

The next picture was lightly processed in Affinity Photo 2 to remove most of the vignetting. There’s a subtle greenish background in the center and the vignetting on the Pixel 8a is not quite symmetric, though the Milkway runs down the left side of the photo so it will be naturally brighter than the right side.

The processed views will be used for markups, though for zoom-ins on the galaxies and clusters we’ll use the original as with the simple processing done here some detail will be lost.

The Pixel 8a can save photos in raw (DNG) mode but I overlooked enabling that; some JPEG compression artifacts will show up when zooming in.

For a discussion about star trails and why they show up in some portions of the Pixel 8a images please see this article.

The first step when hunting for little deep space goodies in these pictures is having a reasonable idea of what constellation(s) are in the picture you took. There are ways to do plate solving on-line that automate the process, but that takes the fun out of it. It also helps to know what might be both bright and large enough to actually make out in the cellphone pictures.

Cellphones have a wide field of view and pixels cover 1 or 2 arc minutes of sky. Dim stars are going to be spread out in to a circle a few pixels in diameter from the a combination of factors including the very small lens diameter, marginal coating on lens surfaces, and general light scatter. There are very good reasons why a real camera lens of decent diameter costs more than a typical cell phone.

This means larger bright nebula, galaxies, and clusters are probably going to be able to be captured by your phone. Objects in the Messier catalog — which start with the letter M, are generally a good place to start.

In the case of this picture the constellations at the top were known to be Cassiopeia on the left and a portion of Pegasus on the right.

This meant that in this part of the sky M31 (Andromeda Galaxy) would be visible as it’s large (around 1 degree angular width) and bright (naked eye visible from a dark site).

I like using Stellarium https://stellarium.org/ as a guide for both figuring out what I’m looking at and for planning observing sessions. The desktop version is free and holds large star catalogs totaling 220 million stars, which is helpful to have when zooming in.

The first step is to make a copy of the picture and using your photo editor (the ones here are edited with Affinity Photo 2) add constellation lines and labels so you can figure out where in the sky you are looking.

This is best done with two monitors, one with Stellarium running and the other with your picture. Start with the most obvious bright stars in the picture and match to what you see in Stellarium. This also works for being outside at night; it can be easier if you’re in a light polluted location as then only the bright stars are visible.

Once the constellations are known, objects in Stellarium can be looked for in the cell phone photo. Over time you’ll develop a sense about what and where to look for deep space objects. In the interest of keeping this article of reasonable length we’ll just present the next picture identifying the objects as a fait accompli and then look in more detail at the objects.

After M31 the galaxy M33 (Triangulum Galaxy) is the other obvious smudge. We want to make sure we can match it to what Stellarium shows.

M33 is about 3 million light years away versus 2.5 million for M31. However M33 is smaller, only 60,000 light years across where as M31 is about 150,000 light years across. M31 appears at an angle to us and M33 is almost face on.

By zooming in in M33 we can compare the stars captured by the phone with what Stellarium shows to verify we have identified the correct object.

Looking at the grouping of stars we see to the left of M33 and comparing to the objects shown in Stellarium we can determine that it’s the open cluster NGC 752. At 1400 light years away it’s close enough to be spread out considerably in angular size; this spread could be aided by its age of 1.3 billion years which means it has started to drift apart compared to young open clusters.

The other two open clusters in the picture are NGC 869 (7460 light years away) and NGC 884 (7600 light years away), collectively known as the Double Cluster in Perseus. They appear within the Milky Way’s disc so there are considerably more stars in the background and foreground compared to M31 or M33. Despite their distance their young (14 million years old) bright stars make them naked eye visible from a dark site.

The double cluster, galaxy M31, and NGC 752 all are visible in binoculars from typical suburban light polluted areas. M33 and the last three objects are too dim or small to show with binoculars unless from a truly dark sky site.

The next object of interest is open cluster M34.

For a challenge try and locate the open cluster NGC 663 in Cassiopeia. Its smaller size makes it difficult to find and therefor it’s not included in the count of the four obvious open clusters.

We now turn back to M31 to find the last two galaxies in our quest: M32 and M110, which are two satellite galaxies of M31. M32 is about 8000 light years across and M110 about twice that at 17000 light years across.

Unlike the other deep space objects that stand out in the picture as being something that’s not just stars, dwarf galaxies M32 and M110 would be very easy to overlook if you didn’t know what they were.

The same steps used on this picture can be used with any astro-photo you happen to take with your cell phone. There are portions of the sky with more large/bright objects than other portions. Generally any area that includes the disc of the Milkway Galaxy will have a lot of objects to find. For example the Sagittarius region is a very rich one to go deep space object hunting.

You do need to know your starting point, i.e. the constellations your phone was aimed at. Stellarium can help you figure that out — assuming you know how to find the cardinal directions. If that’s a struggle too then get a compass — or use the one your phone has!

As you accumulate more experience and knowledge in finding things you’ll be able to quickly find objects. Following up with binoculars or even a small telescope will let you see the object for real and not just an image on the screen.

No AI used — I don’t need help hallucinating.

Also important is living far enough south that the nova is above your horizon. Canada farther north from the US border is mostly out of luck.

It’s interesting to see a new (to us, in real life the white dwarf behind it is several billions of years old) star pop into visible existence. If you happened to save some older pictures of that part of sky you can compare them to show the change.

Constellations have Latin names, so Lupus (wolf) becomes Lupi for the possessive (the wolf’s) form.

The first step is to have clear skies. In the Boston area we’re quite literally going weeks between a single clear night; smoke from the Canadian wildfires has made the lack of clouds moot on some of those nights. The more or less continuous cloud cover has hampered ideas I’ve had for article topics. Until Friday night, with a still brightening nova, clear skies, and being on Cape Cod (Massachusetts, USA) with easy access to a darkish beach to see down to the horizon to the south.

Nova occur in a binary star system where one of the members has evolved to the point of being a white dwarf and is orbited by a companion that’s close enough that the white dwarf can draw material from the companion. When enough hydrogen builds up on the surface of the white dwarf it goes off in a run-away thermonuclear explosion.

In some cases this can happen periodically; we’ve been waiting a year now for the star T Coronae Borealis to erupt in a nova and possibly becoming bright enough to see from even moderately light polluted skies.

There’s several dozen nova per year in the Milky Way galaxy, most are too faint to see as they’re far away or obscured by the many dust clouds in the Milky Way.

About once every one to two years a nova happens close enough and becomes bright enough to be visible without a telescope from a dark sky location, and perhaps once every 20 years that one becomes bright enough to be seen from suburban locations away from light polluted city centers.

V462 Lupi was discovered on 12 June 2025 and appears to be at peak brightness around June 21st. While it may become too dim to see at your location without optical aids (binoculars would work) it can be captured by a cellphone in astrophotography mode for at least several more weeks. Capturing pictures over time will let you see it fade out as newer phones should record down to magnitude 10 or better.

Pre-nova the star’s brightness was magnitude 22.3, and reached around magnitude 5.6, an increase of 16.7. In a previous article the visual magnitude scale used by astronomers was illustrated; all we need to know is that one magnitude step (smaller means brighter) is a factor of 2.5.

For V462 Lupi this means it’s now 2.5 ^ 16.7 = 4.4 million times brighter than a month ago!

Here in Massachusetts V462 Lupi appears at most 7 degrees above the horizon, which means there is a lot of air to look through versus looking straight up. Stars this low in the sky are dimmed by about one magnitude (factor of 2.5 times) as well more atmospheric turbulence means the images are not as sharp, which also makes the stars look dimmer. Along the southern US border V462 Lupi would be around 20 degrees above the horizon and the dimming less than 1/2 a magnitude.

Now the fun part: comparing two pictures to find V462 Lupi. If you want to figure it out for yourself you can download the next two pictures.

First is a picture taken in May of 2023.

The second is taken on June 20th 2025 at a location about a mile away from where the first was taken from. A different phone as well as newer camera software means the unprocessed images have different coloration and vignetting. The second picture was taken with the Beastgrip lens discussed in an earlier article. That add-on lens doesn’t help much with light gathering as the phone’s aperture is too small, but it does provide magnification.

It took 5 attempts to get a picture that wasn’t photo-bombed by a plane on its way to Europe.

If you would like to find V462 Lupi in the 2025 picture then stop here.

If you want to see the results of finding the nova then read on.

Looking at pictures of large parts of the sky can make it difficult to get oriented. The next two pictures add constellation lines and label the star HR Lupi, which is about half a degree south of the nova.

Also known as HR5624 or HD133880, this young star is variable optically as well as exhibits unusual radio activity.

Visually comparing the brightness of nova V462 Lupi with surrounding stars and looking up their magnitudes in Stellarium, it would appear that the brightness is around magnitude 5.4 to 5.5. This ignores dimming from the airmass which dims the nova and the comparison stars by almost 1 magnitude from Cape Cod.

Here’s a zoomed in view around HR Lupi.

Here’s an animated GIF that compares the 2023 picture with the 2025 picture.

In addition to the direct cellphone picture a Seestar S50 (electronic telescope) was used to take a picture of the nova. There’s hundreds of articles on the S50 out there so we’ll skip the details here.

Since the S50 is controlled by a cellphone and you can only view images electronically it will be considered fair game for Astronomy with a Cellphone, though it is something extra. OTOH a lot of the prior articles used the add-on lenses so there’s precedent to go beyond just your cellphone. There’s a less expensive but almost as good — in terms of image quality — model called the S30.

HR Lupi is around magnitude 5.8, and nova V462 Lupi looks a little brighter than that.

The Pixel astrophotography mode also makes a small animation out of the 16 second exposures it takes to stack to produce the final 4 minute exposure. The short (1 second) video clips can be stacked end-to-end to make a longer video; it also helps to cut the frame rate in half.

To take a smooth movie you need to start the next astrophoto immediately and that wasn’t done. Also the Pixel tends to make the first frame exposure differently so you sometimes see a brighter frame.

The boats are being illuminated by light spill from a well lit house (if it wasn’t Chatham you would call it a mansion) a few hundred meters from the beach area.

After taking the nova pictures a few more were taken of the general Milkway area in Sagittarius. A sporadic meteor in Scorpius was captured, which was also seen visually.

Author’s note: No AI was used for any part of this; I can screw things up just fine without the help of a computer.

Three objects made a chance alignment tonight. Despite the closeness in the night sky the reality of their distances and sizes is quite different from what a simple “common sense” view of our ancient ancestors might have thought.

The weather has not cooperated in a long time with articles I’ve been working on. Rather than continuing to wait for cloud free nights just going to go with what I have. Global climate change is getting worse and for New England it means even more clouds in the future so continuing to wait isn’t going help.

Despite clouds the waxing gibbous moon, Jupiter, and the red giant star Aldebaran were all visible in a row tonight. It’s also an interesting contrast in distance. Currently the moon Is 228,000 miles away, Jupiter is 398 million miles away, and Aldebaran at 65 light years, or 382,100,000 million miles which is about a million times farther away than Jupiter.

It takes us about 6 years to send a spacecraft to Jupiter, to reach Aldebaran at a million times farther away would take longer than the human lineage has existed on Earth.

These three objects also contrast in size. The moon clocks in at 2,160 miles in diameter. Jupiter is around 87,000 miles at the equator. Aldebaran, being a star similar in mass to our own Sun but having exhausted it’s hydrogen supply in its core is now in its red giant phase. It has an estimated diameter 45 times our sun’s, or 39 million miles. In our solar system this would extend about 2/3rds of the way out to Venus. Our sun will distend like this in about 4.5 billion years.

The funny greenish orb in the picture is an internal reflection of the moon in the camera lens. The green color is typical of antireflective coatings applies to lenses to lower light loss in the optical path. Phone optics can suffer from many problems and this is not unexpected for a picture like this. The moon is very bright compared to the other things in the picture so even a small percentage of reflected light from the lens will be visible. If the moon had been centered in the frame then the reflection would have overlapped the overexposed moon and not been visible.

This picture was taken with a Pixel 8 mounted on a tripod and a 3 second exposure in the phone’s night mode.

For the next few weeks the moon will take turns appearing near bright Mars, Jupiter, Saturn, and Venus. Can you spot any 3 in a row alignment with bright stars? How about 4 in a row?

In the previous parts the performance of the Pixel 3a was compared with the Pixel 7. In this last part the performance improvement of adding an external lens will be evaluated against the Pixel 3a with a lens as well as the native Pixel 7.

Completing this process will help in future evaluations of what astronomy activities might suitable for a Pixel 7.

The Pixel 7 includes a new Astro filter that can enhance some astrophotos taken with the phone, like the Milky Way picture in the first article on the Pixel 7. On pictures of star fields it seems to enhance the background brightness to a degree that the overall visual appeal of the picture is less.

To keep things simple the pictures presented here do not include the Astro filter. Ursa Major will be used for comparison as it was overhead (May 2023) and used for the comparisons in part 2.

[I am *that* far behind in terms of material vs. articles written. I’m clearly not one of those people that can crank out articles every few days. Or maybe they use AI stuff? This is all me writing here, some might say no intelligence, artificial or otherwise.]

The pictures were taken from Chatham Massachusetts, which nominally has Bortle 4 skies towards the east (Atlantic Ocean) and Bortle 5+ to the west. (For those not familiar with it, Bortle 1 are skies unaffected by light pollution, and Bortle 9+ is Times Square) Here’s the Pixel 3a with Sirui lens. Straight off the phone we see pretty good contrast between the stars and background.

Here’s the same area, taken concurrently with the Pixel 3a picture above. Sky brightness, particularly towards the center, is very noticeable now. Likewise in the corner the vignetting is also more obvious. Despite the increased background brightness we get the sense that the Pixel 7 captured a lot more stars.

To save flipping back to the earlier part, here’s the Pixel 7 without the lens. The vignetting is obvious there so as a first guess we won’t blame the Beastgrip lens. Comparing with the Pixel 7 with lens picture above it’s obvious that the lens has extended the sensitivity.

We’ll pick the area around Mizar/Alcor (middle star in the handle) for a more detailed comparison. It’s near the image center where optical distortions should be less so should allow for a truer comparison.

As a refresher, Mizar is a binary star with an estimated 5000 year orbit that is easily split in a small telescope. The two visual components are spectroscopic binaries (i.e. too close to separate in a telescope but their binary nature can be discerned by looking the the spectrum of the stars), making it a quadruple star system. The latest Gaia dataset places Alcor (itself a binary) closer to Mizar (perhaps one-half light year) than prior measurements (over a light year) and possibly gravitational bound to it. The whole system is a little over 80 light years away.

The comparison

Looking at the comparison between the 7 and 3a (with lenses) there’s really no comparison in terms of what the Pixel 7 can do. 4x integration time and a more modern image sensor, plus the slightly larger lens, leave the 3a in the dust.

The objectionable difference is the 7 does not subtract the sky background from the image as well as the 3a seems to. Given that most people live in light polluted skies this seems like a rather significant oversight in terms of capability to produce visually good astronomy images right from the phone.

Of more interest is what magnitude objects can we get down to with the Pixel 7 + lens combination?

Here’s the comparison of the area between Mizar and UMa 81 without magnitude labels, and with magnitude labels from Stellarium marked on the Pixel 7 capture.

From the last image it seems reasonable to say the Pixel 7 with the Beastgrip lens can get down to magnitude 11, after that it becomes a judgement call to say it clearly captured the stars. If we compare with the Pixel 7 without the lens (see Part 2) the addition of the lens seems to have gained us a little over one magnitude of detection capability. We also gain from the magnification in that we can separate closely spaced stars a bit better.

There is a bit of a mystery here; the star with the red magnitude 12.3 label looks like it’s closer in brightness to 10th magnitude. Its J2000 coordinates are 13h31m52.34s/+55°11'45.6", here’s a link to the Simbad entry for it which labels it as TYC 3853–549-1, though shows slightly different J2000 coordinates. Not something I’m familiar with to be able to explain. Doesn’t seem to be on the AAVSO lists. (something to come back to some other day).

A look towards the edges

The right side of the two images look at the same set of stars. We’ll zoom in to the area near 19 (Xuange) and 33 Boötis. The background is similar between the two.

The stars are definitely elongated and not as well focused as the center, it’s clearly visible zooming in the same 340% used in the zoom a few pictures back.

Using Stellarium as our reference, we annotate the Pixel 7 photo with some magnitudes.

The cutoff magnitude near the edge seems to be about 1.5 mag less than in the center. There’s definitely less faint smudges and the ones that can be made out all are matched with stars < 10.6 magnitude where as in the center we could sort of make out > 12 magnitude.

This is something to keep in mind should we decide to estimate stellar magnitudes; we would need to stick to comparison stars from the same area of the pictures and absolute magnitudes would be expected by the same amounts we’ve estimated for the limits.

Closing thoughts

Now that we have a sense of how dim objects can be and be in range of either the Pixel 7 or Pixel 7 with an add-on (Beastgrip) lens (mag 10 vs. mag 11) it would be interesting to see how accurate the brightness values are as visual brightness can be used to measure variable star periods and exoplanet transits.

There’s a couple of things that could be making the results a bit worse than ideal. Seeing conditions may have been different in the lens of vs. lens off scenario. The pictures were taken the same night and within a few minutes of each other, but New England weather is anything but constant.

We don’t know anything about the stacking algorithm; occasionally images from the Pixel 7 show some noticeable trailing when zoomed in. With no visibility in to what Google’s software is doing we can only make guesses about what might be happening. Alignment error is the obvious thing, we don’t know how/what it picks for reference stars.

The camera is fixed so the stars will appear to move a bit across the sensor during the subframes. Earth rotates at 15 arc seconds/second. For the 16 second subframe (though maybe this number varies, have not found a clear statement on this) that means the stars move 4 arc-minutes.

Mizar and Alcor are separated by 12 arc minutes. Looking at the pictures clearly they are not smeared out 4 arc minutes. The pair at the one o’clock position relative to Mizar are not separated; they are 2 arc minutes apart.

How the Pixel 7 avoids trailing is not clear; shorter subs help but there’s diminishing returns as noise becomes dominate.

More than one phone review website says the Pixel 7 has optical image stabilization; one could imagine some very clever algorithms to track out the earth’s rotation across each sub frame. That idea would have to be wrong; if the phone did that it would fail with the lens attached (different apparent movement) or when the phone is attached to a motorized scope (no movement).

Across all of these pictures star images are rarely sharp — how much of that is seeing conditions versus the phone’s autofocus not really being right for this? The Google camera app is set to far focus, but phone cameras have overtravel to allow for factory variation at infinite focus. There’s no adjustment possible in the Google photo app.

A mystery to me, if anyone knows what the phone and/or software is doing to avoid smearing the stars please put it in the comments.

The same was hoped to be true for the Pixel 7. The aperture size looks bigger; it’s assumed that an external lens could help increase sensitivity. In the next article the performance of the Pixel 7 with the lens will be examined on actual astrophotography in detail.

This post will look at the Beastgrip setup that was used to provide an external lens for the phone.

For those that read the older article for the Pixel 3a or are familiar with the Moment phone case and (Sirui) lens described there, the first question might be “Why didn’t you reuse that lens with the Pixel 7?”

That was the plan, but after getting the Pixel 7 discovered Moment didn’t offer a case for the 7, and after contacting Moment support found that plans for the Pixel 7 were on hold as the existing lenses apparently didn’t work well with the newer optical path design of the Pixel 7.

Moment has since announced new cases and new T series lenses that will work with the Pixel 7, but at the time this article was written (Oct ’23) not shipping and when the Beastgrip was purchased (Feb ’23) not even announced.

What is the Beastgrip system?

Full details are on their website; their approach is definitely a more complex one in mounting the phone to a large frame that has numerous attachment points for sound, lighting and control accessories.

While the frame mount makes it more less universal, it lacks the convenience of the Moment system where you can just add the lens to your phone with the simple bayonet mount built in to the custom case. The Beastgrip mount is something you have to plan to use versus the Moment case/mount is always on your phone and you can just pop on the lens, which is small enough to fit in a jacket pocket. For iPhone Beastgrip does offer a smaller frame style mount specific for lens use.

The Beastgrip mount provides a 37mm threaded lens attachment point, and with enough adapters it presumably would allow use of any lens that would be optically compatible with the cell phone.

For the uses planned here, the only lens of interest was the Beastgrip 1.7x telephoto lens.

Beastgrip Pro frame — close but no cigar. It’s still the only choice.

The basic idea is you “pop” your phone in to the spring clip area, screw on the lens of choice, and you’re good to go.

Unfortunately the reality of it has not been that. A couple of pictures of the frame, this time taken by me.

The lens location in the Y (vertical) direction is adjusted by loosening the two thumb screws and sliding the holder up and down until its aligned. Aligning in the X (vertical) direction is done by just sliding the camera body left or right.

The obvious question being how do you know you have the lens holder optically centered over the phone lens? This seems to be oversight number one: not providing alignment marks on the lens holder or even better a snap on piece with reticle marks to ensure the optical path is centered.

From trial and error it does seem like a small misalignment — like perhaps less than 0.5mm, doesn’t affect system performance. Given that exact alignment could be solved with a US $0.25 piece of plastic this seems like an obvious omission for a > $100 product.

Oversight number two is likewise just as obvious: while the Y direction is locked in to place with the two thumbscrews, there is no adjustable stop to lock the phone’s X position. Every time you put the phone in the Beastgrip holder it has to be realigned in the X-axis to get the phone and add on lens centered.

Having to futz with the alignment every time the Beastgrip setup is used, as well as lack of a way to ensure the centering is correct, really makes the Moment system look like a much better thought system with regards to lens attachment. The Moment system doesn’t support adding accessories or the convenient handgrip; we’re strictly evaluating the lens attachment.

When using the Beastgrip to take pictures or video, the frame does make the process much better — easier to hold the camera level and no fear of fingers in front of the lens.

Number three is killer

The first two oversights, lack of an alignment mark or reticle, and no adjustable stop in the X direction, while annoying, aren’t fatal flaws, the third one really is and makes one wonder who exactly signed off on this design.

Perhaps you noticed in the earlier pictures the Pixel 7 is naked, not in a case. For this clumsy user, that case is an Otterbox Defender Pro, which is a hard plastic frame with a rubber over frame.

The phone’s Z axis position, i.e. how close or far away it is from the lens, is determined by the body of the clamp area that holds the phone to the frame.

The Beastgrip comes with a flat shim installed in the clamp area, adding the shim sets the phone back further from the lens.

After removing the shim, if a phone is in its case the phone is held too far from the lens. For the Pixel 7, the only way to get the Beastgrip frame and lens system to work is to remove the phone from its case.

The Otterbox case is designed to protect the phone, not easily come apart. This is a 100% avoidable problem had the Beastgrip been designed with the phone holder a few more mm towards the front, and shipped with an assortment of shims to set the height correctly. Certainly it would have added minimal cost to avoid users having to remove their Pixel 7 phones from their case.

But wait, is it really that bad?

Yes.

Open these pictures in their own browser tabs so you can zoom in on them.

With the intentional misalignment some vignetting is visible, as well as the tree branch focus is softer. A shift in the Y direction seems to have a bigger effect, but some of that could be the 4:3 aspect ratio is going to show more problems on the left and right side before it shows on the top and bottom of the picture.

6mm (about 0.25") is a pretty obvious misalignment. The good news is that a small optical path centering error doesn’t seem to create noticeable errors.

OTOH the lack of a lock or stop for the X axis means that you can unintentionally slide the phone left or right when using it. Checking alignment after each shot would be overkill; when using this setup at night and not wishing to use lights that might destroy one’s night vision, the uncertainty of the Beastgrip holding alignment is a concern.

One last gotcha

The Pixel 7 has a focus system that uses infra-red (IR) light on the front that does not shut itself off in astrophotography mode —this is not needed since in astro mode the camera is set to infinite focus.

The IR can reflect off the lens and end up in your picture.

The rubber shield on the Beastgrip almost covers the sensor; you may need to add some IR blocking material. Sometimes the rubber turns under a bit, sliding the phone back and forth can help to get the rubber to be flat.

This isn’t a fault with the Beastgrip other than it would be great if the provided shield could deal with this. In a perfect world the phone wouldn’t waste energy turning the emitter on when not needed.

At some point in using the phone the Pixel 7 does stop turning on the IR emitter and this reflection has not really been a serious problem.

The verdict?

If you’ve sensed the disappointment that this product is so close in some ways but lacks that final “well how does it work in the real world” finesse then that pretty much sums it up. For now, the Beastgrip is currently the only reasonable choice for adding a basic lens to the latest generation (2023) phones.

The real determination of the Beastgrip’s value will be in the phone’s performance when taking astrophotos, which is the subject of the next post.

That area of the Milk Way is the largest and brightest section as we’re looking in towards the center where the density of stars is higher as well as the center galactic bulge can be seen sticking up on either side of the disk portion that we see edge on from Earth.

From comparing the two pictures we could get a feeling that the Pixel 7’s astrophotography capability is improved from that of the Pixel 3a.

To be useful for our astronomy investigations we would like to not only understand that the 7 is better than the 3a, but also know in the absolute sense what its capabilities are. Mostly this is looking at the dimmest stars that it can record as well as the closest spacing of two stars that it can still resolve as being two objects.

Ultimately the fact that the phone compresses the images — even in RAW mode — limits what data can be determined from them as the (apparent) compression may change the brightness, size, and color balance. The JPEG compression algorithm was never meant to handle images of stars and nebula with any sort of accuracy.

To compare the 3a and 7 directly, both phones were set up on tripods and aimed at the same portion of overhead sky in Chatham, MA. This location is the easiest one for me to access that has reduced light pollution compared to the Boston suburbs of my home. Even this location on Cape Cod is suffering from increased light pollution in the past 10 years; the Milky Way used to be very obvious and well defined.

At the time when these pictures were taken (May 2023) Ursa Major was directly overhead. With its bright stars as a clear reference point it’s a good constellation to pick for comparing with desktop planetarium software like Stellarium.

The next 3 pictures are best viewed full screen, the best way to do this is to right click on them and select “View image in new tab” or whatever is equivalent in your browser. If expanded by clicking on them in Medium they can not be zoomed in to the full resolution.

Comparing the 3a with 7 picture, there are more stars visible. The sky background from the 7’s picture is brighter, and the vignetting is obvious. Though some of the central brightness may be related to the phone’s image processing software, so far nobody seems to have measured the 7’s performance to that level of detail.

For more about vignetting, and to see why it’s a problem for low f ratio lenses like those in cell phones, there’s a good article on the Edmund’s Optics website.

The phone’s Astro effect filter, which aided the Milky Way picture in the prior article, makes the apparent vignetting even worse on a stars only picture.

What is lacking on the Pixel 7 for in phone processing is a useful background gradient removal tool. At least two options are needed, one for an overall constant and the second for a radial removal. Some gain correction for vignetting could also be helpful.

Looking in more detail

Zooming in on the handle of the dipper portion of the asterism, the difference between the two phones becomes very obvious.

The stars from the 3a appear unnaturally colored, and by comparison the 7 appears almost monochrome.

At an 800% zoom in we can compare stars shown in the phone’s capture with the the stars plotted in planetarium software like Stellarium.

Taking the prior photo and labeling the stellar magnitudes where the convention of not including the decimal point is followed.

For the Pixel 3a magnitude 8.0 stars just barely show. With the Pixel 7 stars up to mag 10.5 are obvious and the limit seems to be a little more than mag 11. The difference in brightness between mag 8 and 10.5 is a factor of 10 (for an introduction to stellar magnitudes see this earlier article).

The 4 minute exposure time of the Pixel 7 should provide a 4x improvement, so the other 2.5x improvement must be coming from the image sensor on the Pixel 7, with perhaps a slight amount maybe from the lens on the 7 gathering more light as it appears to have a slightly larger diameter than the Pixel 3a’s lens.

Estimating the vignetting

We can look at the RGB values of the background to try and understand more about what the phone is doing. We’ll sample a star free area in the upper left, middle, and lower right, using paint.net software. For the Pixel 3a image:

Using the V from the HSV color space (V = Value, roughly like the luminance or overall brightness) we can see there’s a significant difference across the image. A similar set of measurements for the Pixel 7:

The Pixel 7 background is twice as bright in the center as the Pixel 3a. What we can’t be sure of is if that center brightness is some artifact of the phone’s processing or it’s the result of actual vignetting of a sky background that is not totally dark and made brighter by an exposure 4 times as long. Presumably the Pixel 7 does need to scale the image brightness to prevent over saturation of the stars.

Looking at the upper left, the mag 6.5 star HIP48742/HD86012 measures in the Pixel 7 capture as:

where we see it did not saturate the pixels (i.e. RGB values of 255 or close to it). Towards the center of the captured frame another 6.5 magnitude star is selected, HIP1703/HD119992 (actually listed as mag 6.45).

This star seems to have saturated the detector in the center, so its’ not out of the question to assume that the brighter center of the Pixel 7 capture is in part due to vignetting from the Pixel 7’s optical path.

To investigate further we would need to take a picture of a uniformly lit piece of white paper. This is a step that astrophotographers call capturing a flat-field, as it can be used to correct for the image sensor showing uneven illumination even though the image source is all at one brightness. For now we’ll kick that can down the road — flat fields are not easy to create, particularly here where we would prefer to have a very dim source so we could capture it in the phone’s astrophotography mode to figure out what, if anything, the phone’s software is doing.

Star colors

It was already noted that the Pixel 3a pictures seem to have overly colorful stars, where as the Pixel 7 results seem close to monochrome. It was noted that for a lot of the stars captured with the Pixel 3a the color in the captured picture, in some cases, corresponded to the star color from the publish B-V color scale for the star.

There are also a lot of stars that the Pixel 3a rendered with a greenish tint, which isn’t possible in real life.

Using a color camera for accurate star imaging is problematic at several levels. Astronomers gather data using monochrome camera and specialized color filters where the characteristics of the camera, the filters, and the optical path is fully documented.

Such information is not readily available for cell phone cameras, but despite that we can still use our cell phones to make some measurements and observe astronomical phenomena.

Next up

An external lens was used to improve the sensitivity of the Pixel 3a. That lens couldn’t be used with the Pixel 7, so a new one was purchased. The next part compares Ursa Major between the two phones using their respective lenses.

To be notified when that article is published you can subscribe to this channel.

In this multi-part article series we’ll start with a simple qualitative look at a picture of the Milky Way. In part 2 we’ll do a more quantitative look of the 3a and 7 by taking a picture of the same portion of the sky at the same time. The last part repeats that experiment but with aftermarket lenses attached to the phones.

To get a sense of how the 3a and the 7 compare, here’s the first picture I took of the Milky Way with the 3a back in February 2020. It was taken from a dark sky location (Negev Desert, near Mitzpe Ramon, Israel, Bortle class 3 skies).

Next is a recent picture taken with the Pixel 7a taken from a not as dark location — foreground terrain is clearly lit up by local light pollution. (Chatham MA, USA, Bortle class 4/5). More of the brighter parts of the Milky Way are above the horizon in this second picture so it’s not quite an exact comparison. To get a better sense of the differences viewing the pictures full screen is suggested; a zoomed in portion is also included after the next picture.

Even at the default article image size that Medium uses there are obvious differences. The star colors in the 7a picture look more natural, and overall the picture looks like it has more detail.

Both phones show vignetting in astrophotography mode. It assumed the phone software tries to compensate for that, and with day time shots there’s enough light to render it a non-issue. In astrophotography mode it doesn’t seem to be fully corrected, which is probably good as with light polluted skies the results might be worse than no correction.

A closer look

We can see the differences between the 3a and 7 better by clipping out a section with Antares on the right (and M4, the little fuzzy patch to the right of Antares) and 51, 41, and 42 Ophiuchi are the three bright ones on the left. This subsection spans about 17 degrees.

The odd star colors from the Pixel 3a are more obvious in the zoom in. Looking at the stars themselves they are much smaller and look better focused, as well as there are clearly more visible. Though the bright stars seem to be a little oval and other pictures show what is either some trailing or misalignment in the stacking process. Noise in the Pixel 7 appears to be less as well.

The more stars and less noise improvements would be expected if we could do a native 4 minute exposure with the Pixel 3a. The increased brightness of the background sky would also be expected with the longer exposure, though some of that is also expected from more light pollution where the Pixel 7 picture was taken.

The dust clouds of the Milky Way that obscure the background stars look sharper in the Pixel 7 picture, though they do not appear much brighter.

M4, the little fuzzball to the right of Antares, does not appear brighter in the Pixel 7 picture. While the longer exposure time does make diffuse objects brighter relative to the background noise, the phone’s stacking software also has to reduce the brightness as four times as many photons are being captured for the image, but you don’t want the bright stars to saturate.

M4 is the closest globular cluster to Earth, at about 6000 light years away. Antares is a red supergiant star, with a diameter around 1 billion km (approx 600 million miles). If placed in our solar system it would extend out to somewhere between the orbit of Mars and Jupiter. These are just two examples of the dozens of interesting objects revealed in the cell phone picture.

For an image processed in the phone with no user intervention, and just with the phone on a tripod, these results are still in the amazing category. For those that do specialized astronomical imaging the raw camera images typically don’t look like much, and a lot of processing steps are needed to get the near Hubble quality that grace personnel astronomy websites now.

Image sensor comparison

These numbers are found on several websites.

Pixel 3a:
4Kx3K sensor (12.2 Mpixels), 1.4u pixel size. f/1.8 lens with 28.3mm equivalent focal length. Sony Exmor IMX363 sensor.

Pixel 7:
8Kx6K 50MP sensor with Quad Bayer color filters is roughly equivalent to a 4Kx3K sensor. 1.2u pixel size. f/1.85 lens with 25mm equivalent focal length. Samsung GN1 sensor. The Samsung website states that 4 pixels are combined, so the equivalent pixel size would be 2.4u.

Determining noise and sensitivity of the two parts would need an in-depth look at the sensor datasheets; it’s probably a safe bet that the newer Samsung part is more sensitive and less noisy than the older Sony part.

In phone processing

The Pixel’s photo app now includes an Astro enhance feature for astrophotos. This option only shows up on the phone and not in the normal Google photo web page editor options.

When applied to the prior Pixel 7 picture the contrast is improved quite a bit.

The first picture, without enhancement, is close to what could be seen visually for the Milky Way, though with more dim stars in the photo. The enhanced picture is what we might see if it wasn’t for the absurd amounts of light pollution that now exists pretty much everywhere; a view our great grandparents would have had from anywhere but city centers.

The zoomed in view on the enhanced photo shows better detail of the dust clouds, as well as the brightness of the dim stars has been increased.

Since this enhanced mode is part of the Pixel 7 phone’s basic software it may be used in future articles, though the image description will make it clear if the phone’s enhance feature has been used. Also it’s reasonable that Google might change the feature in later releases, so two pictures from different dates might look different for that reason as well.

With the Pixel 3a it did seem like later software versions somewhat cut down on the false star colors observed in the early versions.

Milky Way structure

For those that haven’t contemplated what our home galaxy actually looks like, it is described as a barred spiral when viewed from above — or below — as there’s no up or down in space.

Since our solar system is in the plane of the disk that holds the spiral arms, and our view across the arms is blocked by dust and large numbers of stars, it’s difficult to be certain of the exact morphology beyond 10,000 light years out from Earth.

What are all those stars?

Living in our light polluted world, it’s easy to get confused when you see a picture that is closer to what we should all be seeing. For those that have gone to a true dark sky site (Bortle class 1 or 2) it can be disorienting at first. That disorientation turns to annoyance at the idiots in charge (bad street lighting, most businesses - with car dealers getting a special place in hell, homeowners that think their house looks good lit up like a penitentiary, etc.) for allowing this pointless and avoidable environmental destruction to happen.

As noted in earlier articles, using a picture you’ve captured and then comparing it against a desktop planetarium software package is a great way to expand your understanding of the night sky. Here’s the same portion of the sky from Stellarium with constellations and some star names displayed.

When trying to match photos to the simulated image, start with bright stars (like Antares) or unique asterisms (like the teapot in Sagittarius). Keeping the same orientation and image scale between the photo and simulated view is also helpful. On wide field views the image scale may be off in the outer third as both the phone and the planetarium software have to make tradeoffs to render a horizon to horizon dome on to a flat monitor with limited size.

Up next

In part 2 a more quantitative investigation of the Pixel 3a versus Pixel 7 performance is made.

If you can’t remember that then it’s also easy to derive from remembering that there are 365 calendar days in a year and 24 hours in a day. As an approximation, a chosen star should appear in the same spot in the sky at the same time once per year. 24 hours divided by 365 days works out to about 4 minutes/day.

A year is not exactly 365 days

Approximate was used in the above as the length of Earth’s rotation period does not work out to an integer number of Earth rotations before the Earth is back to the same position around the Sun relative to a fixed reference point.

The Julian year is defined as 365.25 days of 86,400 seconds each and is the basis for astronomy calculations, but does not have a simple relationship to Earth’s orbit around the Sun or the exact position of where the Sun or other star appears in the sky after one Julian year.

The Sidereal year is the more familiar time period and is the time it takes for the Earth to come back to the same position around the Sun and it’s average value is 365.256363004 days in length. Average because the time it takes to go around the Sun (i.e. 360 degrees of orbit) varies over time in part because the barycenter (the gravitational center of the solar system) varies over time. Unless you’re a Qmoron-flat-earther, in which case nothing goes around anything and it’s all a conspiracy created by alien lizards...

Details for the constellations and other things that can be seen in the video are at the end of this post.

In the prior post, Cygnus is in the northeast and therefore looks like it’s rising and arcing to the north. By comparison, Scorpius is toward the southeast and will follow an arc to the south in its travel through the Northern latitude’s night sky.

This can be illustrated in planetarium software such as Stellarium. In the Northern hemisphere the equatorial grid is centered near Polaris, which is why it is also called the pole star. Its position does not appear to change during the night (it traces out a circle only 1 degree in size) which makes it a good reference for knowing which way north is.

Identifying things in the picture

The Pixel 7 can integrate a 4 minute exposure to create an image that shows more stars than the unaided eye can see. Pre 2023 posts in this series were made with a Pixel 3a which is not as light sensitive as the Pixel 7. For the Scorpio time-lapse and the picture below a Beastgrip 1.7x lens was used with the phone, which further increases the sensitivity of the phone’s imaging. A future post will detail the setup and performance of the phone + lens combination.

If you zoom in on the full size image you will notice slight trailing of the stars. This would appear to be from the length of the subframes the Pixel 7 uses and that it’s not aware a lens has been added — nor can you tell the phone a lens is added. This would make the stars appear to move more in a subframe period than without the lens.

There’s no right or wrong way to imagine the figures that the stars represent in Western culture. There was considerable variation in constellation boundaries and star names until formalized by the International Astronomical Union in 1930 in to 88 constellations and a specified list of star names.

Scorpius is one of the historical constellations of the zodiac. If you want to annoy someone when they ask what sign you were born under, say Ophiuchus, as that’s where the Sun is from late November until mid December. Ophiuchus is part of the zodiac now, i.e. the Sun appears in that constellation along with 12 other historical ones. Part of Ophiuchus can be seen in the upper left of the caputured image.

Other goodies in the picture

The most obvious star in Scorpius is Antares, a red supergiant star. This means its near the end of its stellar life cycle, and at an estimated mass of 11 to 14 times that of our own Sun, would if placed in our solar system extend out past the orbit of Mars. Its large surface area makes it appear brought in our sky even though its 550 light years away. At that distance our own sun would be invisible to the eye except in a larger amateur telescope. Anatares is actually a binary star, its companion is too close and too dim to be seen except in a larger telescope.

The next object of interest is the globular cluster M4. At only (estimated) 7200 light years away, it’s the closest globular cluster to Earth. At 12.2 billion years old it’s also one of the oldest object captured in this picture.

A second globular cluster, M80, also shows in this picture. At 32,600 light years away it’s the farthest away object in this picture and as a result appears more like a star than the fuzzy appearance of M4. Its greater distance makes measurements more difficult but it’s probably comparable in age to M4. For an interesting history of this object and associated pictures see this article by Ethan Siegel.

One of the interesting features Google added to the camera’s native astrophotography mode is the phone will condense sub exposures of the overall 4 minute exposure down to a 1.5 second long video.

It’s pretty easy to see that the stars have changed their position in the sky even in 4 minutes but the little snips leave you wanting more. There’s no way to automatically create longer videos, but you can push the picture button every 4 minutes and then concatenate the resulting files later. Windows 10’s default Photo program was used to create the clip, and speed was reduced by 50% to create a 15 second clip from 20 minutes of elapsed time.

Taken from Cape Cod, Massachusetts looking East out over the Atlantic ocean, we get a great view of Cygnus rising. Unfortunately a few planes on their way to Europe photo-bombed the clip. Suggest viewing on a PC monitor/full screen to see the individual stars. YouTube sometimes seems to sometimes start the clip in standard def, you may need to open it directly in You Tube (why is this stuff so fragile?).

One immediate observation of the Pixel 7 is it’s not holding white balance across the sub-frames. Unfortunately the phone’s camera software lacks manual controls to deal with this and other problems. It’s still pretty amazing what the Pixel 7 can do and we’ll dive deeper in to that in some future articles.

If you’re having a tough time figuring out the constellations, here’s an annotated picture. The Pixel 7 makes 4 minute exposures. There is some light pollution in this area of Cape Cod but you can see a hint of the Milky Way despite it being relatively low to the horizon.

The phone takes short subframes because if it just did one 4 minute exposure the resulting picture would show lines instead of pinpoint stars. By keeping each exposure to 10 or 15 seconds (not clear what the exact number is) the apparent motion of the stars is small enough that they don’t appear as streaks. The phone software then stacks sub-frames by aligning them on top of each other so that the star images can add together. This also helps average out the noise from the image sensor. There’s other processing that the phone performs but so far there’s no documentation provided on the details.