Will Barton

Sunday, Aug 16, 2026

On Sustainability

It’s been a hot minute since I checked in on the W3C’s web sustainability work. I briefly joined the working group before other events interfered with my professional life. Given that big tech has thrown their sustainability goals out the window, I want to do what little I can in my little corner of the internet. So I went back down the sustainability rabbit hole to see if I can estimate what this website emits and how I can make improvements.

I got to a place where I could calculate this site’s emissions at an average of 0.11g CO2e per page. But that doesn’t tell the whole story, because my photo galleries pages are a lot heavier (more and bigger images). The pages under /photography are 0.15g CO2e on average, but the “heaviest” is 1.454g CO2e.

So, I want to go through how do we get to those numbers, because there are a couple different sites to check that don’t tell you a whole lot. I also want to how to reason about them, how to make sense across a site and particular pages.

What’s in a number?

W3C’s Sustainable Web Interest Group has a description of the version 4 methodology for calculating the average emissions per page view.

Very briefly, they’re using known values for operating energy consumption for data centers, network, and user devices, and deriving energy consumption from the production of the data centers, network, and user device hardware as a portion of global hardware manufacturing. Those numbers for (operational and “embodied” energy consumption) are then amortized across the total data transferred across the internet, and converted to emissions using a “grid intensity” factor.

The data for operational energy consumption comes from 2022 for data centers and networks (IEA 2022) and 2023 for user devices 2020 (Malmodin 2023). The data for embodied energy consumption comes from 2020 (Andrae 2020). Total data transfer across the internet is as of 2022 (ITU 2024). Grid intensity comes from 2022 (Ember 2022).

If these dates for those values made you go “Wait!” in this 2026, give me a minute, I’ll get there.

Generally, this is pretty tidy — it’s hard extrapolate a useful measurement from, for example, me vising a website on my phone when I charged on a sunny day (with a mostly solar energy mix). By taking top-line numbers for operation and production energy and grid emissions, and then amortizing them over total internet data transfer, you get a relatively straight-forward estimate.

Or you would have in 2022 or so.

The dates matter

All of those top-line numbers would be… highly complicated by the generative “AI” bubble.

You might arrive here from a search page that automatically generated an LLM response whether you wanted or needed one. You might be using a browser that’s downloading gigabytes to your device without your consent or knowledge. You might even not have arrived here directly at all, you may have asked a chatbot a question and it navigated here and scraped the content on its own in the process of answering.

The Data center power consumption story is even more complex. While we can choose green data centers and/or hosting providers, we can’t separate whether their consumption is going to serve data to humans, to chatbot in response to a human, or to an AI crawler adding the data to training data. And we’d need top-line consumption numbers that separate out those green providers’ data centers with data centers that are outrageous (in some cases illegally) dirty.

Thinking about traffic, the total data transferred across the internet has both soared and been taken over by AI (Cloudflare said 57.4% of all web traffic back in June).

But it doesn’t stop there! We don’t just accept that all this traffic, climate cost, and consumption of websites is worthwhile. Defenses against the generative AI crawling, scraping, exist, and have their own climate costs. Cloudflare’s Labyrinth, for example, adds links that are intended to send generative AI crawlers through an endless maze of links to stop them from scraping or crawling a website that doesn’t want to be.

The data that’s starting to emerge backs all of this up, with generative AI projected to increase CO2 emissions by up to ~5% of 2024 global energy emissions, and drive an increase in the carbon intensity of everything they touch. There are some attempts to come to grips with this impact, such as Watershed’s AI emissions framework, trying to understand and measure generative AI emissions. This work could lead to a place where we can start untangling all of this.

Wow, that’s bleak

If this feels bleak, it is, and it’s where the tech industry’s relentless push for something that people generally don’t seem to want.

I have a lot of thoughts about generative AI that don’t fit neatly into this particular essay, which was inspired by the question “how sustainable is my website?” You probably have a sense of what those are (though maybe not to strength of that feeling).

My website is for humans, by a human. I don’t expect it to be used by machines, and I do what I can to make that difficult. I am, by nature of serving public content on the internet and being a human who uses technologies that have (somewhere in their supply chain) use or enable use of generative AI, implicated.

A digression into theory

A quick detour into theory, because I am using the word “implicated” specifically to refer to the concept of an implicated subject from Michael Rothberg. To be an implicated subject is to inhabit or benefit from systems of harm that we do not directly control or contribute to, and may even be harmed by them ourselves, while being implicated by them in the harm of others. But implication is not individual culpability, we are responsible for our own actions, even if those actions are constrained by the systems of harm.

So how do we act? Donna Haraway implores us to “stay with the trouble”, recognize that our responsibility is also response-ability (a capacity to respond), and that it is a continuous, collaborative making-with rather than making-for, with the kin we find occupying the same space, in different positions of implication. To paraphrase Joel Olson, we bring our biography, abilities, and commitments, to those of others staying with the same “trouble”.

And we’re back

That’s all the theory for now. I genuinely value sustainability, as do others within the web sustainability and small web movements. This makes us kin with, among others, those fighting pollution from AI data centers. What I can do is to ensure that my public presence on the internet is as sustainable as possible.

For my little site, the values as gram of CO2e has been useful as a relative value and to discover what I can reasonably target for optimization. I doubt the realism of the absolute numbers in 2026, but I do know that going from an average of 0.70g CO2e to 0.11g is a meaningful difference.

Saturday, Apr 4, 2026

Sometimes you need to walk a tree

Traversing Wagtail StreamFields

During a talk that I gave at Wagtail Space 2024 with my colleague Chuck Sebian-Lander, Auditing Wagtail Content (slides), I introduced a library we had worked on to help us audit. It walks through StreamFields to construct Django QuerySet-like objects specifically for understanding the shape of the content you have. But two functions in that library are super useful outside the audit context, so I’ve broken them out into their own library, wagtail-traverse.

They’re relatively straight forward, and operate on either a model (to determine possible block structure), or a page from the database (to determine what blocks are in use and their content).

As the README says, there’s traverse_block(block) and traverse_value(page_instance.field), with traverse_block() walking a StreamField’s block definition on a model and yielding back the block’s path, and its instance, and traverse_value() walking the StreamField’s value on a page and yielding back the block path and the value’s BoundBlock instance.

There are a couple of hypothetical examples in that README on a MyPage model:

from wagtailtraverse import traverse_block, traverse_value

streamfield = MyPage._meta.get_field("body")
for child_name in streamfield.stream_block.child_blocks:
    for path, block in traverse_block(
        streamfield.stream_block.child_blocks[child_name]
    ):
        print(f"{path}: {block.__class__.__name__}")

page = MyPage.objects.get(slug="example")
for path, bound_block in traverse_value(page.body):
    print(f"{path}: {bound_block.block.__class__.__name__} = {bound_block.value}")

Concretely, this came out of our need at CFPB to understand the content we have deep in our highly nested blocks, so it could be modified manually.

But it is also useful when crafting data migrations for StreamFields. Wagtail’s StreamField migration operations use block path to address the blocks within a field that get migrated. So if you want to know all the paths an ImageBlock might occur in a BlockPage, traverse_block() can handle that.

from wagtailtraverse import traverse_block
from blog.models import BlogPage

for child_name in streamfield.stream_block.child_blocks:
    for path, block in traverse_block(
        streamfield.stream_block.child_blocks[child_name]
    ):
        if block.__class__.__name__ == "ImageBlock":
            print(f"{path}")

This should give you all possible paths of ImageBlocks, both directly as children of the “body” field as well as any that are nested in other blocks. If you’re writing a migration that performs a StreamField operation on ImageBlock, this is a good way to sanity check all the paths you’ll need to migrate.

If in that same case you’re doing data manipulation and you need to know, for example, which pages have alt text in their ImageBlocks and what it is set to (because maybe you’re performing a data migration on it), you can do something similar with traverse_value():

from wagtailtraverse import traverse_value
from blog.models import BlogPage

for page in BlogPage.objects.all():
    for path, bound_block in traverse_value(page.body):
        if (
            bound_block.block.__class__.__name__ == "ImageBlock"
            and bound_block.value["alt_text"] is not None
        ):
            print(
                f"Page {page.pk} at {path} "
                f"has image {bound_block.value['image']} "
                f"with alt text: {bound_block.value['alt_text']}"
            )

I have found this pattern to be highly useful in determining what content deep within StreamFields to migrate and how.

The API is somewhat clunky: right now both functions require you know how to address deeply within StreamFields and their blocks and bound block values. I’d like to lighten that load a little bit over time. But for now, the pattern is documented in the README and this blog post, and I hope others find it useful.

This package can be installed via PyPI:

pip install wagtail-traverse

Tuesday, Dec 23, 2025

Using Sigma's Mount Conversion Service

I am a huge fan of Sigma’s optics and cameras, but I recently decided to consolidate some of my camera gear and switch to a Sony E mount camera (α7R V). But I wanted to continue using some of the Sigma lenses I have and love.

For a number of years Sigma has offered a mount conversion service to convert Sigma lens you already own from one mount to another, as long as the lens is offered in the destination mount. It’s an intriguing prospect in the abstract, being able to switch systems and not rebuy all new lenses. However, there doesn’t seem to be a lot of information out there from people who have had experience with it. I decided to give it a try, and write about my experience, breaking down the price I paid, the time it took, and to compare alternatives and provide some rationale for why I personally decided to use the service in late 2025.

Update: see below for an update based on my experience of converting another lens, my 24-70 II.

In forum discussions of the service, there’s inevitably the question (usually phrased emphatically as a statement of fact) of whether one would be better off selling the lens and rebuying it in the new mount. After all, the cost of the conversion service is not insignificant (and also not readily available publicly). So, the first order of business is to understand the cost of conversion and how it compares to simply selling the lens and buying it again new.

I wrote to Sigma USA, as their mount conversion page suggests, to get quotes on the couple of lenses I was interested in converting before sending them. Once I sent them in, I got a service order with actual estimate. It’s worth pointing out there was a difference between the email quote and the estimate once Sigma had the lenses in-hand.

LensEmail QuoteActual
Sigma 35mm f/2 DG DN$230$250
Sigma 100-400 f/5-6.3 DG DN$250$350

I also did some price research. I looked at how much similar condition copies were going for, if I were to sell the lenses and re-buy. I also got quotes from MPB for good measure, as the lowest-effort, lowest-return option for selling them. This way we can compare the cost of conversion to the cost of the lens new, used, and what I would likely sell it for. The table below has those estimates, as well as the new price.

LensNewMPB QuoteRough Used
Sigma 35mm f/2 DG DN$700$265$450
Sigma 100-400 f/5-6.3 DG DN$1000$330$650

The mount conversion is not cheap, but it’s not terrible (around 1/3 of the price of the lens new), but it is nothing compared to the cost of selling to a company like MPB and rebuying new, so I didn’t consider either of those options. But if I were to sell the lenses and then rebuy in roughly the same condition used, the different is probably just minor overhead costs per-lens.

This is the thought behind a lot of the sentiment around this service from people responding to forum discussions: that selling and rebuying the lens used is basically cost-neutral.

However selling and rebuying used on classified type of forums and sites has non-monetary costs like time and effort, and carries some risk with it. It is generally not my favorite activity. And that assumes you can find similar condition lenses in the alternative mount, which is not always the case. The Sigma 35mm f/2 seems in short supply in any mount. It seems like a lot of churn to engage in, especially for lenses I already have.

Sigma 35mm for L-mount from the mount side
Sigma 100-400mm for L-mount from the mount side

There’s also the “why get these converted when you could buy a ___ lens instead”, and on that question, I find both of these lenses to be the best balance of what I care about (on the spectrum of size/weight/image quality) for what they are in both mounts.

The mount-conversion service

So… I proceeded with the conversion service! For science (and convenience). I filled out the online repair form specifying a mount conversion from L-mount to E-mount, printed the resulting page, packaged up the lenses as well as I could, and shipped them to Sigma America (insured for the new replacement cost). I included only the front and rear lens caps, and kept the hoods, and kept both the rubber ring and tripod shoe ring for the 100-400.

Box with bubble wrap

Looking at the tracking delivery time, it took approximately 3 hours after Sigma received the lenses to provide me with an invoice (this is a turn around time that gives me a lot of confidence if I ever need to have any Sigma gear serviced… which I haven’t to date).

It also took 1 week, door to door, for the two lenses. I sent them on a Tuesday, received the invoice on Wednesday. The following Monday afternoon I got notice that they were being shipped back, and had them delivered Tuesday. When I got them back, I was eager to take a look to try to understand what had been done.

Before I get to that though, there is something I was curious about with the process. I assume the back part of the lens barrel (which appears to me to be a solid piece) has to be replaced since L-mount and E-mount have different depths and diameters. On both lenses, this appears to be one solid piece. On the 35mm, the serial number itself is on this piece, so I was curious if they were replacing this, does it get re-engraved on the replacement part of the barrel? Is the serial number different? What does this mean for warranty registration, etc? The serial number is in the middle of the lens on the 100-400, so if the mount-end of the barrel is being replaced I wouldn’t expect it to change.

Old serial number on the 35mm barrel on the lens mount side
New serial number on the 35mm barrel on the lens mount side

For these two lenses, this “service” simply exchanged my L-mount lenses for E-mount lenses. When I received them, I noted that the serial numbers on both lenses were different. Sigma confirmed when I asked about warranty implications (if I need warranty service, I just have to reference the order number for this service). They said otherwise it would’ve been a 2-month trip to Japan to change the mounts on the lenses. My assumption is that they don’t talk too in-depth about this to preserve options based on what they have available.

I’m actually quite fine with this? Effectively I paid 1/3 the cost of the lenses new to exchange it for a different mount, and continue receiving warranty coverage for the lenses.

For lenses I love, that I bought a while back, that I want to continue using on the new system I’m moving to? I think that’s a pretty good deal considering the level of effort and risk involved (virtually none), and the turnaround time (during the holiday season too).

Would I do it again?

I’m not completely leaving L-mount yet, but if I do, I have a couple other Sigma L-mount lenses that I would consider bringing with me (the 24-70mm f/2.8 Art II). I would definitely use this exchange service again.

I can’t promise that my experience would be universal, but since there’s not really any first-hand information out there about this service, it felt important for me to write about my experience. If you, like me, your priorities lie more along the low-effort and low-risk sides of the triangle, rather than the low-cost side, then I think it makes a lot of sense. The quick turnaround made it even better — I was fully prepared for it take a couple weeks. If you, unlike me, are looking for the lowest-cost way to swap, maybe with more effort or risk, this isn’t it.

Hopefully, dear reader, you found this useful.

Update

About a month after I wrote this up, I did use the service again, to convert my Sigma 24-70 DG DN II Art lens from L-mount to E-mount. The turn-around time was even quicker (<1 week) than for the previous two lenses, and (and this is why I’m writing an update) the lens mount was converted it was not exchanged for a lens in the destination mount. The E-mount lens I received back has the same serial number as the L-mount lens I sent in, and it was the same minor scuffs on the filter ring. The cost of converting the 24-70mm f/2.8 DG DN II Art was $345.

What I can conclude is that, in the US at least, right now, Sigma may either exchange a lens or convert it as part of their Mount Conversion Service, depending on whether they can do it in their service center or whether it will need to be returned to Japan to convert.

Thursday, Sep 12, 2024

Long term thoughts on the Sigma fp L

Or, how I learned to stop worrying and love the smol

The Sigma fp L is a strange little camera. Little because it’s barely larger than the the lens mount. Strange because it has a 61MP senso and some significant drawbacksquirks that you have to work around. But maybe that’s part of its charm.

I got it because I was intrigued by the size and the relative affordability it brought to that resolution. I’ve come around to thinking the fp L is the best camera for a certain photographic niche that I find myself in. It has exceptional image quality for lightweight walking/hiking travel, landscape and environmental photography. If you can work around those quirks.

The Teton Range and Jackson Lake in Grand Teton National Park

All the reasons not to own one

The fp L has some significant drawbacks. It only has an electronic shutter, no IBIS, and slow sensor read-out. Combine these with the high sensor resolution and I have to be very careful with shutter speeds to avoid rolling shutter, banding in artificial light, and camera shake. There are situations where these are at odds, and in those, the fp L is not the right camera.

The screen doesn’t articulate in any way. This is somewhat unfortunate for type of photography I do with it where I’m out for a hike or walk without a tripod and might want to get low or high. But I also photo regularly with a Leica Q2 and M10-R, and used an X-Pro2 as my primary camera for many years, so it’s something I know how to work around. Still, there may be photos I don’t take because I don’t want to work the angles with my body.

Its startup time is slow, so it’s less suited for spur-of-the-moment shots than for more considered photos. You might think, you’ll just leave it turned on and asleep, except that battery life is pretty terrible. Third party batteries exist, are cheap, and quite small, and the camera will run and charge the battery from USB-C, so it’s fine I guess to carry a bunch of them. It is clearly a compromise for the size of the camera.

And the size is everything; it makes all of these problems worth overcoming.

The Tetons from Cascade Canyon Trail in Grand Teton National Park

So why would anyone want one?

Why do I own it then? A combination of the small size and sensor. It is perfect when I want to keep size and weight to a minimum, like when hiking. It’s 427g with the battery and SD card. My S1R II is almost twice that at 790g, and they’re both L-mount so any lens I use on either adds the same weight. It is only slightly larger than the body of a Sony RX1 series camera. But it’s not just the camera and lens, it’s what other lenses I don’t have to carry with it.

The Cascade Canyon Trail in Grand Teton National Park

Its 61MP sensor reduces the number of lenses I need to carry. It might even mean I take one prime where I would otherwise have used a zoom. If I carry my 20mm lens, I can get a 35mm field-of-view equivalent crop that is still 34MP. I’ll often go out with just the 20mm lens or just my 35mm lens, and that’s enough for a day’s shooting. No photographer should be afraid of cropping, aggressively if needed, and 61MP gives plenty of room to crop with confidence. When I do take a zoom, like the Sigma 24-70 f/2.8 Art, the fp L with the handgrip is astonishingly comfortable to hold for a tiny rectangular box with a ridiculously oversized lens attached.

There’s one more thing that the fp L does for the occasional landscape photographer in-camera that no other camera does (that I’m aware of). It allows extremely low ISO. You can enable an “ISO Sensitivity Expansion” mode that enables ISO settings from 6-80—that’s 4 stops below the base ISO 100 sensitivity. At ISO 6, if I have a small tripod with me I can take shots I’d ordinarily have to get out an ND filter for. And given how little I want to carry, I don’t usually carry filters.

What about the quirks?

The Cascade Canyon Trail in Grand Teton National Park
On the trail

That’s all well and good, but enabling a lighter weight kit doesn’t overcome the issues I described above:

  1. Rolling shutter,
  2. Banding in artifical light
  3. Camera shake.

Not all of these apply to what I use the fp L for. I don’t use it as a general-purpose camera, and I don’t use it for any fast moving subjects (I have the S1R II for that), so rolling shutter doesn’t bother me (mostly).

The other two are the most common I ran into when I started using the fp L.

I occasionally use it in artifical light, and banding happens at the electrical line frequency. In the US this is 60Hz. I use auto-ISO with the “ISO Sensitivity Expansion” mode turned on, and I have a custom mode that sets a manual 1/125s shutter speed. The Sony sensor’s exceptional high ISO noise performance, and the fact that I’d rather get the shot and worry about noise later, means I don’t really worry about the ISO going up to compensate. At 1/125s I do have to be careful though, because of the other issue: camera shake.

Camera shake is my biggest issue. Most of the time I’m not under artificial light, so I have another custom mode that sets an auto-ISO minimum shutter speed of 1/500s. I have found this gives me a good amount of room for the type of shake I see when hand-holding the camera and lenses I typically use (either the large 24-70 f/2.8 Art or the small 35mm f/2 prime). Again I rely on the sensor’s high ISO performance and let the auto-ISO do what it needs to do to get the exposure correct.

Do I wish this camera had IBIS though? Yes.

If…

If you find you can work around all of these without them getting in your way (and that is a big if), the fp L is worth considering. The combination of such a tiny full-frame interchangable lens camera, a high resolution sensor, and superb image quality makes it a camera I keep reaching for.

The Teton Range from the air

Tuesday, Jun 25, 2024

Zen and the Art of Package Maintenance

Just… be kind

At this year’s Wagtail Space US, a conference dedicated to the Wagtail content management system, my friend Tim Allen hosted a panel on (Wagtail) packages, the challenges of maintaining them, and the small joys of being a maintainer. It was a good discussion, and you should watch the full panel if you’re interested (once it’s posted). I have some reflections on I wanted to write up coming out of it.

A photo of four people, the nearest one speaking
Left to right: Tim Allen, Vince Salvino, Jacob Topp-Mugglestone, Will Barton.
Photo by Tom Dyson

I shared the panel with Vince Salvino of CodeRed and Jacob Topp-Mugglestone of Torchbox in addition to Tim from The Wharton School (and an enduring fixture in the Django and Wagtail communities). Either Tim or Vince made the observation that with the four of us on the panel, the maintainers of the most-used Wagtail packages were all in one place. This speaks to larger growing pains the Wagtail community is having. Each of us have organizations that provide active support for us contributing to Wagtail, maintaining packages, etc. Not everyone is so lucky.

A theme of Wagtail Space 2024 was getting more people involved in the Wagtail community, as users, contributors, and maintainers. Tom Dyson, opening the conference, asked how do we continue to grow? How do we build the community? I think package maintainers have an outsized role in this.

I think there are, generally, two times in which other people will interact with our packages:

  1. When they’re trying it for the first time
  2. When they’re updating it (whether for its own sake or because they have to for a Wagtail update)

Most people find our packages by searching for a solution to a particular problem they’re trying to solving. They encounter our package and it either solves their problem or comes close enough that they can work with it.

In other words, we’re potentially blockers to them getting done what they need to.

There will always be someone who struggles in that moment or maybe has a question, and may file an issue or look us up on the Wagtail Slack. They may even be frustrated because their deadline is dependent on someone else, and who knows how engaged an open source maintainer is at any given time with a library that they haven’t had to touch in a while.

The way we treat people in those moments matters, and a good interaction is some of the best community-building we can engage in.

We can start by having good-enough documentation. Almost all documentation could be better but we can put in the work, talk to our users, and make it as helpful as possible. We can engage with our users in question-answering in the Wagtail Slack, in issues, etc. We can even encourage contributions for features or needs our package doesn’t yet support, but there’s danger here: the stereotypical open source interaction of “it’s open source, go fix it yourself.” Most of us in and around open source have been on the receiving end of this sort of thing at one time or another. It sucks. We do not have to pay it forward.

Instead, we can treat our users with kindness and empathy. And if we can’t manage that in a specific moment—maybe we’re having a bad day, we’re overwhelmed with something and this is one more thing—then I think it’s best to wait. Acknowledge their need, and get back to them when you can be kind and empathetic.

Which brings me briefly to one last thought based on a question Tim had of us panelists: how do you avoid fatigue and burnout? Because those lead to dealing with users without kindness and without empathy. Package maintainer-ship is mostly thankless. You rarely hear from the users who are just getting on with using the package successfully.

I can only answer this for myself, and hope it’s broadly applicable. I write software because I want to solve interesting problems. I want to write useful software in the process, so I write open source software. Built into the idea of open sourcing something is the idea that it could be useful to someone else. The fact that they’re trying to use my software is where I find meaning in what I do. Reminding myself of that helps.

Monday, May 20, 2024

Capturing the Eclipse in 2024

Part of Eclipse Photography

After my successful effort to capture the eclipse in 2017 I was hooked. I made plans to see the total eclipse of April 8, 2024 early. I figured I’d take my full astrophotography rig since it totality would be driving distance from my home.

That was until I went out to practice setting up during daylight. Getting good enough polar alignment to allow me to track the sun during daylight was doable, but it was somewhat frustrating. The thing I learned in 2017 was that the solar eclipse is not a photo opportunity, it’s a once-in-a-lifetime experience. I didn’t want to clutter the experience with the tedium of equipment setup and alignment.

What I optimized for was: keeping the photography simple so I can enjoy the experience of totality. I think I did this somewhat less intentionally in 2017, but that became an overriding concern for 2024.

A camera set up on a tracking mount to capture the solar eclipse
My S5II on the Vixen Polarie

So I fell back on my trusty Vixen Polarie, the little star tracker that could. This timeI used my Panasonic S5II, and Sigma 100-400mm F5-6.3 DG DN Contemporary lens. This is very similar to my setup in 2017.

One brief note on the Polarie and tracking mounts in general: I don’t think they’re required for eclipse photography per se, but they definitely help with the whole keeping the photography simple priority. Through a long lens the sun moves a lot in the sky over three minutes, which is how long totality in Shelburne, VT, lasted. A tracking mount removed, for me, the impulse to fiddle with the ballhead and reframe.

A view across Lake Champlain toward the Adirondack mountains where totality has ended
Adirondacks out of the shadow

We picked a location in Shelburne, VT, on the shores of Lake Champlain for a weekend trip. It’s a lovely part of the country, but the Adirondacks across the lake were breathtaking. But never more so than while we were in totality, and I could look across the lake and see them emerging from the moon’s shadow.

For the actual photography, I stuck to my basic playbook from 2017, with a couple of minor changes: I wanted to stop down the lens to f/8 to maximize the sharpness and detail from the lens. To accommodate this and keep my exposures consistent with what I captured in 2017, I upped the ISO to 800 from 400. My Panasonic S5II has significantly better noise performance than my Fujifilm X-T2 did, so I wasn’t worry about any increase in noise.

PhenomenonShutter SpeedISOAperture
Chromosphere1/8000s8008
Prominences1/4000s8008
Corona 0.1 Rs1/1000s8008
Corona 0.2 Rs1/250s8008
Corona 0.5 Rs1/60s8008
Corona 1.0 Rs1/15s8008
Corona 2.0 Rs1/8s8008
Corona 4.0 Rs1/4s8008
Corona 8.0 Rs1s8008
Earthshine2s8008

After going through my data from 2017 I decided to target 1/1000s for any exposures of partiality. But I also decided that with a small field of view and no surrounding context, these images wouldn’t be particularly interesting to me. I don’t find these exposures from 2017 compelling. And I wasn’t interested in complicating my setup to capture more context. So, it’s okay if I only use those exposures to test the setup before totality.

Going through my checklist, I also considered external power supplies (extension cords, massive batteries, the like), but decided against it. These would be one more thing to set up, and one more thing to get right. Easiest just to make sure I have fully charged batteries for the camera and the Polarie.

⚠️ For any photography of the sun outside of totality itself a solar filter specifically designed for observing the sun is absolutely required. I keep seeing advice around the internet to just use ND filters. This is wrong and dangerous.

Here’s my checklist for 2024:

Before eclipse:

  1. Make sure camera battery is fully charged
  2. Make sure Polarie batteries are fully charged
  3. Set up tripod and Polarie
  4. Hook camera bag onto tripod for stabilizing weight
  5. Align tripod and Polarie to celestial north pole
  6. Put camera and lens on Polarie

Set up Camera

  1. Set camera to manual focus
  2. Set camera to manual exposure
  3. Set camera drive mode to self timer
  4. Set self timer drive mode option to 2s
  5. Set camera to exposure bracketing
  6. Set exposure bracketing to 7-1
  7. Set camera ISO to 800
  8. Set lens aperture to 8.0

Prepare for exposure

  1. Set Polarie to sun tracking
  2. Put solar filter on lens
  3. Position camera the sun
  4. Zoom lens to 400mm
  5. Focus the lens on the limb of the sun
  6. Put solar glasses on face
  7. Set initial shutter speed of 1/1000s

Partial eclipe

  1. Expose at 1/1000s
  2. Gawp
  3. Repeat as desired for stages of partiality

Totality

  1. Remove solar glasses from face
  2. Remove solar filter from lens
  3. Expose at 1/1000s
  4. Expose at 1/30s
  5. Expose at 1/2s
  6. Gawp
  7. Expose at 1/2s
  8. Expose at 1/30s
  9. Expose at 1/1000s
  10. Put solar filter on lens
  11. Put solar glasses on face
A solar eclipse
Prominences on the sun in a single 1/8000s exposure

2017 was a year in which the sun was at a low point in its solar cycle — this year was very different, and that was almost immediately apparent. There were massive prominences visible to the naked eye on the lower limb! This was shocking to me, wholly unexpected in the midst of an altogether unusual experience.

And because I kept it relatively simple, I was able to capture it to my satisfaction, achieve my array of exposures, and have plenty of time to take in the full experience of being present in the world under the shadow.

I composited 2024 differently than 2017; in 2017 I was struck by the visibility of the bright star Regulus in the frame with the sun, and that’s what I wanted to highlight with my editing — that star visible during the day while our star was obscured. This time it was that set of prominences that I wanted to highlight.

And here’s the result:

A solar eclipse
Composite of the total eclipse of April 2024

I’m already planning 2026.

Friday, Mar 1, 2024

Capturing the Eclipse in 2017

Part of Eclipse Photography

I hadn’t had opportunities to observe a total solar eclipse before 2017, but I knew I wanted to try my hand at photographing it. And photograph it I did!

Before that though, I did some research and captured notes and a simple, fine-grained order of operations, because I didn’t know what I was doing and it’s not something you can really try over again if you fail. Plus, I actually wanted to enjoy the moment without fiddling with my camera, and having a concrete plan helps with that.

A solar eclipse
Composite of the total eclipse of August 2017

There are a few different images I wanted to try to capture, because there are multiple aspects of the sun that are visible during an eclipse that aren’t normally. In brief, we can observe the sun’s chromosphere, any prominences it might have in the moment, the corona, and, at totality, we can see earthshine on the moon. These can all be composited together to form an image like the one above.

For this eclipse, I was traveling with my tripod, Polarie tracking mount, and Fujifilm X-T2 and 100-400 lens, so figuring out the exposures for each of these phenomenon would happen in terms of the ISO+shutter speed+aperture settings best for that combination of equipment. Jerry Lodriguss has a fantastic write-up that forms of the basis of what I decided to do.

Based on this I settled on using an ISO of 400, an aperture of 5.6 (the widest that lens was at 400mm), and chose shutter speeds from there.

PhenomenonShutter SpeedISOAperture
Chromosphere1/8000s4005.6
Prominences1/4000s4005.6
Corona 0.1 Rs1/1000s4005.6
Corona 0.2 Rs1/250s4005.6
Corona 0.5 Rs1/60s4005.6
Corona 1.0 Rs1/15s4005.6
Corona 2.0 Rs1/8s4005.6
Corona 4.0 Rs1/4s4005.6
Corona 8.0 Rs1s4005.6
Earthshine2s4005.6

This became my sequencing for during the eclipse: I would take an exposure at 1/8000s, then 1/4000s, 1/1000s, etc, all the way up to 2s, then do the same in reverse, from 2s down to 1/8000s. To eliminate the number of things that could go wrong, I decided to manually set the shutter speed (with the X-T2’s physical dial, perfect for this use-case) rather than relying on any automation.

The other decision I made was to use exposure bracketing in camera at ±1 stop. So, I would set the shutter speed above, and the camera would take 3 exposures. This wold maximize the range of possible data I’d have to work with.

I also wanted to get some photos before totality as the moon is moving across the sun’s face. For those there’s a lot more time to experiment.

⚠️ For any photography of the sun outside of totality itself a solar filter specifically designed for observing the sun is absolutely required. I keep seeing advice around the internet to just use ND filters. This is wrong and dangerous.

The core decision I stuck to here was just to follow the same exposure list, with bracketing, I had for totality. Again, this gives me a maximum range of exposure data and is an opportunity to practice.

As I said at the beginning of this post, what I wanted was a straight-forward, fine-grained order of operations I could follow on the day to ensure I was spending as little time as possible fiddling with my camera, and the most time enjoying the experience of the eclipse with my family. Here is the list I ended up with:

Before totality:

  1. Put new batteries in Polarie
  2. Set up tripod and Polarie
  3. Align Polarie
  4. Put new battery in camera
  5. Put camera and lens on Polarie
  6. Put solar filter on lens
  7. Attach cable release
  8. Set manual focus
  9. Set ISO 400
  10. Set BKT drive mode
  11. Set 5.6 aperture
  12. Put solar glasses on face
  13. Position on sun
  14. Focus
  15. Set shutter speed 8000

Partiality:

  1. 8000
  2. 4000
  3. 1000
  4. 250
  5. 60
  6. 15
  7. 8
  8. 4
  9. 1s
  10. 2s
  11. Repeat as desired

Totality:

  1. Remove solar filter
  2. 8000
  3. 4000
  4. 1000
  5. 250
  6. 60
  7. 15
  8. 8
  9. 4
  10. 1s
  11. 2s
  12. Gawp
  13. 1s
  14. 4
  15. 5
  16. 15
  17. 60
  18. 250
  19. 1000
  20. 4000
  21. 8000
  22. Put solar filter on lens

Overall, this made for a satisfying and enjoyable eclipse, with results I’m pleased with. I did knock the manual focus ring on the lens slightly out of focus when removing the solar filter for totality, and so my totality shots are that little bit out of focus. But that’s fine. Other people got sharp, in focus photos of this eclipse. But this is mine, where I was, and with my family as we observed it.

Wednesday, Mar 15, 2023

First new light with an new(ish) setup

I wrote about my return to using my German Equatorial Mount, refractor, and Canon DSLR in my last blog post. Since then I’ve refreshed both my mount and my reactor. In many way’s it’s the opposite of the lightweight setup I wanted a year ago, but what I realized I wanted was simpliciy. This gives me that, of a sort:

  • Losmandy GM-8 German equatorial mount with Gemini 2
  • Astro-Tech AT115EDT 115mm ED triplet refractor with 0.8x reducer/field flattener

The Losmandy mount is substantially heavier (even with their lightweight tripod) than anything I’ve used before. But it’s modular, comes apart easily, and is—in my experience so far—rock solid, feature rich, and a joy to operate.

The same can be said for the 115mm refractor, it is much heavier than my ED80, but what I gain for the added weight is a faster optic (f/5.6 with the reducer) with a larger aperture that can capture more in less time.

My first light with this setup was actually limited. I got 30 minutes of the Horsehead and Flame nebulas before they went behind a tree and clouds rolled in:

The Horsehead and Flame nebulas
The Flame nebula and the Horsehead nebula. 30x60s, Canon 6D Mark II, Astro-Tech AT115EDT, 0.8 focal reducer.

Shortly after that I got to capture a 1:23 of the Orion nebula over a couple nights:

The Orion Nebula
The Great Nebula in Orion. 83x60s, Canon 6D Mark II, Astro-Tech AT115EDT, 0.8 focal reducer.

I’ve had no complaints, no struggles, nothing so far with the mount — it is an utter joy to use. And because it’s upgradable, it could be the last mount I’ll ever own. I’m not sure I’d ever need to upgrade — I don’t anticipate putting anything on it heavier than the AT115EDT at present.

The AT115EDT is a beautiful telescope, with superb optics. I will say, figuring out that the camera rotator actually was two pieces, and you have to take one off to attach the reducer/flattener was a bit frustrating. But that’s a minor, one-time thing.

Look for more from this setup in the future.

Wednesday, Jan 11, 2023

First new light with an old(ish) setup

After my experiments with a lighterweight astrophotography setup ended in a bit of failure, I’ve gone back to a fairly tried-and-true setup for myself:

  • Orion SkyViewPro computerized German Equatorial Mount
  • Orion ED80 APO 80mm refractor
  • Orion StarShoot Autoguider
  • Sky-Watcher 0.85 focal reducer for ED80

I’ve been able to control the telescope and autoguider from a computer (both using laptop and a Raspberry Pi attached to the telescope), but where I’ve struggled in the recent past is in controlling my Fujifilm cameras from that same computer. I would usually use the built-in interval timer separately on the camera.

This has a few disadavantages, first that I can’t check progress, tracking, whether any obstructions have gotten into the frame, etc while the timer is running. Before I switched to Fujifilm cameras, I would use BackyardEOS with my Canon DSLR to control my exposures. It had the added bonus of being able to use the laptop screen to frame, preview, and focus, which works much better than the back screen on any camera.

So, what I decided to do was pick up a second-hand Canon 6D Mark II to dedicate to astrophotography and go back to using BackyardEOS. A dedicated camera I can also get modified for Hydrogen-alpha. This made me happy.

So, I tried to take some photos of a couple easy targets this time of year to see how things go:

The Pleiades
The blue whispies of the Pleiades. 60x60s, Canon 6D Mark II, Orion ED80, 0.85 focal reducer.
Andromeda
The Great Galaxy in Andromeda. 26x60s, Canon 6D Mark II, Orion ED80, 0.85 focal reducer.

Things went well.

The software stack I’m using is, in macOS, PHD2 for autoguiding and SkySafari 6 Pro for controling the telescope; in a Windows VM, BackyardEOS, set up to drizzle with PHD2. For post-processing I’m using Siril and then importing into Capture One for my regular photo workflow.

This worked really well. I have some new targets in mind for the next clear night.

Sunday, Dec 11, 2022

Lightweight astrophotography, a retrospective

Part of Lightweight Astrophotography

I wrote earlier this year that my biggest blocker to regularly engaging with my astrophotography hobby is:

[A] fundamental laziness I have when it comes to setting up and taking down my equipment. This is the biggest blocker I find to just getting out and doing.

This inspired me to rethink: what if I assembled a lighter-weight, more portable, less cumbersome astrophotography setup using my (beefy, carbon fiber) camera tripod, a polar mount (first the Vixen Polarie I’ve had for a while, then the iOptron SkyGuider Pro), maybe throw my autoguider on there, and hey presto! A setup that should be less effort to set up and take down on a little more of a whim, that I could set up and carry out whole. Or that was the theory.

What I’ve found in practice is that… it doesn’t solve the problem I set out to solve: my laziness. This is why I haven’t updated this series with any resulting photos yet, or attempts to travel, or just generally anything. The setup just doesn’t work for me. This isn’t to say it won’t work for someone else, but I’m more used to the conveniences of a computerized German Equatorial Mount workflow, where I can have software do a lot of the work for me.

I’ve thought about putting in a permanent pier, so that I don’t have to mess around as much with the tripod and mount alignment

This might be where I’m going next. Stay tuned.