Apple limits TestFlight to 10,000 testers. Folks who are paying a recurring fee expect to see more frequent updates, so they get priority.
Halide shouldn't be afflicted by the lost photo issue you're talking about. Normally, if you close a lock-screen extension before a photo is written, it won't be in your library. This can be a problem when the user quickly smashes the capture button and your photo queue gets backed up.
Luckily, the unlocked app gets a chance later to access the content of lock-screen capture sessions. So our strategy in Halide is to write the intermediate files to disk as soon as you initiate a capture; this should take a split second. When we have uninterruptible time, later, we make sure the photo finished writing to your library, and if not, we try again.
However, Apple's documentation isn't clear if the OS will "wake-up" your main app to process unfinished photos, or if it relies on the user launching the app. I'll just say that I dog-food Halide every day by capturing photos of my toddler, and I can't remember the last time I lost a shot.
At this point, I don't remember if I was using Halide from the lock screen or not, but I lost tons of photos because apparently I turned the screen off before the photo finished saving to the library. Halide forgot about it, and even after reopening the app, it was gone forever. This would only seem to happen when I turned the screen off quickly after taking the photo.
These days, I use the regular Camera app because I need my damn photos to save. I loved Halide, but that bug made it too unreliable. I get very distressed by data loss and couldn't continue to risk it while trying to develop a way to work around it.
We switched to Apple's StoreKit 2 framework, which uses different internal plumbing. Not only is it the recommended way moving forward from Apple, but it improves the user experience for purchase management.
The issue is that, indeed, it actually validates that the App Store account currently logged in purchased the app. I believe this is intended behavior, as it closes a piracy loophole where you could log into the Apple ID of a friend, download all their paid apps, and then log in to a different account while retaining access.
If other apps don't run into this problem, it's possible they haven't moved on to StoreKit 2. Or the app has their own login independent of app subscription; setting aside the time it would take me to build that, I don't think most people are willing to create an account to try a camera app.
> I'm actually appalled they use the term "AI" considering who they are but that doesn't matter, it's ML.
I agree, and I built it!
The app launched in 2019. Ideally I would have said ML, but it felt like AI had more recognition among non-technical users.
If I launched it in 2026, it would 100% say "ML" not only because it's more technically accurate, but to make it abundantly clear it doesn't generate slop.
As for the subscription, there’s always been a pay-once option for major releases. You just miss out on a free trial, the beta program, and anything we might release with ongoing costs.
If the sunflower and Osaka photos aren’t in HDR, there’s something off with your system. Make sure you aren’t in low-power mode.
Fun fact: the HDR photos are actually looping videos, because that’s way more reliable for browsers and CMS platforms than actual HDR photos. Hopefully this improves in the next few years.
This post came out a few weeks ago. To answer your question in advance, your favorite thing from Mark II that’s missing is eventually coming to Mark III.
I’m right now on a short detour updating our video app, Kino. I couldn’t justify touching it until this Mark III preview was out. After that, look for a few more Mark III preview updates before the big launch.
Not going to lie, it’s been an exhausting 12 months, but I’m genuinely excited for what’s ahead.
I love Kino! I film all my wood carving videos with it for my Instagram profile [1]. I'm so glad to hear it's getting updates this year. If you're open to feedback: I like playing with white balance a lot, and it takes just a little too many steps in the current interface. I think Kino would benefit from having a "remember/preserve white balance" setting, and a gesture for changing it without tapping.
For example, I like how we can swipe down to show the exposure slider, then swipe left/right to adjust exposure. Maybe it would make sense to have these hidden sliders on all 4 sides: right side could show a focus slider, left side could be white balance etc.
In any case, even without that, it's my most used new app of the year. Thanks for giving us such a simple way to get good looking color grading out of the iPhone!
> It seems like a mistake to lump HDR capture, HDR formats and HDR display together, these are very different things.
These are all related things. When you talk about color, you can be talking about color cameras, color image formats, and color screens, but the concept of color transcends the implementation.
> The claim that Ansel Adams used HDR is super likely to cause confusion, and isn’t particularly accurate.
The post never said Adams used HDR. I very carefully chose the words, "capturing dramatic, high dynamic range scenes."
> Previously when you took a photo, if you over-exposed it or under-exposed it, you were stuck with what you got. Capturing HDR gives the photographer one degree of extra freedom, allowing them to adjust exposure after the fact.
This is just factually wrong. Film negatives have 12-stops of useful dynamic range, while photo paper has 8 stops at best. That gave photographers exposure latitude during the print process.
> Ansel Adams wasn’t using HDR in the same sense we’re talking about, he was just really good at capturing the right exposure for his medium without needing to adjust it later.
There's a photo of Ansel Adams in the article, dodging and burning a print. How would you describe that if not adjusting the exposure?
> Film negatives have 12-stops of useful dynamic range
No, that’s not inherently true. AA used 12 zones, that doesn’t mean every negative stock has 12 stops of latitude. Stocks are different, you need to look at the curves.
But yes most modern negatives are very forgiving. FP4 for example has barely any shoulder at all iirc.
I agree capture, format and display are closely related. But HDR capture and processing specifically developed outside of HDR display devices, and use of HDR displays changes how HDR images are used compared to LDR displays.
> The post never said Adams used HDR. I very carefully chose the words
Hey I’m sorry for criticizing, but I honestly feel like you’re being slightly misleading here. The sentence “What if I told you that analog photographers captured HDR as far back as 1857?” is explicitly claiming that analog photographers use “HDR” capture, and the Ansel Adams sentence that follows appears to be merely a specific example of your claim. The result of the juxtaposition is that the article did in fact claim Adams used HDR, even if you didn’t quite intend to.
I think you’re either misunderstanding me a little, or maybe unaware of some of the context of HDR and its development as a term of art in the computer graphics community. Film’s 12 stops is not really “high” range by HDR standards, and a little exposure latitude isn’t where “HDR” came from. The more important part of HDR was the intent to push toward absolute physical units like luminance. That doesn’t just enable deferred exposure, it enables physical and perceptual processing in ways that aren’t possible with film. It enables calibrated integration with CG simulation that isn’t possible with film. And it enables a much wider rage of exposure push/pull than you can do when going from 12 stops to 8. And of course non-destructive digital deferred exposure at display time is quite different from a print exposure.
Perhaps it’s useful to reflect on the fact that HDR has a counterpart called LDR that’s referring to 8 bits/channel RGB. With analog photography, there is no LDR, thus zero reason to invent the notion of a ‘higher’ range. Higher than what? High relative to what? Analog cameras have exposure control and thus can capture any range you want. There is no ‘high’ range in analog photos, there’s just range. HDR was invented to push against and evolve beyond the de-facto digital practices of the 70s-90s, it is not a statement about what range can be captured by a camera.
> The sentence “What if I told you that analog photographers captured HDR as far back as 1857?” is explicitly claiming that analog photographers use “HDR” capture,
No, it isn't. It's saying they captured HDR scenes.
> The result of the juxtaposition is that the article did in fact claim Adams used HDR
You can't "use" HDR. It's an adjective, not a noun.
> Film’s 12 stops is not really “high” range by HDR standards, and a little exposure latitude isn’t where “HDR” came from.
The Reinhard tone mapper, a benchmark that regularly appears in research papers, specifically cites Ansel Adams as inspiration.
"A classic photographic task is the mapping of the potentially high
dynamic range of real world luminances to the low dynamic range
of the photographic print."
> Perhaps it’s useful to reflect on the fact that HDR has a counterpart called LDR that’s referring to 8 bits/channel RGB.
8-bits per channel does not describe dynamic range. If I attach an HLG transfer function on an 8-bit signal, I have HDR. Furthermore, assuming you actually meant 8-bit sRGB, nobody calls that "LDR." It's SDR.
> Analog cameras have exposure control and thus can capture any range you want.
Sorry man, you seem really defensive, I didn’t mean to put you on edge. Okay, if you are calling the scenes “HDR” then I’m happy to rescind my critique about Ansel Adams and switch instead to pointing out that “HDR” doesn’t refer to the range of the scene, it refers to the range capability of a digital capture process. I think the point ultimately ends up being the same either way. Hey where is HDR defined as an adjective? Last time I checked, “range” could be a noun, I think… no? You must be right, but FWIW, you used HDR as a noun in your 2nd to last point… oh and in the title of your article too.
Hey it’s great Reinhard was inspired by Adams. I have been too, like a lot of photographers. And I’ve used the Reinhard tone mapper in research papers, I’m quite familiar with it and personally know all three authors of that paper. I’ve even written a paper or maybe two on color spaces with one of them. Anyway, the inspiration doesn’t change the fact that 12 stops isn’t particularly high dynamic range. It’s barely more than SDR. Even the earliest HDR formats had like 20 or 30 stops, in part because the point was to use physical luminance instead of a relative [0..1] range.
8 bit RGB does sort-of in practice describe a dynamic range, as long as the 1 bit difference is approximately the ‘just noticeable difference’ or JND as some researchers call it. This happens to line up with 8 bits being about 8 stops, which is what RGB images have been doing for like 50 years, give or take. While it’s perfectly valid arithmetic to use 8 bits values to represent an arbitrary amount like 200 stops or 0.003 stops, it’d be pretty weird.
Plenty of people have called and continue to call 8 bit images “LDR”, here’s just three of the thousands of uses of “LDR” [1][2][3], and LDR predates usage of SDR by like 15 years maybe? LDR predates sRGB too, I did not actually mean 8 bit sRGB. LDR and SDR are close but not quite the same thing, so feel free to read up on LDR. It’s disappointing you ducked the actual point I was making, which is still there even if you replace LDR with SDR.
What is confusing about the sentence about analog cameras and exposure control? I’m happy to explain it since you didn’t get it. I was referring to how the aperture can be adjusted on an analog camera to make a scene with any dynamic range fit into the ~12 stops of range the film has, or the ~8 stops of range of paper or an old TV. I was just trying to clarify why HDR is an attribute of digital images, and not of scenes.
You opened this thread arguing that Ansel Adams didn't "use HDR." I linked you to a seminal research paper which argues that he tone mapped HDR content, and goes on to implement a tone mapper based on his approach. This all seems open and shut.
> I’m happy to rescind my critique about Ansel Adams
Great, I'm done.
> and switch instead to pointing out that “HDR” doesn’t refer to the range of the scene
"Surprisingly, daytime shots with high dynamic range may also suffer from lack of light."
"In low light, or in very high dynamic range scenes"
"For high dynamic range scenes we use local tone mapping"
You keep trying to define "HDR" differently than current literature. Not even current— that paper was published in 2016! Hey, maybe HDR meant something different in the 1990s, or maybe it was just ok to use "HDR" as shorthand for when things were less ambiguous. I honestly don't care, and you're only serving to confuse people.
> the aperture can be adjusted on an analog camera to make a scene with any dynamic range fit into the ~12 stops of range the film has, or the ~8 stops of range of paper or an old TV.
You sound nonsensical because you keep using the wrong terms. Going back to your first sentence that made no sense:
> Analog cameras have exposure control and thus can capture any range you want
You keep saying "range" when, from what I can tell, you mean "luminance." Changing a camera's aperture scales the luminance hitting your film or sensor. It does not alter the dynamic range of the scene.
Analog cameras cannot capture any range. By adjusting camera settings or attaching ND filters, you can change the window of luminance values that will fit within the dynamic range of your camera. To say a camera can "capture any range" is like saying, "I can fit that couch through the door, I just have to saw it in half."
> And I’ve used the Reinhard tone mapper in research papers, I’m quite familiar with it and personally know all three authors of that paper. I’ve even written a paper or maybe two on color spaces with one of them.
I'm sorry if correcting you triggers insecurities, but if you're going to make an appeal to authority, please link to your papers instead of hand waving about the people you know.
Hehe outside is “HDR content”? To me that still comes off as confused about what HDR is. I know you aren’t, but that’s what it sounds like. A sunny day has a high dynamic range for sure, but the acronym HDR is a term of art that implies more than that. Your article even explains why.
Tone mapping doesn’t imply HDR. Tone mapping is always present, even in LDR and SDR workflows. The paper you cited explicitly notes the idea is to “extend” Adams’ zone system to very high dynamic range digital images, more than what Adams was working with, by implication.
So how is a “window of luminance values” different from a dynamic range, exactly? Why did you make the incorrect and obviously silly assumption that I was suggesting a camera’s aperture changes the outdoor scene’s dynamic range rather than what I actually said, that it changes the exposure? Your description of what a camera does is functionally identical. I’m kinda baffled as to why you’re arguing this part that we both understand, using hyperbole.
I hope you have a better day tomorrow. Good luck with your app. This convo aside, I am honestly rooting for you.
> Hehe outside is “HDR content”? To me that still comes off as confused about what HDR is.
"Surprisingly, daytime shots with high dynamic range may also suffer from lack of light."
That's from, "Burst photography for high dynamic range and low-light imaging
on mobile cameras," written by some of the most respected researchers in computational photography. It has 342 citations according to ACM.
First sentence: "Tone mapping is a technique used in image processing and computer graphics to map one set of colors to another to approximate the appearance of high-dynamic-range (HDR) images in a medium that has a more limited dynamic range."
> Why did you make the incorrect and obviously silly assumption that I was suggesting a camera’s aperture changes the outdoor scene’s dynamic range rather than what I actually said, that it changes the exposure?
Because you keep bumbling details like someone with a surface level understanding. Your replies are irrelevant, outdated, or flat out wrong. It all gives me flashbacks to working under engineers-turned-managers who just can't let go, forcing their irrelevant backgrounds into discussions.
It's cool that you studied late 90s 3D rendering. So did I. It doesn't make you an expert in computational photography. Please stop confusing people with your non-sequiturs.
What does the lack of light quote prove? That’s a statement about color resolution, not range, and it uses “high dynamic range” and not “HDR content”. I think you’ve missed my point and are not listening.
Yes tone mapping is used on HDR images. It just doesn’t imply HDR. SDR gamma is tone mapping, for example, which the Wikipedia link you sent explains. Your claim is that Adams use of tone mapping is evidence that he is capturing “HDR content”. The paper you sent doesn’t use that language, it doesn’t ever say Adams was doing tone mapping, it says they develop a tone mapping method inspired by Adams’ zone system that extends the idea into higher dynamic range.
You’re using your own misunderstanding and mis-interpretation of my comments as evidence that they’re wrong. Hey I totally might be wrong about a lot of things, and sure maybe I’m completely non-sensical, but you certainly haven’t convinced me of that. I haven’t had trouble speaking with other people about HDR imaging, people who are HDR experts. All I’m getting out of this so far is that some people react very badly to any hint of critique.
From my perspective, I’m also only hearing bumbling errors, errors like that HDR is an adjective, that LDR doesn’t exist and nobody uses it, that using “range” is incorrect when I say it but not when you do and “window of luminance values” is better, and that Ansel Adams was doing HDR imaging.
Ben, we’re having a bona-fide miscommunication, and I wanted to fix it but I’m failing, and it feels like you’re determined not to fix it or find any common ground. In another environment we’d probably be having a friendly, productive and enlightening conversation. I’m sure there are some things I could learn from you.
In theory PQ specifies absolute values, but in practice it's treated as relative. Go load some PQ encoded content on an iPhone, adjust your screen brightness, and watch the HDR brightness also change. Beyond the iPhone, it would be ridiculous to render absolute values as-is, given SDR white is supposedly 100-nits; that would be unwatchable in most living rooms.
Bad HDR boils down to poor taste and the failure of platforms to rein it in. You can't fix bad HDR by switching encodings any more than you can fix global warming by switching from Fahrenheit to Celsius.
I don't think that's the colloquial meaning. If you asked 100 people on the street to describe HDR, I doubt a single person would bring up ITU-R BT.2100.
HDR has a number of different common meanings, which adds to the confusion.
For example, in video games, "HDR" has been around since the mid '00s, and refers to games that render a wider dynamic range than displays were capable of, and use post-process effects to simulate artifcats like bloom and pupil dilation.
In photography, HDR has almost the opposite meaning of what it does everywhere else. Long and multiple exposures are combined to create an image that has very little contrast, bringing out detail in a shot that would normally be lost in shadows or to overexposure.
Photography’s meaning is also about a hundred years older than video games; high(er) dynamic range was a concern in film processing as far back as Ansel, if not prior. That technology adopted it as a sales keyword is interesting and it’s worth keeping in mind when writing for an audience — but this post is about a photography app, not television content or video games, so one can reasonably expect photography’s definition to be used, even if the audience isn’t necessarily familiar.
Halide shouldn't be afflicted by the lost photo issue you're talking about. Normally, if you close a lock-screen extension before a photo is written, it won't be in your library. This can be a problem when the user quickly smashes the capture button and your photo queue gets backed up.
Luckily, the unlocked app gets a chance later to access the content of lock-screen capture sessions. So our strategy in Halide is to write the intermediate files to disk as soon as you initiate a capture; this should take a split second. When we have uninterruptible time, later, we make sure the photo finished writing to your library, and if not, we try again.
However, Apple's documentation isn't clear if the OS will "wake-up" your main app to process unfinished photos, or if it relies on the user launching the app. I'll just say that I dog-food Halide every day by capturing photos of my toddler, and I can't remember the last time I lost a shot.