47:31
After two long years of R&D, the Foam Buff camera test is finally here, and we're kicking it off by putting the Galaxy S26 Ultra and the iPhone 17 Pro Max through our seven launch modules, with each module built to answer questions that honestly we couldn't definitively answer before. Stuff like, which phone actually shoots better photos in low light when you have a moving subject, or which one zooms smoother across his lenses, and by how much? which one is more color accurate in its photos and videos, or even which one can track a subject better. And it turns out answering these kind of questions objectively is really hard since you have to strip out every variable that isn't the camera itself. Like take a look at these two
photos. It's the same phone in the same conditions, yet they look different. And that's because they were held side by side the way we normally do in a field test, which, you know, usually doesn't matter a lot when the subject is far away. But anytime a subject is closer, that small gap can dramatically change the shot. So to correct for it, we use computer vision, custom calibration charts, and lasers with our robots aligning every lens on every phone to the same point in 3D space down to a square centimeter. Now, the next big challenge we faced was hand trimmer. In the field, your left hand can shake a little bit differently than your right hand. And if that happens the moment you
press the shutter, it can hand a phone an unfair advantage. And you know, while we could have our robot just stay there completely motionless, nobody actually uses their phone in the real world that way. So instead, we have our robot simulate the exact same tremor on every phone, and we even automated the shutter tap, keeping things both true to life while still being perfectly fair. Now, the next challenge was the subjects themselves. So people in frame will blink and naturally drift from shot to shot, which ruins repeatability. So, our solution is to use realistic silicone heads on a customuilt motion platform which either lets us hold the subject perfectly still or we can move it with
millimeter level precision. And then the rest of the stuff we just control outright. We measure and adjust the lighting for every scene with an automated system. We run multiple trials which is super important when it comes to computational photography with us even resetting on a gray card in between. And whenever possible, we use software analysis to actually score the phones instead of relying on opinion alone. Now, this has been just a massive undertaking for me and the team. Not only did it take us longer than we originally planned it to take, but it also ended up being a lot more expensive than I thought it would be, which actually makes this the perfect time to thank today's sponsor, Zapier. So,
Zapier MCP is one of the easiest ways to unlock the power of AI models like Chat GBT and Claude by letting them actually use the tools in your workflow. Since while a lot of AI platforms already have native connectors, those can be limited in terms of the tools they support, the permissions you can set, and the actions they can perform. But with Zapier MCP, not only do you get access to 9,000 plus integrations, you also get fine grain control over what AI can and can't do, which is important since you don't want AI going out and deleting your data. Now, normally whenever I have to prep for a project like a Galaxy S battery test, it involves me having to bounce between stuff like notion, Google
Sheets, ClickUp, Slack, web research, and more. But since I have Claude connected to Zapier MCP, I can just describe the project in natural language to Claude. So, create a project for an ultimate Galaxy S battery test. Pull up our historical Galaxy S21 through S26 Ultra battery rankings from Google Sheets. Organize them into a table for context in notion. look for any significant battery related software updates since launch, create a ClickUp task for our researcher to go out and conduct the trials on each device, and then send a Slack message to our editor to shoot B-roll once testing starts, where instead of having to coordinate all of that myself, Claude can go out and use Zapier as the orchestration
layer and handle the workflow across all of those connected tools, which means I get to spend less time organizing projects and more time actually testing. You can check out what I built on Zapier and learn more at the first link down below. All right, so this first module was kind of our holy grail in low light motion because if there was any one thing that leads people to delete a photo more so than anything else, it's probably blur, which usually happens when you're taking a photo in a challenging lighting condition and your subject moves, whether that's a kid, a pet, or just anybody who won't stand still. And the results we got were
really interesting. So, starting with the ultrawide camera at a slow easy speed, while neither phone was able to produce a tacsharp photo, most of the shots they produced were still good enough where like if the moment really mattered, you probably wouldn't delete them. When we moved up to the medium speed, the gap did start to grow. On the iPhone, 85% of the shots landed in the keeper category, meaning it got a score above 40 points. Whereas on the Galaxy, it was only 50% of the shots where, you know, some of them look good, but the other half were pretty blurry. And then by the time we got to the faster hard speed, well, both phones struggled. But the good news is once we move over to
the main camera, both phones got dramatically better. At easy speed, both phones were doing really well. The Galaxy was actually producing more shots above the 55 point threshold, which I consider to be like genuinely sharp. At medium speed, the Galaxy outperformed once again with six of its shots crossing that 55 point threshold. And then when we cranked it up to the hard speed, which is more like when a kid quickly turns their head, we did see the iPhone pull away where all 20 of it shots scored above 40, whereas only 50% of the Galaxies did. But the story completely flipped once we switched over to telephoto. So even at the easy speed, we were only able to get one shot on the
iPhone to score above 40. On the Galaxy though, all 20 of the shots landed in that keeper category with two of them even scoring above 55. At the medium and high speeds, both phones did struggle with very few of the shots landing in that keeper range. So, you know, overall the story is mixed. On the ultrawide camera, the two phones were kind of neck andneck. On the main camera, the iPhone definitely had the edge, especially at that higher speed. And then when it comes to telephoto, well, there's no contest. I mean, the Galaxy's telephoto camera can actually get you a usable shot in low light, whereas on the iPhone, obviously, that's much less likely. All right, so in this next test,
we're looking at how good each phone is at keeping focus on a fastmoving subject, which isn't something that you're always going to run into. But if you're filming your kid on a swing or recording your pet running around, then it can make a difference. And starting with the main camera, both phones did pretty well. In their best runs, the iPhone kept about 91% of its frames in focus compared to 89% on the Galaxy. What was more interesting though was the consistency of the phones, where on its worst run, the iPhone still stayed close to its best result. While the Galaxy saw a bigger hit, dropping down to about 81% of its frames in focus, which you know
is still good, but it does suggest that the iPhone is a little bit more reliable when tracking a subject. And that was especially true when we tested the portrait or cinematic modes where the iPhone was able to still keep 90% of its frames sharp while on the Galaxy it varied dramatically where on some of its runs it did pretty well but then on its worst runs it had fewer than 30% of its frames actually sharp which might be due to the portrait processing taking longer to react to the changing scene. Moving over to the ultraide cameras though neither phone was particularly impressive. The iPhone hovered in the mid50s while the Galaxy was just a few points behind in the low50s. So, you know, the edge goes to the iPhone in
that department. But honestly, neither one really stood out. But then when we switched over to telephoto, well, this is where the Galaxy separated itself once again with it keeping more than 85% of its frames in focus while the iPhone was closer to 55. Now, we're not comparing the same exact focal length here, so that is worth keeping in mind. But even accounting for that, the Galaxy's telephoto camera was just clearly doing a better job at keeping its subject in focus. So, you know, overall, the takeaway is kind of the same as we had in the low light motion module with the iPhone having the slight edge in ultrawide, a bigger advantage when it comes to the main camera, and then the Galaxy pretty much dominating
when it comes to telephoto. All right, so next up is zoom smoothness. And I think this one is important because nothing ruins video faster than having choppy lens transitions as you zoom in and out. So to test it, we have our robot perform the exact same zoom gesture on both phones, moving from the ultrawide all the way to the 5x telephoto, while our software tracks the subject frame by frame. And you know, overall, both phones did pretty well. But across every single trial, the iPhone did manage to score higher. And when you look at the footage and especially the overlay here, the reason is pretty clear. The Galaxy has that extra transition at 3x that it has to do, which creates that noticeable detour
on its way to 5x. Whereas on the iPhone, its motion path is a lot straighter. And you know what I found especially interesting was how consistent the iPhone was. Where across its trials, its score barely moved at all. Whereas on the Galaxy, there was a bit more variation with some of its runs producing larger snaps during those lens transitions than others. It's not a huge difference, but still a clear win for the iPhone in zoom smoothness. Here in color accuracy, we're looking at two things. First, we want to see how faithfully each one can reproduce the colors in a scene compared to our calibrated reference. And then second, we want to see how consistent those colors remain as you switch between a phone's different cameras, since some
phones can dramatically shift colors when you go from something like the wide to telephoto. So to measure it, we took over 70 samples across two lighting conditions, matching every patch on our color chart and matching patches on our model's face to account for any local tone mapping. And starting off with photos, the two phones were actually surprisingly close. On the ultra wide camera, both phones matched the chart pretty well. But when we actually looked at the subject, you can see how local tone mapping changes things where across our trials, the skin tones on the iPhone were a little closer to our reference and how the scene actually looked. But when we moved over to the main camera,
the story flipped a little. The chart results were still close and you could see the colors matching like a and her shirt. But this time, the skin tones on the actual model were more accurate on the Galaxy with the iPhone adding some extra warmth. And you know, while some people may actually prefer that look, that's not what we're measuring here. We're measuring accuracy and compared to the reference, you can see that the Galaxy is a little bit closer to reality. By the time we got to the telephoto cameras, the phones were basically tied. The chart results were extremely close and the skin tones were also very similar. But where things started to separate was video. Across every lens we tested, the Galaxy was staying noticeably closer to our
calibrated reference. The iPhone was consistently pushing the brightness and saturation further in video, and it's especially true in skin tones. But you can even see it on the color chart with like the colors just looking brighter and more saturated. But the good news for both phones actually is when it came to color consistency across its own lenses, there were only small differences from camera to camera and averaged out across the three focal lengths that we tested. The two phones were basically tied when it came to consistency. Of course, this is just one scene. There are basically an infinite number of scenes that you could potentially take out there that obviously we couldn't test. But one thing that we did want to push further
was how these phones handle different skin tones. So for this module, the setup is pretty simple. We're taking photos and videos on each phone using three different scenarios. Once with only our lighter skin model in frame, once with only our darker skin model, and then once with both of them together. And the results were pretty interesting. So starting with the lighter skin model by itself, it was very similar to what we saw in the color accuracy module with both phones doing fairly well in photos while the Galaxy was generally more accurate in video. But when we switched over to only having the darker skin model in frame, the story completely shifts. For whatever reason, the Galaxy just starts pushing
the brightness up on the model much more aggressively. It doesn't show up in the skin tones on the color checker, but you can tell it's doing local processing on the actual silicone head here, and it's worse off for it. And, you know, this wasn't just a one-off. We saw the same behavior in all our samples and across all three cameras. And the thing is, this doesn't appear to be a hardware limitation because as soon as both models are back in frame together, the Galaxy went right back to being the more color accurate camera. But maybe the most interesting thing that we found was when we switch over to video, the Galaxy did perfectly fine all the way through both on the lightkinned model and on the
dark skin model with it even doing better than the iPhone. So clearly when it comes to photos, the Galaxy was just overprocessing and that's kind of been the trend that we've seen with smartphones nowadays. Sometimes phones go a little bit too far with their computational photography to the point where sometimes the photos you get are actually worse instead of better. Next up is image stabilization. And this is one of those areas where I've never been able to tell when one phone is truly better than another since without proper controls, you never really know if you're seeing a difference in stabilization or if it was just your hand moving differently. But now we can use the same exact motion path on each phone and then track the marker in the
scene to see exactly how well each phone stabilizes. And starting off with the main camera. First, we're doing a standing tremor. This is probably the most realistic scenario for when you're just holding your phone in place while recording a video with a couple of little jerks mixed in to reflect when your hand twitches. And you know, at least when it comes to normal tremor, the two phones look almost identical. But where there is a slight difference is while the iPhone does a good job at absorbing the initial jerks, it does do this small little overcorrection afterward that the Galaxy doesn't. When we switch over to the walking motion, this time the iPhone pulls ahead.
Although it's only bite a little bit. Honestly, if it wasn't for all this tracking and analysis, I probably wouldn't be able to tell the difference. And it was a similar story when it came to running, where the iPhone just looked a little bit steadier overall. But the biggest difference between the main cameras came when we enabled their respective action modes, where this time the iPhone did score significantly higher with it not only handling the up and down motion better, but there's also just less wobble compared to the Galaxy's image. Switching to the ultra wides, the two phones were really close with only a slight edge for the Galaxy across the three contexts. But honestly,
the biggest difference between all the cameras were in telephoto where the Galaxy just handled everything better. Like even in the standing tremor, which again is the most realistic scenario we're testing, it just does such a better job at handling those jerks. I mean, the iPhone, not only does it jerk more, but it also has this unpleasant wobble to it. any walking condition, it was kind of the same story with the Galaxy just doing such a better job at handling all that motion. And then when it came to the running motion, which honestly is a little bit ridiculous, the iPhone was all over the place, while the Galaxy surprisingly was doing pretty good given how fast we're shaking the
thing. So, you know, overall I would say the two phones are really close across the wide and the ultrawide, but its performance in telephoto does give the Galaxy the edge, at least when it comes to stabilization. All right, so the last module we have is resolution and sharpness. And this one is still very much a work in progress since we haven't quite dialed in our scoring system yet. But starting with the main cameras at their default settings, you can see how the iPhone's 24 megapixel output is giving it an advantage. You can really see it in the smaller bar patterns as well as in the seaman stars. In fact, when we flip on the heat map, the darker the green, the sharper it is. But moving
over to the ultrawide cameras, there was a pretty significant drop from both phones. And honestly, it's not too surprising since, you know, they're not using the very best sensors despite having those high megapixel counts where, you know, the iPhone technically has a small advantage, but I think the phones are pretty close. But then we go to telephoto and once again, the Galaxy starts pulling ahead. You can really see it in the bar patterns with it just being able to resolve finer detail. And that's despite being a 12 megapixel photo compared to the iPhone's 24. I don't want to dive any deeper into this module since, you know, it isn't quite fully ready yet, but I did want to give
you guys a quick preview. But anyways, that is it for me in this video. As you can probably imagine, this has taken a huge amount of work for us to build, and we're still just getting started. So, if you like the video, a big thumbs up would be highly appreciated. But other than that, thank you for watching and as always, I'll see you in the very next
