It's a bit of a gimmick. These phones just don't have the noise performance to make log video work outside if very very specific conditions. I had it also on my old LG V30 and it only was remotely useful in full sunlight (and since we're talking about very low processing, not much changed since then).
This is inevitable because the noise floor is just too high to have a large usable dynamic range unless illumination is high.
Combined with video compression it's just not great. It's not really even unique to smartphones, many DSLRs/MILCs when they first started supporting log video had similar issues, but obviously it's going to be much worse for a smartphone.
The sensor size is the limiting factor here re: noise performance and dynamic range, but the iPhone does better than most with some ML and other “computational photography” tricks for denoising that other small sensor cameras don’t have. And ProRes is a great codec that doesn’t really have compression artifacts at high bitrates.
It’s not going to replace your Alexa or even full frame dSLR but can be useful and is a welcome evolution.
ProRes is actually worse at the same bitrate when compared to H.265 or AV1, and sometimes even worse than H.264. The advantage is that it's easier to edit because it's all intraframe, not that it has higher quality - it's worse. Even intraframe HEVC is going to be better than ProRes. H.265 really is going to give you consistently better pSNR and SSMI than ProRes, and that shouldn't be surprising - interframe compression is better for quality.
The only advantage here is slightly better denoising, yes. But denoising doesn't change the dynamic range limit, it just hides the noise visually. The SNR is actually going to be lower.
This is not true. We've compared RAW video from smartphone cameras to a couple of professional cameras and the difference is not as much as you might think.
I don't understand what you're trying to say. Yes, in absolutely perfect light, if it has the focal length you need, it's going to be good. But that's a very particular situation, we both know that as soon as the ISO comes up it's not even close, and soon enough shooting in log then becomes a gimmick.
Have you done any comparisons that you can point to? The iPhone 15 hasn't even been out all that long. The video in the sibling comment notes that the iPhone does surprisingly well in terms of dynamic range, for example (7:12).
There are two limiting factors in dynamic range. Those are the dynamic range of the sensor itself (which is and has been great, even on my V30), and the difference between the highlights and the noise floor - and the noise floor is fixed. The iPhone 15 doesn't have a meaningfully lower noise floor than the iPhone 14 and it's only barely better than the V30 - or any other smartphone with a BSI sensor and a large aperture.
The video is exactly consistent with what I'm saying. All examples are either in broad daylight or strong studio lighting.
As far as I can tell, you haven't done a comparison in low light conditions of the iPhone 15 against any other video camera. Of course one would expect a camera with a larger sensor to do better in low light, all else being equal. But all else is not equal. Professional video cameras aren't stacking multiple exposures to construct a single frame, and don't have anything like the computational power of the iPhone 15. So it would be much more interesting to see a real comparison than to hear people repeating abstract theoretical points over and over again.
> Professional video cameras aren't stacking multiple exposures to construct a single frame
We are talking about video.
> and don't have anything like the computational power of the iPhone 15
They have far more, because the processing is done in post with powerful workstations. For stills you have a point because of multiframe techniques you can't easily do in post, but that doesn't work at all for video.
> repeating abstract theoretical points over and over again
You don't have to. There are dozens of practical comparisons done on dozens of smartphones over the past years. No one has really done it on the iPhone 15 (yet) because it just came out, but there is no reason why it would be different. People have done the comparison with the iPhone 14 and even without log formats results are far worse in anything but perfect light, even compared to hobbyist-grade video cameras costing less than the phone. This will be even worse for log video by the nature of the logarithmic transformation.
The iPhone does multiple exposure HDR in video too. That's why the dynamic range is so good.
>They have far more, because the processing is done in post with powerful workstations.
If you are shooting RAW video, sure. Otherwise a significant amount of processing has to be done on the camera.
By all means link to any comparison that you think is relevant. But if it doesn't involve an iPhone 15, it doesn't tell us much about the iPhone 15. It especially doesn't tell us much about a video mode that's only available on the iPhone 15.
> The iPhone does multiple exposure HDR in video too. That's why the dynamic range is so good.
This doesn't do anything for the SNR dynamic range limitation. Smartphone sensors nowadays are going to be limited by noise for dynamic range very rapidly. Multiple exposures in video mode reduces total exposure and is only worth it with more-than-perfect lighting. It also means you can't guarantee a 180* shutter angle so will need to disable it for smooth movement.
> If you are shooting RAW video, sure. Otherwise a significant amount of processing has to be done on the camera.
The only additional preprocessing is debayering and color transformation, none of which prevents the type of processing we're talking about. It doesn't have to be in-camera.
> By all means link to any comparison that you think is relevant. But if it doesn't involve an iPhone 15, it doesn't tell us much about the iPhone 15. It especially doesn't tell us much about a video mode that's only available on the iPhone 15.
Log video is not a video mode that's exclusive to the iPhone 15. Various phones have had it since 2017. The only difference is the denoising and sharpening and that's a known quantity.
See https://www.youtube.com/watch?v=vUn68iFMNXY. Remarkably similar. Big difference seems to be unavoidable differences arising from a small sensor like bokeh and DOF, and especially the consistency of colour from the ARRI which makes grading a breeze. But the fact that iPhone footage can be intermingled with 100$k camera footage (after grade) without being noticeable is shocking
Why should it be? The V30 came out in 2017 and people made exactly the same videos about it, about how you could intermingle the footage with an Alexa or a RED. You can, yes, only in perfect light and if the focal/aperture works for the scene. That's rare, and that's why it hasn't caught on.
This is inevitable because the noise floor is just too high to have a large usable dynamic range unless illumination is high.
Combined with video compression it's just not great. It's not really even unique to smartphones, many DSLRs/MILCs when they first started supporting log video had similar issues, but obviously it's going to be much worse for a smartphone.