12/10/2023 0 Comments Pixel 3xl dirt 3 wallpapers![]() ![]() At 4:07, the photos from both devices look almost identical in terms of exposure, but highlights are again slightly better maintained on Mate(Pixel blows out "IDEA" and turns it white) 3:23 tells you the same story, better highlight management and more colour information preserved on the right side of the frame on Mate. If you pause at 3:16, you can notice right away that Mate 20 Pro brings out more details out of the background while restraining the light coming from the street lamp better than Pixel 3 XL. moreSure, here's the comparison that I found on Youtube. S Yu, I'm curious where it is you've seen higher DR from night mode and actual shadow advantage? Lin. OTOH LR's HDRDNG is highly prone to generating this same type of artifact. I haven't seen those in XS samples either but I've noticed that SmartHDR handles movement artifacts exceedingly well in general, the main reason it could be left on permanently to replace a regular single exposure. The blue tint around highlights should be residual artifacts from vibrations and differences in framing during exposure, you're right that I haven't seen this in Huawei's samples but then Huawei has many other artifacts. Each artifact is identically exposed, indicating that each global exposure containing them is also identical, therefore Huawei's night mode also "captures multiple underexposed images, merge them together and lifts up the shadows", just like what you said about GCam's night sight. Huawei's night mode has been shown in many instances-and I believe we've gone over this before-to generate artifacts of moving objects with the exact same exposure. Huawei does not do a "multi-exposure HDR" in the sense of bracketing, and even in this line of argument bracketing should only improve shadows(where it doesn't have an advantage), not highlights(where it sometimes has an advantage). "since it's based on a single exposure" I'm not really getting that, base what on a single exposure? I've seen no indication whatsoever in my experiences with the GCam that it prioritizes one exposure over others for anything other than the framing. ![]() I'm not really gonna go into too many specific details of that video because of the fundamentally flawed methodology, I looked for a sample pack of OOCJPGs, in case he's smart enough to include one, but no there isn't one. This is largely agreeing with you but I stress that I see a far bigger shadow advantage than highlight disadvantage. Regarding pure highlights, I believe that GCam's night sight tends to expose to the right, preserving more shadow(again you need proper samples to actually assess that), but that may slightly sacrifice highlights. If your argument is ultimately that night mode has a DR advantage over night sight *only at sub-1MP*, then I don't disagree, I haven't really looked into that. If phones still had ~1MP output then videos would suffice in most instances. That's why I keep saying low res video comparisons are meaningless gimmicks in the age of 10MP+ outputs. And 940x750 isn't even anywhere near 1MP. Further downsample to 1.5MP and the phone's output may look exactly as clean as the camera's, provided the same FoV and metering. ![]() Think a good camera with a clean output of 24MP at say ISO3200, and a phone which is really really noisy at 24MP, ISO3200, if you downsample to 6MP, both would look cleaner at 1:1 zoom, but the camera's improvement would be negligible while the phone's would be much more obvious, you're burying the noise floor deeper. Know that downsampling(discounting the video compression at this stage) by definition improves (shadow) DR because it effectively introduces a different cutoff for the acceptable SNR captured. You just don't have enough resolution to compare texture and actual shadow retention, you're comparing 940x705 photos with video codec compression and a second youtube upload compression and Huawei could get away with a lot of shadow smearing which suggest insufficient DR rendered invisible with this severe compression. moreYou gave the exact same video as another guy a few weeks ago, and to that guy I already replied that the samples are far too small to draw conclusions. Nick Tagataka, Sure, here's the comparison that I found on Youtube.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |