A small factor like 2048/1920 is difficult to achieve as you basically need a very very large filter kernel to effectively remove just the frequencies you don’t want. “Resampling an image works well for large factors. We’re oversampling by a lousy 14%. To quote one guy on the topic: I happen to agree with that, and Charles Poynton made an eloquent, 5-hour long argument to that effect at a recent talk at the HPA retreat.īut in this case we’re oversampling not by 200% or 300%. It’s become conventional wisdom that oversampling is a good thing. But that’s not a big deal, right? You just press the magic resize button, and there’s no penalty, right? 2048>1920 is not enough Oversampling Why? Because your fancy, non-standard 1152 lines of resolution will have to be scaled down to 1080 lines of resolution. Putting that extra 14% of spatial resolution into your DI can actually damage the picture that your audience sees. Likewise, if you’re finishing in 1.85 and “2k”, your actual deliverable is 1998×1080. If you are finishing in a 16×9 aspect ratio, and you’re finishing in “2k”, you’re actually going to deliver … 1920×1080. As far as DCI, SMPTE, your SDI cables, and Digital Cinema projectors are all concerned, “2k” still means 1080 lines of resolution or less. Yep: none of those containers have more than 1080 lines of resolution. To wit:ĭid you notice anything there? Deliverables are 1080 lines or less. Juggling your post finishing rasters has always been a mild pain. If you are doing your final post finishing in 2048×1152, you are actually hurting your picture, and you should stop at once.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |