I hope Brandon and team decide correct this mistake and put islands where necessary!!! I have some ideas for how to work it in :) I'll put them at the bottom of the description because technically Oathbringer spoilers (joke) (i dont actually think they need to change anything) (roshar is perfect) Nasir can't keep getting away with this - what's next, will Kindle serve us pages in JPG format?
"Ackchually frame 65 at 25fps is exactly 2.6 seconds in, which should line up with frame 156 at 60fps" I wish it were so. But it's not. I'm using 1-indexed frame numbering (for no reason other than that ffmpeg defaulted to it when I was extracting the original video a few days ago). So, when I say "frame 1" I mean the very first frame; there is no frame 0. If we shift it to 0-indexed, then the first frame is frame 0, and now our divisibility tests are accurate. By that metric, we're talking about frame 64, which would be frame 153.6 (zero indexed). Not an integer. And there's a second problem on top of that. Flohack misunderstood how POV-Ray counts frame numbers. They put "-KFI0 -KFF125". The problem is that POV-Ray is inclusive in its frame counts, so this resulted in 126 frames, not 125. When converted from PNG sequence to Theora .ogv file, the interpreted framerate was 25. This puts the duration at 5.04 seconds. I believe Flohack intended to create a 5 second video of 125 frames, but didn't notice what had happened - the output ranges from "julia000.png" to "julia125.png" so they may have gotten confused and thought that that was 125 frames in total, while really it's 126. For this reason, creating an equivalent animation at 60fps is doubly impossible, because frame #65 was interpolated between 0 and 125, not 1 and 125 (there are 126 distinct steps, not 125). For my half/half slice view of wikimedia versus my recreation, I accounted for this by doing (clock*303/302.4), and rendering from 0 to 302 inclusive, and allowing the last 0.6 frames of the Wikimedia to repeat, since the alignment is off by 0.4 frames by the end by the wall clock, but not by the POV-Ray clock, since I adjusted it. This made the frames line up, at the cost of the resulting 60fps video not having an integer frame count from the first to last aligned frame to the original.
"VP9 also uses DCT" Yes, but not in the same way. VP9 has dynamic DCT sizes up to 32x32. There's also a difference between using a DCT to encode the difference between a P-frame prediction and the true frame, and using a DCT for I-frames.
Assets:
Links:
"I gave Isaac the outline of this world that became Roshar. (Based on an iteration of a Julia set, though for a while I played around with making the whole continent a cymatic shape.)"
Okay so. Remember Kaza's chapter in Oathbringer? The islands to the north of Aimia USED to be there, but they've been soulcast into smoke long ago, like how Kaza did to a few nowadays. This could be revealed in a flashback in Stormlight 5. For the southern one, I don't have quite as punchy of a suggestion. It might be necessary to retcon the entire battle of Thaylen Field - to change the orientation to be south-to-north instead of north-to-south. I've looked through and this can actually be accomplished just by changing three words! First, on page 1017 "before heading to the south" - change south to north. Second, on page 1099 "the north side of the battlefield" - change north to south. Finally, on page 1142 "still encompassed the north side of the landing" - change north to south. That should do it in-text, but regarding the maps themselves, perhaps it could be chalked up to faulty cartography?
0 Comments