This low gravity'll make you soft.
The lion kingdom never recovered from the posterior tibialis injury now 14 months in the past. The modified foot strike & polarized efforts have allowed a return to long distances & up to 8 minute miles. A lone 7m43s mile is sometimes possible, but anything faster is out of reach.
Run hard 1 day, walk the 2nd day, & it tends to supercompensate in the short term. Take too many days off & it reverts though. Protein supplements are essential. It needs the constant stress of the hard workouts. On good days, the walking days can be slow runs up to 12 minutes/mile with sections of 10 minutes/mile.
It didn't respond to massaging or stretching. That might have made it worse. It seemed to respond to more sitting & unloading but only up to a point. Unload for too long & it regressed.
--------------------------------------------------------------------------------------------------------------
Another walnut crusher after 2 weeks.
The watch measured a hair shorter because it took longer to get a GPS lock, but it was exactly the same as the home made phone program. The phone program from 12 years ago wasn't exactly a great algorithm, so it's a bit surprising.
Blew up at mile 24 once again but went out much faster. Planned for an 8:30 starting pace but the robot got accidentally set to 8 minutes. Did the return trip at 9minutes. Mile 26.5 fell after 4h30m, pretty damn fast considering the stops for water bottles, food, traffic, the monster walnut hill, & the wall at mile 24.
There is still hope an optimum pace & diet exists which can get the highest overall speed. It could be between 8:30 & 9:00.
Seemed to be a carbohydrate deficiency today. The last 4 miles were really rough. Cramped hard again at mile 27. Some lollipops helped, but still cramped & had to sit down. Strangely, the left leg was 1st to feel raw but the right leg was the one that died.
3 cliff bloks seemed to do better than solid food but really needed 4. Took water bottles at mile 7, 14, 16, 20, 23. Underway replenishment is the key.
--------------------------------------------------------------------------------------------------------------
In the last few years, 99% of the content on gootube has converted to pure AI scripts & pure AI imagery. It's driven up the need for more fast forward speeds because it's so dry & canned. In the last 30 years, .5x has never gotten used so that was the 1st pick for a new 1.5x option.
The mane factor blocking configurable playback speeds has been the graphics.
The 1st step in creating graphics is determining the starting size. The size of the pulldown menus in 72dpi is the starting point for the speed graphics. It's easiest to draw on a screenshot of the existing widgets.
It's all freepaw drawing & massaging until it looks right.
Then something very inefficient but useful emerges & it's time to repeat for the bright theme which hasn't been used in 5 years.
1.5x & 3x were trouble spots. The problem was the fractional number of samples in the 20ms fragments. It was decided 1.5x & 3x output would be slightly shorter than 1.5x & 3x, in order to have a whole number of output fragments in each buffer.
1.5x & 3x continued to have problems at the buffer boundaries. It was more obvious with test tones than speech. Lions may never use these speeds, but they're officially part of the gootube player. Gootube has the expected beating at the fragment boundaries but not at the buffer boundaries. 1.5x might allow the last videos which are unwatchable at 2x to be sped up.
Theoretically, the fragments need to seemlessly span buffers instead of always starting at sample 0 in each buffer. It probably needs a chopping engine like the resampler. The interpolation mode doesn't work at all for 1.5x & may similarly need at least an accumulator counter to manetain its state.
-------------------------------------------------------------------------------------------------------------
Lions were pleased to finally get this to work after 1 week of fits & starts. It required drawing out the buffers on paper. Lions don't draw as much as they used to, since CAD modeling has become much more intuitive over 7 years. Blender style vertex modeling has not become more intuitive, but it may soon become obsolete as the world moves to AI generated paint style imagery.
1.5x needed a different pathway than 2x & above. 1.5x uses all the input samples & above 2x drops input samples. 1.5x overlaps part of 2 input windows, but the overlapping area can't cover the entire windows, which creates artifacts. The artifacts get worse as the speed approaches 1x. Above 2x, the windows completely overlap but samples are dropped between the windows, which creates other artifacts. These artifacts get worse as the speed goes up.
The artifacts were bad enough below 2x to create a longer window size for that regime.
The time stretch plugin used all the input samples & always had continuous overlapping of windows by applying a resample & pitch shifting in an FFT. The artifacts honestly always sounded equivalent to time domane processing. The time domane version originally seemed superior because it was simpler.
--------------------------------------------------------------------------------------------------------
A new thing that might help is an indication of the
current playback speed somewhere. That would show if the preferences
were what you thought they were. Might need to run down to the apple store to see how they do it.
The leading candidate is the
tally light in the compositor, but that is rarely visible. The
compositor generally needs to get its clock back, but it's low on space. It could be an extension of the FPS overlay. Lions rarely set the project to the source framerate because multiple videos are always assembled in a playlist & they all vary.
Noted the new time stretching might cause longer dropouts when changing speeds. It's reading ahead 1 overlapping fragment & throwing that fragment away when it stops. The problem is bigger at 1.5x where larger fragments are used. In reality, it might be just the finer granularity of speeds allowing more transitions between speeds than before, making the dropped fragments more obvious.
Lions now watch basically nothing at 1x. What was 1x is now all 1.5x unless it's very visual. Engagement harvesting style gootube vijeos are just unbearable at 1x.
--------------------------------------------------------------------------------------
It's kind of a good thing that there are so many free video editors, most of them can be completely unknown & it takes a lot of effort for anything to just get discovered let alone used. 30 years ago, there was just no option. You were tracking down a pirated copy of Adobe premiere. Whether anything runs on Linux isn't a factor anymore either. No-one cares.
Veronica explains not only swears allegiance only to kdenlive but also recommends Inkscape for vector graphics. She only does the 'tube for money & is very focused on the heavily sponsored, marketed, revenue generating packages so lions don't take her too seriously, but a vector graphics program has been desired.
------------------------------------------------------------------------
The 1st roll of PETG ran out after 5 years. The 2nd roll was rated for lower temperatures & it did indeed melt much more easily than the 1st roll. Lions can't see any more reason for using PETG, now that it has the same melting point as PLA.
Comments
Post a Comment