LucasHoller

Hey smart people, there are somthing i have been wondering for a while. It have been established and widely accepted that downscaling/upsampling 4k to 1080 result in a 8bit 4:4:4 image, but does this work for 1440p aswell? Will a 4k image upsampled to 1440 result in a 8 bit 4:2:2 image? Logiclly it makes sence by comparison to the upsampling for the 4k to 1080p but i dont think i trusty own logic in this case ? So what do you think? And just a bounus question: if yes, could you make a 4k 4:2:2 image to a 1440p 4:4:4 image with the same logic? Thanks. I know its a complicated debate but i wanted to hear your oppinion.

yeah would be very nice if someone on mac with >4 cores and CC could test if they have the same problem

it's a fine theory but, seen as i can utilize 100% CPU in both Handbrake and resolve, it cannot be a hardware bottleneck in the system as a whole. It can only either be a problem with the software or with my computer and that specific software. and the problem with comparing to your tests, is that you do not have more than 4 cores on any of your systems, which seem to be when the problem accurs ? i would will try to see if i know anyone who has a >4 core system but don't think many of my friends do - if any.

yes i have a program/os SSD and a DATA ssd. As for my settings when exporting these are the exact settings: h.264 3840x2160 framerate 25 profile: high level: 5.2 max render depth VBR, 2 pass target bitrate 80 maximum bitrate 100 mazimum render quality time interpolation: frame sampling. would send a picture but don't know how on this site i have triede doing only 1 pass uncheaking max render quality uncheaking max render depth but the most CPU usage i have seen is 35-40 % and i can see that i do not use all my ram.

i got it working 100% on handbrake, but i think i'm still gonna go for ram upgrade + resolve as i think that has much better GPU (CUDA) support and by that will be faster. ofcource i don't know untill i have tried it.

will try that spider.

yes i have. i know that premiere has problems with h.264 and CPUs with more than 4-6 cores, so i think that is the problem. and about resolve, the data for the CPU to crunch is feed to it by the ram, so i am 99% sure more ram would make it possible to use 100% cpu constantly, for transcoding from prores to h.264.

Science has been done ! maybe other people with the same problem can use this workflow too.

did the tests and the problem is solved! the test was a 33 second long clip in proress 422(hq) (3.1 Gb) rendered to h.264 100 mbps bitrate, 4k. (with a prores copy of the edit i'm only gonna be using around 60 mbps normally for youtube as i have the prores as my high quality copy) the results is as follows: premiere 05:47 (25-30% CPU usage, ram usage around 4 gb) Handbrake 01:44 (60-65% CPU usage, ram usage around 6-7 gb) resolve (can't test, the test is clearly bottlenecked by my ram but seem like it it would eat my system whole if i had more ram ! YES! it works!) the CPU usages jumpes to 100% straight away but within seconds it slows down when my ram is maxed out. i am going to upgrade my Ram and use resolve and untill that use Handbrake.

oh handbreake cropped my video for some reason, and after untoggeling it i could get the 2160p. testing speed now.

i can see that linus uses sorenson squeeze for his h.264 encoding to get better multi-core preformace from his 36-core mashine, but that kind of software is damn near as expensive as my computer...

maybe the windows version can go higher ?

just did that was the x264 encoder link i got and that have 3840x1597 as maximum reselution for h.264

i would love to use AME but simply cant accept it taking 10x the amount of time i should to encode.

the x264 encoder you send also have a max reselution of 3840x1596.