Honestly that’s all that, but when will we really be entitled to updates of models like Artemis, Proteus, Gaia? because it’s been a very long time that we always have the same versions… and still no updates… why don’t you make any updates on the models? I don’t understand… you quote us in your schedule that there will be a model manager with all the old and new models, well then where is this manager? nowhere… where are these model updates? nowhere… performance and everything is a good thing but wouldn’t it be so much to see next to that to do the rest? I will test the beta tomorrow. good day to all
This version is a bit slower than the last beta for my GTX 1060 Ryzen 2700x combo when doing a manual proteus upscale from 720p to 1080p and exporting Prores LT.
v3.1.1.1b upscales at 2.8fps (2.2fps in low power mode)
v3.1.2.0b upscales at 2.6fps (1.8fps in low power mode)
Leftover comment from the previous beta version.Stabilization doesn’t finish. It appears to do the analysis phase ok and the says it can’t complete. Currently running the same job through the latest release v3.1.1 with no problem.
It appears Apollo Soft and Sharpen was removed with the release of 3.1.1 stable and the removal of the feature remains in 3.1.2.0b, however there hasn’t been any official notes on it’s removal (to my knowledge).
Is the removal of Apollo Soft and Sharpen intentional? If so, is there a reason for this?
I can’t process on the 3090 system. Single stream works fine (sometimes), multiple streams reliably crashes. Screen fades to black as if the display driver is restarting.
Scrubbing through the file when editing the trim range while a stream is processing also seems to crash it.
I see slight improvement in processing speed (Proteus Auto: about 5% faster than in 3.1.1, less cpu load 50% now, 70% before, GPU at 98%). Nothing else of importance to be reported !
The 6900xt system works as intended. No issue with processing or scrubbing.
I still can’t do more than 2 streams in parallel though, which Suraj previously suggested that the system is running out of RAM. If I’m monitoring the usage manually though, it never exceeds 26GB or 81% when attempting 3 streams, but the 3rd one always crashes.
I think that is ok with your system setup. On my Ryzen 5800X with RTX 3060 I can only run 2 videos in parallel what loads CPU and GPU to their limits while Memory load still is very low.
Yes, nothing maxed out in that scenario. Plenty of CPU, RAM and VRAM available so I’d be curious to find out what it is about the 3rd stream that it doesn’t like.
I think we probably need to get used to no longer trying to run multiple processes of TVAI. After all, TVAI used to be essentially broken, for the longest time, in that it couldn’t fully saturate our CPU/GPU. Now that they’ve fixed that, maybe it’s time the realization kicks in that TVAI simply uses your hardware to the max, even on a single process (the way it should).
My notion is still, that TVAI is the weakest link in your entire chain (like, for me, VapourSynth won’t crash, nor x265, but TVAI might). And that, as a result, the ‘weakest’ element in the chain should run the shortest. (That is true, regardless of ‘weakness’, btw: if you run any process continuously for 20 days, there’s a higher chance it will crash somewhere along the line than when it just runs for an hour).
So, 2 should really be the max. But I prefer 1 myself, at which TVAI already pushes my i9 12900K to full load, and the GPU to ca. 80%.
Yes, meanwhile me too I prefer to run only 1. However when that will finish after midnight I start a second one before I leave.
It would be helpfull when a sequenc of videos would start one after the other automaticlly after a predecessor has finished.
