Export can be opened in external players or editors regardless of the Auto Recovery setting (previously needed it to be turned on)
Fixed Gaia failing on Mac
Fixed model download failure
Shows warning if output resolution is not supported by the selected encoder
Fixed Apollo model failing with Auto device setting
Improved zooming sensitivity in the preview
Fixes for Slowmo + change FPS does not work
Fixed processing failure due to t=nan
Fixed deinterlacing being too slow when also upscaling
Fixed video timeline shows zero duration when switching videos
Fixed duration reported incorrectly
Fixed trim resets when switching video
Fixed filters getting reset when switching from preview to export
Fixed ETA not reporting days
Fixed looping of preview while processing
Fixed misc. export preview issues
GUI crashes reduced
Fixed Auto audio not working
Fixed Transcode not allowing extra options in the encoders file
Misc. fixes and improvements
Known Issues
Preview mismatch for certain VOB files
Some videos can still have some seeking issues in Preview/Export
Color space / HDR issues
Keyboard navigation missing
Intel codec not working on some machines
Some quality mismatch between 2.6.4 and 3.0
Multiple selection for source videos is missing
Output bitrate is not reset to Preferences bitrate when importing a new file
Different view mode is still missing
Failure can happen when resizing to a big resolution like 4k or 8k without a warning if machine runs out of memory
Speed can be slower than 2.6.4 with some models on some machines. We’ll be working on speed optimization in the next weeks
Proteus with Auto + Stabilization may cause artifacts
Preview with image sequence doesn’t work or plays at 25fps
Problems and Solutions:
Make sure Max. Memory Usage in the Preferences is set to a high number
If the Gaia model fails on Mac, try re-installing the app using the DMG file
If export fails, try using Transcode/None/Auto in Audio Settings in Preferences
Please provide your feedback on the following items:
Apollo: Apollo model is trained to better handle non-linear motions and, in some cases, reduce motion blur while doing slow-mo or FPS conversion. Please check the performance in terms of quality and speed and let us know.
– Note: We do not recommend using the Apollo model only to reduce motion blur.
Stabilization: Stabilization introduced a new feature called “Reduce Jittery Motions” which iteratively reduces the inter-frame jitter-like motions. The intention is to improve stabilization when there is 3d rotation, parallax, lens distortion, etc. involved. Please check the stabilization quality with/without this feature and let us know.
– Note: We recommend turning on Enhance filter when using this feature since the resulting video could be blurry after stabilization.
preview does not remember zoom factor for next preview
preview of specific single frames preferred
this could speed up generating preview up to factor 60 on 60 fps footage
the preview while playing often pauses and jumps back to frame one
cut video start and end with specific frame numbers preferred
the application does not take into account the Windows scaling
running a 2nd instance not working anymore: out of memory
in this version the vertical scroll bar on the far right scrolls in too small steps too slow when scrolling via mouse
the mouse wheel zoom happens in strange steps. preferred: start with 100 % and then rise/lower by 5 % each step: 5% 10% 15% 20% and so on
an old issue appread again where a black box begins to eat the output image. Interesting: Just the moment where the tension cores were included again the black box issue re-appears:
Maybe you missed it, but you can already change the aac bitrate to a custom one, and even have several of them, by editing the encoders file in programData, and create a custom entry or several of them in the preference or the custom output. i didn’t tried it yet, so don’t know if it’s working as expected. it’s there since the last early access version. there is a topic about it :
so you can create a default in preference like H264 like AAC 192kbs, or one with 320, one with copy etc… available as custom output.
Thanks for fixing the allow recovery issue with video editor. videos load ok in vegas pro now !
the - :
the export has some issue once processed. pictures move from beginning to end, and some other stuff.
comestic suggestion : adding noise together with grain at the bottom of the filter would be better. now we must on a 1080p 100% dpi screen, scroll to edit each time proteus.
the allow recovery turned on didn’t fixed the fact that there is still no “in progress” during the export, much check that better.
got a out of memory while trying to do 2 process at the same time, one of the process stopped. much check on an another card with better Vram to be sure.
just got the same error as you (in preference with the several process feature).
This is a more generalized question for the development team. Why is there a lower GPU usage when upscaling than there is when using original(100 percent) or even downscaling? I get the same results no matter the input file resolution. I’ve seen the same type of behavior even back in 2.6 versions. I’m curious more then anything, not criticizing. Trying to understand how the GPU’s and AI coding work at a very top level.
Decent uplift in speed from 3.0.0-2 with my RTX 3060Ti. 1080p - 1080p(100 Percent) scaling, proteus estimated settings. Same estimated settings and all other settings used in 3.0.0-2 and 3.0.0.5.b. 0.30spf for 0-2 and 0.25 spf for 0.5.b. For a 2 minute video completion time was ~21 minutes for .0-2 and ~16 minutes for 0.5.b. The last thing that is odd to me is the average GPU usage as reported by HWInfo64 for each run was very different. For .0-2 it was 82 percent but for the faster 0.5.b it was only 69 percent. GPU memory usage was the same as was CPU and system memory.
I did pixel peek the 2 processed videos and they are the same. The way I could tell the difference was in the audio. While both were set to auto the 0.5.b file audio was converted from 192Kb to 128Kb while .0-2 did not convert the audio. That file reported 192Kb. Again .0-2 was the slower processing but just copied the audio it appears.
The apollo model is not working or names files incorrectly? Changing fps from 24 to 60 (without touching the slowmotion tab) names files with label CHF3, not APO.
Nice the TensorRT models, now it’s faster than v2.6 on my RTX 3090 and finally almost a full GPU usage (75-90%) where before it was staying under 45%.
For example SD-DVD GAIA-HQ X2 I always had about 0.10/0.12 spf with v2 and now it’s 0.07
What happened with Appollo ? If we only want a slo-mo of 50% or just do the trick to double the original frame rate ? It’s now called 4-8x slo-mo and at first there are only options to 400% up 800%, I had to manually set 200% and now it’s get memorized and it seems to work properly though. But this can be disturbing and may think Appollo is only for 4-8x.
Also could it be possible to have the ability to switch the upscale-Slomo order ? If the source is already quality then it could be better to do the slomo first on the original and then the upscale. It should be much faster than doing the slomo after the upscale. Of course if the source is bad or low res better do the slomo at the end with clean enhanced images
About quality mismatch between 2.6.4 and 3.0, will this be fixed because sometimes there’s a huge difference and V2.6 output quality is still far superior on all models and it’s a real downside.
For example we all know small faces is the real weakness of VEAI for the moment and they are now worse on V3. I will try to post some images comparisons when I will have the time.
