suraj
March 23, 2023, 9:59pm
1
Hi All,
Here is an early version of app that should improve performance for all machines. The degree of performance increase will vary. This version should also allow for better support running multiple filters and multiple processes.
Download: Win | Mac | Mac Update | Ubuntu
Please upload problem videos and logs here: Submit Files
NOTE: Users experiencing system crashes on Intel Mac Pros, can you please try this version. If the machine is still crashing can you please send us the logs and apple crash reports from the console app.
Changes from 3.2.0.3.a
Fixed Chronos/Chronos Fast produces gray results on RTX GPUs
Fixed issue with processing failure on intel Mac
GUI updates from 3.1.10 and beyond
Changes from 3.1.0.0.L (Ubuntu users) :
CLI arguments now match Windows/Mac (tvai instead of veai)
Web-login flow is now supported
Better handling of dependencies for .deb package
Updated to Video AI 3.2.0.4
Performance improvements
Thank you for testing
PS: Since this is an alpha version, it will not interfere with Beta or Release versions on the machine.
6 Likes
I can confirm this issue is resolved now, thanks!
@mikmod1
1 Like
Excellent speed in this version, the best Iâve seen so far! A massive boost on Proteus and Artemis!
Topaz Video AI Alpha v3.2.0.4.a
System Information
OS: Windows v11.2009
CPU: AMD Ryzen 9 7950X 16-Core Processor 63.139 GB
GPU: NVIDIA GeForce RTX 4090 23.59 GB
GPU: AMD Radeon(TM) Graphics 0.47446 GB
Processing Settings: device: 0 vram: 1 instances: 0
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 40.59 fps 2X: 16.29 fps 4X: 4.59 fps
Proteus 1X: 35.48 fps 2X: 15.55 fps 4X: 4.43 fps
Gaia 1X: 15.26 fps 2X: 9.9 fps 4X: 4.94 fps
4X Slowmo Apollo: 28.79 fps Chronos: 29.22 fps Chronos Fast: 36.24 fps
v3.1.9 for comparison:
Topaz Video AI v3.1.9
System Information
OS: Windows v11.2009
CPU: AMD Ryzen 9 7950X 16-Core Processor 63.139 GB
GPU: NVIDIA GeForce RTX 4090 23.59 GB
GPU: AMD Radeon(TM) Graphics 0.47446 GB
Processing Settings: device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 29.62 fps 2X: 11.88 fps 4X: 2.97 fps
Proteus 1X: 21.38 fps 2X: 10.67 fps 4X: 2.89 fps
Gaia 1X: 15.79 fps 2X: 9.81 fps 4X: 3.71 fps
4X Slowmo Apollo: 22.02 fps Chronos: 30.37 fps Chronos Fast: 33.13 fps
mikmod1
March 24, 2023, 2:56am
8
Thank you for the heads up! Gotta test it ASAP.
Edit: Indeed, gray frames are a song of the past.
1 Like
TPX
March 24, 2023, 8:36am
9
Please check for Quality too, not just numbers.
1 Like
JakSpoon
March 24, 2023, 10:10am
10
@suraj
Any plan to support the new feature of Nvidia STUDIO Drivers v531.41 - âIncreased the maximum number of concurrent NVENC sessions from 3 to 5 on supported GeForce GPUsâ?
How much NVENC hardware encoding is used for video uspaling? Can the speed of upsaling be improved with this new function?
topaz257
March 24, 2023, 10:51am
11
TVAI uses a maximum of one NVENC sessions per process and a maximum of 4 parallel processes. Therefore, before those drivers, u couldnât run more than 3 parallel TVAI processes with NVENC encoding. With those drivers u can now run 4 parallel processes and an additional NVENC encoding in some other software if wanted.
Be aware, that the NVENC quality is not the best. It seems to be optimized for streaming rather than high quality video encoding. I even found the AMD H265 GPU accelerated encoding of my Ryzen 9 7950X CPU of better quality (less banding artifacts and preserving more fine details) and similar speed. Software H265 running on a CPU can be even better but at a much lower speed compared to AMD- or NVIDIA accelerated GPU encoding.
1 Like
TPX
March 24, 2023, 2:02pm
12
Topaz Video AI Alpha v3.2.0.4.a
System Information
OS: Windows v11.2009
CPU: AMD Ryzen Threadripper 3960X 24-Core Processor 127.88 GB
GPU: AMD Radeon PRO W6800 29.618 GB
Processing Settings: device: -2 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 9.73 fps 2X: 6.88 fps 4X: 2.67 fps
Proteus 1X: 9.88 fps 2X: 6.75 fps 4X: 2.42 fps
Gaia 1X: 4.58 fps 2X: 3.08 fps 4X: 2.06 fps
4X Slowmo Apollo: 12.79 fps Chronos: 5.69 fps Chronos Fast: 8.96 fps
Topaz Video AI Alpha v3.2.0.3.a
System Information
OS: Windows v11.2009
CPU: Intel(R) Core(TM) i7-7820X CPU @ 3.60GHz 63.693 GB
GPU: NVIDIA Quadro RTX 5000 14.795 GB
Processing Settings: device: 0 vram: 1 instances: 0
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 10.83 fps 2X: 6.47 fps 4X: 2.06 fps
Proteus 1X: 9.88 fps 2X: 6.11 fps 4X: 1.99 fps
Gaia 1X: 3.61 fps 2X: 2.42 fps 4X: 1.5 fps
4X Slowmo Apollo: 10.01 fps Chronos: 4.01 fps Chronos Fast: 12.81 fps
Isnât this the Video AI Performance Alpha thread? Please point me to the Video AI Quality Alpha thread
But yes I agree, quality is very important, but unlike benchmarking there is no easy standardized way to test quality (and quality is also highly subjective).
Which leads me to think, maybe there should be a quality test function built in to Video AI?
Something like an assortment of included short clips that can be processed and easily compared frame by frame for each version?
Thoughts?
1 Like
TPX
March 24, 2023, 4:24pm
14
My point was that the performance increase could change the quality.
1 Like
Understood, but I still like the idea of a built in quality âcheckâ function.
mikmod1
March 24, 2023, 5:01pm
16
Oh, so NV became aware of the patch , which lifts this limitationâŠ
Thereâs a slight issue with NVENC encoding (probably) in this alpha, when I choose a static Mbit/s value, like:
Sometimes (intermittently) Iâll get an âUnknown errorâ, if I try to preview with those parameters:
ffmpeg "-hide_banner" "-nostdin" "-y" "-nostats" "-t" "15" "-ss" "6805.006416666666" "-i" "X:/old smrody/filmy/The Lord of the Rings The Fellowship of the Ring EXTENDED (2001) [1080p]/The.Lord.of.the.Rings.The.Fellowship.of.the.Ring.mp4" "-flush_packets" "1" "-sws_flags" "spline+accurate_rnd+full_chroma_int" "-color_trc" "2" "-colorspace" "2" "-color_primaries" "2" "-filter_complex" "tvai_cpe=model=cpe-1:filename=W\\:/Temp/230172669/temp/tvai_558296595_cpe.json,scale=w=2592:h=1080:flags=lanczos:threads=0" "-f" "null" "-"
&&
ffmpeg "-hide_banner" "-nostdin" "-y" "-nostats" "-t" "15" "-ss" "6805.006416666666" "-i" "X:/old smrody/filmy/The Lord of the Rings The Fellowship of the Ring EXTENDED (2001) [1080p]/The.Lord.of.the.Rings.The.Fellowship.of.the.Ring.mp4" "-flush_packets" "1" "-sws_flags" "spline+accurate_rnd+full_chroma_int" "-color_trc" "2" "-colorspace" "2" "-color_primaries" "2" "-filter_complex" "tvai_stb=model=ref-2:filename=W\\:/Temp/230172669/temp/tvai_558296595_cpe.json:smoothness=6:rst=0:wst=0:cache=128:dof=1111:ws=32:full=1:roll=0:reduce=0:device=0:vram=1:instances=0,tvai_fi=model=apo-8:slowmo=1:rdt=-0.000001:fps=60:device=0:vram=1:instances=0,tvai_up=model=prob-3:scale=0:w=2592:h=1080:preblur=0.0390746:noise=0.0799586:details=0.278966:halo=0.0398348:blur=0.269519:compression=0.526377:device=0:vram=1:instances=0,scale=w=2592:h=1080:flags=lanczos:threads=0,split=2[out1][out2]" "-map" "[out1]" "-c:v" "tiff" "W:/Temp/230172669/temp/_tvai_560951152/%06d.tiff" "-map" "[out2]" "-c:v" "h264_nvenc" "-profile:v" "high" "-preset" "medium" "-pix_fmt" "yuv420p" "-b:v" "5M" "-map_metadata" "0" "-movflags" "use_metadata_tags+write_colr " "-map_metadata:s:v" "0:s:v" "-map_metadata:s:a" "0:s:a" "-map" "0:a" "-c:a" "copy" "-metadata" "videoai=Stabilized full frame with smoothness 50. Slowmo 100% and framerate changed to 60 using apo-8. Enhanced using prob-3 with recover details at 27.8966, dehalo at 3.9834799999999997, reduce noise at 7.99586, sharpen at 26.951900000000002, revert compression at 52.637699999999995, and anti-alias/deblur at 3.90746. Changed resolution to 2592x1080" "W:/Temp/230172669/previews/The.Lord.of.the.Rings.The.Fellowship.of.the.Ring.EXTENDED.2001.1080p.BrRip.x264.YIFY_1_stab_apo8_prob3.mp4"
Iâm using latest, 531.41 Studio driver:
2 Likes
Imo
March 24, 2023, 5:34pm
17
the whole windows froze after some time of processing with 3.2.0.4.a. on the other hand the 3.1.9 ran rock solid for a whole day without any problems!
rjo
March 24, 2023, 9:06pm
18
nice speeds all around, except for 4x slowmo which was faster for me in v3.2.0.2.a
Topaz Video AI Alpha v3.2.0.4.a
System Information
OS: Windows v10.2009
CPU: Intel(R) Core(TM) i9-10900K CPU @ 3.70GHz 63.89 GB
GPU: NVIDIA GeForce RTX 4090 23.59 GB
Processing Settings: device: 0 vram: 1 instances: 0
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 26.69 fps 2X: 10.26 fps 4X: 2.98 fps
Proteus 1X: 20.28 fps 2X: 9.54 fps 4X: 2.87 fps
Gaia 1X: 12.18 fps 2X: 8.1 fps 4X: 3.59 fps
4X Slowmo Apollo: 17.28 fps Chronos: 25.74 fps Chronos Fast: 26.1 fps
Topaz Video AI v3.1.10
System Information
OS: Windows v10.2009
CPU: Intel(R) Core(TM) i9-10900K CPU @ 3.70GHz 63.89 GB
GPU: NVIDIA GeForce RTX 4090 23.59 GB
Processing Settings: device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 19.45 fps 2X: 7.14 fps 4X: 1.92 fps
Proteus 1X: 14.85 fps 2X: 6.13 fps 4X: 1.88 fps
Gaia 1X: 13.02 fps 2X: 7.84 fps 4X: 2.97 fps
4X Slowmo Apollo: 17.23 fps Chronos: 24.41 fps Chronos Fast: 25.96 fps
Topaz Video AI Alpha v3.2.0.2.a
System Information
OS: Windows v10.2009
CPU: Intel(R) Core(TM) i9-10900K CPU @ 3.70GHz 63.89 GB
GPU: NVIDIA GeForce RTX 4090 23.59 GB
Processing Settings: device: 0 vram: 0.95 instances: 0
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 24.16 fps 2X: 10.76 fps 4X: 2.84 fps
Proteus 1X: 19.21 fps 2X: 9.4 fps 4X: 2.76 fps
Gaia 1X: 12.3 fps 2X: 8.04 fps 4X: 3.57 fps
4X Slowmo Apollo: 24.32 fps Chronos: 27.28 fps Chronos Fast: 29.91 fps
Topaz Video AI v3.1.9
System Information
OS: Windows v10.2009
CPU: Intel(R) Core(TM) i9-10900K CPU @ 3.70GHz 63.89 GB
GPU: NVIDIA GeForce RTX 4090 23.59 GB
Processing Settings: device: 0 vram: 0.95 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 19.99 fps 2X: 7.05 fps 4X: 2.09 fps
Proteus 1X: 14.37 fps 2X: 6.31 fps 4X: 1.9 fps
Gaia 1X: 13.33 fps 2X: 7.71 fps 4X: 3.03 fps
4X Slowmo Apollo: 19.07 fps Chronos: 24.63 fps Chronos Fast: 25.9 fps
suraj
March 24, 2023, 9:09pm
19
4X slow mo on 4090 was not working correctly. That could explain the difference in performance.
2 Likes
