Like the video flickers when you play it in another player?
The whole TVAI application flickers for me and has since TVAI 3.0.0. I don’t really care so I’ve never mentioned it.
Does it work?
Just updated my graphics driver and rendered / exported , the flickering has stopped.
Additional info….the original clip had no flickering. Did not flicker within TVAI. I also switched within OMEN gaming hub to a duel driver, gaming and apps appropriate driver, along with setting to app makes the choice instead of my specified driver. So far so good.
Mike
I experienced 2 software lockups while in the new app and adding video files to the existing batch, and a hard reset after 40 mins processing on my first use of Proteus V4.1.0. Just FYI, I hope it helps someone. Version 3.3.8 with IRIS1 and Proteus3 was the most stable for me, no crashes.
AMD 5700x
AMD 6650 Driver 23.12.1
32 gigs of ram / Windows 11 latest build.
This has to be the absolute worst update feature I’ve ever come across. I shoot DPX all the time, and not only does the conversion to TIFF completely wreck the image and make it ungradable, it also takes an unbearable amount of processing time and a massive amount of space.
Am I missing something here? Why was native DPX support removed and replaced with this? It worked flawlessly and gave beautiful upscales that was a dream to grade, with no processing time and eating of space at all.
I’m curious to understand the reasoning behind this move. No matter what, it doesn’t work, and it has rendered the software completely useless for me.
Interesting ONLY with these settings my old VHS MP4 gets better if set to Auto default or estimate manual forget it . easy to remember and it works but very counterintuitive
For me i can only use NXF1 in second enhancement the first enhancement requires me to set interlaced then Proteus v4 or iris v2 LQ
But I agree I get a more polished output actually I am astonished how NXFv1 on top op Proteus v4 improves the stone stamped mp4s with the settings from Serious out of the Box .
By the way NXF11 can do the recover details clean up better than Artemis just pull the sliders a bit more to the right Denoise and Blurr. The more stone stamped your MP4s VHS the more pull it to the right. So far I am totally blown away with Protv4 With NXF1 just using deblur and denoise.
For me IRIS lQ2 and manual and 0
and NXF1 scale 4 is most impressive to clean up my old 1994 VHS incredible cleanup

I have one question I see the command but miss the sharpen setting from NXFv1
ffmpeg “-hide_banner” “-t” “9308.02” “-i” “C:/usbsf8008/New folder/20210101_044553B.mp4” “-sws_flags” “spline+accurate_rnd+full_chroma_int” “-color_trc” “6” “-colorspace” “5” “-color_primaries” “5” “-filter_complex” “bwdif=mode=0:parity=-1:deint=0,tvai_up=model=iris-3:scale=4:preblur=0:noise=0:details=0:halo=0:blur=0:compression=0:blend=0.11:device=0:vram=0.94:instances=1,tvai_up=model=nxf-1:scale=1:w=720:h=576:preblur=0.14:noise=0.02:details=-1:halo=-1:blur=-0.92:compression=-1:estimate=8:blend=0.09:device=0:vram=0.94:instances=1,scale=w=720:h=576:flags=lanczos:threads=0,colorspace=ispace=5:space=5:primaries=5:trc=6” “-c:v” “h264_nvenc” “-profile:v” “high” “-pix_fmt” “yuv420p” “-g” “30” “-preset” “p7” “-tune” “hq” “-rc” “constqp” “-qp” “18” “-rc-lookahead” “20” “-spatial_aq” “1” “-aq-strength” “15” “-b:v” “0” “-map” “0:a” “-map_metadata:s:a:0” “0:s:a:0” “-c:a” “aac” “-ac” “2” “-b:a” “320k” “-map_metadata” “0” “-map_metadata:s:v” “0:s:v” “-movflags” “frag_keyframe+empty_moov+delay_moov+use_metadata_tags+write_colr” “-metadata” “videoai=Deinterlaced and enhanced using iris-3; mode: manual; revert compression at 0; recover details at 0; sharpen at 0; reduce noise at 0; dehalo at 0; anti-alias/deblur at 0; focus fix Off; and recover original detail at 11. Enhanced using nxf-1; mode: relative to auto; reduce noise at 2; reduce large grain at -100; revert compression at -100; dehalo at -100; anti-alias/deblur at 14; focus fix Off; and recover original detail at 9” "C:\te
Team Topaz, question…when applying manual estimated settings for a video, the setting is applied globally across each frame, but when auto settings are applied, is TVAI constantly monitoring and adjusting values frame by frame or does it also apply a single global setting to the entire video based on algorythmic averaging?
Auto and Relative to Auto are constantly estimating new values and applying them. Manual does apply the same settings to the whole video.
Hello,
The new EXR/DPX decoding system is currently set up for linear color inputs. In our next update we will be adding a toggle to choose between conversion and non-conversion for imports.
Could you share the MediaInfo output for one of your DPX files so that we can ensure this colorspace is added in a future update?
Thanks for the quick reply. That’s great to hear.
It’s an RGB 4:4:4 raw signal. If the color space is anything else than just RGB then i think it’s the origianal Sony S-gamut. Gamma curve is the original Sony S-log.
Here’s the DPX file info:
General
CompleteName : E:\DPX\Untitled 14\Untitled 14.00000000.dpx
CompleteName_Last : E:\DPX\Untitled 14\Untitled 14.00001566.dpx
Format/String : DPX
FileSize/String : 12.1 GiB
Duration/String : 1mn 5s
OverallBitRate/String : 1596 Mbps
FrameRate/String : 23.976 fps
Video
Format/String : DPX
Format_Compression : Raw
Format_Settings : Big / Filled A
Duration/String : 1mn 5s
BitRate/String : 1596 Mbps
Width/String : 1920 pixel
Height/String : 1080 pixel
DisplayAspectRatio/String : 16:9
FrameRate/String : 23.976 fps
ColorSpace : RGB
BitDepth/String : 10 bit
ScanType/String : Progressive
Compression_Mode/String : Lossless
Bits-(PixelFrame) :* 32.111
StreamSize/String : 12.1 GiB (100%)
Let me know if you need any additional info.
Perfect, thank you!
In this case we will most likely bypass the new system for downsampling to 16-bit TIFF and process the RGB data directly. Internally, Video AI uses RGB48 for internal processing so we will be able to retain the full data range.
In the meantime, I’d recommend testing DPX → R210 QuickTime. This format is lossless 10-bit RGB and will not be run through the new OpenImageIO-based importer.
Solid update! Initial tests looks amazing with new Proteus.
Please bring back quad comparison!
Is a .webm extension potentially compatible with AV1, or is that only a VB8 and VB9 thing?
Great info. Thank you.
You mean converting my DPX files to R210 QuickTime before opening in Video AI?
I’ve been upscaling my 1080p 10 bit 4:4:4 DPX files to 4K H265 Main10 since I learned that Main10 supports 10 bit. After checking with Mediainfo now, the files are indeed 10 bit, but I just noticed that the color sampling is only at 4:2:0.
Are there any plans for Video AI to support a 4:4:4 compressed output format like Main10 or Prores 4:4:4? I went through some effort to get a 4:4:4 recorder for my camera, so it’s a bit of a bummer. I know I can upscale to 4K DPX but those file sizes are way to much for me to handle at he moment. Thanks again for the info. Good to know internal processing is done with RGB48.
Is there a way to adjust the strength of Proteus 4? Like with Proteus 2 you could reduce “Revert Compression” and it wouldn’t mess up faces in full body shots so much. With v4 though that doesn’t seem to work anymore.
Also, what’s the difference between “Recover Detail” and “Improve Detail”?
May I ask you one question? When should I adjust the recovery detail option? It is set to 20% by default
I’ve gotta say, my heart goes out to the guys who have to look at that long list of bug reports.
Out of curiosity, has text seen any improvements with this new IrisS or Proteus model from 4.0.9? I’m no longer eligible to update my software and I’m hesitant to do so until that’s addressed.
