The fact is that TVAI works very poorly with noise. It distorts the noise profile. Makes some stains and watercolor instead of grain. This is evident when the frame is enlarged. Therefore, the first in video processing should be a professional noise reduction Neat Video 5. He does a miracle!
Further, noise-free video becomes more contrasty, sharp, more color-saturated. And at the end of processing, we can allow TVAI to double the frames for a smoother picture, and make small cosmetic improvements.
And most importantly! No upscale old video. Make improvements only in the original resolution. The TV itself will stretch the picture to the desired size. As a last resort, make the upscale the last separate aisle, no improvements!
1 Like
It depends how to use noise reduction: i always use Neat 5 in Resolve but the best results are on HD or FHD videos. Using Prot V3 manual i get a punchy detail for old movies and Super Scale enhanced in Resolve does the “finishing” cause it doesn’t amplify
the vignette effect. This is the best method for me.
I asked this as well awhile ago and a member of these forums told me to use Process Lasso which allows you to suspend apps. This works WONDERFULLY with TVA and I use it regularly. Just install the apps, find ffmpeg in the list, right click it and go to more - suspend. Done. When you ready to start it back, same place and go to resume.
Thanks for following up with me, can you share the source file with me?
You can securely submit your files(s) to our Dropbox using the link below. Please be sure to send me a note to let me know you sent something.
What encoder are you using? Can you share an example?
Shutter Encoder
? Which encoder are you using when exporting with Topaz Video AI?
H265 Main 10
1 Like
If I’m exporting videos from DaVinci in the editing codec DNxHR444 10, TVAI doesn’t seem to pay attention to the color processing I did in DaVinci. Noise reduction, sharpness, details “sees”, and perceives color correction only partially.
But that’s all, maybe for the best! Because the final video DNxHR444 10 after TVAI (H265 Main10) is as similar in color as possible to the original video. It seems that the codec DNxHR444 10 suggests that TVAI ignore color correction errors.
But, if you export video from DaVinci to H265, then TVAI completely retains the color gamut. Even if it is very oversaturated, but not correctly painted in DaVinci.
That is, if you upload the same video encoded by H265 and DNxHR444 10 to TVAI, TVAI, in the first case, processes the “surface” layer that we see on the media player. And in the second case, in the case of the mounting codec, TVAI processes some kind of “inside” layer that we do not see on the player…
1 Like
odd, can you share an example with me? You can securely submit your files(s) to our Dropbox using the link below. Please be sure to send me a note to let me know you sent something.
Color shift is something we are working on resolving at this time 
1 Like
More than likely the video maxes out at the bitrate shown during export.
The problem he is all over the place with encoding.
He’s gone from SD footage to a prores codec.
Then after using Resolve he’s again now changed.
Basically going from 8 bit 4:2:0 to 10 bit 4:2:2 to 12 bit 4:4:4 then back finally at 10 bit 4:2:0.
He’s very new to video editing and is still learning the basics.
With that amount of re-encoding I’m not surprised it’s all over the place.
1 Like
You say it all… To many color space conversions will change the original colors. I’m sure it’s possible to output from Resolve using Prores instead of DNxHR444
I sent you the video files.
How?
I tried different options. Please give advice on how to encode the color space correctly so that there is the least loss, but the quality has improved…
Looks like this is for the color shift issue. Is that correct? If so, we are looking into the color shift issue which should be resolved soon.
The bitrate issue is occurring because the bitrate is a reference number. If you ask for more bitrate than the actual amount of information the file contains, the encoder cannot generate any new information. You will be left with the maximum bitrate possible for the file.
For example, if you ask for 60 and get 22, it means there is not enough information more than 22. No matter how big the bitrate is requested, if it is more than 22, the export will only go up to 22.
On the other hand, if you ask for a smaller bitrate, the encoder will try to compress the information (and will lose some details) to match the requested bitrate.
It is quiet impossible to give out the exact bitrate in one pass. So, if you requests 10, you might get 8, or 9, for example.
1 Like
