Topaz Video AI 5.1 - 5.1.2

I do also, but I would very please to learn how dev transform SDR to HDR by including meta data in the output file.

There’s indeed some metadata, but we, community would be very pleased to listen how they do process this in order to know if we use it or not, without having to sample hours or days during the output result.

1 Like

A single 4090 saturate a 32 core / 64 threads CPU.

So what is the limit 128 ? 256 ? Or is it simply badly coded …

There’s absolutely no bandwidth on PCIE, so CPU is doing much more than what it should.

If CPU is used as a memory access, then VRAM should be used much more that it is the case.
And there’s other way to DMA if RAM is the true need.

Hello I had to roll back to version 504. On the last three updates the Nyx filter Enhancement there are no changes regardless of what I do with the settings. My rig, Ryzen 7 3800X, RTX 3060Ti, 32Gb Memory, 750 Watts PSU. Nvme storage OS and backups.

Bc the GPU does need a long time to process the tile.

And the cpu does cut the tiles.

And the main memory is a big cache.

I’m with you on that.
I’m only interested about the upscale models and the interpolations models.
There were no updates for any of them for a long time.

Thanks for trying to fix the open image sequence crash. Whilst I no longer see this crash, it immediately crashes upon me opening the video input settings “edit” dialog - and then clicking ok to close the dialog. I don’t even have to change the settings in the dialog - and it crashes immediately.

1 Like

Lots of problems in versions 5.1… and versions 4 don’t work well.
Εxception is only the export pause function which works well, the rest features in versions 5 causes have operational problems.

Personally, I’ve returned to version 3.5.4 and i will leave there, as topaz ai instead of improving, is unfortunately getting worse and worse…


Constant bitrate is broken in 5.1.2. Selected x265 4k 40mbps output, ended up with 16mbps.
We truly are the alpha testers…

Great find! Looking into this now.

Everything relative to output bitrate is messy

“High” means no compression at all for TVAI
“Medium” means … I even don’t understand
“Constant” means 16 Mbps

There are TRUE tiers known and used EVERYWHERE else, but TVAI chose a way as if we all are unable to choose.
We do need an option for dumb users, and another for high end users. The same as for models with “Estimate”

This is so easy so code, as easy to patch the x.y opening location of app …

We do not currently ship x265 in Video AI. When exporting H.265 outputs, Video AI uses NVENC, AMD AMF, Intel Quick Sync, or Apple VideoToolbox.

When setting a constant bitrate with these encoders, it is not always possible to use a strict bitrate mode. In most cases, we are able to set a target bitrate, but the encoder may still decide to use a lower bitrate depending on the complexity of the scene.


Here’s an example of a 50Mb/s output for NVENC:

"-c:v" "hevc_nvenc" "-profile:v" "main" "-pix_fmt" "yuv420p" "-b_ref_mode" "disabled" "-tag:v" "hvc1" "-g" "30" "-b:v" "50M" "-preset" "p6" "-an" "-map_metadata" "0" "-map_metadata:s:v" "0:s:v" "-movflags" "frag_keyframe+empty_moov+delay_moov+use_metadata_tags+write_colr" "-bf" "0"

Here’s some additional information about the rate control modes and how they function in NVENC.

Please add NVEnc H265 and H264 lossless compression.

As a nubi with the software it could use a bit to be more user friendly and intuitive. ie. easy to save, show where sound settings are if wish to enhance sound quality, picture improvement appears to be good but horribly slow. I have a super fast system and it is telling me 7 hours to do a 54 minute video that is already HD and needs some clean up.

Ah yes, quite right sorry.
4k, H.265, main10, constant bitrate output worked fine in 5.0.4 (and prior) versions though for both my AMD and NVIDIA cards. Now neither card adheres to the “constant” bitrate that I set, falling quite short.

Tested using the exact same 1080p input file and settings on versions 5.0.4 and 5.1.2.
5.0.4 adheres strictly to the constant bitrate, 5.1.2 does not.

Ok thanks, that does help determine where the change came from.

Could you send a zip of your app logs to this Dropbox link? You probably still have 5.0.4 logs stored there that could help step through the differences in encoding mode.

To gather logs, please select Help > Logging and make sure that there is a check mark next to File Logging. Next, recreate the issue and then return to the Help menu and select > Get Logs for Support and attach the “logsForSupport” file to your reply.

You can securely submit your file(s) to our Dropbox using the link below. Please be sure to send me a note to let me know you sent something.

Submit to Dropbox

1 Like

I hate to not be able to offer any comments past…version 2.6.4 is what I reverted back to. V3 had issues, 4 had many more, and now version 5…Export As does not work, the GPU is not used past 30%, and any type of preview is dirt slow…you guys need to truly fix your QA. Just because it compiles does not mean it works.


great… you made it difficult to use again and removed looping, and now we have a really stupid and annoying “processing” or “unprocessed” solid background text bubble blocking our view when trying to view footage we are processing or not. There shouldnt be anything on the video especially not things that obstruct the view.
The sidebar covers the work area and you cannot adjust or start the render when the sidebar is open.
I really am not sure how to express how dissatisfied i am with the current version.

Unfortunately it’s not AMD vs Intel, it’s the single-threaded nature of some FFMPEG filters, which is unavoidable.
TOPAZ team cannot invent it’s own video processing engine and de-facto is limited to FFMPEG as the only means of video processing on the market.

So if you want fast video processing, you should get the fastest single-threaded CPU with fast RAM and best Nvidia card you can acquire.
My home setup is Ryzen 5800x + RTX 4070, my work setup is Core i7-10700 and RTX 3070 8gb, and man 3070 is the bottleneck in my pipeline. In terms of CPU, both are equal in most scenarios, but AMD eats less power.

1 Like

In my experience, there’s no FPS conversion without ehnancing.
Usually I go with Aion + Proteus with Manual settings which I can tune after watching Previews.
Yes, Aion takes much more time than Chronos but yields almost perfect results.