Video AI does not add itself to the path when opening the command prompt, so if you navigate away from the Video AI folder, then FFmpeg will be the one already in the system %PATH% if present (or not found).
I’ll check out the other system variables to see if I can include them in a Windows Terminal profile. Hopefully they are fixed values.
Latest version for Mac. Every time I quit the app, it crashes and gives me a report.
This is not good.
Another “trick”: If you have, say, a 640 x 480 source that has smaller bars on the sides and/or top/bottom, do the following:
Step1: Import it into Virtualdub2
Step2: Select “filters” and add “resize”
Step3: Upscale it to 648 x 486 or 656 x 492 or 664 x 498, depending on how big the bars are that you need to cut away
Step4: Then select filter “null transform” and click on button “cropping”, now crop the video so that the bars are gone and the size fits 640 x 480 and export it with lossless codec.
This is nothing revolutionary very simple, but in my case I always cut away first and then racked my brains as to how much I had to cut away so that the numbers are even and aspect ratio fits when scaling up.
But if you scale up a little bit first and then cut away, you don’t have to do any of the math, everything you have to do is note the scale up numbers.
This does not work internally with TVAI, one of the reasons why I don’t care about the cropping function in TVAI because I don’t use it.
What have you done !!!
I’ve just spent 15 minutes rendering a video at 8ft (I am sure I got better performance than that the last time I rendered 720 to 1080) and you have rendered the audio from the beginning but the video starts about 1 minute in and then I have a minute of blank at the end.
Waste of my time, do I try again and waste another 15 minutes, simple upscale, Iris Auto, 720 to 1080, nothing else, save as HEVC, I do this all the time and now you’ve gone and broke it, thanks.
With those recurring audio problems in TVAI it might be best to first strip the audio from the file and process only the video, after that remux the result with the audio track again - MKVToolnix is a good (and free) tool for that.
@tony.topazlabs
Are there plans for an Aion TensorRT model? Thank you.
I also muxe audio from original source to it and then it doesen’t matter what TVAI does with audio tracks. This does not work when you just upscale a segment of the video, then you have an original audio track that doesn’t match it
Hello, I updated the software along with FFMPEG. I did a clean remove. Restart and clean temp files. I install, then restart and clean temp files. I mention that I did this, because I’ve tried 4X now, and the software crashes the ENTIRE PC every single time I use it now.
All I’m testing is a 540 x 720p file to HD, 30 FPS, 16s long, MP4 file. All I am doing, is uprezzing to Full HD.
My rig: 32-core AMD Threadripper at 4.3Ghz, 256 GB of DDR 5 RAM, and an NVIDIA 3090 RTX with 24 GB of VRAM, with everything on SSDs, running WIN 11 Pro. There should be no reason why the software is failing with the media. It’s not something it can’t handle.
I try other other files I have used successfully before, and no more than 5-10s after it begins, it not only crashes the software, it crashes the whole PC. It’s not the machines or the media, it’s the software.
Please fix. Thanks.
which version of TVAI, which model was chosen which export format, or does its crash on everything? Have you also tried out V5.3.6? Graphic driver, Mainboard Bios updated?
Can you share with us the original clip?
Maybe it’s the decoding problem, which is not something new to TPAI.
My first guess is power supply.
has gone a bit out of fashion but there should still be these burning tests for example “PassMark”, which maximize CPU and graphics card load to determine a system is stable
Sorry I should make myself clearer. The rendered video was the correct length, the audio (it was set to copy anyway) was correct from start to finish, however the video started at the beginning, encoding from approx 1 minute into the footage.
ie. There was video from the beginning of the render, but that video at the beginning started from approx 1 minute into the original footage, so I had video from the start but not from the start of the original video and resulted in the video ending earlier than the audio so there was 1 minute of black video at the end.
Well, in that case just also use an external editor to cut the video into the desired pieces beforehand.
This is new. 
I think the RTX Blackwell delays are connected to Blackwell data center GPU delays.
Because Blackwell data center is much more important to Nvidia.
Blackwell data center faced overheating and design flaws until november 2024, which did cause an delay of three month.
(same source as tpx)
Despite the mostly negative press and consumer reaction, gamers (and scalpers) were camping out at stores for days ahead of the RTX 5090 and 5080 launch last week. There were also chaotic scenes in Japan as people queued to win a chance to buy one of the cards.
The severe stock shortages of the Blackwell cards have led many people to label this a paper launch. Numerous Micro Center outlets were allocated only handfuls of GPUs, especially RTX 5090s. In some cases, just four or five of the flagships were available to purchase.
This is just sad. They learned nothing from the covid 4090 release debacle. It seems the former situation is repeating exactly.
My previous post was taken down. I downloaded the latest update, and now Topaz Video AI not only crashes EVERY TIME, it takes down the entire PC with it. There are no logs to recover because it crashes everything instantly. I was able to run the October 2024 release without issues, but since the release for 6.02, not thing works.
I’m running Win 11 Pro, on an AMD 32-core Threadripper, with an NVIDIA RTX 3090 with 24 GB of RAM, with the entire PC on NVMe. There’s no reason the machine should be choking. The latest versions of the NVIDIA App/Drivers, FFMPEG, and Win 11 Pro have also been installed.
I’ve been wanting to try all the latest improvements, but it’s been a quarter now, and I still can’t use it, at all.
I have a 1,200 watt PSU. I can generate 4K video with Stable Diffusion and FLUX on this machine, no problem. There’s plenty of airflow. The system worked fine with 5.3, no reason why the software should draw so much more power that it blows out the PC.
It’s a simple, small, vertical clip, in MP4 format, that was 720p. All I wanted to do was uprez to HD. I have done TONS of these with the prior release, no problem.
Agreed. That should be fine, unless it’s a poor quality one.
