Well, considering the negative reviews on 2.4.0, I’ll stick with 2.3.0.
PS, I think I’ve solved my recurring BSOD problems by using VEAI. I did a clean install of my latest NVIDIA driver and temporarily changed a setting in the control panel while waiting for the NVIDIA team to fix the bugs.
I had to switch it to “auto select” mode. Before, he was on NVIDIA Geforce RTX 2080 because before I had no BSOD problem.
And for 2 days in a row now, I haven’t had a BSOD. And I do at least 2 video enhancements with VEAI + final rendering with Adobe Media encoder in one day.
One issue I’m having with the Chronos V2 model. I’m converting a few videos to 59.94 fps, but I find that even though I see them all defaulting to 59.94 fps, if I don’t reselect 59.94, when I start processing they will revert to 29.97 fps. Shouldn’t the frame rate stay the same once I start processing? Thanks!
I just checked, and the Proteus V2 and Chronos V2 model files from that location are the exact same versions as they were when they were first uploaded, from before the 2.4.0 beta was available. (In other words, the file checksums are the same… they are the exact same model files.)
Usually, model files change when the final version of VEAI is published, due to additional training.
So either the model download URL’s have changed, or Proteus/Chronos V2 have not changed since they were first published/uploaded.
Like others, I am staying with 2.3.0 for now, due to the removal of JSON editing and inability to use all available versions of model files.
This release seems to be a dud full of development time for things no customer asked for and actually seems to be a regression. I hope this direction is quickly changed.
I really don’t care so much about JSONs per se, the only two things I used (and probably most people I suspect) were the keepColor and interlacedFrames fields (allows progressive mode on Dione TV/DV). If those were options or tick boxes on an “advanced” or “expert” option panel that would be fine, really.
I checked the 2.4.0 install, and it does appear that the Proteus / Chronos V2 model files have not changed… they are the same ones that were initially uploaded, before the 2.4.0 beta cycle began.
(Not saying that’s a bad thing, just pointing this out.)
for the devs, here is the kind of crash i get temporary (not all the time, i could run VEAI with no issue during 4hours, and now, did a new batch rendering and it crashed on the first video of the batch.
The error is “cvidapps is not responding”
it crashed at the end of the processing of the first video (frame 134 on a 135 frames video).
the windows 10 system log tell this :
the application Topaz Video Enhance AI.exe version 0.0.0.0 stopped to work with Windows and has been closed.
ID process : 2e04
Application adress : C:\Program Files\Topaz Labs LLC\Topaz Video Enhance AI\Topaz Video Enhance AI.exe
Repport ID : 629a793c-f779-4e67-9476-913cd1151dca
Kind of block / crash / stop (not sure of the word in english) : Unknown
viktorz3008
(Z...z...z 😴 I'm sleeping. And I love Sumire 😍.)
71
I’m not sure why a lot of users are having crashes and errors. I have been upscaling a really long movie for 3 days straight without any issue. I really don’t understand. Did you guys uninstall the old versions using Control Panel before installing the new version?
I don’t get it either. I just finished 48 hours of processing running 2 instances with 6 videos batched in each. I was using Proteus V2. I was also using the computer for other work while this was happening. I don’t have a great machine, a Ryzen 9 3900x, B450 board, GTX 1660 super, 64GB ram and a 2tb NVMe. I’m also running Windows 11 Beta. I always do a clean Nvidia Studio driver install. I do not run any of the extra parts of the Nvidia Studio driver set, just the driver itself. I also make sure to completely uninstall Topaz before installing a new version. Everything seems rock solid as far as VEAI not running slow or crashing.
I find that after update to Video Enhance 2.4 models will not download. I have tried uninstall and deletion of both local and roaming data followed by new install but, issue still exist.
I have tried this with 3 different PCs that I have and issue is same all machines. These all have latest Windows 10 64 bit release and all have current NVida drivers.
Video Enhance 2.3 still seems to work. Advise you do not update to version 2.4 until Topazlabs address this issue.
1 Like
viktorz3008
(Z...z...z 😴 I'm sleeping. And I love Sumire 😍.)
78
Did not happen on my end. So this might not be a bug.
We had trouble with version 2 today, and then we found out there was a version 2.4. We downloaded it and are unable to get it out of the preview mode when we try to use the models. Also the models DID NOT SAVE after downloading from file>manage ai models. We have a ticket with customer support and they’re supposed to respond to us. Please let us know if any of you guys have found a fix for this issue!
I can confirm that I am also experiencing an error with the ETA for videos when processing - it appears to estimate double the actual time required.
I’m currently upscaling a series of approximately hour-long 16:9 25fps video from around 400p to 2160p using Proteus V2. The ETA at the start of processing is given at around 14 hours, with a processing speed varying from 0.26 sec/frame to 0.30sec/frame - but the job actually completes at closer to 7 hours.
The total frame count of the video I’m currently processing is 86,929. If we multiply that by an average of 0.29 sec/frame, we get 25,209.41 seconds - 420 minutes - approximately 7 hours. This is roughly consistent with the time it is taking me to process these videos.
So it would appear that the sec/frame estimate is correct, but the ETA calculation is wrong.
Topaz Video Enhance 2.4 destroyed video and audio!
I was using 100% size denoise/deblock on my UHD master SDR/HDR (rec 709, DCI and Rec 2020 video files). I could not believe but the app screw up audio and video at once:
The source audio was uncompressed PCM 24bit 48kHz → now it’s AAC LC 96kbit/s !!! it sounds awful. Who gave you the permission to change the source sound file format?
The video is denoised BUT the app erased/stripped all the Color Metadata!!!
*No color primaries, *
*no transfer functions, *
*no YCbCr Matrix, * no Code Points.
The video looks bad, completely wrong because of lack of metadata.
What is going on?!?!
I did 4 days denoise session… All goes to the trash. No way to fix it.
No way to use Video Enhance in any professional workflow.
