I’m curious what StaxRip will do with this! 
For the SL Sharp/Fast models, I noticed, that at first, the model processes the frames, but then it zeroes the GPU usage just for a couple of seconds, then proceeds to actually encode the video (cause the preview showed up at the ‘end’ of that small blue progress bar only).
You are suggesting this will tamp down some of thehigh detail?
I am experiencing the exact same issue. Only seems to occur on Starlight Sharp for me.
logsForSupport.zip (3.1 MB)
I have the same problem with Starlight Sharp.
1 Like
Try converting your file from MKV to either MOV or MP4 and then running through the app and see if that works. The team is looking into MKV container issues.
2 Likes
The Team should also check the last frame thing again.. there is still something strange.
1 Like
There should still be a log entry for the app launching and connecting to the server for accessing the models. Can you check and send those to the support team at help@topazlabs.com?
This is in the works.
3 Likes
Thanks Steve, let me get this information over to the devs for review. The CLI workflow as removed from the app when we updated to Topaz Video and while some users have reported they were still able to run their processes in earlier releases with some minor changes, there were more changes being made that would lead to other issues when trying to run it.
Re install error warning “neuroserver component install did not finish: neuroserver.tar.xz. First issue: SHA-512 mismatch after streaming downoad. Video AI will use built-in backend.” Deleted neuroserver-install-warning.txt, and ran a starlight precise process on a short video with reasonable results. Still not sure about the “built in backend” vs. neuroserver thing - though checking the log after I do see references to “… Info FF Process Output: 3 sys.path: [‘C:\Program Files\Topaz Labs LLC\Topaz Video\neuroserver\Lib\site-packages’, ‘C:\Program Files\Topaz Labs LLC\Topaz Video\neuroserver\DLLs’, ‘C:\Program Files\Topaz Labs LLC\Topaz Video\neuroserver\Lib’, ‘C:\Program Files\Topaz Labs LLC\Topaz Video\neuroserver’, ‘C:\Program Files\Topaz Labs LLC\Topaz Video\neuroserver’]” and I’m not seeing anything in the log that looks like an error message near instances of the text “neuroserver” so I guess it’s working properly?
Hello, I have often noticed this release problem since several versions before. Why with starlight models when you want to customize the video output folder on another diHello, I have often noticed this release problem since several versions before. Why with starlight models when you want to customize the video output folder to another hard drive for example, well it does not save it in the hard drive that I chose but export it in the same video folder? Can you solve this problem with starlight models please thank you
If the Starlight process is running and completing without error then you should be good and Neuroserver is working.
Can you send logs for reference on what is going on and the file paths you are trying to use? That way I can have the devs look into this and see what is going on. You can email them to help@topazlabs.com.
I think I included GUI crashes in there too, which is why I tried CLI.
1 Like
Thanks. Can you explain the error reference to “Video AI will use built-in backend” in the original error message? Is this basically a limp mode for models including Starlight that will attempt to run the models (slowly) without neural server acceleration?
Has this addressed the memory leak issue that was introduced with the last update ?
This is Hyperion 2, alone? That’s a really nice output given the source condition. I like how it handled the garbled copyright text, delicately cleaning it, instead of guessing. Is it stable throughout scenes with different angles, or does brightness change randomly on scene cuts?
1 Like
