Yes, Starlight is quite picky with the resolutions and sometimes seems to choose those itself according to its mood. Some resolutions even just plain fail.
P.S.: This erratic behavior of SL mini I reported above seems to also affect the cloud versions.
Me, and others also, did get videos that looked exactly the same as what we’ve uploaded.
As a ‘Founder’, do I have to wait until my studio transition is complete to access the starlight models locally?? Even after the starlight models are downloaded, following the system check, the local export options are greyed out and only cloud export is enabled. Even for the regular starlight.
Can confirm this happened with me too. Since the local render wasn’t working for some reason, I had to use the cloud export a couple of times out of necessity, and both the times it returned the exact same video but even more compressed! Tried it with both starlight & sharp. Ended up costing me a few credits.
Is the OG Starlight Mini on Studio or is it just Sharp? Because when I click Starlight it doesn’t give an option to the original or v2, and in the file name it just says Starlight Mini, not sharpen etc, so I don’t actually know which version of Starlight is being used.
HI, trying Topaz Video Studio for the first time, outputting to ProRes Std, it’s been at it 20 minutes and still ‘initialising viewer’, no temp file has appeared in my output folder and according to Resource Monitor, neither Topaz nor runner.exe are writing anything to disk anywhere. The progress bar is moving slightly but the estimated time remaining is also going up.
Any ideas what’s going on
UPDATE
Just had error
Current container (file extension) is not be supported by the codec. Try changing the file extension. - Some parameter(s) may not be supported by the codec. Check FFmpeg command.
Container file is .mov, video is ProRes Std, encoding audio to aac 320, always worked before.
UPDATE 2
Tried again using h265 in case ProRes was an issue, after 45 minutes it stopped with “Recoverable Error” and an option to retry, but think this program is dead in the water.
Yet another memorable release from Topaz, thanks guys, so glad I didn’t uninstall version 7
I would think that it would be possible to run the perpetual version on one machine for less-demanding tasks while running the Studio version on another, wouldn’t it?
That is indeed possible and is officially allowed, as confirmed by Mr Wixom. So we are allowed to have 3 devices rendering concurrently: 2 x TVAI and 1 x Topaz Video.
Thank you for this information. I’ve been spending a LOT of time reading community posts, yet somehow missed this. (Maybe it was posted when I was in the ER last night… )
Question for Video AI Users Working with DV Footage on Macs
Hi everyone,
I made a longer post on the Studio thread, but I wanted to specifically ask Video AI users about working with DV footage on Macs. (My setup: 2019 Mac Pro 7,1 and an M4 Studio Max.)
I’m about to join the Founders plan with Photo and Gigapixel, but I stopped updating Video AI at v2.6.4 (June 2022) due to the lack of Nvidia GPU support on Mac. A “wait and see” stance. However, the ‘new’ Video 1.0 will not work on my older, but powerful Intel Mac Pro.
Here’s my situation:
I’ve digitized over 100 DV tapes from the early 2000s music scene.
The footage needs de-noising, stabilization, sharpening, and level adjustments.
To move from 2.6.4 to 8.4.4, I’d need to pay an additional $149 for the Video AI update.
I’ve seen users mention rolling back to versions like 6 or 7.1.4 instead of using the latest builds.
I’ve been in a ‘holding pattern’ since digitizing the DV tapes Summer 2024.
My questions:
With an 8.4.4 license, would I still have access to older installers (e.g., 6 or 7.x) if those versions work better for my Intel Mac machines?
Did 2.6.4 perform well improving DV footage with Video AI on Intel Macs? Or is there a “Golden version” that is preferred by Topaz users?
On Topaz official, it states my upgrade would be to 3.0.0, which I believe is incorrect (should be 8.4.4, right?)
For context, the Northwest Ohio rock documentary is a passion project (archiving local music footage), with no outside funding. As an adjunct instructor on a fixed budget, $149 feels like a lot—especially for what seems like an ‘end-of-the-line’ 8.4.4 version for Intel Macs.
It would be great if Founders with Intel Macs could get a reduced-cost option for the last Intel-compatible license. Perhaps on Black Friday 2025? Here’s hoping.
Thanks in advance for any advice or experiences you can share!
All opinions much appreciated! Kindest regards to all, Walt
only partly with Proteus. Iris supersedes that sometimes considerably (especially the LQ V1 variant).
Not your third point but still: with such sources I do believe that most of the time Starlight mini (or now just Starlight) is the best choice albeit terribly slow. But still it should run on your M4 Max. So this point would speak for the Studio subscription.
some people like to give the Starlight output a little extra sharpness with an added Proteus run - you could easily do this on your Intel Macs while the M4 is sweating with the next SL mini upscale.
You can try to turn that off in settings, just for now to see if the local shows up. There are two tabs in the queue window also, that displays Device and Cloud. I’m not sure if you have to have Device selected or not
After the upgrade and re-logging in I can no longer use FFMPEG direct at the command line. I see output from TopazAuthManager indicating the status is “OWNED” (LOL, Topaz and pwned me). But startup take several minuts, FPS is low and there is a watermark on the output video.
I primarily use this tool via the command line. This is kinda of a deal breaker.
EDIT: issue resolved. It looks like there was some pathing issues with the previous install and I was picking up the old models folder. Nuked it all and re-installed.
I had ChatGPT analyze the user log files. BLUF: The code is messed up in the Topaz Video app. The HD output path being passed to ffmpeg is wired to an MP4 muxer internally even though the UI requires MOV as the only choice for ProRes output. The code is trying to shove ProRes into an MP4 container, so it fails. That’s why SD works and HD doesn’t—this is a regression/bug in 1.0’s export pipeline, not the output settings. It also explains why h.264 and h.265 export successfully…because MP4 is the only possible container allowed by the bug.
Here’s what ChatGPT had to say:
I unpacked your logsForSupport.zip and the latest log shows Topaz ignoring your MOV choice on the HD run and trying to mux ProRes into an MP4 anyway:
prores in stream #0, codec not currently supported in container
[out#0/mp4] Could not write header (incorrect codec parameters?): Invalid argument
Immediately after, Topaz looks for a preview .mov that doesn’t exist and cleans up the temp files.
Earlier in the same session the SD export shows a proper MOV target:
and then dies with the MP4/ProRes header error. In short: the HD path is wired to an MP4 muxer internally even when you pick MOV. That’s why SD works and HD doesn’t—this is a regression/bug in 1.0’s export pipeline, not your settings.
Is it normal that the Starlight Sharp model runs very slowly? I noticed that when the GPU temperature is not very high (around 49°C–50°C), the progress is almost zero. Only when the GPU is running at higher temperatures (around 79°C–83°C) does the progress move forward a little. Why can’t the GPU stay at high temperature so that the progress keeps going?