Okay, the original file is actually quite compressed, not only grainy and low-rez…
SLP using @hi8upscale’s 5-export blend (1440x720p that will be upscaled further to 4K later)
Okay, the original file is actually quite compressed, not only grainy and low-rez…
SLP using @hi8upscale’s 5-export blend (1440x720p that will be upscaled further to 4K later)
4 Likes
Yes, I noticed that. My SLM works but I get instant crashes with Starlight HQ, while Fast2, SLP, and SLS works, but not very stable. Some others have also reported that at least one model no longer works, so it’s not consistent. It’s strange how sometimes one thing stops working, then some reported there is another model that does not working anymore.
Additionally, I’ve noticed that with V1.4 in general, the program has more trouble with input formats (colorspace issue?), and it tends to crash more easily, eats your memory…freeze…
I have two computers—on one of them I reverted to the previous version because I don’t want to deal with a gamble every time when I start a encode, so for all none SLP projects it is better using the previously version.
Well, for me, all the four available Starlight models worked perfectly fine in 1.3.0.
With 1.4.0, none of the five Starlight models work, and I have already tried a clean re-installation. It errors out immediately at the “Loading model” stage. I suspect it has something to do with that Neuroserver thing introduced in this version.
3 Likes
Did you try with different containers?
TV 1.4.0 is said to be quite picky there.
I have an NVIDIA GPU (4090) and still don’t see Starlight Precise 2.5. Do you need something higher than the Legacy Plan?
I’ve tried a few different ones, like webm, h264 mp4, ProRes 422 HQ mov. The processing keeps failing at the “Loading model” stage with the Starlight models. Guess will have to wait for a patch.
3 Likes
what is this 5 export blend?
I kept getting the error using SLP as others have mentioned. In my case I found out that it was because my source videos were in the MKV container. That was the problem for me. I imported an mp4 and that worked.
I hope this is of use to anyone who encountered the same issue as me. It just doesn’t seem to like the MKV container format.
I do hope that the Topaz team are aware of the file container issue and are working on a fix, because apart from that issue I’m genuinely impressed by the results with SLP. Credit where credit’s due SLP is a significant move forward, well done.
3 Likes
in all videos that I enlarge 1x only once using SLP, there is a brightening in the lower part of the video; if I enlarge it 2x, it disappears. ???
x1x
3 Likes
By 1x you mean really 1x or minimum?
Most of my videos are too low-res for 1x to be available..
the resulotion from the video is 1080p , minimum .
…so is Starlight Precise 2.5 simply listed as “Starlight Precise” in the UI? Because that’s all I’m seeing.
Yes it is “AI model = Starlight”
Then under “model variant = Starlight Precise”
Same as you i am on “Legacy Plan”
Works just fine… minus… well… all the things mentioned by people above.
Reported on: https://community.topazlabs.com/t/topaz-sl-precise-rtx-5090-sm80-ampere-bf16/102591
I discovered while running the Precise model is that on my 5090 it’s using an Ampere series SM80 and BF16. There could be an option for FP8 and even FP4 for the 5090, which would greatly speed things up. I couldn’t see it for the other models because they were disabled after I reinstalled.
6 Likes
Check out the original post: Topaz Video 1.4.0 - #258 by hi8upscale
1 Like
Why i can’t run precise on 8gb nvidia card? Since Topaz introduced Neurostream, it should be supported… Plz Topaz Team make it happen!
try also avi uncompressed
Trying to run Starlight mini on my m3 max. Why i’m getting “Local export requires pro or legacy access”
I am OG Topaz user since begining. :c
Admins, can you fix me pls?
Just payed 30 dollars, reneved montly license and get this sadly thing. I want use Starlight mini. Do i should write to support?
2 Likes
SLP on my machine crashed my computer. When I set 60%, 25% and 10% memory used, the computer did not crash, I received a message “out of memory”. However, in each of these scenarios, the memory usage was identical.
