Can someone explain the difference to me? Until now, I thought that the price list was the same for astraapp and Topaz Cloud, but now I see that astraapp costs 10 credits for 10 seconds of processing up to 4K (starlight model). Topaz Cloud, on the other hand, costs 24 tokens for 1 minute of processing (4K / Starlight).
Different variants of Starlight models cost different amounts. Can you share what model variant you are using in Astra?
The Precise v1 and v2, and Fast v2 will be more expensive, but the Mini and Fast v1 should be close to the same cost as cloud processing from the app UI. There can be a slight difference depending on how the cloud server is seeing the source video frame rate and frame count.
1 Like
Oh, I see.
I’m familiar with the prices for now, but if it’s 25 tokens per minute for 4K starlight, I’ll come back here when the prices drop to the minimum level, because right now the prices are insane.
See here: https://www.topazlabs.com/cloud-render?srsltid=AfmBOopK3jVUL1ia6zERjrnsP_orxD0kSFN3Xh-6vWhW9Np0fCOnDg3l —it says 24 tokens per minute.
The cloud credit cost you are seeing there are older estimates, that we are working on updating.
The 24/minute was the original pricing rate for use with the GAN models (Proteus, Iris, etc. )
The original Starlight research model, is the one referenced on the Starlight tab at 150 credits/minute.
I will reach out to the web team to have this updated and looked into.
1 Like
In that case, starlight is completely useless for ordinary hobbyists at the moment.
Is it still limited to 5 minutes and h264 output is CFR=23?
Dude… For some models, there’s a maximum of 18 seconds :D. Doesn’t Topaz Cloud have these limitations?
That’s even worse than I thought 
With certain models the limit is 9000 frames - Starlight Precise 1, Fast 2 and Proteus.
Precise 2, and the creative models the limit is 450 frames.
More information here on the docs page.
thanks for the reply, this gives:
25fps → 6 minutes
30fps → 5 minutes
