Topaz Labs released Starlight Precise 2.5, its latest video upscaling model, as a Partner Node in ComfyUI. The update delivers sharper 720p-to-4K results with fewer artifacts than its predecessor, available at the same per-frame pricing with no workflow changes required.
What Happened
ComfyUI announced the integration on March 27, making Topaz's professional-grade video upscaling available directly within local AI workflows. Starlight Precise 2.5 is a drop-in replacement for the previous SLP-2 model. Users select it from the existing Topaz Video Enhance node dropdown without modifying their pipelines.
The update focuses on three improvements. Fine details are better preserved through the upscaling process, particularly noticeable in 720p-to-4K workflows. Artifact generation is reduced, with cleaner results and less hallucination compared to SLP-2. The model also generates realistic detail when filling in information during upscaling, making output look native rather than stretched.
Template workflows are available on the Topaz Labs GitHub repository. The integration works on both Comfy Cloud and local ComfyUI installations.
Why It Matters for Creators
This puts professional video upscaling inside the same environment where creators already run local AI generation workflows. Instead of exporting video from ComfyUI, running it through a separate upscaling app, and importing the result, the entire pipeline now stays in one place. For creators who generate AI video at lower resolutions to save compute, then upscale for final delivery, this removes a significant friction point. Combined with NVIDIA's recent ComfyUI optimizations, local AI video workflows are becoming genuinely production-ready.
What to Do Next
Update ComfyUI to the latest version and find the Topaz Video Enhance node in the Node Library. Select "Starlight Precise 2.5" from the model dropdown. Pricing matches the previous version, so existing Topaz subscribers pay nothing extra for the upgrade. Start with a 720p-to-1080p test to compare quality against your current upscaling pipeline before committing to full 4K renders.
This story was covered by Creative AI News.
Subscribe for free to get the weekly digest every Tuesday.
