Crystal Video Upscaler

High-precision video upscaling on fal.ai with a simple scale-factor control and exact post-run billing correction.

Model Overview

Crystal Video Upscaler takes an input video and increases its resolution while trying to preserve the original motion, structure, and fine detail. This Nodespell node keeps the surface intentionally small: one input video plus the scale factor that drives the upscale.

Best At

• Upscaling existing clips without turning the node into a full video-editing interface.
• Restoring detail on footage that needs a cleaner higher-resolution export.
• Workflows where the input video should stay structurally faithful while gaining resolution.

Limitations / Not Good At

• Cost scales with output megapixels, duration, and FPS band, so large or high-FPS clips can get expensive fast.
• The selected scale factor must still keep the upscaled output under the model's 5K limit.
• This node is for upscaling, not prompt-driven restyling or generative video changes.

Ideal Use Cases

• Preparing an existing clip for a higher-resolution delivery target.
• Cleaning up source footage before downstream editing or compositing.
• Simple graph workflows that just need a video-in, video-out upscaler.

Input & Output Format

• Input: required video_url plus optional scale_factor.
• Output: upscaled video returned on response.

Performance Notes

• Fal prices this model per megapixel per second, with an additional multiplier based on FPS bands.
• Nodespell starts from an input-video estimate, then corrects the final credits from fal's billed amount after the run completes.