System objective

• Video reconstruction and upscaling from any low-resolution source (240p, 480p, 720p, analogue SD, etc.).

• Output in native 4K resolution with recovery of detail, colour and fluidity.

• Applicable to historical films, television archives, surveillance footage, period cinema, old programmes and home videos.

   

Technology used

• Deep neural networks (Deep Learning) specialised in super-resolution.

• GAN (Generative Adversarial Networks) models for recovering missing details.

• Automatic algorithms for deinterlacing, denoising, noise reduction and colour correction.

• Pipeline optimised for GPUs and automated batch processes.

Main Features

• Intelligent upscaling up to 4K UHD (3840×2160)
• Reconstruction of lost frames and frame rate increase (e.g. from 24 to 60 fps)
• Reduction of compression artefacts (e.g. MPEG2, DivX, H.264)
• Colour restoration with scene and lighting recognition
• Support for legacy formats (VHS, Betacam, MiniDV, DVD, SDI)
  
  

Benefits for the Customer

• Enhancement of archive content
• Monetisation of old catalogues on new platforms (streaming, VOD, Blu-ray)
• Elimination of the need for manual re-editing or restoration
• Automation and scalability for large volumes

Use Cases

• Public and private television stations: restoration of historical programmes
• Film archives: recovery of low-quality digitised films
• Post-production companies: automatic optimisation of legacy footage
• Public and cultural institutions: high-resolution digital preservation

Proven Results

• Up to 400% improvement in perceived visual quality
• Recovery of details on degraded or compressed frames
• Output compatible with HDR and 4K broadcast standards