Vidaio
SN85AI-powered video processing that can enhance, analyse, and transform video at scale
Decentralized AI video processing. Miners compete to upscale low-resolution video and compress high-quality video while preserving visual fidelity. Validators measure output quality using industry-standard metrics. Real users upload real videos and get processed results back.
// Better video, cheaper. By competition.
Vidaio is a subnet where miners process video using AI models for two core tasks: upscaling (making low-resolution video look high-resolution) and compression (making video files smaller without visible quality loss). Validators test miners with controlled benchmarks and route real user uploads through the network.
The simple version: You have a video that looks grainy. Upload it to Vidaio, and 100 miners compete to make it look sharp. Or you have a massive video file. Upload it, and miners compete to shrink it without you noticing the difference. The best processor wins.
Centralized equivalent: Think Topaz Video AI, Adobe's Enhance, or cloud transcoding services (AWS MediaConvert, Mux), but run through decentralized competition rather than a single company's servers.
How it works:
- Miners deploy AI models for video upscaling and compression. They can optimize open-source models or develop proprietary ones. Recent changes restrict emissions to the top 50 miners in each category, concentrating rewards on quality.
- Validators benchmark miners using controlled datasets. For upscaling: downscale a high-res video, send the low-res version to miners, compare their upscaled output against the original. For compression: provide high-quality video, measure the quality-to-size ratio of miner outputs. Quality is measured using VMAF (video multi-method assessment fusion) and PieAPP (perceptual image-error assessment).
- The problem it solves: Video processing is compute-intensive and expensive. Cloud transcoding costs scale linearly with volume. Centralized services charge premium prices and don't let creators own their processing pipeline.
- The opportunity: Video accounts for over 80% of internet traffic. Every platform, creator, and business that handles video needs processing: upscaling legacy content, compressing for mobile delivery, transcoding for different devices. The market is massive and growing.
- The Bittensor advantage: 100 active miners means genuine competition. The dual-metric evaluation (VMAF + PieAPP) prevents miners from gaming a single metric. Organic query support means real video processing, not just benchmarks.
- Traction signals: 722 commits across 20 contributors with steady 1-4 commits per week. Led by Arpan Tripathi. 100 active miners (one of the highest counts in our coverage). Organic video processing pipeline live. Roadmap includes adaptive bitrate streaming and live streaming.
Category: Media and Entertainment | Centralized Competitor: Topaz Video AI, AWS MediaConvert, Mux, Cloudflare Stream, HandBrake
Vidaio targets one of the most universal compute workloads: video processing. Unlike many Bittensor subnets that exist primarily as benchmarks, Vidaio has an organic processing pipeline where real users upload real videos and receive processed results. This is the bridge from "interesting experiment" to "useful service."
Mechanism:
The dual-task design (upscaling + compression) is smart. These are complementary workloads: upscaling is computationally intensive but produces larger files, compression reduces file size but risks quality loss. Miners who can do both well are genuinely useful for a complete video pipeline.
Recent development has focused on quality concentration. The March 12 commit restricts emissions to the top 50 miners in each category, and the March 16 commit weighs emission distribution toward higher-ranked miners. This rewards excellence over participation, a sign of a maturing incentive design.
The codebase is substantial: 722 commits across 20 contributors. Arpan Tripathi drives recent development with steady weekly commits focused on mining quality and emissions tuning. 15 GitHub stars and 21 forks indicate community engagement.
Market metrics show a healthy mid-tier subnet. At 42,306 TAO market cap with 2,295 holders, Vidaio has solid distribution. Gini of 0.597 is one of the lowest in our coverage (well-distributed ownership). Root proportion of 0.198 confirms organic demand.
The 7-day decline of 14% and net outflow of -430 TAO suggest short-term profit-taking. The 30-day (+6%) and 90-day (+8%) returns are positive, showing a generally healthy trend.
- Short-term outflows: -430 TAO net 7-day flow and -14% weekly decline suggest some holders are rotating out.
- Competitive landscape: Video processing is commoditized. AWS MediaConvert, Cloudflare Stream, and other cloud services have massive distribution advantages.
- Anonymous team: No named team members on TAO.app. Arpan Tripathi is the visible GitHub contributor but organizational structure is unclear.
- Quality ceiling: The best centralized video AI (Topaz) has years of model training on curated datasets. Decentralized miners may struggle to match proprietary model quality.
