Interactive Explorer

Virtual Channels &
Channel Factories

Create payment channels between non-adjacent nodes without new on-chain transactions. Explore Perun, Elmo, and Horcrux — the protocols that make it possible.

Perun 2019 Elmo 2022 Horcrux NDSS25 Channel Factories 2018 Zero on-chain txs
Section 1
Why Virtual Channels?
Basic payment channels require on-chain transactions to set up. Virtual channels eliminate this requirement for intermediate connections.
❌ Option A: Direct Channel
Alice and Carol open a direct channel.

Cost: 1 on-chain transaction (fees + block space)
Benefit: Always available, no intermediary
Problem: Expensive if they rarely pay each other
⚠️ Option B: Route Through Bob
Route every Alice→Carol payment through Bob.

Cost: No new on-chain tx
Problem: Bob must be online for every payment
Problem: Collateral locked in Bob's channels during HTLC
✅ Option C: Virtual Channel
Create an off-chain channel between Alice and Carol using Bob's channels as collateral substrate.

Cost: 0 new on-chain transactions
Benefit: After setup, Alice↔Carol pay freely without Bob!
Benefit: 1000 payments without touching Bob
Cost & Availability Comparison
Approach On-chain txs Bob online required? Payments between Alice↔Carol Setup cost
Direct Channel 1 tx Never Unlimited High
Routing (HTLC) 0 new Every payment Each needs Bob Low
Virtual Channel 0 new Setup only Unlimited Very Low
Key Insight: Collateral vs Participation
Routing (HTLC):
Bob participates in every payment actively. His channels relay funds. If he goes offline, Alice cannot pay Carol. Each payment requires Bob's signature.
Virtual Channel:
Bob contributes collateral at setup, then steps back. His capital is "delegated" to the Alice↔Carol virtual channel. After setup, Bob is not needed until close.
Section 2
Channel Factory — The Foundation
Burchert et al. 2018 proposed Channel Factories: a multi-party off-chain contract that lets N participants create N(N−1)/2 channels with a single on-chain transaction.
Click tier to explore:
TIER 0 — BLOCKCHAIN (L1) 4-of-4 Multisig 900,000 sats TIER 1 — FACTORY LAYER (off-chain allocation) Alice 200,000 sats Bob 300,000 sats Carol 150,000 sats Dave 250,000 sats TIER 2 — CHANNEL LAYER (Lightning channels) Alice 200k Bob 300k Carol 150k Dave 250k 100k 80k Alice↔Carol virtual 50k TIER 3 — VIRTUAL CHANNEL LAYER Alice ↔ Dave — Virtual Channel (30k) Intermediaries: Bob + Carol — No new on-chain tx needed
Tier 0 — Blockchain (L1): A single on-chain 4-of-4 multisig funding transaction deposits all participants' funds. This is the only interaction with the blockchain needed. All subsequent channel operations happen off-chain.
1
On-chain transaction
6
Possible channels (4 members)
45
Channels from 1 tx (10 members)
N(N-1)/2
Formula: channels per factory
Section 3
Virtual Channel Protocol — Step by Step
Creating a virtual channel between Alice and Carol with Bob as intermediary. Navigate each step or use auto-play.
Step 1 / 6
Loading...
Section 4
Protocol Comparison
Perun, Elmo, and Horcrux each take different approaches to virtual channel construction. Toggle to highlight each protocol's architecture and trade-offs.
Feature Perun Elmo Horcrux
Nesting depth 1 level Recursive (any) 2 levels
Collateral per hop O(n) hops O(1) O(1)
Bitcoin compatible Partial Full Full
Grace exit No Yes Yes
PTLC-based No No Yes
Arbitrary state (smart contracts) Yes Payments only Payments only
On-chain txs to close 2 O(depth) O(1)
Intermediary offline tolerance Stuck Grace exit Split secret

Perun — Generalized State Channels (Dziembowski et al. 2019)

Perun introduces generalized state channels: not just payments, but arbitrary smart contract logic can execute off-chain. The "virtual channel funding protocol" (VCF) bootstraps a virtual channel using existing on-chain deposit channels.

Alice Ethereum Bob (Hub) Intermediary Carol Ethereum ch A↔B ch B↔C Virtual channel: Alice ↔ Carol (arbitrary app state) On-chain L1 deposit channels ⬆ Off-chain virtual channel (smart contract logic runs here)

Elmo — Recursive Virtual Channels (Aumayr et al. 2022)

Elmo achieves recursive virtual channels: you can nest virtual channels inside virtual channels to arbitrary depth, with O(1) on-chain complexity regardless of nesting depth. The key innovation is the "grace exit" mechanism allowing correct closure even when intermediaries go offline.

Recursive Layer 2 Virtual Layer 1 Virtual Alice Base channel Bob Intermediary Carol Base channel O(1) on-chain · grace exit · Bitcoin-compatible

Horcrux — PTLC-based Bitcoin Virtual Channels (Aumayr et al. NDSS 2025)

Horcrux uses Point Time-Lock Contracts (PTLCs) for Bitcoin-native virtual channels. The "split secret" design distributes trust among multiple intermediaries: a single intermediary going offline does not block the virtual channel.

Alice PTLC Hub1 split-secret s₁ Hub2 split-secret s₂ Carol PTLC Virtual channel (PTLC) — single hub failure resilient
Section 5
Collateral Lockup Problem
When Bob intermediates a virtual channel, he must lock collateral for its entire lifetime. Calculate the opportunity cost and minimum fee Bob should charge.
Virtual channel amount 100,000 sats
Number of hops (intermediaries) 2
Duration 7 days
Intermediary opportunity cost (APY) 5%
INCENTIVE STRUCTURE
Bob locks 100,000 sats per hop for 7 days. During this time that capital cannot be deployed elsewhere. The minimum fee Alice should pay Bob is the opportunity cost of locking this capital.
PERUN vs ELMO COLLATERAL
Perun (O(n)): 200,000 sats total locked
Elmo (O(1)): 100,000 sats per hop (same as single hop)
200,000
Total collateral locked (sats)
96
Opportunity cost (sats)
96
Min fee Bob should charge (sats)
0.096%
Fee as % of channel amount
At 5% APY for 7 days on 100,000 sats: Bob forgoes ~96 sats per day × 7 = 672 sats opportunity cost per hop.
Section 6
Grace Exit Mechanism (Elmo)
What happens when Bob goes offline and can't cooperate on closing the virtual channel? Elmo's grace exit provides a safe fallback.
Block timeline (each cell = ~10 blocks):
Normal Cooperative Close
  1. Alice and Carol agree on final state
  2. Both sign the closing transaction
  3. Bob confirms and countersigns
  4. All channels update atomically off-chain
  5. No on-chain transaction needed (mutual close)
Grace Exit (Bob Offline)
  1. Alice broadcasts latest Alice-Bob state unilaterally
  2. Grace period starts (e.g. 144 blocks ~24h)
  3. During grace: Bob can respond with fresher state
  4. After grace period expires: Alice's state becomes final
  5. Carol's side: parallel grace exit process with Bob-Carol channel
Section 7
Real-World Status
Where are these protocols in terms of production deployment and standardization?

Channel Factories

Burchert et al. 2018

Theoretical proposal for multi-party off-chain channel creation. No production implementation exists. Draft BOLT discussion but not standardized.

Research only

Perun / go-perun

Dziembowski et al. 2019

Active implementation by PolyCrypt GmbH. go-perun library deployed on Ethereum mainnet. Supports generalized state channels and virtual channel subprotocol.

Production (Ethereum)

Elmo

Aumayr et al. 2022

Bitcoin-compatible recursive virtual channels. Research paper, no production implementation. Grace exit mechanism is a key theoretical contribution.

Research only

Horcrux

Aumayr et al. NDSS 2025

PTLC-based Bitcoin virtual channels with split-secret intermediary design. Research paper published NDSS 2025. No production implementation yet.

Research (2024-25)

LN Channel Factories BOLT

Draft proposal

Draft BOLT specification exists for Lightning Network channel factories. Not merged into official BOLTs. Requires significant protocol changes to LN spec.

Draft BOLT
2018
Channel Factories (Burchert, Decker, Wattenhofer) Proposed multi-party channel factories. Foundation for all subsequent virtual channel work.
Theory
2019
Perun: Virtual Payment Hubs (Dziembowski et al.) First UC-secure virtual channels. Ethereum-focused with generalized state support.
Active
2021
go-perun v0.1 released First production implementation of Perun virtual channels on Ethereum.
Deployed
2022
Elmo: Recursive Virtual Payment Channels (Aumayr et al.) O(1) on-chain complexity, grace exit, Bitcoin-compatible recursive nesting.
Theory
2024
Horcrux: Bitcoin Virtual Channels with PTLCs NDSS 2025. Split-secret design, intermediary fault tolerance, O(1) close.
NDSS25