LIP-SYNC QUALITY
SyncNet LSE-C · best in the comparison
PUBLIC PRICE
$/sper output-second
GENERATION SPEED
×faster than the original Aurora
CONTINUOUS DURATION
sUp to one minute · 4.1× the H3 Max Lip Sync limit
One photo becomes a complete performance
REAL AD EXAMPLE · RETAIL SKINCARE UGC
Turn an approved ad frame into a polished speaking performance.
This example starts from a real vertical ad asset with a creator, product, retail environment, and campaign typography already in frame. Aurora 2.0 brings the presenter to life from the supplied voice while keeping the composition ready for a performance ad.
Ad-native input.
Start with the creative your team already approved.
Natural delivery.
Mouth, expression, head motion, and gestures follow the voice.
Fast variation.
Swap the voice track to create new hooks, offers, and localized versions.
Best-in-class lip sync
9.14 LSE-C, the highest result in our side-by-side comparison.
Full-body expressiveness
Facial nuance, eye contact, breathing, head motion, hand gestures, and body language move together.
Exact source voice
Your original audio stays intact, with 99.97% waveform correlation and 0 ms measured offset in long-video validation.
One photo, no setup session
Start from the person, character, styling, and composition you already approved.
Built for any language
The performance follows the supplied voice track, opening the same workflow to global campaigns.
Made for creator formats
UGC ads, podcasts, explainers, localization, virtual spokespeople, and character content from the same simple input.
See the difference on the same inputs
These selected examples compare Aurora 2.0 and Aurora 1.0 on the same first frame, the same voice track and the same written instruction. Use the controls to inspect lip timing, identity, and motion directly.
SAME FIRST FRAME · SAME VOICE TRACK · SAME INSTRUCTION
Quality, speed, and economics
Best overall quality at the lowest public price in the comparison.
On our avatar benchmark, Aurora 2.0 reaches a 90.5 Avatar Performance Score—the strongest overall result in the comparison—at a public price of 7¢ per output-second, the lowest listed public price in the comparison. That is 12.5% below H3 Max Lip Sync and 30% below HeyGen Avatar 4 on fal.
Avatar Performance Score = 50% lip sync + 25% perceptual quality + 25% identity preservation, normalized to a 0–100 scale. H3 Max Lip Sync was generated through fal’s public 768P endpoint and evaluated locally with the same measurement stack.
Aurora 2.0 pairs the strongest overall quality score in the benchmark with a 31.6-second warm median generation time. The latency view uses the same “higher and further left is better” convention as fal’s H3 Max launch chart.
Latency is request-to-result time recorded during the benchmark; provider queue and cold-start behavior may vary. Aurora 2.0 uses warm serving measurements; all quality scores use the same fixed scoring protocol.
A full-generation upgrade to Aurora
Aurora 2.0 upgrades the complete experience: the highest overall quality in the comparison, approximately 2.0× faster generation, and continuous videos up to 60 seconds.
A FULL-GENERATION UPGRADE
The gains you can see and feel
Higher overall quality, twice the speed, and a complete one-minute performance
BEST OVERALL QUALITY
90.5
Avatar Performance Score
Highest score in the comparison
FASTER GENERATION
2.0×
faster than Aurora
63.4s → 31.6s median
LONG-FORM OUTPUT
60s
continuous video
One generation, one complete message
Quality and latency use completed benchmark records; 60 seconds is the public product ceiling.
THE PRODUCT ADVANTAGE
Faster generation. Lower public price.
Speed and price, both compared with Aurora 1.0
FASTER THAN AURORA 1.0
Generation time
Aurora 1.0
63.4s
Aurora 2.0
31.6s
2.0× faster
LOWER PUBLIC PRICE
Public price per output-second
Aurora 1.0
14.0¢
Aurora 2.0
7.0¢
50% lower
Both panels compare Aurora 2.0 with Aurora 1.0: warm median generation time, and public price per output-second at 720p.
Aurora 2.0’s warm median generation time is 31.6 seconds, down from 63.4 seconds for Aurora 1.0. Public pricing is $0.07 per output-second at 720p—half of Aurora 1.0’s $0.14/s.
Long enough to tell the whole story
THE ONE-MINUTE ADVANTAGE
60 seconds. One continuous performance.
Long enough for a complete ad, product story, lesson, pitch, or localized message
H3 MAX LIP SYNC
14.8s max
AURORA 2.0
60s
4.1×
LONGER
in one continuous video
H3 Max: 5–15-second workflow; supplied audio is clipped beyond 14.8 seconds.
Aurora 2.0: up to one minute from one photo and one complete voice track.
LONG-FORM BENCHMARK · 29.2 SECONDS
A complete message from one portrait
Five overlapping generation windows preserve the supplied 28.8-second speech track without trimming it to a short clip.
DURATION VALIDATION · 59.7 SECONDS
One continuous minute
A one-minute validation render with the original waveform preserved at 99.96% correlation. Long-form quality can vary; this run contains one detected camera cut near 54 seconds.
AURORA 2.0 VS H3 MAX LIP SYNC
60 seconds versus 14.8 seconds.
H3 Max Lip Sync’s public workflow accepts 5–15-second videos and clips supplied audio beyond 14.8 seconds. Aurora 2.0 supports continuous videos up to one minute—long enough for the complete message, not just the opening line.
Aurora 2.0 is our highest-quality avatar model, approximately 2.0× faster than Aurora, with the lowest listed public price in the comparison—12.5% below H3 Max Lip Sync at 768p—and supports a 4.1× longer continuous clip.
From a single assetto a complete campaign
Ads and UGC
Turn an approved creator image and script read into a complete one-take performance. Generate hooks, offers, and localized variants without reshooting.
Podcasts and explainers
Let a thought finish. Longer clips make product walkthroughs, lessons, commentary, and narrative content practical in one generation.
Virtual people and characters
Animate real, stylized, or synthetic identities from a single frame while preserving the intended look and original voice.
One photo. One minute. Your exact voice.
Our most capable avatar model yet: best-in-class lip sync, stronger overall quality, approximately 2× faster than Aurora, transparent public pricing, and long enough to tell the whole story.
SyncNet LSE-C measures audio-to-mouth alignment; Q-Align measures perceptual video quality; ID-SIM measures identity preservation; motion smoothness measures frame-to-frame temporal stability. Higher is better for every quality score shown.
All side-by-side quality values use the same fixed evaluation suite and scoring pipeline. The visual gallery is selected from that suite and is not a random sample.
H3 Max Lip Sync was generated through fal’s public 768p endpoint with transcription guidance enabled. Its measured aggregate results are 7.95 LSE-C, 4.80 Q-Align, 0.72 ID-SIM, and 0.51 motion smoothness.
The 9.14 headline is Aurora 2.0’s strongest completed lip-sync result. Q-Align, ID-SIM, and motion smoothness use its balanced quality evaluation.
The speed chart uses recorded request-to-result latency. Aurora 2.0 uses warm serving measurements; provider queue and cold-start behavior can vary.
Current public rates checked for this update: Aurora 2.0 7¢/s; H3 Max Lip Sync 8¢/s at 768p; Aurora 14¢/s at 720p; VEED Fabric 15¢/s at 720p; Veo 3.1 40¢/s; HeyGen Avatar 4 on fal 10¢/s; Kling Avatar Pro 11.5¢/s; OmniHuman 1.5 16¢/s.
Aurora 2.0 is shown at its public price of $0.07/s. Competitor rows use the current public prices for their benchmarked tiers.
The public product ceiling is 60 seconds. The 59.7-second example is a validation render and includes one detected camera cut near 54 seconds.
Sources:
original Aurora launch·
Aurora pricing·
H3 Max speed/quality chart·
H3 Max public limits and pricing·
HeyGen Avatar 4 pricing on fal·
VEED Fabric pricing·
Veo pricing·
Kling Avatar pricing·
OmniHuman pricingAurora 2.0 and H3 Max quality values are from completed Creatify benchmark records. Updated September 29, 2026.
