The presenter stays visually believable overall, but the output has a fused/warped right hand at 1s and 14.9s, which breaks the realism enough to keep it from being cleanly convincing.
What was measured
Avatar realism
Believable presenter; accurate lip-sync; how uncanny or AI-obvious the presenter looks.
decisive for this rankingtransformation
If the presenter looks fake or lip-sync is off, the ad fails at the core UGC avatar job. (3 of 3 judges)
What was given, what came back
Test input: FutureSmart AI – SaaS · mixed · group: ugc-style-video-ads
Input — what we sent
Input, verbatim
Product URL: https://aidemos.com/ Exact script: Hook: Finding the right AI tools shouldn't take hours. Script: I've been using FutureSmart AI to discover and compare AI tools in one place. It helps me find the right tool faster with real use cases, rankings, and detailed comparisons. If you regularly use AI tools for work, it's definitely worth checking out. Output requirements: - Vertical (9:16) - 20–30 seconds - UGC-style AI presenter - Natural AI voice - Captions enabled
A short-form UGC ad prompt for a SaaS product discovery/comparison platform, designed to test whether the tool understands a software product, writes a believable testimonial-style script, and produces a realistic AI presenter video with captions.
Why this input is hard
- · SaaS product understanding
- · Script quality
- · Avatar realism
- · Voice quality
- · Caption quality
- · Overall ad readiness
Output — unretouched
Also checked on this input — same tool, 2 other criteria
Caption quality✗ FailedCaptioning is absent in the FutureSmart run: the researcher checked 8 timestamps from 0s through 14.9s and found zero burned-in text anywhere in the clip.Platform formats✗ FailedThe output only fills about 55% of the 9:16 canvas, leaving large dead white bars top and bottom, so the clip is not a native full-bleed vertical export.
Provenance
- Observation
- 2399684a-6762-4a39-a524-676aeddfa25c
- Evidence run
- 9fdbb84b-4aa0-484d-8f89-5c46608019a5
- Study
- Generate UGC-Style Video Ads With AI Avatars
- Research task
- 86baxegpg
- Tested at
- not recorded
- Source
- first-party
- Evidence state
- verified
- Proof shown
- input + output shown
- Cost / latency
- not captured
- Repeat run
- not captured
- Tester
- not captured
The last three rows are honest blanks, not placeholders — our capture has no field for them yet.
Query this
get_evidence({
tool: "heygen",
scenario: "ugc-style-video-ads"
})MCP · mcp.aidemos.com/api/mcp
Free with attribution.
Same input, same check — 7 other tools
measured on Avatar realism
AKOOL◐ MixedProduces a usable talking-head presenter, but the avatar still reads as somewhat AI-generated and less natural in parts, so realism is only moderate.DreamFace◐ MixedThe FutureSmart avatar is visually believable, with natural skin and hair texture, but a fused-finger hand glitch at about 0:02 breaks the realism.Invideo AI✓ WorkedThe presenter reads as a believable human talking head, with natural blinking and clean hand rendering reported across the render and no flicker or warping noted.JoggAI✓ WorkedGenerates a convincing photorealistic talking-head avatar, with the report noting natural skin, hair, teeth, expressive gesturing, and no visible deformities or robotic feel.Synthesia✓ WorkedThe FutureSmart output used a stable talking-head presenter that stayed centered through the short video, and the report says lip-sync was consistent across the run.Topview AI◐ MixedThe tool can keep a vertical presenter visible throughout the clip, but the presenter still reads as AI-generated: the report says realism was lower than higher-end tools and expression/gesture range was minimal.Vidnoz AI◐ MixedProduces a presentable talking-head presenter, but the avatar still reads as slightly AI-generated rather than fully human; the report calls the realism only moderate.
This evidence is published in
Real inputs and real outputs, no retouching · every cell queryable via API & MCP · aidemos.com