Skip to content
Grill · 20 Nov · Round 2 of 5

omnihuman-15.com landing page review

72
Cooked

Higher than 50 of the 63 other launches on the public record, level with 1 (79%)

“The page proves the tech works but fails to tell the buyer why they need it today.” 

What we read, 20 Nov at 19:19: the page, its pricing page, at laptop and phone width.

The numbers

This page, in numbers.

Its four scores against the public record, and its public grills over time.

Its four axes

omnihuman-15.comthe median launch on the public record
81
median 81
64
median 64
100
median 77
39
median 39
Audience392 of 4 tests pass

Creators, educators and marketers appear only deep in the page; the top never says who it is for or what problem it ends.

  • half:A1Named
  • passed:A2Their words
  • missed:A3Pain before product
  • passed:A4The payer

A3, missed: no mention of pain points before the product.

Choose an axis to read it.

Where 72 sits

50 of 63other launches score lower

0s10s20s30s40s50s60s70s80s90smedian 66

The public record, 22 Nov: each launch counted once, by its latest public grill.

Its public grills

2 publicsince 28 Oct. This one, 72, is the best.

050100477228 Oct20 Novv0.14

Private grills are left out: only the owner sees them.

In their own words

What people looking for this kind of product write, quoted as they wrote it.

  • “It could be interesting if there was an actual face to read them out too.”

    adrianwaj · Hacker News · 10 Nov 2026 · open

Against this page

Its headline

OmniHuman 2.0 AI Video Generator

Buyers ask for "digital avatars", "a face to read them out" and a way to "add VO", while the headline gives a model name and "AI Video Generator".

Found by our critic in public posts; each quote was checked again at its link.

First screen

What a stranger sees first.

1 problem

Before scrolling, on a laptop and on a phone.

predicted, not measured
The page's first screen on a phone, predicted first looks, brighter means more
Phone 390 × 844 · header 38 px
PartScreenTodayFixed
  • The button “Create a video”5.5%2.6%6.1%
  • Headline16.2%31.6%25.0%
  • Subheadline15.4%11.4%8.4%
  • Menu bar4.5%6.2%5.7%

The button “Create a video” gets 2.6% of the first looks on 5.5% of the screen, though it is the one thing a visitor must use; eyes go to the headline. With the fixes it gets 6.1%, about 2 times more.

After five seconds a stranger would say: “Some AI tool that turns a picture and a sound file into a video.” They should be able to say: “Upload one photo and your audio, get a talking or singing video up to 5 minutes.”

Predicted, not measured: a model trained on 1,721 screens people looked at for 3 seconds with eye trackers. On 27 pages it had not seen it ranked the spots people looked at above other spots 83% of the time; “people look at the middle” scores 59%. It can over-weight loud things. The fixes are drawn by us on a copy of your page in our browser; your page is not touched. UEyes (Jiang et al., CHI 2023, CC BY 4.0) · TranSalNet · 20 Nov

Your message

Your message, as a buyer reads it.

4 fixes · 2 to keep

Four questions, in the order a buyer asks them. Each fix quotes your page and ends in something you can do today.

Fatal · they leave before they understandCostly · they understand and still leaveDrag · it works, and could work better

1. Do I get what it is?

Clarity 814 of 5 tests pass.
  1. DragClarity · C2 partialminutesup to +6.6
    “Turn a single image and audio track into a talking, singing, or animated character video”

    The sub-headline describes the process (what it does) rather than the outcome (what the buyer achieves).

    Proposalyou decide
    Replace the sub-headline with the primary benefit, e.g., ‘Create professional presenter videos without a camera or studio’.

2. Is it for me?

Audience 392 of 4 tests pass.
  1. Fix first · 1CostlyAudience · A3 failan afternoonup to +6.8
    “OmniHuman 2.0 AI Video Generator”

    The page starts with the solution; visitors don't see their problem mirrored and may not realize they need the tool.

    Proposalyou decide
    Add a section before the generator listing the pains of traditional filming, such as studio costs or equipment needs.
  2. Fix first · 2CostlyAudience · A1 partialminutesup to +3.1
    “first viewport”

    The target audience is only named in the 'What Can You Create' section, forcing users to scroll to find themselves.

    Proposalyou decide
    Add a ‘For [target audience]’ label above the headline, e.g., ‘For content creators and educators’.

What people type

Yours“ai video generator”, the phrase in its headline, is the most searched of the five.

“ai video generator” on Google, week by week

Shaded, the last 3 months: −13% on Google · Bing agrees: +10%

Nov 2025Nov 2026
  • “ai video generator”81✓ headline
  • “make a photo talk”under 1no
  • “talking photo ai”1no
  • “image to talking video”under 1no
  • “ai avatar from audio”under 1no

Google Trends: the five phrases in one search, worldwide, the past 12 months, so they share one scale (100 = the busiest week of any of them; shares, not counts). Bing Webmaster Tools: the exact phrase, worldwide, an average week. Read 20 Nov.

Against this page

Its title

OmniHuman 2.0: Make Photos Talk, Sing & Move Online

Its headline

OmniHuman 2.0 AI Video Generator

The headline uses “ai video generator”. The title uses none of them.

3. Do I want it?

Offer 644 of 5 tests pass.
  1. Fix first · 3DragOffer · O4 failminutesup to +10.4
    “pricing page: Select the credit pack that fits your workflow”

    There is no urgency or incentive to buy now, leading visitors to bookmark the page and never return.

    Proposalyou decide
    Add a limited-time launch discount or a bonus credit bundle to the pricing page.

4. Do I believe it?

Proof 100Nothing to fix here.
  • Keep: The 'Result notes' in the examples section provide specific, verifiable benchmarks for quality.
  • Keep: The pricing page is transparent with a clear refund policy and no recurring subscriptions.
Why you

Why you, not them.

not answered

Products a buyer may meet while looking for what you sell, read the same day, and the free option most of them pick.

Starts atProof from outsideSays what it is
HeyGenCreator $29/mo ($24/mo billed yearly)“4.8 / 5 from 1,000+ reviews” (G2), “Trusted by 85% of the Fortune 100”partly
Magic HourCreator €21/mo (€13.2/mo billed yearly); credit packs from $10“Trusted by 5.2M+ creators and 2K+ teams”, Trustpilot 4.4/5yes

Found by searching your category, read on their own pages the day of this grill. Others a buyer may meet: D-ID, Hedra, Synthesia, ElevenLabs.

The option most buyers pick: Filming the video yourself with a phone or webcam, or not making the video at all.

Most creators and teachers already own a camera, so a paid credit pack has to beat a free recording they can make today.

  1. CostlyWhy younot in the score

    Say why this beats filming it yourself

    The only contrast with filming, “without filming every segment”, sits deep in the use cases, and nothing says why to pick this over another avatar tool.

    A visitor who can film for free, or who already knows other avatar tools, needs a reason to choose this one before scrolling.

    Proposalyou decide
    One image and one audio track become a video up to 5 minutes long, with lip sync, emotion and body motion, and no filming.
Three visitors

Where each one hesitates.

simulated · not real people

Our critic reads your page as three different strangers would. A simulation from the page alone: not real people, not your analytics.

The teacher

An educator who wants video lessons from scripts already written

Hesitates at the headlineA1

“OmniHuman 2.0 AI Video Generator”

This reads like a model name for AI hobbyists. I can’t tell if it is meant for a teacher like me.

Would stay if: A line at the top that names educators and creators, before the model name (A1).

The HeyGen user

A creator who has already paid for another talking-avatar tool

Hesitates at the subheadlineC2

“Turn a single image and audio track into a talking, singing, or animated character video”

My current tool does this already. What do I get here that I don’t get there?

Would stay if: A subheadline that says the gain: body motion and videos up to 5 minutes, not only lip sync (C2).

The newcomer

A small business owner who has never made an AI video

Hesitates at the label above the headlineA3

“Image + audio to video”

I can see what goes in, but not why I would want a video made from a photo instead of just filming.

Would stay if: A short part before the generator that names the problem it removes: filming, a camera, retakes (A3).

The pattern: All three meet the tool before they meet a reason: the page never says who it is for or what problem it ends (A1, A3), and the subheadline describes how it works, not what they gain (C2).

Measured, not scored

What passed, and 7 to tidy.

We read the page as AI and search do, and measure speed, links, accessibility, privacy and the technical basics. Beside the score, never in it.

  1. DragContrastnot in the score
    Make the faint text darkerShow the fix
    Too faint to read in 16 places (accessibility check).
    Proposalyou decide
    Darken that text (or lighten it on a dark page) until it reaches 4.5 : 1 against its background.
  2. DragAccessibilitynot in the score
    Fix 5 more accessibility problemsShow the fix
    Elements must only use permitted ARIA attributes · Aside should not be contained in another landmark · Main landmark should not be contained in another landmark · Document should not have more than one main landmark
    Proposalyou decide
    Fix them in the order listed: each names the element it found.
  3. DragImagesnot in the score
    Shrink an oversized imageShow the fix
    The file is 1536 × 1024 px, shown at 56 × 37: 27× more than needed.
    Proposalyou decide
    Export it at 112 px wide (twice its shown size).
  4. DragPhone speednot in the score
    Unfreeze the page on phonesShow the fix
    Frozen for 0.4 s after it appears; aim for 0.2 s.

    Taps in that time do nothing.

    Proposalyou decide
    Load less JavaScript before the first screen is usable.
  5. DragPrivacynot in the score
    Set cookies only after consentShow the fix
    __client, __client_uat, __client_uat_H3I86gP3, __cf_bm, _cfuvid

    Worth checking, not a verdict: in the EU most cookies need consent first.

    Proposalyou decide
    Set them only after the visitor agrees.
  6. DragTechnicalnot in the score
    Fix 8 technical settingsShow the fix
    HSTS: missing · CSP: missing · frame: missing · nosniff: missing · referrer: missing · permissions: missing · security.txt: 404 · DMARC: missing
    Proposalyou decide
    Each is a setting on your host or DNS, named in the line above.
✓ Everything else passed3,019 words read without JavaScript · a title of 55 characters · a description · a 1448 × 1086 preview image · robots.txt lets every AI crawler in · laptop 1.3 s · 537 KB · no layout shift · 13 links work · terms, privacy all there · 3 of 11 technical checksShow all results
Largest paint, desktop
1.3 s
target ≤ 2.5 s
Layout shift
0
target ≤ 0.1
Links on the page
0 broken
13 checked
Accessibility
6 issues
elements must only use permitted aria attributes; elements must meet minimum color contrast ratio thresholds; aside should not be contained in another landmark
Frozen on a phone
0.4 s
after it appears; taps do nothing. Target ≤ 0.2 s
Weight on a phone
537 KB
52 requests

The look

Font families3: -apple-system, ui-sans-serif, ui-monospace
Text sizes24, 0 to 54 px
Filled button styles8
Images far from their shown size4

Search and sharing

Words without JavaScript3019
Title55 characters
Descriptionyes
H1 in the HTML1
Preview image1448 × 1086
Site name“OmniHuman 1.5”
Structured data1 block
AI crawlers in robots.txtall allowed

Privacy and the law · worth checking, not a verdict

Cookies before any click__client, __client_uat, __client_uat_H3I86gP3, __cf_bm, _cfuvid, UM_distinctid
Third parties on loadc.cnzz.com, v1.cnzz.com, z12.cnzz.com, z3.cnzz.com
Imprintno link found
Privacy/subsidiary/privacy-policy · 200
Terms/subsidiary/terms-of-service · 200

Technical

HTTPS and redirect301 → https://www.omnihuman-15.com/
certificatevalid to 2027-01-17 (58 days), Google Trust Services
HSTSmissing
CSPmissing
framemissing
nosniffmissing
referrermissing
permissionsmissing
security.txt404
SPFv=spf1 include:_spf.mx.cloudflare.net ~all
DMARCmissing
Launch day

Ready for Show HN and Product Hunt?

2 to fix first

What those crowds check first, and drafts written from your page's own words. Edit them before you post; we never post for you.

Can I try it without an account?No, the generator button says “Sign In to Generate”.fix first
Who made it?The model is credited to ByteDance, but no person or team behind this site is named.fix first
Does the page say what it is?Yes, the headline says “AI Video Generator” and the label says “Image + audio to video”.ready
Is the price shown?Yes, the linked pricing page starts at $10 for 25 credits.ready
Drafts for Show HN, Product Hunt and X“Show HN: OmniHuman 2.0 – Turn one image and an audio track into a talking video”Show the drafts
Show HN titleShow HN: OmniHuman 2.0 – Turn one image and an audio track into a talking video
Product Hunt taglineOne image + audio becomes a talking or singing video
First commentI built this site so anyone can run OmniHuman 2.0, ByteDance's audio-driven avatar model, in the browser. You upload one image and an audio track of up to 5 minutes, add an optional prompt for actions or camera, and get a video with lip sync, emotion and body motion. [your reason] [your question]
X postOmniHuman 2.0 is now on the web. Upload one image and an audio track. Get a talking, singing or animated character video, up to 5 minutes, with lip sync, emotion and body motion. Optional prompts guide gestures and camera. Credits are returned if a generation fails.
The score

Your score in detail.

How 72 adds upevery axis and its weight, the level, every test, every grill of this pageShow the sum
Clarity35%
81
+28.4
Audience18%
39
+7.0
Offer29%
64
+18.6
Proof18%
100
+18.0
The sum 71.9, so
72
8 points to Well done

Level 3 of 4 Cooked

all open: 100
now 72
0406080
4
left to fix
+28
points in reach
100
could reach · Well done
Badges
  • ✓First grill
  • ✓Cooked — reached 60
  • Well done — no round left on this launch
  • ✓Before and after — finish round 2
Every point from a test

The four axes

Every point comes from a test. Open a row, or read the rubric.

Clarity35% of the score
81
C1
OmniHuman 2.0 AI Video Generator
pass
C2
Turn a single image and audio track into a talking, singing, or animated character video
partial
C3
OmniHuman 2.0 AI Video Generator
pass
C4
Turn a single image and audio track into a talking, singing, or animated character video
pass
C5
Create a video ↗
pass
4 passed · 1 partial · 0 failed → 81. Weighted at 35%, this contributes 28.4 of the overall 72.
Audience18% of the score
39
A1
social media, education, storytelling, marketing
partial
A2
Create Social Media Videos from Audio
pass
A3
no mention of pain points before the product
fail
A4
addressed to creators and educators
pass
2 passed · 1 partial · 1 failed → 39. Weighted at 18%, this contributes 7.0 of the overall 72.
Offer29% of the score
64
O1
Pricing link in navigation
pass
O2
Starter, Creator, Pro Studio tiers with credit counts
pass
O3
pricing page: $10 for 25 credits
pass
O4
no launch price or time limit on pricing page
fail
O5
pricing page: 7-Day Refund Window
pass
4 passed · 0 partial · 1 failed → 64. Weighted at 29%, this contributes 18.6 of the overall 72.
Proof18% of the score
100
P1
OmniHuman 2.0 Video Examples
pass
P2
Result notes: The bottle keeps a consistent overall shape
pass
P3
Upload an Image / Upload Audio interface screenshots
pass
P4
audio-driven AI avatar video generation model from ByteDance
pass
4 passed · 0 partial · 0 failed → 100. Weighted at 18%, this contributes 18.0 of the overall 72.
28.4 + 7.0 + 18.6 + 18.0 = 71.9 → 72.
Full page read including pricing page.
2 grills

Every grill of this page

  1. 47Undercooked28 Oct 2026
  2. 72Cooked20 Nov 2026+25You are here

Your launch, read the same way.

Free. No card.

Ready to launch? The Launch Kit: your agent submits you to up to 85 free directories · $49 →