DEV Community: MakeHumanAI

We turned our document-tools site into a faceless video studio — here is the stack

MakeHumanAI — Mon, 09 Nov 2026 16:39:30 +0000

MakeHumanAI started as a set of free document tools — photo resizers, PDF compressors, signature makers. This week it grew a second half: a faceless YouTube video creation service. You send a topic; we return a finished video — script, voice, clips, captions, thumbnail.

Here is how the production line actually works, because the stack is the interesting part.

The pipeline

Script. A large open model writes the script in a documentary structure — hook, one idea per scene, plain sentences a voice can carry. A 10-minute video becomes roughly 70 short scenes, each with its own narration line and a visual brief.

Voice. We run text-to-speech through HeyGen's Voice Turbo API while it is free, with a local Kokoro model as the permanent fallback. Every scene is generated as its own audio file, which makes retries and revisions cheap — fix one scene, not the whole track.

Clips, not slides. This was a hard rule from day one: no photo slideshows. Each scene gets real motion footage, matched to what the narration is saying at that second, pulled from free stock libraries through their APIs, then graded with one consistent cinematic look — same contrast, same warmth, a light vignette — so stock from five different cameras still feels like one film.

Captions and assembly. Captions are chunked to short phrases, timed from the real audio durations, and burned in. Assembly is ffmpeg end to end: per-scene segments, a concat filter join (stream-copy concat drifts timestamps — ask us how we know), a quiet generated music bed under the voice, and title and end cards.

The business model, briefly

The sample on the service page is free and short — about a minute — so anyone can judge quality before paying. Full videos run on subscriptions: $49/month for two videos, $399/year for twenty-four. Payment is crypto-only (USDT or BTC) with a small live calculator on the page that converts the plan price at the current BTC rate, and orders come through an on-site form — no accounts, no checkout platform taking a cut or holding payouts for a week.

What we learned building it

Verify every stock clip with your eyes. Search relevance is loose; our first sample had a sparkler, a fantasy underwater city and the Great Wall of China in a video about Rome. Tag filters plus a thumbnail review fixed it.
Generated audio formats vary per scene (sample rate, mono vs stereo). Normalize before you join anything.
A demo is worth more than any description. The service page converted from "trust us" to "watch this" the moment the sample went in.

The free tools are still the front door. The studio is what is behind it now.

Your Signature Doesn't Need an App: Making a Clean Digital Signature in the Browser

MakeHumanAI — Fri, 06 Nov 2026 04:39:54 +0000

Every form has moved online, but signatures never really did. You get three bad options:

Type your name in a fancy font. Fast, but anyone can tell it is a font. It does not look like your signature, because it isn't.
Sign on paper, photograph it, paste it in. Real, but the photo is grey, crooked, and carries the whole desk with it — paper texture, shadows, the line of the notebook.
Install an app. Now a simple signature costs you an account, a subscription screen, and your file sitting on someone's server.

We run a small free toolkit (MakeHumanAI) built around exactly these form problems, and the signature tools are the most used things on the site. So here is the workflow that actually works, the same one we tell our own users.

Step 1: Start from your real handwriting

A signature people trust is the one you already make with a pen. Draw it — on paper, or with your finger/mouse on a canvas.

If you draw it, draw it bigger than feels natural and sign the way you sign a cheque, not the careful version. The careful version always looks staged.

Our free signature generator gives you a canvas to draw on and download the result as a transparent PNG. No account, and the drawing never leaves your browser — the canvas is rendered locally, which matters more than people think for a thing as personal as a signature.

Step 2: Clean the background

If you started from a photo or a scan, the signature itself is fine — everything around it is the problem. Grey paper, shadows, and ruled lines make a pasted signature look pasted.

Run it through a signature background remover. What you want out the other side:

Pure transparent (or clean white) background
The ink strokes kept dark and complete — cheap removers eat the thin parts of letters
No halo around the strokes

Hold it up against a white document page. If you can see a box, it is not clean yet.

Step 3: Resize it to what the form actually asks for

This is the step everyone skips, and it is why uploads get rejected. Government and job portals are oddly specific: a signature image often has to be under 20 KB, sometimes in a fixed box like 140×60 pixels. Your clean PNG might be 300 KB.

Shrink the file, not the signature: resize your signature to 20 KB and check the number before you upload, not after the portal rejects you. Portals rarely tell you why a file failed — the size limit is the first thing to rule out.

Step 4 (optional): When you don't have a signature style yet

Some people genuinely do not have a settled signature — students applying for their first documents run into this all the time. An autograph maker lets you try your name in different handwriting styles, pick the one that feels like you, and then — this is the important part — practise that one on paper until it becomes your real signature. The digital version should follow your hand, not replace it.

What to avoid

Signature packs and font downloads that give everyone the same "handwritten" name. If two people's signatures are identical, neither one proves anything.
Uploading your signature to random converters. A signature is a credential. If a tool processes it on a server, you have handed a copy of it to a stranger. Prefer tools that work in the browser — check whether the page still works with your connection throttled or offline after it loads; if it does, your file stayed home.
Over-cleaning. A signature smoothed into perfect vector curves looks printed. Keep a little of the pen's mess. That mess is the authenticity.

The whole thing takes two minutes

Draw or photograph → clean the background → resize to the limit the form demands → upload. Four small tools, no installs, no accounts, and your signature never sits on a server.

That is the entire trick. The forms got digital years ago; your signature can too, without turning into a font.

Your PDF Converter Should Not Need Your Files

MakeHumanAI — Wed, 04 Nov 2026 02:35:56 +0000

Here is a question most people never ask when they use an online file converter: where does my file actually go?

With most big converter sites, the answer is: onto a stranger's server. You upload a bank statement, a signed contract, a passport scan. It is processed somewhere, stored for some amount of time (the privacy policies use words like "temporarily" and "up to 24 hours"), and you download the result and hope the deletion promise is true.

We built our converters at MakeHumanAI the other way round, and this post explains why — and why the trade-offs are worth it.

The whole conversion happens in your browser tab

Modern browsers are quietly powerful. Libraries like PDF.js, SheetJS and Mammoth can parse and rebuild documents entirely in JavaScript, on your own machine.

So when someone uses our PDF to Word converter, this is the entire journey of their file:

They pick the PDF from their device.
The browser reads it locally and rebuilds it as a document.
They download the result.

No upload. No server queue. No copy of the file sitting in a data centre in another country. The network tab stays silent because there is nothing to send. You can literally load the page, switch off your internet, and the conversion still works.

The same design runs through the rest of the set:

PDF to Excel — pulls tables out of statements and invoices into real spreadsheet cells, locally.
PDF to Image — renders pages to JPG or PNG on the device, which is why it is fast even on a phone.
The Word and Excel converters going the other way use the same rule: your file, your machine.

Why this matters more than "free" and "no sign-up"

"Free, no sign-up" is on every converter homepage on the internet. It costs nothing to say.

Privacy in this category is different because the files are not random. Nobody converts a PDF of a birthday card. They convert:

Bank statements, to total them up in Excel
Signed contracts, to change one clause
Salary slips, ID documents, visa paperwork

These are exactly the documents you least want copied onto an unknown server as the price of a ten-second conversion. Local processing removes that price entirely — there is nothing to leak, sell, or subpoena, because there is no copy.

The honest trade-offs

Client-side conversion is not magic, and we will not pretend otherwise:

Very large files are limited by the device. A 200-page scanned PDF will strain a cheap phone in a way it would not strain a server.
Scanned PDFs stay scans. Turning a photographed page into editable text needs OCR, which is a different (and heavier) job than rebuilding a document that already holds real text.
The first page load carries the libraries. A few hundred kilobytes of JavaScript, cached after the first visit, in exchange for never uploading your documents. We think that is the right swap.

A ten-second test for any converter

Before you upload a sensitive document anywhere, do this: open the site, open your browser's network tab (or just watch your upload indicator), and convert a test file. If megabytes flow up to a server, your document now exists somewhere else. If they do not — if the work happens quietly in the tab — you keep the only copy.

Documents are personal. The tools that touch them should act like it.

We build free document tools at makehumanai.com — converters, photo and signature resizers, and form-photo checks that run in the browser, without sign-ups and without uploads.

How to Turn a Phone Photo into a PDF (No App, No Sign-Up)

MakeHumanAI — Tue, 03 Nov 2026 16:03:38 +0000

Every form, portal and office desk eventually asks for the same thing: "Please attach a PDF."

The document is sitting in your phone gallery as a photo. Maybe it is a signed page you photographed, a certificate, an ID card, or three pages of a contract you snapped one by one. The portal does not want JPGs. It wants one PDF, in the right order, small enough to upload.

Here is the way we do it, and the mistakes we see people make every week.

The quick way (browser, nothing to install)

Open a photo-to-PDF tool in your phone browser — we run one at MakeHumanAI, and it is free with no sign-up.
Add the photos in the order the finished document should read. First photo becomes page one.
Convert and download. One PDF comes out with all pages inside.

That is genuinely it. The part people get wrong is not the conversion — it is the photos and the settings around it.

Get the photo right first

A PDF made from a bad photo is just a bad document in a nicer coat.

Light: Photograph near a window or under a plain ceiling light. Shadows across the page survive into the PDF.
Angle: Hold the phone directly over the page, not tilted from the side. Tilted text stays tilted.
Background: Put the page on a plain, darker surface so the edges are easy to see and crop.
One page per photo. Do not photograph two pages side by side and hope the reader zooms in — they will not.

Order matters more than people think

Portals and offices read PDFs the way you hand them over. If page 2 of your contract lands before page 1, a busy clerk may not go hunting. Before you convert, look at the sequence once: first page first, attachments after, ID copies where the form asks for them.

Most tools let you drag photos into order before converting. Use that minute.

The file is too big — now what?

Phone photos are large, and a five-page PDF can sail past a portal's upload limit (many form portals cap uploads at 200 KB or 500 KB). Two fixes:

Retake or resize oversized photos before converting, or
Shrink the finished file with a compressor — we keep one at Compress PDF to 200KB for exactly the government-form limits we kept running into.

Check the portal's stated limit first and aim just under it, not as small as possible. A document squeezed to 40 KB "to be safe" can turn small text into mush, and a mushy document gets rejected by a human even if the portal accepted the upload.

When a single photo, not a PDF, is the problem

Sometimes the form does not want a PDF at all — it wants the photo itself at an exact size, like 50 KB. That is a different job: resize the photo, do not convert it. We wrote the tools in that order because the confusion is so common: photos at exact sizes go through a resizer, pages that must travel together go through the PDF converter.

A short checklist before you submit

[ ] Pages in the right order
[ ] Smallest text on the page is readable when you zoom in
[ ] File size is under the portal's limit (but not crushed)
[ ] File name looks human: Application-Form.pdf, not IMG_2041.pdf

Do those four things and the boring document step stops being the reason applications stall.

We build free document tools at makehumanai.com — photo and signature resizers, PDF converters and form-photo checks, all without sign-ups. If a portal limit has ever ruined your evening, we have probably built a tool for that exact limit.

DV Lottery 2027 registration is about to open — and most photo mistakes are made in the first hour

MakeHumanAI — Wed, 28 Oct 2026 03:16:39 +0000

The DV Lottery (Diversity Visa) program has been paused since December 2025. Under a US court order, the government is now expected to announce the DV-2027 entry dates by October 29, with registration opening around November 1 for at least 30 days. The official dates will be published on dvprogram.state.gov — that is the only place to enter, and nobody else can enter for you.

We run a free photo checker for DV entries, so we see the same pattern every season: people rush in on day one, upload a photo that "looks fine", and get disqualified months later without ever knowing why. The entry itself takes ten minutes. The photo rules are where entries quietly die.

The photo rules that actually matter

600 × 600 pixels, JPEG, under 240 KB. Not "about square" — exactly square.
Plain white or off-white background. No shadows on the wall, no patterned curtains, no outdoor shots.
Head sized between 50% and 69% of the photo height, eyes looking straight at the camera.
No glasses. This changed years ago and people still get caught by it.
Taken within the last 6 months. An old visa photo from two years ago is a disqualification risk.
No editing or filters. No beautify mode, no background erasers, no AI enhancement. A slightly imperfect natural photo is safer than a polished one.
Babies and children follow the same rules — eyes open where possible, nobody else visible in the frame, no pacifier covering the face.

Three things to sort out before the window opens

Your passport. Reporting around this year's program points to a passport requirement and a small $1 entry fee paid to the US government on the official site. If anyone asks you to pay them the entry fee, walk away — the government collects it itself, on its own site, or not at all.
One entry per person. Duplicate entries disqualify everything. A husband and wife can each enter once, with the other as spouse — that is legal and doubles a family's chances the right way.
Your confirmation number. After you submit, you get one number. It is the only way to check your result in May. No email ever arrives. Screenshot it, write it down, send it to yourself.

Check the photo before you rely on it

We built a free DV lottery photo checker that measures the things above — exact size, file weight, format — and fixes the fixable ones in your browser. Your photo is not uploaded to a server; the check happens on your own device. There is also a registration guide with the expected dates and countdown that we will update the moment the official announcement lands.

Rushing is normal. Everyone does it on day one. Just rush with a photo that passes.

5 PDF conversions we made free, unlimited, and private (they run in your browser)

MakeHumanAI — Mon, 26 Oct 2026 06:33:38 +0000

5 PDF conversions we made free, unlimited, and private (they run in your browser)

We build MakeHumanAI, a small toolkit of free document tools. The converters were the hardest part to get right, because the obvious way to build them is also the worst way for the user: upload the file to a server, convert it there, and hope the person trusts you with their documents.

Most people converting a PDF are converting something private. A contract. A payslip. A visa application. So we built our converters the other way around: the conversion happens in the browser, and the file never leaves the device.

Here are the five we use the most ourselves:

1. PDF to Word — for when someone sends you a "final" PDF and then asks for one small change. makehumanai.com/pdf-to-word

2. Word to PDF — the reverse trip, for sending a document that should arrive looking exactly the way you left it. makehumanai.com/word-to-pdf

3. PDF to Image — when a form, portal or application insists on JPG pages instead of a PDF. makehumanai.com/pdf-to-image

4. Excel to PDF — spreadsheets print terribly; a PDF freezes the layout before you share it. makehumanai.com/excel-to-pdf

5. PDF to Excel — for pulling a table out of a PDF report back into something you can actually sort and filter. makehumanai.com/pdf-to-excel

No sign-up, no daily limit, no watermark on the result. That was a deliberate choice — the tools sit next to our photo and signature resizers that people use for government and job application forms, and those forms are stressful enough without a paywall in the middle.

If you try them and something breaks, we genuinely want to know. Small tools live or die on the details nobody reports.

— The MakeHumanAI team

DV Lottery 2027 Registration Opens November 1 — Dates, the New $1 Fee, Passport Rule and Photo Specs

MakeHumanAI — Fri, 23 Oct 2026 13:19:30 +0000

The US Diversity Visa (DV) Lottery — the "Green Card Lottery" — has been paused since December 2025. That is about to change. A US federal court order issued on October 14, 2026 requires the State Department to open DV-2027 registration by November 1, 2026, and to keep it open for at least 30 days. An official announcement with the final dates is expected by October 29, 2026.

Here is what is known so far, what is only reported (not yet official), and how to get your entry — especially your photo — right the first time.

The key dates

Event	Date
Program paused	Since December 2025
Court order requiring registration to open	October 14, 2026
Official announcement expected by	October 29, 2026
Registration opens by	November 1, 2026
Registration stays open	At least 30 days

Treat November 1 as the working date, but always confirm the final window on the only official entry site: dvprogram.state.gov.

Reported new requirements: $1 fee and a passport scan

News reports on the restarted program describe two changes this year:

A $1 entry fee — a small fee where entering used to be completely free.
A passport scan — applicants may need to upload a scan of a valid passport with the entry.

Both points are reported, not yet published in a final official announcement. Requirements can change between a court order and the official instructions, so check dvprogram.state.gov before you enter, and follow the instructions published there — not screenshots or social-media posts.

What has NOT changed: entering is something you do yourself

Submitting your own entry on dvprogram.state.gov costs nothing beyond the reported $1 fee. There is no official agent, broker or "commission" system for the DV Lottery.
Nobody can guarantee selection. Selection is a random draw. Anyone charging you for "guaranteed" selection, a "reserved slot" or a higher chance is running a scam.
Never hand over your confirmation number. You receive it when you submit your entry, and you need it to check your result. Anyone who submits "for you" and keeps the number can hold your result hostage. Enter yourself, save the number yourself.
A private service may legitimately help you prepare — for example checking your photo against the official specs — but the entry itself should always be submitted by you, on the official site.

DV photo specs (the part that rejects the most entries)

The photo is uploaded with your entry, and it is checked automatically. The requirements:

Requirement	Specification
Dimensions	600 × 600 px (square, 2×2 inch)
File format	JPEG (.jpg)
File size	240 KB or less
Background	Plain white or off-white, no shadows or patterns
Recency	Taken within the last 6 months
Head size	Head height must be 50–69% of the total image height
Face	Full face, directly facing the camera, neutral expression, both eyes open
Glasses	Not allowed
Editing	No filters, no retouching that changes your appearance

The most common failures are a head that is too small in the frame, a non-white background, and a file that is far over 240 KB straight from a phone camera.

You can check a photo against these rules before you enter with the free DV Lottery photo checker — it validates the size, dimensions and composition in your browser (nothing is uploaded) and can fix a photo to an exact 600×600 file. The full dates, rules and entry walkthrough are collected in the DV-2027 registration guide.

Pre-entry checklist

[ ] Valid passport ready (scan may be required this year — confirm on dvprogram.state.gov)
[ ] Photo: 600×600 px, JPEG, ≤ 240 KB, white background, taken within 6 months
[ ] Head fills 50–69% of the photo height, no glasses, no filters
[ ] Full names and dates of birth exactly as in official documents
[ ] Spouse and ALL children under 21 listed, as the instructions require
[ ] Entry submitted by you, on dvprogram.state.gov only
[ ] Confirmation number saved by you, in more than one place
[ ] One entry only — entering more than once disqualifies you

Prepare the photo and documents now, wait for the official announcement, and enter once, carefully, in the official window.

MakeHumanAI is a private tools website and is not affiliated with the US government or the State Department. Always confirm the final DV-2027 dates and requirements on dvprogram.state.gov before submitting.

US Visa Photo Requirements (DS-160) — The Exact Specs and How to Fix Rejections

MakeHumanAI — Fri, 23 Oct 2026 13:13:10 +0000

Every year, thousands of US visa applications hit an avoidable wall: the photo. The DS-160 form validates your uploaded photo automatically, and if it fails the technical checks, you cannot proceed until you fix it. Here are the exact requirements — and how to repair a rejected photo without Photoshop or a photo studio.

The official DS-160 photo specs

Requirement	Specification
Dimensions	Minimum 600 × 600 px, maximum 1200 × 1200 px (square, 2×2 inch)
File format	JPEG (.jpg) only
File size	240 KB or less
Background	Plain white or off-white, no shadows or patterns
Recency	Taken within the last 6 months
Head size	Head height must be 50–69% of the total image height
Eye position	Eyes must sit at 56–69% of the height from the bottom
Face	Full face, directly facing the camera, neutral expression, both eyes open
Glasses	Not allowed (since Nov 2016), except documented medical reasons
Head coverings	Only for religious/medical reasons, and the face must remain fully visible
Editing	No filters, no retouching that changes your appearance

Why photos get rejected

Wrong file size — A phone photo is usually 2–5 MB. The limit is 240 KB.
Head too small or too large — The 50–69% head-height rule is the single most common failure. A normal portrait usually has the head at 30–40% of the frame.
Non-white background — Cream walls, blue studio backdrops and outdoor shots all fail.
Shadows — A shadow on the wall behind you or across the face triggers rejection.
Glasses glare — Even when glasses sneak in, glare on the lenses hides the eyes.

How to fix a photo yourself (free, in the browser)

You do not need an app install or a studio visit for the technical parts:

Crop with a guide, not by eye. Use a framing guide that shows the head oval and eye lines so you can drag/zoom the photo until the head fills 50–69% of the frame. The DV Lottery photo checker uses the same head/eye geometry (the State Department applies near-identical composition rules), and its Smart Crop frame exports an exact 600×600 photo.
Validate against the DS-160 preset. The free Form Photo Validator checks dimensions, file size and format against the US visa preset and auto-fixes what it can, then lets you enhance a dark photo before downloading.
Compress last, not first. Always finish cropping before compressing — compressing first and cropping later multiplies quality loss. If the file is still over 240 KB after cropping, a targeted resize to an exact KB size keeps the dimensions intact.

One privacy note worth caring about: your passport/visa photo is identity data. Browser-based tools that process the image locally (on your device, with nothing uploaded) are safer than random "free converter" sites that keep your photo on their servers. Check for that before uploading a face photo anywhere.

Quick pre-upload checklist

[ ] Square, at least 600×600 px
[ ] JPEG, ≤ 240 KB
[ ] White background, no shadows
[ ] Head fills roughly 50–69% of the height
[ ] No glasses, no filters, taken within 6 months

Get those five right and the DS-160 uploader accepts the photo on the first try.

If you are applying for the DV Lottery instead, the composition rules are the same but the program has its own checker flow — see the DV Lottery photo checker. Always confirm the final specs on the official travel.state.gov website before submitting.