DEV Community: rogin yang

Preparing a 100 mm photo PNG for laser engraving: size, dithering and import checks

rogin yang — Fri, 09 Oct 2026 03:01:29 +0000

I maintain this tool.

I maintain RasterGo, a free browser tool for preparing photo PNGs for laser engraving. This guide covers image preparation, not a tested power/speed recipe or a guarantee of an engraving result.

Start with physical size

For a 100 mm square image at 300 DPI:

pixels = round(mm / 25.4 × DPI)

round(100 / 25.4 × 300) = 1181 pixels per side.

That specifies the digital image size. It does not prove that 300 DPI is best for every laser, material or focus setting. Upscaling also cannot restore detail missing from the original photo.

Crop and adjust before dithering

Use a crop that keeps the face and important details clear at the final size. Adjust tones while comparing the original and processed image. A busy background can obscure a subject after dithering. Avoid treating a dramatic preview as proof of a good physical result.

Decide which application handles the image

Dithering turns tones into a pattern of dots. If you export an already dithered image, check the image mode when importing it into your laser software; applying another tonal conversion can change the pattern. LightBurn documents its image modes here: https://docs.lightburnsoftware.com/latest/Reference/CutSettingsEditor/ImageMode/ . Follow the mode appropriate for your own workflow and test it.

Slate needs a material test

Slate usually reveals a lighter mark against a darker surface. Whether your pipeline needs inversion depends on the input and software settings. Check the preview and a small test piece rather than assuming that one invert switch works universally. Digital material simulations on RasterGo are labeled; they are not photographed customer engravings.

Export, import and confirm

In RasterGo, choose the image, set size and DPI, compare the crop and tone adjustments, then export the PNG. The image processing stays in your browser. On import, confirm the physical width/height and inspect the result before running the machine. The tool also supports a small batch ZIP workflow.

I am looking for practical feedback from makers: did the exported file keep the intended size, and which preparation step still required manual work? Free beta; no calibrated machine settings or physical engraving result are being claimed.

https://rastergo.com/?utm_source=devto&utm_medium=social&utm_campaign=launch-2026-10&utm_content=rastergo-devto-en

Building a passport/ID photo maker that never uploads the photo (MediaPipe in the browser)

rogin yang — Mon, 05 Oct 2026 16:49:15 +0000

Most "passport photo online" sites ask you to upload a selfie and then charge to remove the watermark. I wanted to see how far a browser-only version could go: crop to the official spec, replace the background and export print-ready files without the image ever leaving the phone. Here is what turned out to matter.

1. Specs are about head height, not zoom

Official rules describe the head height (chin to top of head) and the margin above it:

Photo	Size	Head	Source
Brazil 3x4	30 x 40 mm	≈ 30 mm (70–80%)	ICAO guidance published by ANAC (gov.br)
US visa	2 x 2 in; DS-160 file 600x600 px, ≤ 240 KB	25–35 mm (50–69%)	travel.state.gov
Canada visa	35 x 45 mm	31–36 mm	canada.ca (IRCC)
Schengen	35 x 45 mm	32–36 mm	EU Visa Code / ICAO 9303

Once you know the chin and the crown in pixels, the crop is just arithmetic:

function computeCrop({ chinY, crownY, centerX }, spec) {
  const pxPerMm = (chinY - crownY) / spec.headTargetMm;
  const w = spec.wMm * pxPerMm, h = spec.hMm * pxPerMm;
  return { x: centerX - w / 2, y: crownY - spec.topMm * pxPerMm, w, h };
}

2. The chin is easy, the crown is not

MediaPipe's FaceLandmarker gives 478 points. Point 152 is the chin, but the highest point (10) sits mid-forehead. What worked for me:

De-rotate the image using the eye line first.
Run a person segmenter on a window around the head and take the first row, from the top, where the mask covers a central band of the face. That row is the top of the hair.
For specs that measure to the skull (Canada, ICAO), clamp that value. Otherwise big curly hair shrinks the face below the allowed range.

3. Segmentation: download size vs. quality

selfie_segmenter is 250 KB and good enough for a flat white background. selfie_multiclass_256x256 is 16 MB and handles curly hair much better. I load the small model first, show a result, and then fetch the big one in the background unless navigator.connection reports 2g/3g or saveData. Running the segmenter on the crop region instead of the full photo makes the 256-px mask far sharper.

4. Getting the print size right: write DPI into the JPEG

canvas.toBlob('image/jpeg') doesn't set a useful density, so print shops print the file at the wrong size. Patch the JFIF header bytes directly:

b[13] = 1; b[14] = dpi >> 8; b[15] = dpi & 255; b[16] = dpi >> 8; b[17] = dpi & 255;

5. A4 PDF without a library

A one-page PDF with the JPEG embedded as /DCTDecode and each copy placed with a cm operator (mm × 72 / 25.4) takes about 40 lines of JS. An A4 sheet with 12 mm margins and 3 mm gaps fits 30 copies of a 3x4 photo.

6. Feedback before printing

MediaPipe blendshapes (eyeBlink*, mouthSmile*, jawOpen), plus two cheap measurements (nose offset from the eye midpoint for a turned head, cheek luminance difference for side lighting), are enough to warn people about closed eyes, smiling, a turned head or one-sided light before they print.

Code and data

The specs (each with its official link) and the engine are open: https://github.com/MMMystery/foto-documento-specs (MIT code, CC BY 4.0 data). The live tool built on it, in Portuguese for Brazil, is free and has no watermark: https://fotoapta.com/

Disclosure: I built FotoApta. This post was drafted with AI assistance and reviewed; all numbers come from the official sources linked above.

Fiz uma API grátis (sem chave) para calcular salário líquido, INSS e IRRF de 2026

rogin yang — Mon, 05 Oct 2026 13:21:49 +0000

Todo mês de janeiro a mesma coisa: sai a portaria nova do INSS, o salário mínimo muda, e quem mantém sistema de folha, planilha de RH ou simulador precisa caçar os números em PDF do governo. Em 2026 ainda veio a Lei 15.270/2025, que zera o IRRF até R$ 5.000 e cria uma faixa de transição até R$ 7.350, com uma fórmula própria (R$ 978,62 − 0,133145 × rendimento).

Juntei tudo num registro de regras com fonte oficial e data de conferência em cada valor, e coloquei um motor de cálculo na frente. O resultado é uma API pública:

curl "https://contaclt.com/api/v1/irrf?salario=6000"

Responde em JSON, sem cadastro, sem chave e com CORS liberado (dá para chamar direto do front-end). Para R$ 6.000 sem dependentes: INSS de R$ 641,51, imposto pela tabela de R$ 564,85, redução de R$ 179,75 e IRRF de R$ 385,10.

Endpoints

Endpoint	Exemplo
`salario-liquido`	`/api/v1/salario-liquido?salario=3500`
`inss` (empregado, autônomo, plano simplificado, baixa renda)	`/api/v1/inss?salario=3500&categoria=individual`
`irrf`	`/api/v1/irrf?salario=6000&dependentes=1`
`seguro-desemprego`	`/api/v1/seguro-desemprego?salario1=3000&salario2=3000&salario3=3000&solicitacao=1&meses=18`
`saque-aniversario`	`/api/v1/saque-aniversario?saldo=5000`
`das-mei`	`/api/v1/das-mei?atividade=servicos&faturamento=52000`
`empregada-domestica` (folha + guia DAE)	`/api/v1/empregada-domestica?salario=1621`
`tabelas`	`/api/v1/tabelas`

Três detalhes que pegam muita gente

INSS é por faixa, não alíquota única. Para R$ 3.500 são 7,5% sobre R$ 1.621,00 + 9% sobre R$ 1.281,84 + 12% sobre R$ 597,16 = R$ 308,60.
O eSocial trunca, não arredonda. Na folha da doméstica com salário mínimo, o INSS sai R$ 121,57 (não 121,58) e a DAE fecha em R$ 445,76. Se o seu sistema arredonda, vai dar um centavo de diferença.
Desconto simplificado x deduções legais. O IRRF usa o que for mais vantajoso: R$ 607,20 fixo ou INSS + R$ 189,59 por dependente + pensão. A redução da Lei 15.270 entra depois disso.

Dados abertos

As tabelas (com fontes, vigência e data de conferência) estão em JSON no GitHub, licença CC BY 4.0: https://github.com/MMMystery/tabelas-clt-2026 . Documentação completa da API: https://contaclt.com/dados/api/

Encontrou divergência com a fonte oficial? Abre uma issue com o link do ato que eu corrijo. Sugestões de endpoint também são bem-vindas.

Transparência: eu mantenho o ContaCLT, que usa esses mesmos dados nas calculadoras.