Halftone Generator

Loading the halftone generator...

Turn a photo into a deterministic AM-style halftone made from circles, squares, diamonds or lines. The browser samples the selected image on a regular grid and changes each mark according to local tone, with controls for cell size, angle, contrast, gamma, smoothing and inversion. Choose a single-color result, retain a sampled color in each cell, or create an RGB-rendered CMYK-style simulation. Processing and the two-hour working project stay in browser-local state rather than an image-conversion service. Review the result at different zoom levels, then download a raster PNG or an SVG containing editable vector-dot geometry.

How to make a halftone from a photo

  1. 1

    Choose an image

    Select a conventional JPG, PNG or WebP file no larger than 20 MiB. The browser also checks decoded dimensions and working size before processing.

  2. 2

    Pick a color mode and shape

    Use monochrome, sampled color or the CMYK-style simulation, then choose circles, squares, diamonds or lines for the repeated marks.

  3. 3

    Set the grid

    Choose the cell size in pixels and rotate the sampling grid with the screen-angle control. Smaller cells retain more image detail and create more vector elements.

  4. 4

    Tune the tones

    Adjust contrast, gamma and smoothing, invert the tone mapping when needed, then generate the result. Change one control at a time to understand its effect.

  5. 5

    Review and download

    Inspect highlights, shadows and edges at full size. Download PNG for a fixed raster result or SVG for scalable, editable dot geometry.

What AM-style halftoning means here

A halftone represents gradual tones with discrete printed or displayed marks. In traditional amplitude-modulated screening, the marks sit on a regular grid while their size changes: small dots leave more background visible in highlights, and larger dots cover more area in shadows. A European Patent Office description of traditional AM screening makes the same fixed-spacing, variable-size distinction. Adobe’s PostScript Language Reference describes a halftone screen in terms of frequency, angle and a spot function.

This generator applies that visual model to browser-decoded pixels. It is deterministic: the same image and settings produce the same dot geometry. It is not a stochastic or FM screen, a generative AI redraw, or a raster image tracer. The program does not recognize faces, products or important edges; it only samples pixel color and tone.

The cell size is measured in source/output pixels. It controls spacing between sample positions, not a physical press ruling. Commercial screen frequency is measured in lines per inch (lpi), while image resolution uses pixels per inch (ppi) and output devices are commonly described in dots per inch (dpi). Adobe explains these separate quantities in its guide to printer resolution and screen frequency. Without a declared physical print size, a pixel cell size cannot truthfully be converted into lpi.

Modes, shapes and tone controls

Control What it changes What it does not guarantee
Monochrome mode Draws one foreground color against the selected background A calibrated ink density or press separation
Sampled-color mode Uses a sampled source color for each grid mark Exact reproduction of the source image gamut
CMYK-style mode Simulates overlapping cyan, magenta, yellow and black-style marks in browser color A profiled CMYK file, plate set or contract proof
Cell size (px) Changes grid spacing and the number of marks A physical lpi, ppi or dpi value
Screen angle (°) Rotates the regular sampling grid Freedom from moiré on every printer
Contrast Expands or compresses differences between lighter and darker samples Recovery of clipped or missing source detail
Gamma Redistributes midtones before mark size is chosen ICC color management or monitor calibration
Smoothing Averages local variation before sampling Semantic noise removal or edge recognition
Invert Reverses which tones produce larger marks A photographic negative with press-specific behavior

Circles give the familiar clustered-dot appearance. Squares produce firmer corners, diamonds rotate that geometry, and lines emphasize the screen direction. Shape changes the visual texture as well as the amount of background visible at a given size. Very large marks can merge in dark regions; very small marks can disappear when a raster result is reduced or printed.

Choosing settings deliberately

Start with a medium cell size and the monochrome circle preset. Increase contrast when a flat photo produces similarly sized marks. Reduce contrast when highlights vanish or shadows become nearly solid. Gamma is most useful for shifting midtones without moving the lightest and darkest endpoints as aggressively. Smoothing can calm sensor noise, skin texture, fabric or foliage, but too much can remove small features.

Screen angle is the orientation of the regular grid. A 45° screen is common for monochrome work, but it is a convention rather than a universal requirement. Adobe warns that an already screened image can interact with a printer’s own halftone and produce a moiré pattern; see its current halftone screen guidance. For commercial color printing, angles and ruling depend on the press, substrate, RIP and separation workflow. Adobe therefore advises users to confirm frequency and angle with the print provider.

Useful review points include:

  • faces, text and narrow edges that may be lost between sample cells;
  • pale areas where marks become too small to survive raster export or printing;
  • dark areas where neighboring marks touch and close background gaps;
  • regular texture in the source that may interfere visually with the halftone grid;
  • the SVG element count, which grows quickly as cell size decreases.

RGB, CMYK and print expectations

Browser canvas and ordinary SVG colors are rendered through browser color systems, commonly sRGB. The CMYK-style option is an artistic four-channel simulation rendered for browser display; it is not a color-managed conversion to a printer condition. The International Color Consortium explains that reliable conversion from source RGB to printer CMYK depends on an output device profile and the intended printing conditions. A browser preview also cannot prove trapping, overprint, dot gain, plate registration or the smallest mark a specific press can hold.

Output Contains Best use Important limit
PNG A fixed raster rendering of the generated pattern Sharing, slides, web graphics and predictable pixel dimensions Enlarging it does not create new detail; browser canvas export is not a press profile
SVG Vector circles, squares, diamonds or line geometry Scaling the generated marks and editing them in vector software It vectorizes the marks, not the original photograph; dense settings can create a large file

The browser’s canvas.toBlob() method creates a raster Blob, with PNG as the required fallback format. SVG 2 defines scalable graphic shapes such as the circle element. Those standards support useful downloads, but neither makes the result press-ready. For production work, open the download in a color-managed design application, set the final physical size, and ask the print provider about the correct profile, screen ruling and proofing process.

File limits, local project storage and privacy

The compressed input must be 20 MiB or smaller. Compressed size is not the same as memory use: after decoding, a photo occupies multiple bytes per pixel and processing needs additional working buffers. The tool therefore also rejects images beyond its decoded-dimension and processing caps. If a valid file is refused, resize it or export a smaller copy before trying again.

The W3C File API defines user-selected File and Blob objects that web applications can read. This tool reads the selected image, creates a bounded normalized working image, generates the result and stores the working/result blobs plus settings in browser-local project storage for up to two hours so the four-step funnel can survive page changes. Starting over clears the active tab’s project. An isolated copy opened in another tab remains separate and expires within two hours. Expired or invalid local records are discarded.

The selected image, decoded pixels and generated files are not server-side properties, are not placed in funnel URLs and are not uploaded to an external image-processing service. That statement does not mean the webpage is offline or makes no requests. Normal document, asset, analytics and funnel-step requests still occur, carrying ordinary page or campaign data rather than the image bytes. Browser storage remains subject to the browser profile, private-browsing behavior, quotas and user controls.

Canvas pixel reads and exports are also protected by the platform’s origin-clean rules. The HTML canvas security model blocks pixel extraction from unsafe cross-origin image sources. User-selected local files avoid that particular remote-image workflow, but corrupt files, unsupported encodings and browser resource limits can still stop decoding.

Accessibility and practical use

Halftoning is a visual transformation, so no generated pattern can replace an accessible description of the source image. The interface provides named controls, keyboard operation and textual progress/error states; the preview should be treated as a visual comparison, not as the only carrier of essential information. The applicable principles include WCAG 2.2 text alternatives and keyboard operation and its guidance for programmatically exposed status messages.

For a final print, produce a small proof at the intended size. Check whether the lightest marks survive, whether dark marks fill in, and whether another screening stage introduces an unwanted pattern. A successful screen graphic on one desktop printer is not proof that the same file will reproduce identically on a newspaper press, screen-printing mesh or commercial offset workflow.

Frequently Asked Questions

The marks occupy positions on a regular grid and change size according to the sampled tone. That fixed-spacing, variable-size behavior follows the visual principle of amplitude-modulated screening. The tool is a browser effect, not a printer RIP or certified prepress screen.

Monochrome draws one selected foreground color, sampled-color mode takes a color from the source at each cell, and CMYK-style mode simulates cyan, magenta, yellow and black-style marks. The last option is still rendered in browser color and is not a profiled CMYK separation.

No. Cell size is pixel spacing in the generated artwork. Lines per inch require a known physical output size; ppi describes image sampling, and dpi usually describes output-device addressability. Confirm physical ruling and resolution with the print provider.

A 45° angle is a common starting point for monochrome work, but there is no universal best angle. Source texture, multiple color screens and the printer’s own screening can interact. Commercial jobs should use angles specified by the print provider or RIP workflow.

Choose PNG for a fixed raster image with predictable pixel dimensions. Choose SVG when you want scalable, editable dot geometry. The SVG contains generated marks rather than a vector reconstruction of the original photograph, and very small cells can create a large file.

No. It is an RGB-rendered visual simulation and does not contain a printer-specific ICC conversion, validated plates, trapping or proof conditions. Use a color-managed production application and the print provider’s profile and screening requirements for press work.

The selected file can be up to 20 MiB. Decoded-dimension and working-size caps also apply because a compressed photo can expand to many times its file size in browser memory. Resize very large images if the interface reports a safety-limit error.

The image and generated files are processed in the browser and are not sent to our servers or an external conversion service. A bounded normalized working blob, result blob and settings may remain in browser-local project storage for up to two hours. Start over clears the active tab’s project; isolated copies opened in other tabs expire within two hours. Normal page, asset, analytics and funnel requests still occur without the image bytes.

Related Tools

Tool available in other languages