Apply a sepia filter to image RGB pixel samples

Apply a sepia filter to decoded image pixels without uploading a file or relying on an image codec. Provide an ordered array of RGB samples and receive corresponding warm brown-toned samples calculated with the standard sepia transform matrix. Every transformed channel is rounded and clamped to the valid zero-to-255 range. Complete channel validation prevents incomplete or malformed pixels from quietly changing the result, while deterministic processing makes the converter suitable for production pipelines, browser checks, teaching examples, and exact automated tests.

Prepare complete RGB pixel samples

Place decoded pixels in the samples field as a nonempty array of objects. Every object must contain the channel names r, g, and b, and every channel must be a finite numeric value between zero and 255 inclusive. Integers are conventional, but fractional input is accepted because some image pipelines produce intermediate floating-point samples before final encoding.

The converter checks every record rather than assuming that a missing component means zero. Keep samples in raster order if you intend to rebuild an image afterward, because output order exactly matches input order. Split larger rasters into fixed-size chunks, submit them in a controlled order, and concatenate returned sample arrays in that same order.

The capability works with RGB samples, not PNG, JPEG, AVIF, WebP, or other encoded bytes. Decode a source file first and preserve width, height, row stride, orientation, and any alpha channel separately so those values can be combined with the transformed samples during encoding.

Understand the standard sepia transform

Each output pixel comes from three weighted sums. Red is calculated as 0.393 times r plus 0.769 times g plus 0.189 times b. Green uses 0.349 times r plus 0.686 times g plus 0.168 times b.

Blue uses 0.272 times r plus 0.534 times g plus 0.131 times b. This familiar sepia matrix changes both luminance and channel balance, producing highlights dominated by warm red and green while reducing blue. The converter rounds each weighted sum to the nearest whole number and then clamps it to the inclusive channel range from zero through 255.

Clamping matters because the rows of the matrix can yield values above 255 for bright input pixels. Without it, downstream byte buffers could wrap, reject the value, or display an unintended color. Processing is deterministic and pure: it does not use a network service, trained model, random source, current time, hidden color profile, or mutable request cache.

Integrate and verify the converted result

Read the returned samples array as a positional replacement for the submitted RGB array. The count field should equal the number of input records; compare it before placing samples back into a tile, scanline, thumbnail, canvas buffer, or full raster. If the original image has transparency, keep its alpha values in a parallel buffer and restore them when encoding, because this contract intentionally transforms only red, green, and blue.

Bright white is especially useful for verifying that all three computed values clamp to 255 instead of exceeding the byte range. Pure colors reveal accidental coefficient or channel-order changes, and neutral values show the characteristic warm separation between returned channels. Treat validation failures as problems with decoding or request construction rather than transient failures to retry unchanged.

A missing-channel message means the source buffer or mapping logic must be corrected. A range or finite-number error commonly indicates an unnormalized HDR value, NaN from earlier arithmetic, or a typed-buffer conversion mistake. Correct such inputs deliberately before resubmitting so the sepia output remains reproducible.