ArchiveOptics

Negative Converter · Film conversion & restoration for macOS

Forty years of family film, waiting in a shoebox. Bring it back to life.

Negative Converter turns scanned black-and-white and color negatives into finished photographs. Not with a naive color flip — with a conversion engine modeled on how film prints actually render density, tuned against hand-corrected reference conversions, and honest enough to preserve the warmth of the light that was really there.

In development · Coming to the Mac App Store Get release updates
Negative Converter's split view showing the converted positive portrait on the left and the original orange-masked color negative on the right, with a queue of film scans and conversion controls
Split view: the finished positive beside the orange-masked negative it came from. One conversion engine, no guesswork.

The conversion

A convincing positive begins with the character of the film.

Color negative film hides its picture behind an orange mask, crossed dye layers, and a density curve designed for a printing process that no longer exists on most desks. Negative Converter's engine works the way a good print did: it analyzes each scan's actual densities, removes the mask per channel, and remaps tones through a film-print-style curve — calibrated against real, hand-corrected reference conversions, not a generic inversion.

Color balance is found in the scan itself: the engine neutralizes from the image's own grey evidence, and on-device face detection anchors skin tones toward natural values. Deliberately, it does not sterilize every frame to grey — a warm room stays warm, evening light stays evening light. Film type is detected automatically; a roll of mixed black-and-white and color scans just works.

Negative Converter showing a forty-year-old color negative converted to a positive: a boy in a yellow shirt reaching into spring grass, the dark film rebate still framing the picture, with a queue of scans and adjustment panels
Forty-year-old color negatives from a family vacation, freshly scanned and converted — the film rebate still framing the moment.

The darkroom

Refine like a printer, not like a spreadsheet.

Auto-conversion gets each frame most of the way; the darkroom takes it the rest. Full tone control from black point to highlights. Temperature, tint, and per-channel color balance. Three-way color grading wheels for shadows, midtones, and highlights — which double as split toning for black-and-white. Lightroom-style crop with free-angle straighten that can never leave blank corners in an export.

Fixed one frame from a roll? Copy its settings and paste them across the entire selection in a single undo step. Every slider updates a live preview instantly, and a full-quality render settles in the moment you pause.

Negative Converter showing a strip of film with sprocket holes: a toddler beside a Minnie Mouse character from a forty-year-old family vacation, with tone, color, and color grading wheel controls open in the inspector
A forty-year-old family vacation on the light table — sprocket holes and all — with tone, color, and grading wheels, and a live histogram of the actual rendered positive.

Restoration

Decades-old dust, healed by machine learning that lives on your Mac.

Film that has waited forty years in a shoebox arrives with dust, scratches, and blemishes baked into every scan. Paint over them with the heal brush and an on-device inpainting model synthesizes replacement detail that matches the surrounding grain — no cloning smudges, no soft spots, no uploads. Below: a real blemish on forty-year-old film, and the same spot one brush stroke later.

Negative Converter's toolbar with the Heal tool selected and its tooltip reading: Heal brush — paint over dust and scratches
The heal brush: paint over dust and scratches.
A 1:1 crop of forty-year-old film grain with a white dust blemish circled by the heal tool
Before: a dust blemish on the scan, circled at 1:1.
The same 1:1 crop after healing: the blemish is gone and the film grain continues uninterrupted through the healed area
After: one stroke later — the blemish is gone and the grain runs straight through the repair.

Denoise runs a dedicated real-noise model with a chroma-only mode that removes color noise while keeping the film grain, because grain is the photograph. A floating 1:1 detail loupe renders full-resolution previews in a fraction of a second, so you judge every decision on the detail that matters — an eye, a texture — before committing.

Negative Converter's 1:1 detail loupe magnifying an eye in a scanned portrait with heavy film grain, denoise set to zero
Before: the 1:1 loupe on a grainy scan, denoise off.
The same eye in the 1:1 detail loupe with denoise applied — color noise gone, detail preserved
After: color noise gone, real detail kept — judged at full resolution before committing.

Output

Preview fast. Export perfectly.

Editing happens on a fast proxy; every export re-reads the original scan and re-runs the complete conversion at full resolution through a 16-bit, color-managed pipeline with no 8-bit shortcuts anywhere. The same scan and the same settings produce the identical file, byte for byte, today or in ten years.

MASTER

16-bit TIFF

A full-resolution archival master with embedded ICC profile — the file you keep forever.

DELIVERY

Maximum-quality JPEG

Full-size or resized sharing copies at deliberately generous quality settings, with EXIF carried over from the source.

RECIPE

Settings sidecar

Every conversion decision saved as readable data beside the export — reopen, refine, or reproduce any frame, any time.

Batch-export whole selections with per-image progress and never-overwrite naming, or send a fresh 16-bit TIFF straight to Photoshop or any editor with Open Positive With… — always the full render, never a preview bitmap.

N

The negative scan is never rewritten.

Your scan is the artifact — the closest thing to the original negative you'll ever have in a file. Negative Converter reads it, renders from it, and writes its results elsewhere. Nothing you do in the app can damage it.

Under the hood

Built native, measured honestly.

  • Platform Native macOS for Apple Silicon — Swift, Core Image, Core ML
  • Input Camera-scanned RAW (tested at 45.7 MP), RAW+JPEG pairs, TIFF, and JPEG scans
  • RAW handling Deterministic linear decode — every camera “enhancement” zeroed for honest analysis
  • ML models Bundled on-device inpainting and denoise models — commercially licensed, never cloud
  • Pipeline 16-bit/float end to end; GPU-accelerated rendering
  • Reproducibility Same scan + same recipe = byte-identical output, guaranteed by design

And when the roll is converted?

PhotoLibrarian catalogues the results — keywords, ratings, people, and all — alongside the rest of your archive.

See PhotoLibrarian →