Tiling — large format assemblable

You've scanned a photo, you've corrected it with Calibration Flow, and you want to print it large — say, 90 cm wide. Your printer only handles A3 (30 × 42 cm). What do you do? You could crop the photo to A3 size and sacrifice three-quarters of your image. Or you could pay a specialized lab for large-format work and lose continuity in your printing chain. Tiling offers a third path: slice your negative into A4 and A3 tiles, print each tile, and assemble them edge-to-edge with no interruption. It's called a mounting board. Calibration Flow computes the possible grids automatically, proposes the best ones for your photo, and generates the ZIP of negatives ready to print.

#What you see

You open the tool via the Large format button on the negative page. You drop your source photo — a JPG, PNG, HEIC, or TIFF image up to 50 MB. Calibration Flow first enumerates the possible grids: for your photo, how many rows and columns of A4 or A3 are needed to cover between 30 and 120 cm of the longest side, while keeping DPI ≥ 90 (the minimum resolution before the image pixelates)? It proposes up to 6 candidates.

For each one, the app tests 6 mounting patterns: simple aligned, brick (offset tiles), all-A4 (uniformity), focal (protects the photo's core), symmetric, branded (composition around the subject). After deduplication, it returns up to 12 compositions ranked from best to least best.

You see a gallery: mini-previews of each composition, with colored badges showing DPI quality (green ≥ 240, orange ≥ 150, red < 150). Each composition shows you the total tile count and the percentage of cropping needed.

Next, you optionally select an assembly style: Studio (artistic values), Tableau (symmetry), Cross-device (contrast), Reserve (protects the focal point), Board (optimal packing), Frieze (orientation). This changes the scores and can reorder the proposals. You pick your preferred composition, see a live preview in positive or calibrated negative, and export. Calibration Flow renders each tile in OffscreenCanvas (one at a time, without allocating a huge final canvas), applies your calibration curve and negative inversion, compresses everything to PNG, stacks the files in a ZIP STORE, and offers it for download. The recipe is saved for later reconstruction.

#Why it matters

For anyone doing large formats without a specialized lab. Manually assembling tiles means calculating the crop, placing cut marks, crossing seams. Tiling removes the calculation and gives you a geometrically sound mounting from the start.

For anyone who wants to respect their calibrated negative. Once you have the perfect cyanotype curve on your photo, you don't want to crop arbitrarily before printing. Tiling preserves your image's tonal intent — it never resizes, it fragments intelligently.

For anyone experimenting. With 12 compositions to see at once, you test quickly: do you care about symmetry, or do you prefer to maximize size? Are A4 tiles enough, or do you need A3 to reduce seams? Tiling answers in seconds.

#When you don't need it

You're working on a calibration target. Tiling is designed for photographs, not for 25-patch gray targets. A target split into tiles loses its metrological integrity — seams interrupt measurements and the continuity of reference curves. For targets, scan in one monolithic shot.

You have a very atypical image: abstract photogram, raw negative scan, discrete color palette. Tiling enumerates grids based on spatial geometry. On an image with no clear structure (a uniform fade, for instance), it proposes arbitrary slicing. Fall back to manual tools.

You can't print without margins (full-bleed). Physical assembly requires borderless printing. If your printer isn't capable of it, you'll have to trim 3–5 mm of white margin from each seam, risking image interruption. Verify this detail before cutting a tiling layout.

#Worth remembering

#The test

Load a portrait or landscape photo of 60+ cm long side into Calibration Flow. Access tiling, select three different compositions, observe the DPI badges and tile counts. Export a composition, open the downloaded ZIP, and count the PNG files: the number should match the announced tiles (e.g., 2 rows × 3 columns = 6 PNG). Open a random PNG: it should contain a coherent excerpt of your photo in calibrated negative (inverted, without artifacts). Then try exporting in positive — if your curve is well calibrated, you should recognize your photo split into a grid.