Transparency arrived in PDF 1.4 and immediately created a compatibility problem that has never fully gone away. Drop shadows, soft edges, blend modes and partial opacity are all straightforward to describe and genuinely difficult to output on devices built around opaque ink. Flattening is the process of converting live transparency into opaque artwork that looks the same — and the artefacts it produces, white lines and rasterised text among them, are among the most confusing print problems there are.
What live transparency is
Before PDF 1.4, every object was opaque: drawn objects covered what was beneath them. The transparency model added:
- Constant alpha — an object at 60% opacity.
- Soft masks — a greyscale image controlling opacity per pixel, which is how drop shadows and feathered edges work.
- Blend modes — multiply, screen, overlay, and the rest, specifying how an object's colour combines with the backdrop.
- Transparency groups — sets of objects composited together before being composited with the page, so a group's internal blending is isolated from the background.
All of this is live: the file describes the intent, and the renderer computes the result. That is exactly what you want, and it is also why the result depends on the renderer.
Why flattening exists
Two reasons, one historical and one operational.
Compatibility. PostScript, which drives a great deal of print equipment, has no transparency model. A PDF containing live transparency sent to a PostScript RIP has to be converted somewhere. PDF/X-1a, still requested by many printers, forbids live transparency outright.
Predictability. A flattened file renders identically everywhere because the compositing has already been computed. A file with live transparency depends on the RIP implementing the model correctly — and RIP implementations have historically varied, particularly on overprint interacting with transparency.
Flattening replaces the transparent objects with an equivalent arrangement of opaque ones: regions are divided into atomic areas where the composited appearance is constant, vector regions are recoloured to the composited result, and areas too complex to express as vectors are rasterised into images.
The artefacts
Everything people dislike about flattening comes from that division into atomic regions.
White stitching lines. Flattening carves the artwork into adjacent vector shapes that meet exactly at their edges. Mathematically they abut perfectly; in a renderer with antialiasing, the boundary can display as a hairline. These lines frequently appear on screen and not in print, which makes them maddening to diagnose — the honest first response to "there are white lines in my flattened PDF" is to check whether they exist on paper before chasing them.
Rasterised text. When text overlaps transparent artwork, the flattener may convert the affected text to an image. It stops being selectable and searchable, prints at the flattener's raster resolution rather than the RIP's, and looks slightly softer than the text around it. This is the artefact with the most consequences: it breaks accessibility, extraction, and often the visual match with adjacent text.
Colour shifts, particularly where transparency interacts with spot colours or overprint. The flattener has to decide what a multiply blend over a spot colour becomes, and the answer is sometimes a process approximation. See spot colours and overprint in PDF.
File size explosion. Rasterising a large area at 300 ppi in place of a few vector objects can multiply the file size. A 4 MB design file becoming a 60 MB flattened one is unremarkable.
Object count explosion. Even without rasterisation, splitting artwork into atomic regions can turn a handful of shapes into thousands, which slows every subsequent operation on the file.
The flattener settings
Adobe's flattener presets expose the trade-offs directly:
Raster/Vector Balance (0–100). At 100 the flattener preserves as much vector data as possible, rasterising only where it must. At 0 it rasterises everything, producing a simple, large, resolution-fixed file. Keep it at or near 100 for print quality; drop it only when the vector output is so complex that the RIP chokes on it.
Line Art and Text Resolution (default 300 ppi, often set to 1200). This is the resolution used when text or line art does get rasterised, and it is the setting most worth raising. Rasterised text at 300 ppi looks visibly soft next to real text; at 1200 ppi the difference is hard to see.
Gradient and Mesh Resolution (default 150 ppi, often 300). Applies to rasterised gradients, which tolerate lower resolutions because they have no hard edges.
Convert All Text to Outlines. Ensures consistent text weight across flattened and unflattened areas — at the cost of making all text unsearchable and inaccessible. Occasionally the right answer for a poster; almost never right for a document.
Convert All Strokes to Outlines. Prevents thin strokes from varying in weight after flattening. Adds objects.
Clip Complex Regions. Reduces stitching artefacts at the cost of more complex clipping paths.
Adobe's stock presets — Low, Medium, High Resolution — set these together; High Resolution is the correct starting point for anything going to press.
How to avoid flattening entirely
The best fix is not to flatten.
Export as PDF/X-4. It permits live transparency and layers, and every current RIP handles it. If your printer accepts X-4 — most do in 2026 — supply it and let their RIP do the compositing at output resolution, which is better than any flattening you can do beforehand. See PDF/X print format explained.
Ask the printer rather than assuming. "Do you want X-1a or X-4?" is a ten-second question that determines whether this whole topic applies to you.
Do not flatten twice. Flattening an already-flattened file compounds the artefacts. Keep the live-transparency master and flatten from it as a derived output.
Working with a file you must flatten
If PDF/X-1a is required:
- Use the High Resolution preset, with line art and text at 1200 ppi.
- Flatten from the source application (InDesign, Illustrator) rather than from the finished PDF where possible — the source has more information about the artwork and produces cleaner results.
- Check the result in Overprint Preview at high magnification, particularly at the edges of transparent objects and wherever text overlaps artwork.
- Check whether text got rasterised: try to select it in the flattened file. Text that will not select has been converted to an image.
- Simplify the artwork if the flattener is struggling: fewer overlapping transparent objects, fewer blend modes stacked on top of each other, transparency confined to areas away from text.
Acrobat's Flattener Preview (Tools → Print Production → Flattener Preview) highlights exactly which objects will be affected, before you commit. It is the diagnostic equivalent of Output Preview and equally underused.
Transparency outside print
Flattening is a print topic, but its effects surface elsewhere:
- Old viewers and mobile renderers occasionally handle live transparency poorly, producing black boxes where a soft shadow should be.
- Converting to older formats or to image formats forces a flatten, with the same artefacts.
- Accessibility suffers when text is rasterised: it disappears from the structure tree and from extraction entirely. See tagged PDF vs untagged PDF.
- File size matters on the web, and a flattened file can be far larger than the original. Related: reduce PDF file size.
Summary
Flattening converts live transparency into opaque artwork by dividing the page into atomic regions and rasterising what cannot be expressed as vectors — producing stitching lines, softened rasterised text, and much larger files. The best mitigation is to avoid it: export PDF/X-4 and let the RIP composite at output resolution. When you must flatten, use the high-resolution preset with text at 1200 ppi, flatten from the source rather than the PDF, and check in Flattener Preview which objects will be affected before you commit.
