The single most common mistake in preparing images for print is confusing screen quality with print quality. An image that looks sharp, detailed, and professional on a monitor can produce a noticeably soft or blocky result on paper, not because anything went wrong in the printing process, but because the file never had enough pixel information to meet print requirements in the first place.
Understanding this gap, and knowing how to close it, is one of the more practically valuable technical skills in visual communication.
The Number That Matters: 300 PPI at Output Size
Print resolution is measured in PPI, pixels per inch, at the intended physical output dimensions. The standard for professional print is 300 PPI. This is not an arbitrary number; it corresponds to the resolution at which pixel structure becomes invisible to the unaided eye at normal reading distance.
What this means in practice: a photograph that will be printed at 20cm × 25cm needs at least 2362 × 2953 pixels to reproduce at full quality. Take an image with 1200 × 1500 pixels and print it at that size, and you’re working at 150 PPI, visible softness, especially in areas of fine detail, sharp edges, or small text.
The calculation is simple. Multiply the intended print dimension in inches by 300 to get the required pixel count on that axis. An A4 sheet (approximately 8.3 × 11.7 inches) at 300 PPI requires 2480 × 3508 pixels minimum.
Most people’s source images fall short of these requirements more often than they expect, which is where upscaling enters the workflow.
Two Approaches to Upscaling, and Why They Produce Different Results
Traditional resampling, the kind that happens when you drag a scale handle in Photoshop or Illustrator, or use the Image Size dialog with Resample turned on, fills in new pixel values by averaging the colours of neighbouring pixels. The mathematical terms for different methods of doing this are bicubic, bilinear, and nearest-neighbour, but the practical result of all of them at significant upscale factors is the same: a larger image that looks softer than the original, because the added pixels contain invented data rather than real visual information.
AI upscaling works from a different premise. Rather than averaging existing pixels, a neural network trained on large datasets of images learns to predict what additional detail would plausibly exist at higher resolution. Run an image through the upscale image tool in Pixelcut, an AI photo editing app, and the model analyses the content – edges, textures, gradients, fine structure – and generates new pixels based on learned patterns of how high-resolution versions of similar content typically look. The result at 2x or greater upscale factors is measurably sharper than traditional resampling, particularly in the areas where traditional methods produce the most visible softness: fine edges, textured surfaces, and detailed backgrounds.
This isn’t magic. The AI is making informed predictions, not recovering information that wasn’t captured. But for most practical use cases, the predictions are good enough that the output is usable at print dimensions where the original wasn’t.
When Upscaling Is the Right Call and When It Isn’t
Upscaling makes sense when the shortfall between your available pixel count and your required pixel count is within a reasonable range. A 2x upscale on a clean, well-exposed original produces reliably good results. A 4x upscale on a sharp, high-quality source is possible with AI tools, though the quality ceiling of the output is determined by the quality of the input.
What upscaling can’t do is manufacture detail that was never there. A blurry image becomes a larger blurry image. A heavily compressed JPEG with visible artefacts produces a larger image where those artefacts may be enhanced rather than removed. A screenshot at 72 PPI taken from a web page is not a suitable source for a poster, regardless of what upscaling is applied to it.
The Fogra Research Institute for Media Technologies, which develops and maintains the print industry standards widely used across European print production, specifies input resolution requirements for different print applications and substrates. These specifications provide a useful reference point for determining whether a given source image can meet requirements after upscaling, or whether the gap is too large for the technology to bridge acceptably.
A Practical Workflow for Upscaling Before Print
1. Establish your target resolution. Calculate the pixel dimensions you need for your print size at 300 PPI. For poster formats, presentation graphics, or display print, confirm the required DPI with the print provider, as some applications accept 150 PPI for large-format output viewed at a distance.
2. Measure the shortfall. Compare your source file dimensions to the target. A 2x shortfall (source is half the required pixel count on each axis) is well within AI upscaling territory. A 4x or greater shortfall is possible but produces diminishing returns.
3. Process through the AI upscaler. Upload to Pixelcut, download the enhanced file. The AI sharpening is applied during the upscale, so you get both the dimension increase and the edge recovery in a single step.
4. Evaluate at 100% zoom. View the upscaled file at actual pixel size in an image editor and assess the edge quality in the most detail-critical areas. This is the most reliable check before sending to print.
5. Convert colour mode if required. AI tools output RGB. Print workflows typically require CMYK. Perform the conversion in your image editor and check for any colour shifts, particularly in saturated hues, before final export.
Format and Colour Notes for Print Output
TIFF is the preferred format for print output: lossless compression, wide support in professional print applications, and no re-compression artefacts. If the print provider accepts PSD or high-quality JPEG, confirm the specific compression settings they require.
For colour conversion, the ICC profile your print provider specifies, often FOGRA39 for coated stock in European print, or a proprietary profile for specialist substrates, should be applied during conversion rather than letting the RIP at the print provider handle it. Colour-managed conversion on your end gives you a reliable preview of the printed result before the file leaves your hands.