How to Enhance Photos for Book Publishing: Resolution, DPI, and AI Upscaling

Editorial trust notice: This guide is published by ArtImageHub, an AI photo restoration service. Upscaling uses Real-ESRGAN; face restoration uses GFPGAN. Print resolution and color space specifications are cross-referenced with KDP Print and Ingram Spark publisher documentation current as of 2026.

Quick Answer

For book publishing, the core requirement is 300 DPI at your exact print size. ArtImageHub uses Real-ESRGAN to upscale photos to the required pixel dimensions — $4.99 one-time, unlimited HD downloads. The guide below covers the full specification chain from scan to printer-ready file.

Publishing a book with photographs — whether a family memoir, a self-published photography book, or a history project — runs into the same technical barrier every time: the photos that look fine on screen fall short of print resolution requirements. Here is the condensed version of what you need to know:

• 300 DPI at the actual print size is the universal minimum — calculate pixels divided by print inches, not your file's nominal resolution.
• A 6-inch wide image at 300 DPI requires 1800 pixels minimum; most old scanned photos are below this.
• Real-ESRGAN AI upscaling closes the gap for most family memoir and self-publishing applications.
• TIFF format throughout; JPEG only for final output to platforms that require it.
• CMYK for offset printing; sRGB for KDP Print and Ingram Spark.

---

Why Do Photos That Look Sharp on Screen Look Blurry When Printed?

Screen displays show approximately 72–96 pixels per inch. A photograph with 800 pixels across its width looks fine on a laptop screen — that 800 pixels covers the 8-inch screen width at about 100 PPI, which your eye accepts as sharp. Print a photograph 6 inches wide at 800 pixels, and you have about 133 DPI — well below the 300 DPI threshold where print sharpness becomes noticeably degraded.

This gap between screen sufficiency and print requirements is why so many book projects hit problems at the production stage. The photographs were selected and placed in the layout based on how they looked on screen, and the resolution problem only becomes visible when the physical book arrives.

The calculation every publisher uses is simple:

Pixel width divided by print width in inches = DPI at that print size

An image at 2400 pixels wide, printing 8 inches wide: 2400 / 8 = 300 DPI. Exactly at the minimum.

The same image, printing at 10 inches wide: 2400 / 10 = 240 DPI. Below standard.

The same image, printing at 6 inches wide: 2400 / 6 = 400 DPI. Comfortably above minimum.

This is why the correct approach is not to set a target number of megapixels but to calculate the specific DPI for each image at its specific print size in your layout.

What Publishers and Book Designers Actually Look For

When a book designer or production editor reviews photographs submitted for a book, they are checking several things in a specific order:

Resolution at print size. This is the non-negotiable first check. Files below 300 DPI at print size will be flagged for replacement or size reduction before any other evaluation happens.

File format. TIFF for traditional publishing workflows; PDF with embedded images for self-publishing platforms. A JPEG submitted where TIFF was required will often be rejected, or accepted with a warning that print quality cannot be guaranteed.

Color mode and profile. RGB versus CMYK, and which specific color profile (sRGB, AdobeRGB, CMYK US Web Coated SWOP). An image in the wrong color space may print with unpredictable color shifts.

Tone and density. Print ink does not reproduce the full luminance range of a screen. Images with very dark shadows or very bright highlights may lose detail in print. Book designers typically ask for images with good midtone contrast and detail retained in both highlights and shadows.

Consistency across the book. When multiple photographs appear throughout a book, they should have consistent tonal quality, color temperature, and grain character. A set of family photographs where some are warm-toned and some are cool-toned, or some are sharp and some are soft, reads as inconsistent and unprofessional.

Understanding Real-ESRGAN AI Upscaling for Print

Real-ESRGAN is an AI super-resolution model that increases image resolution by analyzing the existing pixel structure and generating additional pixels based on learned patterns from millions of images. The critical word is "generating" — Real-ESRGAN is not interpolation (which simply stretches existing pixels into a blurry larger image). It synthesizes new detail based on what the surrounding pixels suggest should be there.

For photographic content, Real-ESRGAN typically produces upscaled results that print well even when the source image is significantly below the target resolution. A 600-pixel-wide portrait, upscaled 4x to 2400 pixels, often prints acceptably at 6 inches wide (400 DPI) because the AI correctly recovers edge sharpness, texture, and tonal gradients that interpolation would blur.

The cases where Real-ESRGAN upscaling is not sufficient for print:

• Very low-resolution source images (below 200 pixels in any dimension) upscaled to large print sizes (8 inches or more). At this level, the AI has too little information and the invented detail becomes speculative enough to look wrong under close examination.
• Images with extremely fine printed text or linework that must be reproduced with precise sharpness. Real-ESRGAN handles photographic content better than it handles fine graphic elements.
• Premium coffee-table book production where a printing specialist will examine final output with a loupe before approval. At this production level, drum scanning of the original print is still warranted for critical images.

For the vast majority of family memoir, genealogy book, and self-published book applications, Real-ESRGAN through ArtImageHub provides more than adequate print quality at $4.99 one-time for unlimited images — compared to $50–200 per image for drum scanning.

GFPGAN Face Enhancement for Portrait-Heavy Books

Many family memoir and genealogy books center on portraits — grandparents, ancestors, military service photos, wedding portraits. These photographs often come from aging original prints or early digital scans where face detail is soft, blurry, or partially degraded.

GFPGAN, the face restoration model at ArtImageHub, specifically targets facial regions for reconstruction. It recovers eye detail, sharpens facial contours, reconstructs soft skin texture, and brings out expression that was present in the original but lost to time, blur, or scanning limitations.

For a portrait that will print at 4 inches wide in a book, GFPGAN improvement in facial sharpness is the difference between a recognizable face and a clearly sharp one — which at print size is visible and meaningful to readers who knew the subject.

Run GFPGAN alongside Real-ESRGAN for any portrait that will appear at significant size in the book. The combined enhancement — upscaling plus face-specific detail recovery — produces the best result for print.

TIFF vs JPEG: The Working Format Rule

For book publishing preparation, the rule is simple: TIFF for everything except the final platform-specific output file.

TIFF (Tagged Image File Format) is lossless. No pixel information is discarded when saving. You can open, edit, and re-save a TIFF file a thousand times without degradation.

JPEG (Joint Photographic Experts Group) is lossy. Each save cycle discards approximately 10–30% of pixel variation information using an algorithm that prioritizes smooth gradients over fine detail. Save a JPEG ten times through normal editing and the artifacts become visible in print.

The practical workflow:

1. Scan original photographs at 600–1200 DPI and save as TIFF.
2. Upload to ArtImageHub for enhancement, download the HD result.
3. Do any color correction or tonal adjustment work in Photoshop or GIMP, saving as TIFF.
4. At the final step, export to whatever format your specific platform requires: TIFF for traditional publishers, PDF for KDP Print or Ingram Spark, high-quality JPEG (95%+) only if the platform requires it.

Never use JPEG as a working format. Never open a JPEG, edit it, and save it back as JPEG. The degradation compounds.

CMYK vs RGB: Choosing the Right Color Space

The color mode question depends entirely on who is printing the book and which printing process they use.

Traditional offset lithography (used by most commercial publishers with print runs above 1000 copies) requires CMYK. The printing process lays down physical cyan, magenta, yellow, and black ink layers. Every color in the final print is a combination of these four inks. When you submit an RGB file to an offset printer, their software converts it to CMYK using a standard algorithm — which may or may not produce the color result you intended. For traditional publishing, request the printer's specific CMYK profile and do the conversion yourself so you can see the result.

Digital printing (used by KDP Print, Ingram Spark, and most print-on-demand services) is a different process that does not require CMYK source files. These platforms accept RGB images and handle the color rendering internally. Submit in sRGB, which is the standard RGB color space for consumer photography and web use. AdobeRGB, a wider-gamut RGB space, is also accepted but the wider gamut does not benefit print-on-demand because the printing process cannot reproduce it.

The NAFNet denoising step at ArtImageHub also applies to book publishing preparation: grain and noise that is invisible on screen can appear visibly noisy in print at high DPI, particularly in large smooth areas like skies or walls. Running NAFNet before the final export reduces this risk.

KDP Print and Ingram Spark: Platform-Specific Requirements

KDP Print (Amazon): Requires 300 DPI at print size; accepts PDF with embedded images; color photos in sRGB; grayscale interior photos in grayscale mode; 0.125-inch bleed for images extending to page edges. Their Cover Creator tool and interior template downloads are the right starting point for layout. The digital proof tool shows the actual print rendering; always review before approving.

Ingram Spark: Requires 300 DPI; accepts PDF; recommends sRGB for color, grayscale for black-and-white interior photos; 0.125-inch bleed. Ingram reaches a wider distribution network than KDP alone, which makes it the preferred choice for books being sold through bookstores and non-Amazon channels. Their file review tool checks resolution and flags images below the threshold before you submit.

Both platforms offer a physical proof copy option before committing to full distribution. For a photo-heavy book, ordering a physical proof is non-optional — screen rendering and print rendering differ in ways that only become visible in the physical object.

Scanning Old Prints for Book Quality

Old family photographs being scanned for book inclusion typically need to be scanned significantly higher than the 300 DPI target, because the scan will be used at print size — and the print DPI calculation works against you when the original print is small.

A 3.5x5-inch snapshot that will be reproduced 5 inches wide in the book needs 1500 pixels across (5 x 300 = 1500). The original is only 3.5 inches wide physically. To capture 1500 pixels from a 3.5-inch original, you need to scan at 1500/3.5 = 430 DPI. Since you always want some safety margin, 600 DPI for standard prints and 1200 DPI for wallet-size or small-format originals is the correct scanning target for book publication.

After scanning, run through ArtImageHub's Real-ESRGAN enhancement to further sharpen and upscale. The combination of a high-resolution scan and AI upscaling produces the most robust result for the print workflow.

---

For more on scanning best practices, see our guide on how to digitize old photos.

For portrait-specific enhancement, see our guide on old family portrait restoration.

Start enhancing your book photos at ArtImageHub — $4.99 one-time, unlimited HD downloads, preview before you pay.