Why Do AI-Restored Old Photos Print Like Oil Paintings? A Practical Guide to Paper Stocks, Dot Gain, and Sharpening

The main reason ultra-high-resolution old photos generated by AI look dark and muddy when printed is 'dot gain' caused by paper ink absorption, which amplifies the overly sharpened edges reconstructed by the AI

To bridge this gap, I usually advise clients to apply the 'MINDS Three-Step Prepress Check' framework: ① Identify the ink absorption characteristics of the paper stock, ② Restrict the Total Area Coverage (TAC) in dark areas, and ③ Dial back artificial sharpening by one level

Dot Gain: A physical phenomenon where ink dots printed on paper expand beyond their original file settings due to paper fiber absorption and printing pressure. This is particularly severe on uncoated paper (such as wood-free paper) and tends to swallow up details in the shadows

Over the past few months, we have worked on several brand commemorative books where clients provided their only copy of an out-of-print physical catalog and asked designers to use AI software to generate details out of thin air for low-resolution old photos

They look incredibly sharp on screen, but sending them directly to the printer for reproduction will inevitably result in heavy patches of solid ink

AI restoration works by aggressively boosting micro-contrast, inserting extremely light and dark pixel blocks adjacent to each other to create the illusion of sharpness

When these dense color blocks meet highly absorbent paper like wood-free paper, physical ink bleeding causes adjacent dots to merge

Tiny dark dots originally meant to define hair strands or suit fabric textures suddenly expand into solid black patches, making the print look like a cheap, plastic oil painting

On uncoated paper, a dot gain of 20% to 25% is a very common physical limit. This means a 75% dark gray on screen will end up close to 100% solid black in print

To rescue shadow details, the first step is to control the Total Area Coverage (TAC)

Images exported from AI software are mostly in RGB format. When converted to CMYK without proper adjustments, the shadow areas often spike to a heavily saturated ink level of 300% or even 340%

When handling mid-to-high-end fully customized commercial printing, the first thing the press operators at MINDS do upon receiving a file is inspect the ink density

If wood-free or lightly coated paper is used, I will request the design team to limit the TAC to under 240%

The practical approach is to open image-editing software and manually define the CMYK values of the darkest point, reducing the proportions of Cyan, Magenta, and Yellow, and relying on Black (K) to support the shadow structure

This significantly reduces the amount of wet ink on the paper, allowing the image to retain its restored gradations even after drying

With ink coverage sorted out, the next step is addressing the side effects of over-sharpened edges

Just as prepress practice requires trapping to prevent paper stretching from causing thin lines or reversed-out text to show distracting white gaps, we must proactively intervene on the harsh edges generated by AI

My habit is to first apply a slight Gaussian blur to the AI-restored image to soften the harsh, over-processed pixels

Then, based on the final output size and screen ruling (such as 150 lpi or 175 lpi), I apply Unsharp Mask (USM) sharpening again

This process essentially strips away the artificial details created by the AI for screen display, replacing them with a true sharpness tailored for halftone printing

If your team is building this kind of automated prepress conversion workflow but running into roadblocks, consider consulting with the MINDS Knowledge Academy advisory team to integrate these physical variables directly into your SOPs

The final pitfall lies in the logic of color mode conversion

Just like when an AI-generated image is converted to grayscale for two-color risograph printing and turns into a muddy mess, converting vintage tones of old photos from RGB to CMYK also requires precise color separation

If the rich, nostalgic brown produced by AI is left to the system's default distribution across all four CMYK colors, it will print with a muddy, dirty look

Experienced prepress professionals use GCR (Gray Component Replacement) technology to extract the muddy CMY gray components and replace them with clean black ink

This not only maintains the warm tone of the old photo, but also prevents edge blurring caused by misregistration during high-speed printing

・Sharp details generated by AI will turn into muddy solid black on absorbent paper due to dot gain

・For uncoated stocks like wood-free paper, make sure to keep the Total Area Coverage (TAC) in shadow areas under 240%

・Soften the over-processed edges from the AI, and re-apply USM sharpening based on the printing screen ruling

・Make good use of GCR (Gray Component Replacement) to substitute clean black ink for muddy CMY gray, preserving image depth

The physical limitations of traditional printing are not obstacles to AI, but rather a touchstone for evaluating the maturity of image processing workflows. For printers and software developers alike, integrating paper absorption coefficients and dot gain into automated image-processing pipelines is the only way to truly bring the stunning quality of digital restoration to physical paper