Ink wash is one of the hardest media to bring into a digital file. The whole point of the technique is the way water carries pigment into pale, uneven gradients, and that is exactly what a scanner struggles to hold onto. Most people end up with a file where the beautiful soft greys have collapsed into flat patches, and the paper texture has picked up a grey cast that was never there in reality.
If you work with wet media like this, a sensible AI image transformation pass can recover tone range that a straight scan throws away, as long as you prepare the input properly first.
Scan for tone range, not for contrast
The instinct is to crank contrast while scanning so the blacks look strong. Resist it. Scan flat, close to what you see, and keep the histogram bunched in the middle third. You want maximum tonal information in the file, even if it looks weak and grey on screen. Everything downstream depends on having that range available.
Scan at 600 dpi minimum if you plan to print larger than the original. Scan in 16-bit greyscale or full colour even for a black-only piece. Colour capture picks up the warm tone of rice paper that pure greyscale discards, and you can always decide later whether to keep it.
Deal with the paper before you touch the ink
Rice paper and cotton rag both have visible fiber and a slightly uneven surface. That texture is a feature when you print it and a problem when it is uneven. If your scan shows brighter patches from inconsistent lighting, correct those first with a soft, wide radius adjustment. Do not sharpen yet. Sharpening a gradient adds noise that will be very hard to remove after any further processing.
If the paper has a yellow cast you want removed, sample it deliberately from an unpainted corner and neutralise that value globally. That single step fixes most of the colour problems people try to solve with saturation sliders.
Where a transformation pass earns its keep
The specific thing processing does well here is rebuild gradient continuity. In a wash, pigment edges are not hard lines; they are soft blooms where the water pushed the ink outward. A scan often clips those blooms into visible stair-stepping. A careful pass smooths that out, and the result prints with the graduated feel the painting actually has.
Keep the pass restrained. Push it too far and you get a suspiciously clean image where every wash looks machine-made, which defeats the purpose. Compare against the physical painting, not against the screen. Hold the print next to the original under the same light and see whether the soft edges behave similarly.
Sharpen for paper, not for pixels
Screen sharpening and print sharpening are different jobs. What looks crisp on a monitor often prints muddy, and what looks slightly soft on screen prints cleanly. Set your sharpening at final output size, not before resizing. For ink wash specifically, use a modest amount with a larger radius rather than a small aggressive radius. Fine detail was never the point of the medium.
Test on the real substrate early
Do a small proof print before committing to a full run. Ink wash files that look perfect on a screen can print too dark because absorbed light on paper behaves differently than emitted light on a monitor. One A5 proof on textured stock will tell you more than any amount of monitor adjustment, and it costs almost nothing compared to reprinting a run.
Keep a record of every setting you used per piece. Wash paintings vary enormously depending on how much water was on the brush, and a setting that rescues a heavy dark piece will wash out a light airy one. A short note with the file saves you re-solving the same problem in three months.
None of this is complicated, but the order matters: flat scan, neutralise paper, then process, then sharpen at output size, then proof. Do it in that order and your transformation workflow will hold onto the thing that makes your painting worth printing in the first place, which is the way the water moved.