Fix Photo Lighting with AI: Shadows and Color

Identify exposure and color problems separately

Brightness and color are different decisions. A portrait can be dark but have correct color; an indoor image can be bright enough but too yellow. Inspect the subject, the lightest area and an object that should be neutral, such as a white shirt. Decide which problem matters first. Splitting the adjustments lets you compare the exposure change before judging a new color balance, rather than accepting several changes together.

Brighten a dark subject without washing out the background

Tell the model which area needs lifting and which bright areas need protection. A request to brighten the entire photograph may weaken the contrast or turn a detailed sky into a pale patch. Keep the direction of the light and some natural shadow. Compare facial features and texture with the source: AI may create plausible detail in areas that were too dark to show clearly.

Brighten the person’s face slightly and lift the jacket shadows. Keep the sky’s visible detail, the original sunlight direction and natural contrast. Do not change facial proportions.

Soften harsh highlights and shadows

Name the highlight or shadow by location. Strong overhead light often creates forehead shine, eye-socket shadows and a dark area below the chin. Reducing those effects does not mean removing all signs of directional light. Ask for a restrained correction and review the face and background together. If the subject appears illuminated by a different light source after editing, return to an earlier version and narrow the request.

Soften the bright forehead highlight and lift the shadow below the eyes a little. Preserve skin texture and keep the light coming from above and to the left.

Correct a yellow, blue or green color cast

Describe the unwanted cast and identify neutral objects to protect. White balance should not make skin look gray or turn a warm room unnaturally blue. Mixed lighting can be harder because daylight and indoor bulbs affect different areas. Start with a neutral correction, then add warmth as a separate preference if needed. Check white clothing, walls and metal reflections as well as the main subject.

Reduce the green cast from the indoor light. Keep the white shirt neutral and the person’s skin tone believable. Preserve the existing exposure and background.

Preserve texture when asking for cleaner lighting

Lighting edits can become retouching edits if the prompt is too broad. Include the details that should remain: pores, freckles, fabric texture, hair and wrinkles. “Clean and perfect” may encourage unwanted smoothing. For a portrait, review eyes, teeth and the outline of the nose after the correction, even when those areas were not named. The edited face should still match the source person.

Compare alternate corrections in History

Keep a balanced correction and test a warmer or moodier version on a separate branch. View the full scene as well as a close crop; contrast that looks pleasant on a face can leave the background flat. Choose the result for its intended use rather than the strongest visible change. A profile photograph, a family print and a product listing may need different amounts of warmth or contrast.

Understand what AI cannot recover faithfully

Completely clipped highlights and very dark regions may contain little usable information. A generated correction can fill those regions with plausible texture, but that is not the same as recovering the original camera detail. Use the best source available and keep realistic expectations for blurred or damaged images. For controlled exposure correction without generated changes, a conventional photo editor with curves and local adjustments may be more suitable.

Frequently asked questions