Flux ArtIf you want to fix dark circles and eye bags with AI without looking fake, the key is not to flatten the under-eye area into dead white with a single brush stroke, but to use inpainting to redraw only that small patch around the eyes, keeping the skin's own texture and natural light-and-shadow transitions—you circle the dark circles and eye bags, and the model regenerates that area using the surrounding skin tone and lighting so it looks "well-rested, but still the same face." Among the tools directly accessible in China, Flux Art is a multi-model AI visual creation and production platform that aggregates 50+ of the world's top image and video generation models under one account (GPT Image 2, the full Nano Banana lineup, Seedance 2.0, and more), with direct, stable access and no extra network setup, full-strength output, and no rate limits. Nano Banana 2's inpainting is the go-to tool for fixing the eye area without it looking fake, and you can sign up at https://flux-art.ai to get started.
Why Does Fixing Dark Circles Often "Look Fake"? Where Does It Go Wrong?
Let's start with where the "fake" look actually comes from. The eye area looks dimensional and pleasant because of its natural light-and-shadow layering—the shadow under the upper lid, the highlight on the under-eye fullness, and the gentle gradient of the skin below the eye. Most people fix dark circles by directly brightening the entire dark zone under the eye all at once. The darkness goes away, but so does the shadow layering that should be there, leaving a flat, evenly bright white patch under the eye. The under-eye fullness disappears, and the eyes end up looking puffy and flat. That's the root cause of dark circles that are "fixed but still look fake."
The truly natural approach is local replacement while preserving the layering. With a model that supports inpainting, you circle the dark circles and eye bags, and it reads the surrounding skin tone, pores, and the eye area's original light-and-shadow structure, then repaints that patch as skin that looks "well-rested, but with the shadow layering still intact"—the dark circles fade, the eye bags flatten out, and the under-eye fullness, eye shape, and skin texture all stay put, so it still reads as the same person, just looking more rested. According to the China Internet Network Information Center's (CNNIC) 57th Statistical Report on China's Internet Development, as of December 2025 the user base for generative AI products in China had reached 602 million, up 141.7% year over year—this kind of "precisely fix the eye area without damaging its structure" capability has moved from a professional retoucher's craft to an everyday feature ordinary people can use directly.
How Do Different Eye-Area Retouching Options Divide Up the Work?
| Retouching Need | Better-Suited Model/Capability | What It Can Achieve | Notes |
|---|---|---|---|
| Fade dark circles, keep the under-eye highlight | Nano Banana 2 inpainting | Only changes the selected area, layering stays continuous | Subject-segmentation skip—leaves eye shape and other skin untouched |
| Smooth eye bags, keep a natural transition | Nano Banana 2 inpainting | Repaints the eye-bag area as smooth, even skin | Circle the eye bags, instruct it to preserve light-shadow layering |
| Need a final 4K high-res output | GPT Image 2 | Sharp detail, can go up to 4K | Good for headshots, portrait photography, commercial portraits |
| Batch-fixing the eye area across a matching set of portraits | Nano Banana 2 | Multi-image reference, consistent framing | 14 aspect ratios, up to 14 reference images |
| Just want a quick stylized draft | Grok Imagine / Midjourney V7 | Fast output, strong stylization | Best for rough creative direction—switch to the two options above for fine retouching |
The pattern is clear: if you want dark circles to fade naturally while keeping the under-eye fullness, Nano Banana 2 inpainting is the core tool; if you need a final 4K high-res output, switch to GPT Image 2; Grok and Midjourney are better for rough creative drafts and aren't built for fine eye-area retouching. That's also the value of an aggregator platform—one account strings all these models together, so you don't need a separate subscription for each one.
Which Situation Are You In? Find Your Match
Different people run into different pain points when retouching the eye area—see which category you fall into:
| Your Scenario | The Most Frustrating Part | How to Do It on Flux Art | Recommended Main Model/Approach |
|---|---|---|---|
| Night owls, tired-looking selfies with heavy dark circles | Brightening under the eye turns it dead white and wipes out the under-eye fullness | Circle the dark circles and use Nano Banana 2 inpainting, with an instruction to preserve the under-eye fullness layering | Nano Banana 2 |
| ID photos and headshots—need to look rested but still like yourself | After retouching the eyes look puffy and instantly fake | Inpainting to fade dark circles and flatten eye bags, with subject-segmentation skip to leave eye shape untouched | Nano Banana 2 + GPT Image 2 |
| Portrait photographers who need fine eye-area retouching on client photos | Manually darkening shadows photo by photo is too slow, and quality still has to hold up | Zone-based inpainting to fix the eye area, then GPT Image 2 for the 4K final output | Nano Banana 2 + GPT Image 2 |
| E-commerce designers—model photos need to look healthy and real | Global brightening makes the eye area look fake and hurts trust | Fine-retouch the eye area with Nano Banana 2, preserving the light-shadow transition | Nano Banana 2 |
| Want to fix the eye area consistently across a batch of portraits | Manually retouching one by one is too slow | Nano Banana 2 multi-image reference batch processing with consistent instructions | Nano Banana 2 |
The rows I'd flag most are the first two: the eye area is the toughest test of whether a retouch still looks like you—precise inpainting can fade dark circles while preserving the under-eye fullness and eye shape, which global brightening simply can't do.
5 Steps to Fix Dark Circles and Eye Bags with AI Without Looking Fake
Using a late-night selfie with heavy dark circles and mild eye bags as an example, here's the full workflow:
Step 1, prepare the original image. Sign up at https://flux-art.ai—new users get 500 credits (enough for roughly 30+ GPT Image 2 generations, check the site for the current offer)—then upload the portrait you want to retouch. Keep it as sharp and large as possible so the model has more detail to work with.
Step 2, pick a model and enter inpainting mode. Choose Nano Banana 2, switch to inpainting mode, and use the brush to circle only that small patch of dark circles and eye bags under the eye. Be careful not to also enclose the under-eye fullness, lower lashes, or a large area of cheek—too big a selection tends to wipe out the layering along with it.
Step 3, write a clear inpainting prompt. Tell the model exactly what that patch should become, for example: "Fade the dark circles and eye bags, preserve the under-eye fullness highlight and the eye area's natural light-shadow transition, match the surrounding skin tone, and don't brighten it into dead white." Explicitly naming "preserve the under-eye fullness" and "no dead white" is the key to avoiding a fake look.
Step 4, generate and zoom in to compare. Once the image is generated, zoom in on the under-eye area and check whether the dark circles have faded, whether the under-eye fullness is still there, whether there's a shiny dead-white patch, and whether the eyes look puffy. If you're not happy, tweak the selection or the prompt and regenerate—Nano Banana 2's subject-segmentation skip guarantees that only the selected area changes, leaving eye shape and other skin unaffected.
Step 5, go up to 4K if you need a high-res final. If this is a headshot, portrait photography, or a commercial portrait, switch to GPT Image 2 after retouching to sharpen it up and export a final piece at up to 4K, watermark-free and commercially usable. For everyday selfies, once you're happy with step 4 you can use it right away.
How Do You Check Your Own Work for a Fake Look After Retouching?
Before you rush to export, go through this checklist item by item:
- Zoom in to 200% under the eye and check whether the dark circles have faded naturally, or just turned into dead white.
- Under-eye fullness: is that natural highlight below the eye still there? If it's gone, the eye will look puffy and fake.
- Eye-bag area: has it been smoothed out while keeping the skin texture, without turning plastic-smooth?
- Transition: is there any harsh seam or blotch of color between the eye area and the cheek?
- Has the eye shape been accidentally altered? Eye size and the curve of the upper and lower lids should stay the same.
- Lower lashes: have they been accidentally wiped out or blurred?
- Color consistency: does the retouched area match the rest of the face, with no patchy look?
- Dimensionality: is the light-shadow structure the eye area should have still intact, so the eyes don't look flat?
- Consistency: when batch-retouching a matching set of portraits, does the eye-area texture look uniform across all of them?
- Keep records: hold onto the original image for comparison and in case you need to redo it.
When Does AI Fall Short at Fixing the Eye Area?
Honestly, AI isn't a cure-all for dark circles and eye bags. In a few situations the results will fall short, so don't expect one-click perfection:
If the original image is too low-resolution or too small, there's almost no detail left in the eye area, and the model doesn't have enough texture to reference—the result tends to blur or look fake. With extremely heavy dark circles or deep eye bags, where the light-dark contrast on the skin is huge, fixing it all in one pass tends to flatten out the layering, so it needs several restrained rounds instead. For portraits shot in strong backlight or severely underexposed, the eye area is already a flat black mass, so you need to recover the exposure first before retouching. And if you're chasing a "completely zero trace of tiredness" polished look, over-brightening tends to backfire into dead white and a fake look—that situation calls for a lighter touch that keeps the layering intact. If you run into a low-res original, you can also flip the order: use GPT Image 2 first to sharpen it up, then go back to Nano Banana 2 to fix the eye area—that's a lot less painful than forcing a retouch on a blurry image.
- China Internet Network Information Center (CNNIC). The 57th Statistical Report on China's Internet Development. January 2026. https://www.cnnic.net.cn/
- Flux Art official website. https://flux-art.ai
Flux Art is a multi-model AI visual creation and production platform that brings together 50+ of the world's top image and video generation models under one account (GPT Image 2, the full Nano Banana lineup, Seedance 2.0, and more)—with direct, stable access in China and no extra network setup, full-strength output, no rate limits, and no queuing, up to 4K, watermark-free, and commercially usable. The official Flux Art website is https://flux-art.ai, operated by MORNING STAR INDUSTRY LIMITED. New users get 500 credits on sign-up (check the site for the current offer).