PIXQUILL
Free credits on the house

10 credits on sign up, 5 credits daily.

NEW

Flux 3 Klein and Kling 4.0 are live. Free credits to start.

✦ AI IMAGE EDITOR

MARK THE SPOT, EDIT THE PHOTO

Mark the exact spot, say the change in plain words. The rest of the photo stays put.

DEMOWatch the mark decide
Street portrait of a person in a red jacket before AI editingThe same street portrait with the jacket edited to deep green, everything else unchanged

PROMPT ▸make the jacket in [Box 1] deep green

TRY IT NOW

BASE PHOTO

POINT, BOX OR BRUSH: YOU NAME THE TARGET
MARK
ONLY THE MARKED AREA CHANGES
EDIT
EVERY RESULT BECOMES THE NEXT BASE
ITERATE
EVERY VERSION KEPT, UP TO 2K OUTPUT
SAVE
SAMPLES

ONE MARK, ONE SENTENCE

Everyday edits, each driven by one mark and one prompt. The rest of the photo stays put.

Room interior with a plain sky visible through the windowThe same room with a sunset sky edited in outside the window

PROMPT ▸swap the sky outside the window to a sunset

THE SKY SWAP

One brush stroke over the glass, one sentence. The room, the frame and the reflections hold still.

Product photo of a light-colored mug on a tableThe same product photo with the mug edited to matte black

PROMPT ▸make the mug at [Point 1] matte black

THE COLOR CALL

A point is enough for small targets. The model reads the object, not the pixels around it.

Street storefront with a plain painted signThe same storefront with the sign edited into a glowing neon sign

PROMPT ▸turn the storefront sign in [Box 1] into a neon sign

THE SIGN FLIP

Box the sign and restyle it in place. Perspective and lighting follow the street.

TRY IT NOW
ROUND BY ROUND

EVERY RESULT IS THE NEXT BASE

Two rounds on one photo. The output of round one is the input of round two, and every version stays in the strip.

  1. Base street portrait before any AI editV0 · BASE

    The upload. Marks and prompts attach here.

  2. Version 1: the jacket edited to deep greenV1

    make the jacket in [Box 1] deep green

  3. Version 2: the sky edited to a sunset on top of version 1V2

    swap the sky to a sunset

TRY IT NOW
USE CASES

ONE EDITOR, EVERY JOB

PRODUCT SHOTS

Recolor packaging, swap a prop, restyle the backdrop, no reshoot. Try: 'make the box in [Box 1] matte black'.

PORTRAITS

Fix the shirt color, tame a busy backdrop, warm the light on one side. Try: 'soften the background behind the subject'.

INTERIORS AND LISTINGS

Test a wall color or swap the sky through the window before anyone lifts a brush. Try: 'paint the wall in [Box 1] sage green'.

SOCIAL AND THUMBNAILS

Punch up a flat frame or restyle a sign so the series matches. Try: 'make the title on the sign bright yellow'.

MARKETING AND ADS

Iterate one hero image per channel instead of booking another shoot. Try: 'swap the background to a studio gradient'.

MOOD AND GRADE

Describe a mood and regrade the scene around your subject. Try: 'grade the scene like faded 70s film'.

WHY THE MARK

WORDS ALONE GUESS. MARKS DECIDE.

Text-only editors guess what 'the sign' means. This one asks you to point first.

TEXT-ONLY EDITING (PROMPT-ONLY TOOLS)

  • 'The flower' in a field of flowers is anyone's guess
  • Ambiguous prompts reroll the whole photo, not the spot
  • Every retry is a fresh run, a fresh wait, a fresh cost
  • Unmarked areas drift while the model regenerates

MARKED EDITING (PIXQUILL)

  • The point, box or brush stroke names the target exactly
  • Only the marked area changes; the rest of the photo holds still
  • Each result becomes the next base, so retries branch, not restart
  • The exact price of the round shows before you generate
THE STUDIO

EVERY TOOL, ONE BALANCE

Editing is usually one step of the job. The rest of the studio shares your uploads and credits.

OPEN →Image AgentText and references in, finished images out. The studio's front door.OPEN →Video AgentPrompts, frames and references in, finished videos out. 10+ models, up to 4K.OPEN →Image UpscalerUpscale small or soft images to crisp 4K.OPEN →Object RemoverBrush over what shouldn't be there. It won't be.OPEN →Background RemoverCut the subject out and get a transparent PNG back.OPEN →Clothes ChangerChange the outfit in a portrait without touching the person.
HOW IT WORKS

LOAD, MARK, EDIT

  1. DROP A PHOTO HERE, OR CLICK TO BROWSE

    PNG / JPG · UP TO 5 MB

    LOAD THE PHOTO

    Any JPG or PNG up to 5 MB. It lands on the canvas as V0, the base of every round.

  2. Street portrait of a person in a red jacket before AI editing

    MARK THE SPOT, THEN DESCRIBE IT

    MARK AND DESCRIBE

    Drop a point, drag a box or brush an area, then write the change: 'make the wall in [Box 1] brick red'.

  3. Street portrait with the jacket edited to deep green after one round

    street-portrait-v2-2k.jpg✓ 2K · JPG · V2

    EDIT IN ROUNDS

    Runs take 30 to 60 seconds. The result comes back as the new base; branch off any earlier version any time.

TRY IT NOW
FEATURES

A REAL EDITING DESK

POINT, BOX AND BRUSH

On Seedream 5.0 Pro, the default, marks compile into interactive-edit coordinates. The model edits where you pointed, not everywhere.

CITABLE MARKS

Every mark is numbered and stable: '[Box 1]', '[Point 2]'. Cite them in the prompt and the target stops being a guess.

VERSION STRIP

The original plus every round, one click apart. Jump back three versions and branch from there.

DRAFT AT 1K, FINISH AT 2K

Every round is priced by model and tier, from 2 credits up. Explore cheap, finish at 2K. The console shows the price first.

ZOOM TO 400%

Pixel-peep with the pan tool and zoom slider; marks keep their precision at any magnification.

HONEST BILLING

The round's price is on screen before you generate; failures on our side refund automatically.

TIPS

EDIT LIKE A PRO

  • 01One change per round beats a paragraph of changes. You can always iterate.
  • 02Cite your marks in the prompt: 'the jacket in [Box 1]' leaves nothing to guess.
  • 03Brush for areas (a stain, a sky), boxes for objects, points for tiny targets.
  • 04Draft at 1K while exploring; switch to 2K for the final round.
  • 05A wrong turn is cheap: jump back in the version strip and branch off the last good version.
  • 06Need an object gone with the scene rebuilt behind it? That is the Object Remover's job, not this one's.
ABOUT

WHY POINTING BEATS PROMPTING

Text-only editing has an ambiguity problem: 'the sign' in a street of signs, 'the flower' in a field. The AI Image Editor removes the guessing. You mark the target on the canvas, the mark becomes coordinates in the model's native interactive-edit format, and the change lands where you aimed.

Because every result returns as the next base photo, editing becomes a conversation instead of a lottery. Draft at 1K, finish at 2K, and branch off any earlier version when a direction stops working.

The default engine is Seedream 5.0 Pro, the one model that reads point, box and brush coordinates natively. The picker holds 20+ alternatives from Google, OpenAI, Alibaba, Ideogram and more for prompt-driven edits. The editor is built for changing what is in the photo, not for wiping it: when the job is pure removal or a clean cutout, the Object Remover and the Background Remover share your uploads and your credits.

FAQ

STRAIGHT ANSWERS

▸ How do I edit a photo with AI here?

Upload the photo, mark the spot with a point, box or brush stroke, describe the change in plain words, and generate. A round usually takes 30 to 60 seconds, and the result lands back on the canvas as the new base version.

▸ Why mark the spot myself?

Because you know which sign, which flower, which jacket. One mark removes the ambiguity that makes text-only editors reroll the whole photo.

▸ How do the points and boxes work?

With the default Seedream 5.0 Pro, marks you drop on the canvas are converted to normalized coordinates in its interactive-edit format and appended to your prompt automatically. You can also cite them in words ('[Box 1]', '[Point 2]'), since the numbering is stable. The other models in the picker edit from your prompt alone.

▸ What can I edit?

Anything you can point at and describe: recolor objects, swap skies and backdrops, restyle signs, change materials, regrade the mood. One area per round works best, and rounds stack.

▸ What can't it do?

Full-scene rewrites (a new photo wearing the old one's outfit) are the Image Agent's job, and pure object removal with the scene rebuilt belongs to the Object Remover. The editor is strongest when most of the photo should stay exactly as it is.

▸ Is the AI Image Editor free to try?

Yes. New accounts start with free credits and the daily check-in adds more, so you can edit real photos before spending anything. The console shows the exact price of every round upfront.

▸ What does each round cost?

One result per round, priced by model and resolution: from 2 credits on Flux 3 Klein to 10 for Seedream 5.0 Pro at 2K, with flagships higher. The console shows the exact price before you generate, and failed rounds on our side refund automatically.

▸ What files can I upload, and what do I get back?

Upload any JPG or PNG up to 5 MB. Results come back at 1K or 2K. The format follows the model: JPG or PNG where the model offers the choice.

▸ How long does a round take?

Usually 30 to 60 seconds. The result lands back on the canvas as the new working version, and the version strip keeps every round one click away.

▸ Can I undo marks?

Yes. Undo and redo cover brush strokes, points, boxes and clears, ten steps deep. The trash button clears everything after a confirmation.

▸ Which model powers the editor?

Seedream 5.0 Pro is the default, the one model with native interactive editing: it reads point, box and brush coordinates directly. The picker also holds 20+ image-to-image models, from Nano Banana to GPT Image to Flux to Ideogram, for prompt-driven edits at their own prices. Output runs at 1K or 2K, and the console prices every round before you generate.

▸ Does it work on my phone?

Yes. Nothing to install: upload from your phone, the edit runs on our servers, and the result downloads back.

▸ Can I use the edited photos commercially?

Photos you generate are yours to use, including in commercial projects, within our terms. Rights in anything you upload stay where they were, so only process photos you own or have permission to use. Details live on the Copyright page.

READY

YOUR NEXT EDIT IS ONE SENTENCE

Upload a photo, mark the spot, and describe the change. New accounts start with free credits.

EDIT A PHOTOSEE PRICING