AI Image Editor: Edit Any Photo with AI by Typing the Change
VdoBloom’s AI image editor changes an existing photo from a plain-language instruction. Upload an image, type what should change (“replace the background with a beach at sunset”, “make the jacket red”, “remove the cup”), and the AI returns an edited copy that keeps the rest of the picture intact.
It works as a full AI photo editor in the browser: 41 editing models on one page, including GPT Image 2.5, Nano Banana 2.1 and Pro, Seedream 4.5, 5 and 5.0 Pro, Flux Kontext, Qwen Image, Kling IMAGE O1 and Luma UNI-1 Max. Depending on the model you can combine up to 16 reference images, paint a mask over the exact area to change or erase, expand the canvas in any direction, vectorize an image, or export at 1K, 2K or 4K.
It is free to try: new accounts get 10 credits, and edits start at 1 credit (Muse Image), 3 credits (Nano Banana 2.1 or GPT Image 2.5 at 1K) or 4 credits (Seedream 5 Flash). You pay once per edit, however many reference images you upload. Credit packs start at 300 credits for $19.50 one-time, and one-time credits never expire.
How it works
- 1
Upload your photo
Add any image file up to 10 MB. Multi-image models take extra reference photos too, for example a product shot plus a background, or a face plus an outfit.
- 2
Type the edit
Describe the result you want in one or two sentences. Name the object or area and what it should become. For erase or targeted edits, paint over the area in the browser instead of making a mask file.
- 3
Choose a model and resolution
Leave it on Auto, or pick a model for the job. The credit cost of your choice is shown on the button before you run it.
- 4
Run, compare and download
The edited image appears in a few minutes. Download it, or use it as the next input and chain another edit on top.
Why use this tool
Edit with words, no layers
Swap backgrounds, change colours and clothing, add or remove objects, relight a scene or restyle the whole photo by describing the change.
Up to 16 reference images
GPT Image 2.5 takes 16 images in one edit; Seedream 5 Lite, Nano Banana 2 and 2.1 take 14; many other models take 8 to 10. Use them to merge people, products and scenes into one image.
Paint a mask in the browser
FLUX Erase removes whatever you paint over. Qwen Image 2.1 has a mask mode that changes only the painted area and leaves the rest untouched.
Expand the canvas (outpainting)
FLUX Outpainting adds up to 512 px on each side, so you can turn a portrait into a landscape or give a product shot more room.
1K, 2K and 4K output
Nano Banana Pro, Nano Banana 2 and 2.1, GPT Image 2 and 2.5, and Riverflow 2.0 Pro edit at up to 4K; Seedream 4.5 and 5 Lite have a High (4K) tier.
Refund if an edit fails
If the model provider fails or its content filter refuses an edit, the credits go back to your balance automatically.
Is the AI photo editor free?
You can start free. A new VdoBloom account gets 10 credits with no card required. With those you could run 10 edits on Muse Image (1 credit each), 3 on Nano Banana 2.1 or GPT Image 2.5 at 1K (3 credits each), or 2 on Seedream 5 Flash (4 credits each).
You are charged once per edit request, not per reference image, so a 10-image composite costs the same as a one-image edit on the same model. When you need more, the Lite pack is 300 credits for $19.50 one-time or $15 a month. One-time credits never expire; subscription credits reset each cycle.
Read more: See all credit packs · How much does an AI image cost?
Which AI editing model should I use?
Every model in the editor, with the most images it accepts per edit, its output options and the credits one edit costs. “Basic/High” is a quality tier; where a model has resolution tiers the cost is listed per tier.
| Model | Max input images | Output | Credits per edit |
|---|---|---|---|
| Auto (picks the engine for you) | 9 | Basic 2K / High 4K | 7 |
| Muse Image (Meta) | 10 | 2K | 1 |
| Boogu Image Edit | 1 | Standard | 3 |
| GPT Image 2.5 Flare | 16 | 1K / 2K / 4K | 3 / 5 / 8 |
| GPT Image 2.5 Sunburst | 16 | 1K / 2K / 4K | 3 / 5 / 8 |
| GPT Image 2 | 10 | 1K / 2K / 4K | 3 / 5 / 8 |
| Nano Banana 2.1 | 14 | 1K / 2K / 4K | 3 / 4 / 6 |
| Qwen Image 2.1 Edit (mask, transparent output) | 10 | 1K / 2K | 3 / 5 |
| Seedream 5 Flash | 10 | Basic 1K / High 2K | 4 |
| WAN 2.7 Image | 9 | 1K / 2K | 4 |
| FLUX.2 [dev] | 4 | Standard | 4 |
| Nano Banana Edit | 10 | Standard | 5 |
| Grok Imagine | 1 | Standard | 5 |
| Qwen Image 3.0 Edit | 10 | 1K / 2K | 5 |
| Seedream 5 Lite | 14 | Basic 2K / High 4K | 6 |
| Qwen2 Image Edit | 10 | Standard | 6 |
| P-Image-Edit | 10 | Standard | 6 |
| Qwen Image 3.0 Pro Edit | 10 | 1K / 2K | 6 / 7 |
| Seedream 4.5 Edit | 10 | Basic 2K / High 4K | 7 |
| Flux Kontext Pro | 1 | Standard | 7 |
| Nano Banana 2 Lite | 10 | Standard | 8 |
| Flux Kontext Max | 1 | Standard | 10 |
| GPT Image 1.5 | 10 | Medium / High quality | 10 |
| Qwen Image 2.0 | 3 | Standard | 10 |
| Grok Imagine Pro | 1 | 1K / 2K | 10 |
| Riverflow 2.0 Fast | 10 | 1K / 2K | 10 |
| FLUX Erase (painted mask) | 2 (photo + mask) | Standard | 10 |
| Seedream 5.0 Pro | 10 | Standard / High | 10 / 19 |
| Flux 2 Pro | 8 | 1K / 2K | 12 |
| Qwen Image 2.0 Pro | 3 | Standard | 12 |
| FLUX Outpainting (expand canvas) | 1 | Up to 2048 px | 12 |
| Kling IMAGE O1 | 10 | 1K / 2K | 14 |
| Luma UNI-1 Max | 8 | 1K / 2K | 14 / 20 |
| Recraft V4 Vector (vectorize) | 1 | Vector | 14 |
| Nano Banana 2 | 14 | 1K / 2K / 4K | 15 |
| Nano Banana Pro | 8 | 1K / 2K / 4K | 18 / 18 / 24 |
| WAN 2.7 Image Pro | 9 | 1K / 2K | 18 |
| Riverflow 2.0 Pro | 10 | 1K / 2K / 4K | 18 / 18 / 27 |
| Nano Banana Special (two-pass) | 8 | Standard | 20 |
| GPT Special (two-pass) | 10 | Standard | 20 |
| Recraft V4 Pro Vector | 1 | Vector | 24 |
Read more: GPT Image 2.5 Sunburst edit · Nano Banana 2.1 · Seedream 5 Flash edit · Qwen Image 2.1 edit · All models
Edit types: what each tool in the editor does
Most edits only need a prompt. A few jobs have their own model with dedicated controls:
- Prompt edits (any model): change backgrounds, colours, clothing, weather, lighting or style, and add or remove objects.
- Multi-image composites: put a person into a new scene, place a product in a lifestyle photo, or merge two photos into one, using up to 16 reference images.
- Erase with a painted mask (FLUX Erase): upload the photo, paint over what should disappear, and the gap is filled to match the surroundings.
- Targeted inpainting (Qwen Image 2.1 mask mode): paint the area, describe what goes there, and only that area changes.
- Outpainting (FLUX Outpainting): set 0 to 512 px per side in 16 px steps; the final image must stay at or under 2048 px.
- Transparent output (Qwen Image 2.1): return the edit as a PNG with a transparent background.
- Vectorize (Recraft V4 Vector and V4 Pro Vector): convert a raster image into vector artwork.
Read more: How to use FLUX Erase · How to use FLUX Outpainting · Merge two photos into one · Transparent backgrounds with Qwen Image 2.1
What does an AI photo edit cost? Worked examples
Credits are charged once per edit. Here is what common jobs cost and how many you get from the free credits or a 300-credit Lite pack.
| Job | Model and setting | Credits each | Edits from 10 free credits | Edits from 300 credits |
|---|---|---|---|---|
| Swap a background | Nano Banana 2.1, 1K | 3 | 3 | 100 |
| Quick restyle or colour change | Seedream 5 Flash, Basic | 4 | 2 | 75 |
| Combine 10 product photos into one scene | GPT Image 2.5 Sunburst, 2K | 5 | 2 | 60 |
| Remove an object with a painted mask | FLUX Erase | 10 | 1 | 30 |
| Widen a portrait to landscape | FLUX Outpainting | 12 | 0 | 25 |
| Print-ready 4K edit | Nano Banana Pro, 4K | 24 | 0 | 12 |
What people edit with AI
Typical jobs for the editor:
- Online sellers: clean or replace product backgrounds, place products in lifestyle scenes, make colour variants.
- Creators: change outfits, locations and lighting for thumbnails and social posts.
- Marketers: turn one hero shot into several ad sizes with outpainting, and add headline text to an image.
- Real estate and interiors: restyle a room, change wall colours or add furniture from a reference photo.
- Everyday photos: remove strangers or clutter, fix a dull sky, or restyle a picture as anime, watercolour or 3D.
- Designers: vectorize a sketch or logo, or export a transparent PNG for a layout.
Tips for better AI edits
Clear, single-purpose instructions give the most predictable results.
- Change one thing per edit, then chain edits: background first, then lighting, then small details.
- Name what to keep as well as what to change: “keep the face and pose, change the jacket to navy wool”.
- Use a painted mask (FLUX Erase or Qwen Image 2.1) when the prompt keeps changing parts you wanted left alone.
- For composites, upload the most important image first and refer to images by order in the prompt.
- Test on a low-cost model at 1K, then re-run the edit you like at 2K or 4K.
- Qwen Image 3.0 Edit has an 800-character prompt limit; most other models accept much longer prompts.
Limits of the AI image editor
Worth knowing before you start:
- Files must be images of 10 MB or less each.
- Some models edit one image at a time (Flux Kontext, Grok Imagine, Boogu, Recraft Vector, FLUX Outpainting).
- Photos of real people need their consent, and you confirm this before uploading. Edits that break the content policy are refused and refunded.
- Outpainting stays within 2048 px in total, so very large photos need smaller expansion values.
- AI edits regenerate pixels, so fine details such as small text or jewellery can shift. Mask the area or chain smaller edits to limit drift.
Read more: Why AI refuses to edit some photos
AI image editor or AI image generator: which do I need?
Start here when you already have a photo. If you only have an idea, the AI Image Generator creates a new image from a text prompt with 30 models. Both use the same credits, and a generated image can be uploaded here for further edits.
Read more: AI Image Generator · AI Image Upscaler
Frequently asked questions
Is the VdoBloom AI image editor free?
You can try it free with the 10 credits every new account gets. Edits start at 1 credit with Muse Image, 3 credits with Nano Banana 2.1 or GPT Image 2.5 at 1K, and 4 credits with Seedream 5 Flash. After that, credit packs start at 300 credits for $19.50 one-time.
How do I edit an image with AI?
Upload your photo, type what should change in plain language, pick a model (or leave it on Auto), and click Transform Image. The edited copy appears in a few minutes, and you can download it or use it as the input for another edit.
How much does one AI photo edit cost?
Between 1 and 27 credits per edit depending on the model and resolution, and you pay once per edit no matter how many reference images you add. Auto costs 7 credits; Nano Banana 2.1 costs 3, 4 or 6 credits at 1K, 2K or 4K; Nano Banana Pro costs 18, or 24 at 4K.
How many images can I upload in one edit?
Up to 16 with GPT Image 2.5 Flare or Sunburst, 14 with Seedream 5 Lite, Nano Banana 2 and Nano Banana 2.1, and 8 to 10 with most other models. Flux Kontext, Grok Imagine, Boogu and the Recraft vector models take one image.
Can the AI remove an object from a photo?
Yes. Choose FLUX Erase, upload the photo and paint over the object in the browser; the painted area is filled to match its surroundings. You can also ask any model in words, for example “remove the person on the left”.
Can I edit only part of an image?
Yes. Qwen Image 2.1 has a mask mode: paint the area you want changed, describe what should go there, and the rest of the photo stays as it was. Seedream 5.0 Pro also handles precise local changes from the prompt.
Can AI extend or expand a photo?
Yes. FLUX Outpainting adds new content on any side, up to 512 px per side, with a final size of up to 2048 px. It is useful for turning a vertical photo into a horizontal one or adding space around a product.
Can the AI image editor output 4K?
Yes, on Nano Banana Pro, Nano Banana 2 and 2.1, GPT Image 2 and 2.5, and Riverflow 2.0 Pro, plus the High (4K) tier of Seedream 4.5 and 5 Lite. Other models output 1K or 2K. For an extra boost after editing, run the result through the AI Image Upscaler.
Why did the AI refuse to edit my photo?
Each model provider runs its own content filter and may refuse edits involving real people, copyrighted characters or sensitive content. When that happens on VdoBloom, the credits are refunded and you can adjust the prompt or try another model.
Is an AI image editor the same as an AI photo editor?
Yes. Both names describe a tool that changes an existing picture using AI. This editor works on photos, illustrations, screenshots and AI-generated images alike.