8 Best AI Tools for Turning Photos Into Paintings in 2026

Key takeaways

• Melius is the best AI tool for turning photos into paintings overall: its agent writes the painting prompt, sends the photo to models such as Nano Banana Pro and GPT Image 2.5 and keeps every step editable.

• Krea takes up to four reference paintings, each with its own strength slider, and its upscaler reaches 22K for large prints.

• Adobe Firefly uses your photo as a composition reference with a Strength slider, so you set how closely the painting follows it.

• ChatGPT and Gemini handle a quick one-off: upload the photo and describe the medium in plain words.

• For a print, plan on about 2,400 x 3,000 pixels for 8 x 10 inches at 300 ppi, and upscale the finished painting rather than every draft.

How we ranked

Turning a photo into a painting is mostly a balancing act: push the style too far and the face or pet stops looking like itself, hold back and the result looks like a filtered photo. We ranked tools on how well they keep the subject, how much control they give over the medium (presets, reference paintings and strength settings), the models doing the conversion, the resolution you can print at, commercial rights and how easy they are for a single photo. Prices come from each tool's site as of October 26, 2026.

The eight tools compared

1. Melius: Best overall for painting photos with several models

Melius is the AI-native operating system for creatives, where AI agents for creative work run many of the leading image, video and audio models on one node-based canvas.

What works: Upload the photo, tell Mel, the Melius agent, the medium you want, and it writes the prompt and sends the photo to the image models suited to the job, such as Nano Banana Pro for following a style brief or GPT Image 2.5, which edits with up to 16 reference images. Branches let you compare an oil and a watercolor side by side, and Image Style Analysis turns two to 10 reference paintings into one style for a whole set. Every node stays editable, and a Topaz upscale enlarges the winner up to 4x.

What doesn't: You keep a face recognizable through the prompt, since there's no strength slider.

Best for: Photographers, gift makers and brands painting a set of photos in one consistent style.

2. Krea: Best for weighted reference paintings and large prints

Krea is a multi-model creative suite with its own Krea 2 and Krea 3 image models, a node editor and an enhancer.

What works: Krea 3 edits an uploaded photo from a description at up to 2K, and Krea 2 takes up to four reference images to set a style, each with its own strength slider. The Enhancer upscales 1x to 8x, with an Oil Painting preset that softens edges and adds painterly texture, and Topaz upscaling reaches 22K. Basic costs $9 a month, or $5 billed yearly, and paid plans include a commercial license.

What doesn't: Weighted style references work in Krea 2 generation, not in photo edits, and Krea warns that upscaling a textured painting adds brushstrokes that weren't in the original.

Best for: Artists who want to steer the style from specific reference paintings and print large.

3. Adobe Firefly: Best for controlling how closely the painting follows your photo

Firefly is Adobe's generative AI app, with its own Firefly Image models plus partner models.

What works: Upload your photo as a composition reference and the Strength slider sets how closely the result follows its outline and depth, while the prompt or a style reference sets the medium. Adobe suggests combining styles like "oil painting" and "palette knife." Adobe says images made with its own Firefly models are safe for commercial use, and you can finish the painting in Photoshop. Standard costs $9.99 a month, and free daily generations let you try it first.

What doesn't: The painting workflow is spread across prompt, style and composition panels, and partner models fall outside Adobe's commercial-safety statement.

Best for: Designers who want a fidelity dial and already work in Photoshop.

4. Midjourney: Best for a painterly house style

Midjourney is a subscription image generator known for its own aesthetic, used on the web and in Discord.

What works: The Edit model in V8.1 and V8.2 can convert an uploaded image into a different style, guided by style references, Moodboards and Personalization, with up to four attached images. Midjourney suggests describing the style in your prompt as well. Basic costs $10 a month, or $8 billed yearly.

What doesn't: Images top out at 2048 x 2048 pixels, about 6.8 inches at 300 ppi. Likeness isn't guaranteed, and uploads and results are public unless you pay for Pro or Mega.

Best for: Anyone who cares more about the painted aesthetic than an exact likeness.

5. ChatGPT Images: Best for a quick conversion by description

ChatGPT Images, now on GPT Image 2.5, creates and edits images inside ChatGPT.

What works: Upload a photo, ask for a watercolor or an oil painting and refine it in conversation. ChatGPT may ask follow-up questions about the style, and image creation is available on every tier, including Free. Plus ($20 a month) raises the limits.

What doesn't: There's no strength slider, preset styles or upscaler, and it declines requests in the style of a living artist.

Best for: Anyone painting a single photo without learning a new tool.

6. Google Gemini: Best for free restyles from a reference image

Gemini's image editing runs on Google's Nano Banana models: Nano Banana 2 on the free tier and Nano Banana 2.1 on paid Google AI plans, which can redo an image with Nano Banana Pro.

What works: Give Gemini your photo plus a reference painting and ask it to copy the texture, color or style. It's free to start, and Google AI Pro, at $19.99 a month, unlocks the newer models.

What doesn't: It lacks a strength slider and an upscaler, and which model you get depends on your plan.

Best for: Free experiments with a specific look you can show it.

7. Magnific (formerly Freepik): Best for style transfer with structure control

Magnific is the AI creative suite formerly branded Freepik, which now includes the original Magnific upscaler.

What works: Magnific's API offers style transfer that applies preset or custom reference styles with separate controls for style intensity and structure preservation, and can upscale the result in the same step. A photo editor, relighting and the Magnific and Topaz upscalers sit in the same suite. Premium costs $20 a month, with a commercial license and no training on your data.

What doesn't: The style-transfer settings sit in the API, so the most control goes to people comfortable calling it from code.

Best for: Developers and technical creators who want a structure dial and print-size upscaling.

8. NightCafe: Best for trying it free

NightCafe is a community AI art site with a dedicated Photo to Painting template.

What works: The Photo to Painting template takes a photo, a painting style such as Baroque oil and a model of your choice, including Nano Banana 2.1 and GPT Image 2. You get up to five free creations a day without logging in, and NightCafe assigns copyright of the painting to you if you own the original photo.

What doesn't: Free creations from the template appear in a public feed, and you pick from preset styles rather than a reference painting.

Best for: Trying a few painted versions of a photo before paying for anything.

Prisma, the phone app that popularized art filters in 2016, still offers hundreds of one-tap painting styles. It's quick for social posts but gives you no reference images or strength control.

How to get a better painting from a photo

• Start with a good photo. A sharp, evenly lit image with one clear subject converts best. Heavily compressed JPEGs lose detail that the model needs.

• Name the medium and its marks. "Oil painting, thick impasto, visible palette-knife strokes, warm underpainting" works better than "make it artistic."

• Describe a style, not a living artist. Public-domain movements such as Impressionism or Baroque are fine, and some tools refuse living artists' names.

• Set fidelity before style. Where there's a slider, start with high structure and low style, then raise the style until the face starts to drift. Without one, say what must stay: "keep her face, hairstyle and pose; change only the rendering."

• Use a reference painting when you want a specific look.

• Compare, then upscale once. Generate drafts at 1K or 2K, then upscale only the one you'll print, with a low strength setting so the upscaler doesn't invent brushstrokes.

• Pick a print ratio up front: 4:5 for 8 x 10 and 16 x 20 inch prints, 2:3 for 4 x 6 and 24 x 36.

Frequently asked questions

What is the best AI tool to turn a photo into a painting?

Melius is the best overall, because its agent writes the prompt, runs the photo through several leading image models and lets you revisit any step for a consistent set. Krea offers weighted reference paintings and 22K upscaling, and Adobe Firefly has a Strength slider for how closely the painting follows the photo.

Can I turn a photo into a painting for free?

Yes. NightCafe gives five free creations a day with no login; Gemini and ChatGPT include free image editing; Adobe Firefly and Krea offer free daily generations. Free tiers limit how many images you make, and some make your results public.

How do I turn a photo into an oil painting with AI?

Upload the photo, ask for an oil painting and describe the brushwork, such as thick impasto or palette-knife strokes. Tell the tool to keep the subject's face and pose, compare a few results and upscale the best one if you plan to print it.

Can ChatGPT turn a photo into a painting?

Yes. Upload the photo and describe the painting style you want, and ChatGPT will restyle it and let you refine it in conversation. It's available on every tier, but it has no strength slider or upscaler.

How do I keep the face looking like the person or pet?

Use a tool with a strength or structure control, such as Adobe Firefly, Krea or Magnific, and keep the style setting low. With tools that work from a prompt, say exactly what must stay the same and start with a subtle style.

Can I print or sell an AI painting of my photo?

Check the resolution first: an 8 x 10 inch print at 300 ppi needs about 2,400 x 3,000 pixels. Then check the tool's terms. Melius says: "You own the content you create on Melius and can use, publish, and distribute it commercially, subject to the underlying model providers' terms. Melius doesn't claim ownership of your work and doesn't use it to train AI models." Krea and Magnific include a commercial license on paid plans, and Adobe says its own Firefly models are safe for commercial use.

Can I make my photo look like a Van Gogh or Monet painting?

Yes, most tools will paint a photo in the style of a public-domain artist or movement, such as Post-Impressionism or Impressionism. Some tools refuse living artists' names, so describe the technique, such as swirling brushstrokes or broken color, instead.