AI Painting Generator: Create Art from Text or Photos

Oil, watercolor, acrylic and gouache, from a prompt or a reference image

Describe an original painting or upload an existing image as a Reference. Choose the medium, brushwork, palette, surface texture and visual direction, generate several interpretations, then refine the version you want to keep. The output is a digital raster image built to reproduce painterly cues such as brushstrokes, pigment and canvas texture.

Start creating
10M+ users
Founded in 2022
Oil painting portrait of an elderly woman reading beside a window, thick brushstrokes and canvas texture
Watercolor painting of a mountain valley and river in pale blue and green transparent washes
Acrylic still-life painting of lemons in a bowl and a blue-patterned jug on a wooden table
Gouache botanical painting of eucalyptus leaves and seed pods on cream paper

How to create a painting with AI

Six steps from a prompt or a photo to a finished painted image.

1. Open Create image

Sign in and open the Create image action, the text-to-image entry point of the workspace. Open it here.

2. Describe it or add a reference

Write what the painting shows, or upload a portrait, pet, landscape or object photo as a Reference and describe the painted treatment you want applied to it.

3. Set the direction and generate

Name the medium, brushwork, palette and surface, pick an aspect ratio, then generate multiple variations in a batch. Refine the strongest result and upscale it once the artwork is approved.

Oil painting of a stormy coastline with palette-knife impasto in ultramarine and grey

Turn a photo into a painting

Upload an image as a Reference, then describe the painting medium and visual treatment you want. Reinterpret a portrait, pet, landscape, house or other photo as oil, watercolor, acrylic, gouache or another painted style. Self portraits, family portraits, couple portraits, pet portraits and wedding portraits all work from a single source photo. References guide the subject and composition, but stronger painterly stylization can simplify or change smaller facial, fur and clothing details.

Watercolor portrait of a brown and white spaniel lying in grass, soft washes on textured paper

Oil painting in the style of Van Gogh, vibrant blooming meadow, thick, swirling brushstrokes depicting wildflowers in a multitude of colors, dynamic composition, visible canvas texture, intense use of complementary colors, energetic and expressive

What an AI painting actually is

Three distinctions worth making before you generate, so you know what the file you get back can and cannot do.

Painting vs. painting filter

A traditional filter primarily changes the visual treatment of the existing pixels. Generative painting workflows can reinterpret texture, brushwork, lighting and smaller details while using the original image as guidance.

A filter recolors and textures what is already there. Edges, shapes and detail stay where the camera put them.

Generative repainting redraws the scene in the chosen medium. That is why output is not a pixel-perfect copy of the source photo.

Explore oil, watercolor, acrylic and gouache

Each medium has its own visual logic, and naming it changes the result. Oil reads as thick brushstrokes, impasto and layered pigment over canvas texture. Watercolor reads as transparent washes, soft edges, pigment blooms and visible paper grain. Acrylic reads as opaque pigment, bold color and clean textured shapes. Gouache reads as matte opaque coverage and smooth shapes on paper. Pastel and mixed media sit between drawing and painting and work as secondary directions.

Oil painting of peonies in a glass jar, thick impasto petals over visible canvas weave

Control brushwork, surface texture and palette

Four dimensions decide whether an image reads as a painting or as a smooth digital illustration. Brushwork: thick impasto, loose expressive strokes, fine detailed strokes or dry brush. Surface: canvas texture, cold-press paper, smooth ground or rough paper grain. Pigment and palette: muted, saturated, earthy, pastel or high-contrast. Edges: soft blended, hard-edged, feathered or outlined. Name them in the prompt instead of relying on a single style word.

Close-up of thick impasto oil paint ridges in blue, orange and white across raw canvas weave

What can you create with an AI painting generator?

Portraits, pet paintings, landscapes, still lifes, botanical studies, architecture and house portraits, abstract compositions, fantasy scenes, editorial artwork and wall-art concepts.

Painted portraits from photos

Self portraits, family portraits, couple portraits, pet portraits and wedding portraits, generated from a source photo used as a Reference.

Oil painting group portrait of two adults and a child seated outdoors in warm ochre and green tones

Landscapes and house portraits

Coastlines, meadows and city scenes, or a photo of a house reinterpreted as a watercolor or oil house portrait for wall art.

Watercolor painting of a two-storey brick house with a garden gate and a mature tree

Still life, botanical and abstract

Fruit and flower still lifes, botanical studies, and abstract compositions built from pigment, gesture and surface texture.

Abstract painting with red, deep blue and off-white blocks crossed by broad curved brushstrokes

Guide your painting with visual References

Use up to 10 reference images in a single generation to guide likeness, artistic style, palette or mood. One image can carry the subject, another the palette and brushwork direction.

Subject reference Use a portrait or pet photo so the painting keeps the likeness and composition of the person or animal in frame.
Palette reference Add a second image purely for color: the painting borrows its pigment range without copying its subject.
Style reference Point at artwork whose brushwork and surface treatment you want as a general painterly direction.
Oil painting portrait of a young woman in profile against a plain wall in teal and coral tones

Keep one painting direction across a series

Save a reusable Style or other visual Element and apply it to additional subjects or scenes instead of rebuilding the direction from scratch every time. Useful for an illustration series, a campaign, matching pet portraits or a landscape set. Generative consistency is guidance, so individual brush marks, palette balance and details can still vary. See the Elements guide for how to save and reuse one.

Save the direction once Store medium, brushwork, palette and surface treatment as an Element you can call into any later generation.
Apply it to new subjects Reuse the same painted language across a set of portraits, scenes or product-adjacent artwork.
Share it with a team Keep one visual direction across everyone producing artwork in the same workspace.
Gouache painting of rolling farmland with hedgerows and a distant barn in sage and ochre

AI painting prompt examples

A reliable formula: medium + subject + composition + art movement or direction + brushwork + palette + surface + lighting.

Oil

Oil reads through thick brushstrokes, impasto, layered pigment and canvas texture.

Oil portrait of an elderly woman reading beside a window, three-quarter composition, realist painting, visible layered brushstrokes, warm earth-tone palette, subtle canvas texture, soft directional afternoon light.

Expressive oil painting of a stormy coastline, heavy impasto, broad palette-knife strokes, deep ultramarine and gray pigments, rough canvas texture, dramatic light.

How to get better AI painting results

Most weak painting output comes from prompts that name a style but never describe how paint behaves. Specify the physical properties instead.

Name the medium and brushwork Say oil, watercolor, acrylic or gouache, then describe the strokes: thick impasto, loose expressive, fine detailed or dry brush.
Define the surface Canvas texture, cold-press paper, smooth ground or rough paper grain change how the whole image reads.
Choose a palette and edges Muted, saturated, earthy, pastel or high-contrast pigment, and whether edges are soft blended, hard-edged, feathered or outlined.
Use a Reference when likeness matters Upload the portrait, pet or building you want recognized rather than describing it in words.
Generate several directions Produce multiple interpretations of the same subject and compare brushwork, palette, medium and composition before committing.
Upscale last Approve the artwork first, then [upscale](/tools/upscaler) so you are not enlarging a composition you plan to change.

Prepare AI paintings for print

Finalize the composition first, then upscale the selected image to the pixel dimensions required by your final print size. Check the printer's resolution, bleed and trim requirements before delivery.

Start from the physical size Decide the finished print dimensions, then work back to the pixel count you need. 300 PPI is the usual benchmark for high-quality print.
Upscale, then inspect Run the approved artwork through [Upscaler](/tools/upscaler) and check brushwork and edges at full size before sending it out.
Match the delivery spec Use [AI Resizer](/tools/ai-resizer) for the target ratio and confirm bleed and trim with the printer. A canvas print reproduces the painted look, not a physical layer of paint.

Frequently Asked Questions