- Home
- AI Image Models
- P-H3-Image
The open image model built from MiniMax H3
P-H3-Image AI Image Generator: Photoreal Up to 8 MP
Get magazine-grade portraits, film stills and posters with exact short text, at up to 8 megapixels, from a one-line prompt.
P-H3-Image is an open image model built from MiniMax H3. It makes photoreal photos, film frames, anime stills and prints, sets short quoted text, and edits a photo or combines up to 8 images into one scene. Pixel Dojo rewrites your request into the detailed form P-H3-Image renders best, so one plain sentence is enough. We made every example on this page on Pixel Dojo, and each one comes with the exact prompt we typed.
1 credit ($0.06) per image at 2K. Included in every plan, from $10/month.
Already a member? Sign in
By Brian Lovett ยท Updated

Why people choose P-H3-Image
Photoreal at 8 megapixels
At 4K P-H3-Image renders about 8 megapixels, where skin texture, beard hair and the weave of a wool coat come through. It is the model to reach for when the picture has to look like a real photograph.
Film stills and editorial looks
Open the prompt with what the image is, a frame from a seventies crime film or a documentary photograph, and the look follows: the color, the grain, the wardrobe and the light.
Short text, spelled right
Put each word in quotes and say where it goes. On our concert poster NIGHT TIDE fills the top and LIVE AT THE PIER sits at the bottom, spelled exactly as typed.
Edits from up to 8 images
Attach a photo as picture 1 and say the change. Add more images as picture 2, picture 3 and so on, and P-H3-Image combines them into one scene.
What you can create with P-H3-Image
We made these images with P-H3-Image on Pixel Dojo at 4K, about 8 megapixels each. Each tile shows its settings and credit cost. Tap Use this prompt on any tile to load it into the prompt box, then press Create with P-H3-Image.

A crime film frame from the seventies. Naming the decade and genre brings the color, grain and wardrobe of the era.
A frame from a 1970s crime film: a man in a tan suit lights a cigarette inside a phone booth on a rainy night, orange sodium streetlight through the wet glass

Every word in quotes, each with a place. Both lines came out spelled exactly as typed.
A two-color screen-printed concert poster in orange and navy that reads "NIGHT TIDE" in bold condensed letters above a large stylized wave, with "LIVE AT THE PIER" in small letters at the bottom, no other text

A studio product photograph. Hard sun makes the glass glow and throws colored light into the shadows.
A studio product photograph of two amber glass perfume bottles on a slab of travertine, hard afternoon sun casting long sharp shadows across a sand-colored backdrop

A documentary photograph. People stay natural and the flame, steam and bulb light carry the moment.
A documentary photograph of a street food vendor tossing noodles in a flaming wok at a night market, customers waiting under a striped awning, steam and warm bulb light

A frame from a Japanese anime TV series, with clean cel shading and a painted sunset sky.
A frame from a Japanese anime TV series: a girl on a bicycle waits at a seaside railway crossing at sunset, the crossing lights glowing red, power lines against an orange sky

An architectural photograph. One light source with a direction gives the room its shape.
An architectural photograph of a concrete reading room with a long oak table, a skylight throwing a sharp rectangle of afternoon sun across the floor, one reader in a grey sweater

An overhead food photograph. At 4K the crumb, crust and salt crystals stay crisp.
An overhead food photograph of a sourdough loaf torn open on a linen cloth, open crumb and blistered crust, a knob of butter and flaky salt, soft morning light
Edit a portrait with one sentence
We gave P-H3-Image our ferry captain portrait from the top of this page and asked for a rain hat and a storm outside.

One photo and one sentence: the face, coat, wheel and cabin kept, a yellow rain hood and a storm added.
Give him a yellow oilskin rain hat and add heavy rain streaming down the windows behind him
- Describe the finished result, not the action.
- Pixel Dojo writes the full edit brief for you, including what must stay the same.
- Leave the aspect ratio on Auto to keep the shape of your photo.
How to use P-H3-Image on Pixel Dojo
- 1
Say what the picture is
Start with one line that names the kind of image: a portrait photograph, a film frame, a screen-printed poster.
- 2
Describe one scene
One subject, one moment, where the light comes from. Put any words in quotes and add no other text.
- 3
Pick the size
Use 4K for the sharpest faces and detail, 2K for most work and 1K for quick ideas.
- 4
Edit what you made
Press Edit on any result to make it picture 1, then describe the change in one sentence.
P-H3-Image pricing: from 0.5 credits per image
P-H3-Image is included in every Pixel Dojo plan, with plans from $10/mo. It uses 0.5 credits per image at 1K, 1 credit at 2K and 1.5 credits at 4K.
Every plan also includes unlimited Relax runs on P-H3-Image at 1K. They use no credits, so you can explore as many ideas as you like.
P-H3-Image questions
What is P-H3-Image?
P-H3-Image is an open image generation and editing model built from MiniMax H3. It makes photoreal images, film stills and posters with short text at up to 8 megapixels, and it edits with up to 8 input images.
What is P-H3-Image good at?
Photorealism: editorial and documentary photos, portraits with real skin texture, film stills, anime frames, prints and posters with short quoted text.
How do I use P-H3-Image on Pixel Dojo?
Open P-H3-Image, write one line about what the image is and the scene, pick a size and press Generate. To edit, press Edit on a result or attach a photo, then describe the change.
How much does P-H3-Image cost?
It is included in every Pixel Dojo plan: 0.5 credits per image at 1K, 1 credit at 2K and 1.5 credits at 4K. Relax runs at 1K use no credits.
Can I use P-H3-Image through an API?
Yes. Call p-h3-image at /api/v1/models/p-h3-image/run with prompt, aspect_ratio, resolution and reference_images. It also works in Canvas and our MCP server.
Make your first P-H3-Image image
Write a prompt above or start from any example on this page. One plan covers every top model, six studios and Sensei, and your credits roll over.
Start with P-H3-Image