Royal court portrait of your dog
The classic: ermine cloak, dark background, old-master light. The Reference card keeps the muzzle, the art direction carries the painting. The gift works because the likeness survives the costume.
10% off monthly plans with code NODE10, until August 31
The family dog becomes a court portrait, then a scene, then it blinks. The reference sheet keeps its own muzzle. The workflow is visible here, without an account.
Real workflow, real results. Pan and zoom freely.
Use this workflowThe pet sheet holds the muzzle, the art direction carries the costume and the paint, the video node animates the portrait. Pan, zoom, open the nodes.
At a glance
A pet portrait generated from a description always gives a dog of the right breed. The problem is that it is not yours: the white blaze is in the wrong place, the ear leans the wrong way, and the gift falls flat because nobody recognises them.
So this whole workflow rests on a reference sheet filled with YOUR photos, front, profile and three quarter. The model refers back to it for every image, and the quirks survive the costume. The art direction carries the painting: the light on the canvas, the dark ground, the craquelure. The two stay separate, which lets you try another world without losing the animal.
Seedream 5 Pro holds a likeness better than most on animals, and Kling 2.5 Turbo is enough to make a painted portrait blink without waking up the material. The guide keeping a character consistent applies to animals too, and the public calculator gives the exact cost before you sign up.
Photograph from three angles
Front, profile, three quarter, in daylight and without a collar. Three photos from the same angle leave the model to invent the other side of the head.
Choose the world
The art direction node describes the painting or photograph you are after: period, light, ground, material. It feeds the portrait and the scene, so they stay in the same world.
Describe the costume
The portrait prompt only talks about the costume and the pose. The likeness comes from the sheet, the material from the art direction: no need to describe them again here.
Animate the portrait
Five seconds are enough: a blink, a tilt of the head, flickering light. Ask for a still camera, or the model thinks it is filming a scene and reanimates the whole painting.
The same canvas, tuned for different needs: each variation is two or three lines of prompt away.
The classic: ermine cloak, dark background, old-master light. The Reference card keeps the muzzle, the art direction carries the painting. The gift works because the likeness survives the costume.
Same chain, different world: uniform, armour, cape. Long-haired cats often need a fourth photo in the card, since the silhouette is harder to hold than on a short coat.
The workflow works with old, even mediocre photos: three angles are enough. For a framed print, generate in portrait format and check the eyes, which are what the family looks at first.
Five seconds with Kling 2.5 Turbo: a blink, a tilt of the head, flickering light. Ask for a locked camera, otherwise the model reanimates the whole painting and the canvas starts breathing.
Pick the format in the Image node before generating: square for a mug, portrait for a canvas. The resolution covers a framed print up to A3.
Shooting your dog from standing height distorts the muzzle, and the model copies the distortion. Get down to their eye level, in daylight, without the collar.
The model will produce a fine example of the breed, not yours. Likeness comes from the photos in the card, never from the written description.
The costume belongs in the portrait prompt, material and light in the art direction. Mixed together, they make it impossible to change worlds without rewriting everything.
Yes, the method does not depend on the species: three angles, a sheet, an art direction. Long haired animals sometimes need one extra photo to hold the silhouette.
Yes, the images come out at high resolution, well enough for a framed print. The format is chosen in the image node before generating.
Two images and a five second video come to around forty cents of credits. The detail is on the calculator, no account needed.
No, older photos are fine as long as they show three angles and the coat is sharp. That is what makes the memorial portrait possible, when only a few images are left.
Three, from three different angles. Ten front-on photos add nothing: it is the variety of angles that lets the model understand the head.
For a framed A3, yes. Beyond that, run the render through a dedicated upscaler: image models output high web definition, not print-shop files.
Five credits per image, twenty-six for a five-second animation. Trying ten different worlds on the same pet comes to fifty cents of credits.
Turn your pet into a character with AI
Use this workflowFree account, no card required. The workflow opens pre-filled in your canvas.