Made for this idea
A soft portrait for a small print, keepsake or quiet profile image. Pale washes work best when the face remains the strongest shape in the composition.
Image to image
Give each reference photo a clear job, record visible identity details, and download a portrait brief for an artist or the AI photo workflow.

Your next useful step
Choose your starting point, edit the instructions, and review the credit quote before generating. Current access conditions are shown in the creator.
Illustrative scenario, not a customer testimonial.
The main photo has the expression you love but hides one ear. Add a second authentic angle that shows that ear, label its role and avoid mixing photos of different animals in one identity reference set.
Watercolor portrait
A soft portrait for a small print, keepsake or quiet profile image. Pale washes work best when the face remains the strongest shape in the composition.
Use even lighting and a plain background. Ask for a restrained palette and leave paper-colored space around the ears; busy scenery can obscure fine whiskers.
Check that watery edges have not erased an eye, merged the ears or moved a distinctive patch. Review at the size you intend to display, then use the print-size calculator if printing.
Upload one to five JPG, PNG or WebP references, each up to 12 MB. Start with a clear face and add another angle only when it explains a marking or body feature.
Say what should change and what should remain: for example, change the setting while keeping the ear shape, eye color and chest patch. Avoid combining an outfit, background, pose and lighting change in the first attempt.
Check the quoted credits before submitting. When the work is ready, compare the face, edges and composition; for video, watch the beginning, middle and end. Download the accepted file from your library before its expiry date.
Image to image
This editable preset guides the model. Review the result before keeping it; a preset is not a guarantee of likeness, exact motion or flawless editing.
Edit the uploaded pet photograph. Soft watercolor pet portrait with visible paper texture. Preserve the pet’s recognizable face, coat markings and proportions. Do not add text or watermarks.Try the other starting point · Text to image →
These are individual tests with licensed photographs or a written prompt. Compare the details before choosing a workflow; each result can differ.


The seated pose, white blaze, dark saddle and white-tipped tail remain recognizable. The face and fur edges are redrawn, and woodland shapes remain despite the request for a plain background. Check these details before using the portrait.
Source photograph: Grzes1966 · CC0-1.0
R07 · 2026-10-14 · fal-ai/nano-banana-2/edit
Create a gentle watercolor portrait from this exact tricolor dog photo. Preserve the white blaze, black saddle, tan ears, white-tipped tail and visible seated pose. Keep the entire existing photo content in view. Use a clean pale cream paper background and soft sage accents. No additional animals, text, clothing, wings, halo, extra limbs or invented props.
Images and videos are compressed for this page. The dimensions shown describe the downloadable original. Video previews use the delivered clip without changing its action or playback speed.
Free previews carry a watermark: images up to 768 px on the long side and videos at 480p. They remain available for 14 days. Original access requires purchased credits when checkout is available; unlocked works are kept for 90 days. Check the actual dimensions and expiry on the completed work.
View pricing and access →Uploads and generated files belong to your private workspace. Generation providers process the material you submit. Sharing is a separate choice; upload only material you have permission to use.
Read the data-handling notice →Upload one to five JPG, PNG or WebP references, each up to 12 MB. Start with a clear face and add another angle only when it explains a marking or body feature.
Check that watery edges have not erased an eye, merged the ears or moved a distinctive patch. Review at the size you intend to display, then use the print-size calculator if printing.
A reference image gives the model information to work from, but distinctive markings, proportions and expressions can change. Text-only creation invents a pet from your description. If exact documentary detail matters, retain the original photo alongside any AI interpretation.
Free previews carry a watermark: images up to 768 px on the long side and videos at 480p. They remain available for 14 days. Original access requires purchased credits when checkout is available; unlocked works are kept for 90 days. Check the actual dimensions and expiry on the completed work.
This editable preset guides the model. Review the result before keeping it; a preset is not a guarantee of likeness, exact motion or flawless editing.
Updated 2026-10-12
A good reference brief says what each photograph contributes. One picture may show the face clearly while another explains a chest patch. Sending many unrelated photos without those roles can make the intended appearance harder to understand.
Name the source files, mark the details that matter and save a brief you can use when making the portrait.
Your entries stay in this browser tab. Download a copy before leaving; this worksheet does not send messages or place orders.
Pick the photo that best represents the expression and age you want. Record the original filename and whether it has been cropped or edited. A clear side view is useful when you deliberately want a profile; it does not establish unseen details on the far side.
Use extra photos to clarify an ear, a coat patch or the body rather than asking the maker to average different ages. Label the feature and the side of the pet it shows. Keep another animal out of the identity set.
Put eye color, muzzle shape and distinctive markings in the identity field. Put costume, background and art style in the creative field. If a detail cannot be confirmed from any photo, mark it unknown instead of supplying a confident guess.
The TXT or JSON download contains your notes, not the image files. Send authorized originals through the channel you choose. Start a dog portrait or cat portrait when the references are ready, then compare the result with this brief.
A useful pet portrait from photo brief identifies one main image and explains the others. The main image controls age, expression and pose. A supporting image might show an ear hidden by shadow or a chest marking outside the crop. Label each file with that role. When images disagree, say which one controls the detail rather than hoping the maker combines them correctly. This is especially useful when family members contribute pictures from different years or when the clearest face is in a less attractive setting.
Keep space around ears, nose, paws and tail until the final crop is decided. A full-body request needs a full-body reference if you want those proportions to come from evidence. If the only good image is a close-up, a head-and-shoulders composition may be the more honest choice. The dog photo guide focuses on muzzle and body shape, while the cat photo guide addresses markings and resting poses. Choose the guidance that matches the photograph you actually have.
An illustrative handoff might list: “Photo A: chosen age and expression. Photo B: left ear in daylight. Photo C: chest pattern only. Keep the pose from A; do not use C’s background.” Add the final orientation and the use, such as a phone image or a portrait for a frame. That brief is easier to follow than “use whichever is best.” Keep original files at their available resolution and send copies for editing. A screenshot of a chat preview may discard information that remains in the original attachment.
When the work returns, compare it with the main reference first. Check whether the supporting photographs were used only for their stated roles. If the result looks younger or carries a different collar, identify the image that may have introduced the change. Revise the brief accordingly instead of adding more unlabelled photographs. To decide how the portrait will be made, use the custom portrait comparison; to preserve an old damaged source before any reinterpretation, visit pet photo restoration.
No. It records filenames and descriptions only.
You can record it, but identify the features it does not show reliably. A generator cannot verify missing detail from a filename.
Use the intended age consistently, or explicitly identify which period the portrait should represent.