Images, video & music

Zinley makes and edits media for you — stills, clips, soundtracks — all by asking in chat. There's no separate studio to open: describe what you want, and it comes back in the conversation.

Images

Generate

  • "Make a wide banner of a foggy pine forest at sunrise, no text."
  • "Draw a simple flat icon of a coffee cup, transparent background."
  • "Three logo concepts for a bakery called Rise — minimal, one color."

Edit an image you already have

Attach the image and say what should change:

  • "Remove the background from this."
  • "Change the shirt to navy and keep everything else identical."
  • "Extend this to a 16:9 crop."

Examine an image

Zinley can read an image closely and tell you what's in it — good for screenshots, receipts, charts, forms, and diagrams.

  • "What's the error in this screenshot?"
  • "Pull the totals out of this receipt."

Find images

  • Image search — "find photos of mid-century desk lamps."
  • Visual search — attach a picture and ask about it: "what plant is this?", "where was this taken?", "find where this image is from."

Video

  • From text — "make a 5-second clip of waves hitting rocks at dusk."
  • From an image — attach a still: "animate this photo, slow camera push in."
  • Merge clips — "stitch these three videos into one, in this order."

Video takes longer than a still. Zinley tells you it's working and posts the result into the conversation when it's ready — you can keep chatting meanwhile.

Music

  • "Write a 30-second upbeat acoustic track for a product demo."
  • "Make a calm lo-fi loop for a focus playlist."

Choosing the model behind each type

Generation quality and credit cost track each other, so you can pick the trade-off you want. Your choices are saved to your account and synced across devices.

The AI Models panel listing Reasoning, Image Generation, Video Generation, and Music Generation, each with its current model.
The model behind each type of media, in one place.

Image models

The Select Image Model dialog listing Gemini and GPT image models with credit costs, Gemini 3.1 Flash Image marked Default.
Trade quality against credits per image.
Model Cost Notes
Gemini 3 Pro Image ~27 credits / image Highest quality; best for text and fine detail.
Gemini 3.1 Flash Image (default) ~14 credits / image Fast, high-quality.
Gemini 3.1 Flash Lite Image ~7 credits / image Fastest and most efficient.
GPT Image 2 ~11 credits / image State-of-the-art generation and editing.
GPT Image 1.5 ~7 credits / image Advanced image generation.

Video models

The Select Video Model dialog listing Veo 3, Veo 3.1, and Veo 3 Fast, with Veo 3 Fast marked Recommended.
Quality versus cost per clip.
Model Notes
Veo 3 Higher quality, costs more.
Veo 3.1 Latest version, improved quality.
Veo 3 Fast (recommended) Faster, cheaper per clip.

Reasoning stays on Auto. The model that does Zinley's thinking is chosen per task and isn't part of this panel — see Modes & models.

Styling your agent's avatar

The picture on your agent card can be restyled without you opening an image editor. Upload a photo and pick:

  • Original — leave it as it is.
  • Studio Ghibli style — hand-painted anime, soft watercolor.
  • 3D render style — a clean Pixar-ish animated bust.
  • Custom — describe the look you want in your own words.

Every style keeps the person recognizable — same face, same hair, same skin tone — and frames the result for a round avatar.

Related