Image 2.5

AI Image & Video Generator

Text-to-image, image-to-image, text-to-video, and image-to-video in one studio. Switch Image or Video, then pick GPT Image 2.5, Nano Banana, Veo, Kling, or Seedance—set ratio, resolution, and references to create.

My Works

0/4000

Quick examples

Model

Aspect ratio

Resolution

Background

Add reference

Result

Preview, download, copy, and rerun. Finished work syncs to My Works.

Example preview

Example preview 1Example preview 2Example preview 3Example preview 4

Examples only — not this run.

Write a prompt and generate. The panel shows examples until your image is ready.

AI image & video generation

How to run text-to-image and AI video: prompts, parameters, workflow

This Create page covers image and video: text-to-image, image-to-image, text-to-video, and image-to-video. Switch Image or Video, then pick GPT Image 2.5, Nano Banana, Veo, Kling, or Seedance—set ratio and resolution to create. Built for posters, ecommerce heroes, portraits, short-form covers, and product clips.

Name the subject, scene, light (or camera moves) first, then pick aspect and quality. For images, use ratio and 1K–4K; for video, duration, resolution, and audio sync. Model deep-dives live on /models/ pages; this page is the all-in-one studio.

  1. Text-to-image & image-to-image prompts

    Structure prompts as subject → scene → lighting/lens → style → quality. Mix precise lighting and camera terms when you need tighter control.

    • Name the subject clearly—avoid vague pronouns
    • Add light: soft window light, neon, golden hour, volumetric light
    • Add framing: close-up, wide shot, overhead, cinematic crop
    • Add style: photoreal, anime, cyberpunk, ink wash, and more

  2. Text-to-video & image-to-video

    After switching to Video, describe camera motion and pacing. Pick text-only, first frame, first+last frame, or multimodal refs based on the model.

    • Text-to-video: describe the shot, action, and moves (dolly, follow, orbit)
    • Image-to-video: upload a first frame (or start/end) and say what stays vs moves
    • Common controls: duration, 720p / 1080p / 4K, audio sync
    • Video families: Veo, Kling, Seedance, Wan, Vidu (details on model pages)

  3. Image vs video parameters

    Use the Image / Video switch first so the model list and controls match the media type:

    • Image: ratios like 1:1 / 16:9 / 9:16, 1K–4K resolution, quality, and references
    • Video: duration, resolution, audio sync, plus text / image / start-end / multimodal modes
    • Image models: GPT Image 2.5 Flare / Sunburst, GPT Image 2, Nano Banana, Seedream
    • Video models: Veo, Kling, Seedance, Wan, Vidu

  4. Recommended AI image & video flow

    Craft the prompt here, then generate on this site. Results are live outputs you can download or save to My Works.

    • Pick Image or Video, then choose a model and write the prompt
    • Add references (or first/last frames) if needed and check the quote
    • Tap Generate image / Generate video and wait for on-site results
    • Browse the Gallery or a model landing when you need references, then rerun here

Create in four steps

  1. 01

    Choose Image or Video

    Switch media type at the top. Image lists still models; Video lists video models and matching controls.

  2. 02

    Write a prompt and pick a model

    Describe subject and scene (add camera moves for video). Choose GPT Image 2.5, Nano Banana, Veo, Kling, Seedance, and more.

  3. 03

    Set ratio, resolution, and references

    Images: aspect and 1K–4K. Video: duration and clarity. Upload refs or start/end frames when needed.

  4. 04

    Generate and save

    Check the quote, tap Generate image or Generate video, then download or sync to My Works.

Practical tips for images & video

01

Structure before adjectives

Write who/what/where first (plus camera moves for video), then style words.

02

Change one variable at a time

When comparing runs, change only ratio, quality, duration, or model.

03

Lock look with refs / first frame

For consistent characters or products, add a reference or first frame plus a clear keep/change note.

04

Check commercial rights

Confirm plan terms before commercial use; avoid unauthorized celebrity or trademark assets.

Related searches

AI image / text-to-image GPT Image 2.5 models Nano Banana Kling video Google Veo Seedance video Wan video Vidu video AI tools Prompt library Gallery Tutorials

AI image & video FAQ

Is this page for AI images or AI video? +

Both. Switch Image or Video at the top first: Image lists GPT Image 2.5 and Nano Banana; Video lists Veo, Kling, Seedance, and more. Controls and the button label follow the selected type.

What is the difference between text-to-video and image-to-video? +

Text-to-video relies on the prompt alone. Image-to-video needs a first frame (some models support start/end or multimodal refs) when you want a still to move. In Video mode, pick text / first frame / first+last / multimodal.

Do I always need a first frame? 720p vs 1080p? +

Image-to-video or start-end modes usually need at least one image; pure text mode does not. Use 720p for drafts; prefer 1080p+ for final ads when the model allows. Longer duration and higher resolution typically raise the quote.

How do I choose Kling, Veo, or Seedance? +

Try them here, then open the matching model landing for capabilities and examples. Unsure? Run the same short prompt on two models and compare motion and clarity.

Does Generate create the final image or video on this site? +

Yes. This page uses the on-site generate API and returns live results you can download or save to My Works. Header “Start creating” opens this workspace.

Flare vs Sunburst on this page — how do I choose? +

Pick Flare for photoreal portraits, product shots, architecture, landscapes, and brand moodboards. Pick Sunburst for anime character sheets, stylized illustration, and manga-like frames. Both use GPT Image 2.5 with different quality defaults. Unsure? Run the same prompt on both — or switch to GPT Image 2 / Nano Banana.

My prompt is unstable — what is a practical fix? +

Try: (1) clearly name subject + action (and camera moves for video); (2) borrow a lighting/lens skeleton from the Prompt Library; (3) lock ratio and quality before changing other knobs; (4) change one variable per run. Helpful add-ons: soft window light, cinematic lighting, 85mm, tracking shot.

How should I use a reference image or first frame? +

References help lock character vibe, product shape, or composition — but still spell out what to keep vs change. Start with fewer images, then add more. For commercial work, confirm you have rights to the asset.

How are credits charged? Do I still pick Relax / Fast / Turbo? +

No speed modes. This page matches the homepage: you see a quote before generate, and the task receipt is the final charge. Models have different catalog prices. If credits run short, the page sends you to top up.

Which aspect ratio should I use for my channel? +

Avatars often 1:1; covers/blog heroes 16:9; mobile feeds/shorts 9:16; print posters may use 2:3 or 3:4. Choose aspect before generating to reduce hard crops later.

Generation fails, stalls, or totally misses the brief — what next? +

Check network/login, then simplify conflicting style stacks. When off-brief, name the subject (and motion) clearly. Compare another model or quality/resolution tier, or a quick example as a control. Billing questions live on the Pricing FAQ.

Keep learning Image 2.5

AI models hub Prompt library Gallery Blog Tutorials Pricing AI tools

Ready for AI images or AI video?

Generate on this site, or start from the Prompt Library / a model landing.

Start creating →