GPT Image 2 Generator for Images and Edits | NENE Video

Create an image from a written brief, or edit your own reference images. Try it in the generator below, with examples and only the controls that fit this model.
Text or reference image · Choose a style and ratio · Credit cost shown first · Saved to History
GPT Image 2 · 3 credits per generation
Credits are refunded automatically if generation fails.
Loading the current model collection
Text or reference image · Choose a style and ratio · Credit cost shown first · Saved to History
Credits are refunded automatically if generation fails.
GPT Image 2 is a practical choice when the deliverable is a still image that needs a clear subject, a defined composition, or a controlled edit. Use the NENE Video workspace to describe a product scene, a campaign visual, an illustrated poster, a concept frame, or a background variation. The page keeps the prompt, optional references, output ratio, resolution, credit estimate, and result review in one flow, so you can start the task without opening a separate model directory.
A useful brief tells the model what must be visible and what may change. Name the subject, placement, camera view, lighting, materials, and intended use. If you are editing a reference, separate the requested change from the details that must remain stable. GPT Image 2 can help with exploration and production drafts, but generated lettering, hands, reflections, logos, and fine product geometry still need a full-size human check before publication.
Keep the first request small enough to review. The steps below match the controls and result states in the generator above.
Choose whether the result is a product shot, poster, portrait, environment, or edit. Put the subject and the intended audience first, then add the setting, framing, lighting, materials, and mood. One clear deliverable gives each variation a useful comparison point.
Upload a reference when composition, identity, packaging, color, or another visual constraint matters. Explain what to preserve and what to replace. A sharp source with visible edges is easier to inspect than a small or heavily compressed image.
Select the aspect ratio and resolution shown for GPT Image 2, read the live credit quote, and submit after the balance check. When the task completes, inspect text, faces, hands, reflections, and object boundaries at full size before downloading or starting a variation.
Choose a use case that fits the model's actual input and output. Treat each result as a reviewable draft until it passes your own quality and rights checks.
Describe the product before the environment when shape, material, and packaging are important. This makes the first draft easier to judge for silhouette and placement. Use later variations to adjust the surface, background, or camera instead of changing every variable at once.
GPT Image 2 can explore a poster layout, illustrated cover, or social composition. Keep copy short and state the approximate placement. Treat generated typography as a draft: verify every character and use a design tool for final typesetting when the wording must be exact.
An edit prompt works best when it identifies the part to change and the part to protect. You can request a new background, lighting direction, wardrobe color, or crop while retaining the main subject. Expect some variation in identity and small details between attempts.
A still image can become a starting frame for a video workflow. Build a readable silhouette, leave enough space for movement, and avoid tiny elements that will be difficult to track. Review the image first, then choose a video model when the next task is motion rather than another still.
The generator is the source of truth for the current controls. Use this guide to plan a request, then confirm the exact quote before submitting.
Choose GPT Image 2 when you want a flexible still-image workflow with both text-to-image and reference editing in the same control panel. Use GPT Image 2.5 when you specifically need its wider range of extreme ratios or higher-resolution choices. Use Grok Imagine Image 2.0 for a compact 1K image and editing flow, or Seedream 5.0 Pro when a product-focused 1K or 2K workflow and explicit reference set fit the job better.
Before approving a GPT Image 2 result, compare the final crop with the intended placement, then zoom into the smallest meaningful detail. Check whether the subject still has the right silhouette, whether the light falls consistently across the materials, and whether background objects create a new claim or trademark issue. If the image will become a video frame, look for a clean edge and enough negative space for motion. Keep the prompt, source permission, selected settings, and review decision together in your project notes so a later variation can be evaluated against the same brief.
All three versions run in the generator on this page; pick one in the model menu. GPT Image 2 is the standard still-image and edit flow, with the resolutions and ratios the menu shows for it.
GPT Image 2.5 adds the widest choice of extreme canvases and resolutions up to 4K, for banners, tall posters and large prints. The price depends on the resolution you pick.
GPT Image 2.6 is the newest. On NENE it makes one 1K PNG per run for a flat 5 credits, takes up to 4 reference images and a prompt of up to 4,000 characters, and is the only version here with a background setting (Auto, Opaque or Transparent) for cut-outs. OpenAI says Images 2.6 is faster than before; it has no 2K or 4K output and no quality selector here.
A simple rule: start with 2.6 for a quick 1K draft or a transparent cut-out, move to 2.5 when you need a very wide or tall canvas or a large output, and keep GPT Image 2 for the standard flow you already know.
These answers describe the current NENE workflow and its review boundaries. Confirm live controls when you submit a task.
Use a related page when the input, output, or review requirement is different. Each link opens the same shared generator with that mo