wan-2.7/pro/edit
WAN 2.7 Pro image edit: high-fidelity prompt-driven edits with 1–4 references, prompt expansion, and the same controls as the standard edit endpoint.
Upload one to ten source stills, then describe the change in plain language. When more than one photo is attached, name each one by its position so the roles stay clear. Leave the frame on auto when the first photo should set the crop.
| Parameter | Required | Type | Default | Range / Options | Description |
|---|---|---|---|---|---|
images* | Yes (*) | Array | Sample image | 1–10 images | Source images in the order the instruction names them. Each side at least 256 pixels. Each image no larger than 4 megapixels. |
prompt* | Yes (*) | String | Example instruction | Must not be empty | What should change, and what should stay. Name uploads as image 1, image 2, and so on. Put exact on-image wording in quotes. |
aspect_ratio | No | String | auto | auto, 21:9, 2:1, 16:9, 3:2, 7:5, 4:3, 5:4, 1:1, 4:5, 3:4, 5:7, 2:3, 9:16, 1:2 | Output frame. auto follows the first image. |
resolution | No | String | 1k | 1k, 2k, 4k | Output size tier. This is the only setting that changes the price. |
enable_prompt_expansion | No | Boolean | false | true, false | Expand a short instruction while keeping the intended edit and the role of each image. |
output_format | No | String | jpeg | jpeg, png | File type of the returned still. png avoids extra JPEG compression. |
A file under 256 pixels on a side, or above 4 megapixels, sits outside the accepted range. jpeg is the default file type. Choose png when you want to skip a second compression pass. Leave prompt expansion off when the wording is already exact, and turn it on only when a short note needs a fuller scene description.
Billing is per finished image. Only the resolution tier changes the amount. Adding more reference images does not add a separate charge. The 1k tier is the lower-cost pass for trying an instruction. The 2k and 4k tiers cost more because the output is larger. A batch multiplies the selected rate by the number of outputs.
| Resolution | Price per image |
|---|---|
1K (1k) | $0.055 |
2K (2k) | $0.132 |
4K (4k) | $0.715 |
WAN 2.7 Pro image edit: high-fidelity prompt-driven edits with 1–4 references, prompt expansion, and the same controls as the standard edit endpoint.
WAN 2.7 image edit: text-guided edits with 1–4 reference images, optional prompt expansion, bilingual instructions, and preset output sizes.
FLUX 3 Image Edit revises existing photos from a written instruction and can combine one to ten references into a single still. It fits product composites, campaign frames, style changes, and layouts that need readable type. Designers and product teams use it when a reshoot would be slower than an instruction.
FLUX 3 Image Edit accepts one to ten source images in one run. Order matters: name them in the instruction as image 1, image 2, and so on, and say what each one contributes. Each image needs at least 256 pixels on every side and must stay at or under 4 megapixels.
FLUX 3 Image Edit is a practical fit for placing a product into a new scene, restyling materials and color, or building a poster from several references. State both the change and the parts that must stay, such as a face, a logo, or the frame. Put exact wording that should appear on the image inside quotation marks.
Based on publicly available information, FLUX 3 Image Edit adds a wider set of output frames and three size tiers, up to 4K, on top of instruction-led multi-image edits. It does not expose a mask or a seed control. For a single tight local change, a mask-based editor may still be the clearer tool.
FLUX 3 Image Edit is built to follow a concrete instruction, including which reference supplies the subject and which supplies the scene. Short, specific nouns work better than a long mood description. If the first pass drifts, turn prompt expansion off and restate what must remain unchanged.
FLUX 3 Image Edit outputs one still at 1k, 2k, or 4k, as jpeg or png. Aspect choices run from ultrawide 21:9 and 2:1 through square and portrait ratios down to 1:2; leaving the ratio on auto follows the first image. There is no duration or frame-rate setting, because the result is a still.
Yes. Prototype FLUX 3 Image Edit in the RunComfy Playground Web UI, then call the same model through the RunComfy API with the same parameters. Generations consume usd or credits. This does not promise a fixed latency or an unlimited concurrency level.
FLUX 3 Image Edit is priced per finished image by resolution only: $0.055 at 1K, $0.132 at 2K, and $0.715 at 4K. Extra reference images do not add a separate charge. A batch multiplies the selected rate by the number of outputs.
RunComfy is the premier ComfyUI platform, offering ComfyUI online environment and services, along with ComfyUI workflows featuring stunning visuals. RunComfy also provides AI Models, enabling artists to harness the latest AI tools to create incredible art.
