HunyuanImage 3.0 Instruct Edit: change one locked still (9cr)

HunyuanImage 3.0 Instruct Edit is an edit. Point at the object. A new text-to-image throws away the still you approved.

HunyuanImage 3.0 Instruct Edit is an advanced instruction-following image editing model by Tencent. It allows users to modify and transform images based on natural language instructions while preserving the original content and structure of the image. Nine credits. Categories: edit-image and image-to-image. requires_image is true. There is no max output resolution listed, no duration list, no audio. If you do not have the locked file, you are in the wrong aisle.

Instruct means a sentence, not a new prompt

"Make a better version of this campaign" is a generate. "Remove the paper cup, keep the rest" is an instruct edit. HunyuanImage 3.0 Instruct Edit is sold as the second thing. The structure of the photograph is the product. Faces, layout, type, the crop you already signed — those stay unless you name them.

Nine credits is the catalog sticker. That is more than a 4-credit all-round image endpoint and less than a motion pass. Spend it on a change. Do not spend it on "try again" energy. Trying again is a text-to-image row, and it will not preserve the structure this model exists to preserve.

The door is the AI photo editor. The ranked shelf is best image editing model. Prompting for this slug: HunyuanImage 3.0 Instruct Edit prompting.

The upload is the whole spec

requires_image: true. max_reference is 3. Aspect ratios: 16:9, 4:3, 1:1, 3:4, 9:16, auto. Match the approved still. An edit that silently changes 9:16 into 1:1 is a new picture wearing the old filename.

No resolution ceiling is listed. Do not promise 4K in a slack thread. Judge the output against the file you sent. If the still is soft, upscale the image before the instruct pass, or accept a soft edit. This row follows instructions; it does not invent missing pixels.

There is no Hunyuan provider hub on this site. Do not look for one. The model page is the roster.

Preserve structure, then stop

Tencent's copy in the catalog is the rule: modify based on natural language, preserve original content and structure. That is inpainting discipline even when you are not drawing a mask. One object per run is the honest batch. Three unrelated changes in one sentence is how structure starts to slide.

When the edit is signed, then you may buy motion on an image-to-video row. Animating the pre-edit file because the edit "mostly worked" is how the cup you removed comes back as a blur.

Nine credits to keep the photograph. A new text-to-image to throw it away. The second option is always available and usually wrong once a human has said yes.

FAQ

Can I use HunyuanImage 3.0 Instruct Edit to create a still from scratch?

Not as this row is specced. requires_image is true and the categories are edit-image and image-to-image. There is no text-to-image category on the slug. Generate elsewhere, then instruct.

Why 9 credits instead of 4?

Because this is the instruct-edit line, billed at 9 credits in the catalog. It is not a 4-credit generate. You are paying to keep structure, not to sample a new picture.

Does it output a set resolution?

No resolution is listed on the spec. Keep the geometry of the source. Do not assume 4K.

What should the instruction look like?

Name the object and the change. "Change the jacket to navy, do not touch the face." Not "fashion editorial, same vibe." Vibe is a generate. Navy is an edit.