Skip to content

MeiGen 3.4.0 — Guided GIF and SVG generation

Latest

Choose a tag to compare

@jau123 jau123 released this 05 Nov 09:23
· 2 commits to main since this release

MeiGen MCP 3.4.0

With the MeiGen provider, call list_models to find the current GIF or SVG image model and its price. Use generate_image for animated GIFs and editable SVG originals.

For example, “生成一张猫眨眼的循环动图” uses the listed GIF model, outputFormat: "gif", and optionally gifOptions: { frameSize: "384x384", frameCount: 8, frameDurationMs: 250 }. This makes a two-second loop. Choose a size and frame count from the model’s live limits. For “生成一个可缩放的矢量徽标”, select the SVG model and outputFormat: "svg". For transparent still images, choose PNG or WebP and imageBackground: "transparent".

For a new GIF or SVG, include optional image guides in referenceImages. The model's max_reference_images in list_models sets the limit, up to two. The local npm server accepts direct image URLs or local file paths. To use a saved MeiGen character, connect to the remote MCP with OAuth, select it with list_characters, and call generate_image on that connection. Each distinct character uses one guide slot. A model with two slots can take one character and one ordinary image. To continue editing an existing GIF or SVG, open it in the MeiGen editor.

Completed GIFs keep their animation. For SVG, imageUrl is the PNG preview and vectorOutput.url is the original. Local downloads save the requested artifact. Existing requests can still be recovered by request ID.