feat(minimax-h3): default workflows; starter-model decision documented (PR 5/5)

invokeai/app/invocations/minimax_h3_ideal_dimensions.py

@@ -0,0 +1,52 @@
+"""Compute the MiniMax H3 canvas for a source image's aspect ratio.
+
+Unlike Wan (which offers several target resolutions), H3 was released for exactly one
+canvas family: a 768 px short edge under a soft area cap of 768x1344, both axes rounded
+to the nearest multiple of 32, aspect ratios limited to 1:4 through 4:1. That policy is
+implemented by :func:`invokeai.backend.minimax_h3.packing.resolve_canvas_size` (a
+first-party port of the released pipeline's canvas resolution); this node is a thin
+workflow-graph wrapper so image-to-video workflows can derive the canvas from the
+uploaded keyframe instead of asking the user to type matching width/height values.
+"""
+
+from invokeai.app.invocations.baseinvocation import BaseInvocation, Classification, invocation
+from invokeai.app.invocations.fields import InputField
+from invokeai.app.invocations.ideal_size import IdealSizeOutput
+from invokeai.app.services.shared.invocation_context import InvocationContext
+from invokeai.backend.minimax_h3.packing import resolve_canvas_size
+
+
+@invocation(
+    "minimax_h3_ideal_dimensions",
+    title="MiniMax H3 Ideal Dimensions",
+    tags=["minimax", "video", "dimensions", "math"],
+    category="video",
+    version="1.0.0",
+    classification=Classification.Prototype,
+)
+class MiniMaxH3IdealDimensionsInvocation(BaseInvocation):
+    """Ideal dimensions for MiniMax H3 from a source image's aspect ratio.
+
+    Applies the released pipeline's canvas policy: short edge 768, soft area cap of
+    768x1344, both axes rounded to the nearest multiple of 32. Only the aspect ratio of
+    the inputs matters. Aspect ratios beyond 1:4 / 4:1 are rejected. Wire from ``Image
+    Primitive``'s width/height outputs and into the width/height of ``Prompt - MiniMax
+    H3``, ``Frame Conditioning - MiniMax H3`` and ``Denoise - MiniMax H3`` (all three
+    must share the same canvas).
+    """
+
+    width: int = InputField(
+        default=1024,
+        gt=0,
+        description="Source image width in pixels.",
+    )
+    height: int = InputField(
+        default=1024,
+        gt=0,
+        description="Source image height in pixels.",
+    )
+
+    def invoke(self, context: InvocationContext) -> IdealSizeOutput:
+        # resolve_canvas_size returns (height, width).
+        h, w = resolve_canvas_size(float(self.width), float(self.height))
+        return IdealSizeOutput(width=w, height=h)