From 3b678549c33c8f025100f16eab814bd308c51ec1 Mon Sep 17 00:00:00 2001
From: Kevin Mi <45493463+kevin-mii@users.noreply.github.com>
Date: Fri, 4 Sep 2026 04:18:02 -0700
Subject: [PATCH] [diffusion] chore: warm up minimax-h3 at the served clip
 shape (#37945)

Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
---
 .../configs/sample/minimax_h3.py              | 36 ++++++++++++++++---
 .../test/unit/test_minimax_h3_admission.py    | 16 +++++++++
 2 files changed, 47 insertions(+), 5 deletions(-)

diff --git a/python/sglang/multimodal_gen/configs/sample/minimax_h3.py b/python/sglang/multimodal_gen/configs/sample/minimax_h3.py
index 4c4d4546b..c78216d5b 100644
--- a/python/sglang/multimodal_gen/configs/sample/minimax_h3.py
+++ b/python/sglang/multimodal_gen/configs/sample/minimax_h3.py
@@ -146,15 +146,41 @@ class MiniMaxH3SamplingParams(SamplingParams):
 
         self.task = task
         self.conditions = conditions
-        self.target = {
-            "short_edge": 768,
-            "aspect_ratio": "16:9",
-            "duration_seconds": 5.0,
-        }
+        self.target = self._synthetic_warmup_target(req, server_args)
         selected_seed = req.seed if isinstance(req.seed, int) else int(req.seed[0])
         req.extra.update(self.build_request_extra(_seed_override=int(selected_seed)))
         self._video_hooks().prepare_for_queue_sync(req)
 
+    @staticmethod
+    def _synthetic_warmup_target(req: Any, server_args: Any) -> dict[str, Any]:
+        """Warmup canvas from ``--warmup-num-frames`` / ``--warmup-resolutions``."""
+        from sglang.multimodal_gen.runtime.pipelines_core.stages.model_specific_stages.minimax_h3.constants import (
+            MINIMAX_H3_RECOMMENDED_SHORT_EDGE,
+            MINIMAX_H3_SUPPORTED_FPS,
+        )
+        from sglang.multimodal_gen.runtime.pipelines_core.stages.model_specific_stages.minimax_h3.task_profiles import (
+            MINIMAX_H3_FINITE_ASPECT_RATIOS,
+        )
+
+        target: dict[str, Any] = {
+            "short_edge": MINIMAX_H3_RECOMMENDED_SHORT_EDGE,
+            "aspect_ratio": "16:9",
+            "duration_seconds": 5.0,
+        }
+        if server_args.warmup_num_frames is not None:
+            target["duration_seconds"] = (
+                server_args.warmup_num_frames / MINIMAX_H3_SUPPORTED_FPS
+            )
+        if server_args.warmup_resolutions is not None:
+            ratio = req.width / req.height
+
+            def distance(name: str) -> float:
+                w, h = map(int, name.split(":"))
+                return abs(w / h - ratio)
+
+            target["aspect_ratio"] = min(MINIMAX_H3_FINITE_ASPECT_RATIOS, key=distance)
+        return target
+
     def project_video_queued_job_fields(self, req: Any) -> dict[str, str]:
         return self._video_hooks().project_queued_job_fields(req)
 
diff --git a/python/sglang/multimodal_gen/test/unit/test_minimax_h3_admission.py b/python/sglang/multimodal_gen/test/unit/test_minimax_h3_admission.py
index 571917e0c..c6c7e4d75 100644
--- a/python/sglang/multimodal_gen/test/unit/test_minimax_h3_admission.py
+++ b/python/sglang/multimodal_gen/test/unit/test_minimax_h3_admission.py
@@ -206,6 +206,22 @@ def test_loaded_weight_partition_admits_only_its_declared_tasks(partition, tasks
         metadata.canonical_task(rejected)
 
 
+def test_synthetic_warmup_target_honors_warmup_flags():
+    def target(num_frames=None, resolution=None):
+        width, height = map(int, (resolution or "896x512").split("x"))
+        req = SimpleNamespace(num_frames=17, width=width, height=height)
+        server_args = SimpleNamespace(
+            warmup_num_frames=num_frames,
+            warmup_resolutions=None if resolution is None else [resolution],
+        )
+        return MiniMaxH3SamplingParams._synthetic_warmup_target(req, server_args)
+
+    assert target() == TARGET
+    assert target(num_frames=345) == {**TARGET, "duration_seconds": 345 / 24.0}
+    assert target(resolution="768x1344") == {**TARGET, "aspect_ratio": "9:16"}
+    assert target(resolution="832x464") == TARGET
+
+
 def test_duration_admission_accepts_released_4_to_15_second_range():
     for duration in (4.0, 15.0):
         target = {**TARGET, "duration_seconds": duration}