Upload app.py

app.py

@@ -100,6 +100,10 @@ FPS, FRAMES_PER_CHUNK, LATENTS_PER_CHUNK = 24, 17, 5
 # 15.083 s, and is refused. 14 is the last whole second that survives the snap.
 MAX_UI_DURATION = 14
 MIN_DURATION = 2
+# The model's absolute floor is LATENTS_PER_CHUNK frames (5 @ 17*0+5). At that length the encoded mp4 is technically
+# valid but short enough that gr.Video / mobile WebView frequently fail to display it, so we also surface the raw
+# frames as stills whenever the take is at or under this threshold.
+SHORT_CLIP_GALLERY_FRAME_THRESHOLD = LATENTS_PER_CHUNK
 # A reference video shorter than 2 s gives the model almost no motion to read.
 MIN_REFERENCE_VIDEO, MAX_REFERENCE_VIDEO = 2.0, 15.0
 # `MINIMAX_H3_MAX_REFERENCE_IMAGES`. The slots are built up front and revealed one at a time, because a demo asking
@@ -1271,6 +1275,14 @@ def generate(
     # route's artifact; the runner's ffprobe gate should see audio_stream_count=0 before accepting the take.
     encode_video(frames, fps=FPS, output_path=path, audio=None)
 
+    # Below the practical playback floor, hand back the raw stills too, since the mp4 itself is frequently
+    # rejected or shown blank by the video element even though encoding succeeded.
+    gallery_update = (
+        gr.update(value=[frame.cpu().numpy() for frame in frames], visible=True)
+        if num_frames <= SHORT_CLIP_GALLERY_FRAME_THRESHOLD
+        else gr.update(value=None, visible=False)
+    )
+
     print(
         f"[ref2va] {[kind for kind, _ in references]} · `{width}x{height}`, {num_frames} frames "
         f"({num_frames / FPS:.3f} s), {int(steps)} steps · conditioner {condition_seconds:.0f}s "
@@ -1280,7 +1292,7 @@ def generate(
         f"{' · LoRA ' + ', '.join(lora_labels) if lora_labels else ''}",
         flush=True,
     )
-    return path, refined, gr.update(visible=bool(refined))
+    return path, refined, gr.update(visible=bool(refined)), gallery_update
 
 
 def p15b_preflight() -> dict:
@@ -1742,6 +1754,13 @@ with gr.Blocks(title="MiniMax-H3 Reference Custom Lora") as demo:
 
         with gr.Column():
             result = gr.Video(label="Видео со звуком")
+            # Short clips (5 frames / 0.2 s) are technically valid mp4s but many players (gr.Video, mobile WebView)
+            # fail to open a file that short — the "Ошибка в отображении материала" symptom. Rather than touch the
+            # denoising or the mp4 encoding, we additionally hand back the raw frames as stills whenever the take is
+            # at or under that floor, so there's always something visible even if the video player chokes.
+            frames_gallery = gr.Gallery(
+                label="Кадры (короткое видео)", visible=False, columns=5, height="auto"
+            )
             # An output, so it can be revealed only for a request that asked for a rewrite.
             with gr.Accordion("Расширенный промпт", open=False, visible=False) as upsampled_panel:
                 upsampled = gr.Textbox(show_label=False, lines=8, interactive=False)
@@ -1819,7 +1838,7 @@ with gr.Blocks(title="MiniMax-H3 Reference Custom Lora") as demo:
     )
     run.click(
         apply_short_clip, [short_clip, duration], effective_duration, api_name=False
-    ).then(generate, request, [result, upsampled, upsampled_panel], api_name=False)
+    ).then(generate, request, [result, upsampled, upsampled_panel, frames_gallery], api_name=False)
 
     # P15B v02 has exactly two public endpoints. They deliberately use File components rather than Image components:
     # Gradio 6 uploads arrive as FileData, and the runner must preserve this exact reference order.