{ "title": "The Architect's Stillness Video Model Sweep", "published_date": "2026-05-09", "run_id": "architects-stillness-20260508T232959Z", "client_id": "codex-architects-stillness-20260508T232959Z", "remote_movie_directory": "/srv/neonflux/shared/chat-assets/movie/architects-stillness-20260508T232959Z", "local_artifact_directory": "/Users/nation/Documents/New project/artifacts/architects-stillness-20260508T232959Z", "source_asset_id": "108e3c7634cd4a77958d716d77395d16", "finished_at": "2026-05-09T02:09:00.242Z", "headline": { "models_tested": 10, "completed": 3, "failed": 7, "best_completed_model": "CogVideoX 5B", "best_completed_prompt_match_10": 3.5, "production_readiness": "The current text-to-video lanes can produce reviewable atmosphere clips, but this prompt exceeds the narrative and object-binding ability of the completed models. Several image-to-video adapters need runtime fixes before creative evaluation." }, "source_keyframe_assessment": { "rating_10": 2.0, "verdict": "The generated source frame captured wet cinematic scale and lone figures, but missed the obsidian dome, exposed cybernetic brain, porcelain doll, golden neural ripple, and reflected-eye endpoint." }, "vitals": { "sample_count": 292, "first_sample": "2026-05-08T23:30:11.518Z", "last_sample": "2026-05-09T02:08:58.363Z", "avg_gpu_util_pct": 85.7, "p95_gpu_util_pct": 96.0, "max_gpu_util_pct": 96.0, "avg_temp_c": 57.8, "max_temp_c": 83.0, "avg_power_w": 31.7, "max_power_w": 93.66, "active_samples": 260, "multi_generator_samples": 24 }, "models": [ { "model": "Wan 2.1 T2V 1.3B", "model_id": "Wan-AI/Wan2.1-T2V-1.3B-Diffusers", "task": "text-to-video", "status": "completed", "latency_seconds": 343.94, "resolution": "832x480", "frames": 49, "fps": 16, "steps": 20, "guidance_scale": 5, "visual_rating_10": 1.0, "prompt_match_10": 0.5, "quality_rating_10": 1.0, "verdict": "Rendered a valid MP4 but the image is effectively black, so it fails the visual and narrative requirements.", "matched_prompt_elements": [], "missed_prompt_elements": [ "obsidian dome", "exposed brain", "child and doll", "soldier cutaway", "reflected eye" ], "failure_class": "", "error_excerpt": "" }, { "model": "CogVideoX 2B", "model_id": "zai-org/CogVideoX-2b", "task": "text-to-video", "status": "completed", "latency_seconds": 209.27, "resolution": "720x480", "frames": 49, "fps": 8, "steps": 28, "guidance_scale": 6, "visual_rating_10": 3.0, "prompt_match_10": 2.5, "quality_rating_10": 4.0, "verdict": "Produces a coherent Gothic/industrial surface with cracked-glass energy, but it is almost static and omits the central story objects.", "matched_prompt_elements": [ "dark Gothic architecture", "cold metal palette", "cracked/fogged surfaces" ], "missed_prompt_elements": [ "brain network", "child/doll action", "remote soldier", "final reflected eye" ], "failure_class": "", "error_excerpt": "" }, { "model": "CogVideoX 5B", "model_id": "zai-org/CogVideoX-5b", "task": "text-to-video", "status": "completed", "latency_seconds": 714.53, "resolution": "720x480", "frames": 49, "fps": 8, "steps": 32, "guidance_scale": 6, "visual_rating_10": 4.0, "prompt_match_10": 3.5, "quality_rating_10": 5.0, "verdict": "Best completed clip: it suggests a cracked dome/cathedral mood, but remains mostly environmental and does not execute the prompt sequence.", "matched_prompt_elements": [ "cathedral/dome silhouette", "high-contrast horror palette", "wet metallic atmosphere" ], "missed_prompt_elements": [ "visible cyborg brain", "gold neural ripple", "child/doll beat", "soldier control beat", "eye reflection" ], "failure_class": "", "error_excerpt": "" }, { "model": "Mochi 1 Preview", "model_id": "genmo/mochi-1-preview", "task": "text-to-video", "status": "failed", "latency_seconds": 20.73, "resolution": "848x480", "frames": 31, "fps": 8, "steps": 28, "guidance_scale": 4.5, "visual_rating_10": null, "prompt_match_10": null, "quality_rating_10": null, "verdict": "CUDA out of memory during model load/generation.", "matched_prompt_elements": [], "missed_prompt_elements": [], "failure_class": "CUDA out of memory during model load/generation.", "error_excerpt": "[perknation@192.168.12.251] {\"ok\": false, \"error\": \"CUDA error: out of memory\\nSearch for `cudaErrorMemoryAllocation' in https://docs.nvidia.com/cuda/cuda-runtime-api/group__CUDART__TYPES.html for more information.\\nCUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.\\nFor debugging consider passing CUDA_LAUNCH_BLOCKING=1\\nCompile with `TORCH_USE_CUDA_DSA` to enable device-side assertions.\"}" }, { "model": "SkyReels V2 DF 14B 540P", "model_id": "Skywork/SkyReels-V2-DF-14B-540P-Diffusers", "task": "text-to-video", "status": "failed", "latency_seconds": 7200.01, "resolution": "960x544", "frames": 49, "fps": 16, "steps": 24, "guidance_scale": 6, "visual_rating_10": null, "prompt_match_10": null, "quality_rating_10": null, "verdict": "Exceeded the 7,200,000 ms HF video runtime timeout at 960x544, 49 frames.", "matched_prompt_elements": [], "missed_prompt_elements": [], "failure_class": "Exceeded the 7,200,000 ms HF video runtime timeout at 960x544, 49 frames.", "error_excerpt": "[perknation@192.168.12.251] HF video runtime timed out after 7200000ms" }, { "model": "LTX Video 13B Distilled", "model_id": "Lightricks/LTX-Video-0.9.8-13B-distilled", "task": "image-to-video", "status": "failed", "latency_seconds": 48.88, "resolution": "704x512", "frames": 65, "fps": 12, "steps": 18, "guidance_scale": 3, "visual_rating_10": null, "prompt_match_10": null, "quality_rating_10": null, "verdict": "CUDA out of memory despite supplied source keyframe.", "matched_prompt_elements": [], "missed_prompt_elements": [], "failure_class": "CUDA out of memory despite supplied source keyframe.", "error_excerpt": "[perknation@192.168.12.251] {\"ok\": false, \"error\": \"CUDA error: out of memory\\nSearch for `cudaErrorMemoryAllocation' in https://docs.nvidia.com/cuda/cuda-runtime-api/group__CUDART__TYPES.html for more information.\\nCUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.\\nFor debugging consider passing CUDA_LAUNCH_BLOCKING=1\\nCompile with `TORCH_USE_CUDA_DSA` to enable device-side assertions. Hint: This model needs a source image. P" }, { "model": "HunyuanVideo I2V", "model_id": "hunyuanvideo-community/HunyuanVideo-I2V", "task": "image-to-video", "status": "failed", "latency_seconds": 300.56, "resolution": "960x544", "frames": 49, "fps": 16, "steps": 18, "guidance_scale": 1, "visual_rating_10": null, "prompt_match_10": null, "quality_rating_10": null, "verdict": "Runtime tensor/indexing failure after source-image handoff; likely adapter/source conditioning incompatibility.", "matched_prompt_elements": [], "missed_prompt_elements": [], "failure_class": "Runtime tensor/indexing failure after source-image handoff; likely adapter/source conditioning incompatibility.", "error_excerpt": "[perknation@192.168.12.251] {\"ok\": false, \"error\": \"index -1 is out of bounds for dimension 1 with size 0 Hint: This model needs a source image. Provide inputImageBase64 or inputAssetId from the public page. HunyuanVideo-I2V is heavy. If this fails, reduce resolution or keep only one active video job at a time.\"}" }, { "model": "SkyReels V1 Hunyuan I2V", "model_id": "Skywork/SkyReels-V1-Hunyuan-I2V", "task": "image-to-video", "status": "failed", "latency_seconds": 5.75, "resolution": "960x544", "frames": 49, "fps": 16, "steps": 22, "guidance_scale": 1, "visual_rating_10": null, "prompt_match_10": null, "quality_rating_10": null, "verdict": "Repository layout mismatch: missing model_index.json for the current diffusers loader path.", "matched_prompt_elements": [], "missed_prompt_elements": [], "failure_class": "Repository layout mismatch: missing model_index.json for the current diffusers loader path.", "error_excerpt": "[perknation@192.168.12.251] {\"ok\": false, \"error\": \"404 Client Error. (Request ID: Root=1-69fe94b5-6cc0a0a4608f58ed4b9d7af7;36e4dfcd-3711-4f12-ab74-86886dfd284d)\\n\\nEntry Not Found for url: https://huggingface.co/Skywork/SkyReels-V1-Hunyuan-I2V/resolve/main/model_index.json. Hint: This model needs a source image. Provide inputImageBase64 or inputAssetId from the public page.\"}" }, { "model": "SkyReels V2 I2V 14B 540P", "model_id": "Skywork/SkyReels-V2-I2V-14B-540P-Diffusers", "task": "image-to-video", "status": "failed", "latency_seconds": 607.73, "resolution": "960x544", "frames": 49, "fps": 16, "steps": 24, "guidance_scale": 5, "visual_rating_10": null, "prompt_match_10": null, "quality_rating_10": null, "verdict": "Tensor channel mismatch in image-conditioning path; model adapter does not match runtime input shape.", "matched_prompt_elements": [], "missed_prompt_elements": [], "failure_class": "Tensor channel mismatch in image-conditioning path; model adapter does not match runtime input shape.", "error_excerpt": "[perknation@192.168.12.251] {\"ok\": false, \"error\": \"Given groups=1, weight of size [5120, 36, 1, 2, 2], expected input[1, 16, 25, 68, 120] to have 36 channels, but got 16 channels instead Hint: This model needs a source image. Provide inputImageBase64 or inputAssetId from the public page.\"}" }, { "model": "Stable Video Diffusion XT", "model_id": "stabilityai/stable-video-diffusion-img2vid-xt", "task": "image-to-video", "status": "failed", "latency_seconds": 7.71, "resolution": "1024x576", "frames": 25, "fps": 7, "steps": 18, "guidance_scale": 2.5, "visual_rating_10": null, "prompt_match_10": null, "quality_rating_10": null, "verdict": "Runtime API mismatch: guidance_scale was passed to a pipeline that does not accept it.", "matched_prompt_elements": [], "missed_prompt_elements": [], "failure_class": "Runtime API mismatch: guidance_scale was passed to a pipeline that does not accept it.", "error_excerpt": "[perknation@192.168.12.251] {\"ok\": false, \"error\": \"StableVideoDiffusionPipeline.__call__() got an unexpected keyword argument 'guidance_scale' Hint: This model needs a source image. Provide inputImageBase64 or inputAssetId from the public page. Stable Video Diffusion XT is image-to-video only. Use a clear source frame and expect short clips.\"}" } ], "recommendations": [ "Add low-memory presets for Mochi, LTX, and 14B SkyReels before public default use.", "Exclude unsupported kwargs per pipeline, especially Stable Video Diffusion guidance_scale.", "Split complex cinematic prompts into keyframe-first workflows with explicit source frames for each major beat.", "Fix image-conditioning adapters for Hunyuan/SkyReels I2V before judging creative quality.", "After SSH timeouts, explicitly terminate the remote process group to prevent stale GPU jobs." ] }