{"id":85121,"plugin_id":"plugin_asdk_app_6a18d4c190a0819186e6b129a09e931e","kind":"skill","collection_source":"plugin_package","comparison_source":null,"observed_at":"2026-11-07T00:09:09.827Z","digest":"192b33d9cab5da16c5016c7b0b5f81799218cf145626a12618cdf4a4bbd82e68","against":null,"payload":{"description":"Audit media assets BEFORE they enter scene cards, boxes or looks - alpha padding, dimensions, aspect, video properties, URL-reuse hazards, and whether a photo in a reference is a masked view of a larger one. Use before authoring scenes with new assets or when a styled card renders with mysterious edges or misalignment.","included_files":[],"name":"asset-audit","skill_md_contents":"---\nname: asset-audit\ndescription: Audit media assets BEFORE they enter scene cards, boxes or looks - alpha padding, dimensions, aspect, video properties, URL-reuse hazards, and whether a photo in a reference is a masked view of a larger one. Use before authoring scenes with new assets or when a styled card renders with mysterious edges or misalignment.\nmetadata:\n  category: workflow\n  version: '1'\n  ported-from: asset-audit 1 at 55.75.0\n---\n\n# Asset audit - check the media before blaming the scene\n\nMost \"the mask is broken / content is misaligned / corners aren't rounded\" reports are ASSET\nproblems. Audit first; it takes a minute and saves a debugging afternoon.\n\n## The checklist\n\n1. **Alpha padding (the #1 trap).** Sticker-style PNGs bake transparent margins into the canvas:\n   `fit: cover` covers the box with the FULL canvas, so the visible content sits inset and any\n   rounding or mask lands in the transparent margin - the visible edge is the asset's own square\n   boundary. Check it by looking: build a one-layer scene whose composition is the asset's own size\n   (from `picsart_media_probe`), `background: \"#ff00ffff\"`, the asset with `fit: \"contain\"`, and\n   look at it with `picsart_scene_contact_sheet` (`resolution` set to the composition size). No\n   magenta inside the frame: clean. Magenta around the subject: a padded sticker; trim it, or use\n   it deliberately as a decal with no look.\n2. **Dimensions and aspect:** `picsart_media_probe` (no download; EXIF-corrected width and height,\n   duration for video). Cover-fit crops from decoded dimensions - a 4:3 photo in a 9:16 card loses\n   its sides; check the subject survives the crop. Also compare the asset's size with its largest\n   placement box: `scale = max(box_w / own_w, box_h / own_h)`, and a scale above 1 is an upscale. A\n   585x1024 photo in a 1080x1920 full-bleed box is stretched to nearly double. Rendering can decode\n   an asset whose dimensions are omitted, but query and selection geometry cannot invent them: for\n   `fit: \"none\"`, declare both `width` and `height` when the layer must be selectable or\n   queryable, or use its intrinsic centre as the default transform anchor. Without them the source\n   still renders at natural size, but `picsart_scene_query_layout` returns no bounds and the\n   fallback pivot is the content origin. Do not add `content.bounds` as a workaround: `fit: \"none\"`\n   has no destination box and rejects `bounds` and `align`. When a scene already declares an\n   asset's `width`/`height`, compare them with the probe: a mismatch means the document describes\n   a different file than the one that will load.\n3. **Video specifics:** duration against the trim window you plan (a trim beyond the end freezes\n   the last frame); first frames may decode dark or blank, so don't trim in exactly at 0 or at a\n   scene cut; embedded audio (mute it unless wanted); reversed playback does NOT reverse audio.\n4. **URL reuse:** fine. The same media URL in two different boxes renders correctly. Do NOT add a\n   distinct `#fragment` per instance: it defeats content dedup, and it silently rewrites a URI that\n   already carries a fragment of its own.\n5. **SVG data-URIs:** encode only `< > space`; raw spaces in interpolated text are tolerated, raw\n   `#` is not (colours as hex-without-# interpolated after `%23`). Declared width and height on the\n   asset matter - undeclared dimensions break mask plus bounds combinations.\n6. **Content vs box semantics:** an asset that is itself a full-bleed photo behaves under any fit;\n   UI screenshots and renders with baked-in margins do not - measure where the content actually\n   starts.\n\n## Is the reference element MASKED?\n\nWhen sourcing assets for a recreation of a reference video, decide for every photographic element\nwhether you are looking at a WHOLE asset or a MASKED VIEW of a larger one. Baking the mask into the\nbitmap gives the right pixels and the wrong scene: the photo can no longer be swapped, the mask\ncannot animate, and the \"asset\" is a screenshot of someone else's render. Narrow slivers cut from\nthe visible part of a card, or a full frame grabbed as a \"photo\", are this mistake.\n\nHow to tell, cheapest test first:\n\n1. **Does content continue past the edge across frames?** A masked view is a WINDOW: as the card\n   moves, grows, or the mask animates, NEW parts of the same photo appear at the edge while the\n   rest stays registered. Sample the element at 3-4 timestamps in its own span\n   (`picsart_media_video_stills`) and check whether its content translates WITH the box (a whole\n   asset moving) or STAYS PUT while the box reveals more of it (a mask sliding over a fixed\n   image). The second is a mask, always.\n2. **Is a feature cut mid-stride?** A hard straight edge slicing a face, a hand, or a horizon\n   mid-feature is a mask boundary; a real asset's edge lands where the photographer put it.\n3. **Does the aspect ratio look impossible?** 238x1095 (1:4.6) is not a photograph anyone shot; it\n   is a viewport. Compare the element's aspect against plausible source aspects (3:2, 4:3, 1:1,\n   9:16).\n4. **Corner geometry:** `picsart_media_inventory_diff` (reference as `a`, your candidate as `b`,\n   same frame size, `times` of interest; it spends credits) reports whether corners are rounded\n   against the local surround. A rounded photographic corner means a mask or a `rounded_corners`\n   look, not a rounded photo. Leave `plate` at `auto` for cards over a uniform background; with it\n   off, a pale-topped card can split into two invented elements.\n\nWhat to author once identified: the FULL asset plus a `mask` (or the look that owns the shape),\nnever a pre-cut bitmap. The mask can then animate and the photo can be replaced. Source ONE\nfull-size asset, not a sliver cut to the mask's current opening: two slots showing one photo at two\nscales bind the SAME URL, differing only by the host layer's scale.\n\nNever take pixels from the reference clip itself as an asset; use what the user supplied, and a\nstand-in only when none was.\n\n## Debug heuristic\n\nA styled card shows square edges but sibling cards are fine: suspect THIS asset's alpha (swap the\nphoto first, the scene second). A masked thing renders tiny or misplaced: mask plus explicit\nbounds on one layer (use a boxed aux composition with a bounds-less masked layer inside). Content\ninvisible at layer start: fade-in or decode timing, start later.\n"},"changes":[],"summary":"First saved snapshot. No earlier version is available for comparison.","summary_kind":"deterministic","summary_metadata":{}}