# Deep research 18: current R04 public record These are the two frozen public assessments, reproduced without new viewing or ranking. Repeated source descriptions do not add source frame credit. Current and prior variants have no supported material gap within the inspected static scope; no superiority or moving-quality approval is established. # Greenhouse portrait sequence ## Result **No supported material gap within the inspected static evidence under the previously fixed mechanism criteria.** This carries forward the completed review; no new viewing, decoding, or ranking was performed for this document. The illustrated subject in orange clothing remains recognizable across a wider green setting, a closer body view, and a face-led close view. The left-to-right one-panel, two-panel, three-panel build preserves the observed reference timing and fixed-window layout. Common edges, light gutters, and outer space remain coherent. The raised face, hair silhouette, collar, held drawing, and other focal shapes are readable in the inspected images. The simplified illustration has less photographic texture and different artwork from the reference. Those differences alone do not establish a failure of the fixed framing and readability criteria. No conspicuous deformation, smear, or panel seam is supported by the inspected detail views. This is not a claim that every fine detail in every frame is defect-free. ## Evidence scale The existing review contains seven still-image views of this content: - Four contact sheets cover all 59 native frame positions at 320 by 180 pixels per frame, a 1:6 linear reduction from 1920 by 1080. - One full native-raster image shows frame 29. - One unscaled right-panel crop revisits frame 29 at 552 by 824 pixels. - One side-by-side endpoint image contains unscaled left-panel crops from frames 0 and 58, each 552 by 824 pixels. The complete annotated image is 1104 by 848. Thus three time positions have some native-raster evidence; the other 56 have contact-scale evidence only. The 59 clock positions contain 56 distinct decoded RGB images. Repeated images, crops, and endpoint revisits do not earn extra unique content credit. The display may fit images, so physical one-to-one display sampling is not certified. The actual file's native clock was verified at 24000/1001 frames per second with 1001-tick frames on a 1/24000 time base. Contact observations place the additions at frames 11 and 19 and the final hold through frame 58. Fine native-scale detail at the addition boundaries was not separately inspected for this content. ## Unresolved questions Endpoint stills preserve the basic framing and pose while showing small changes in the hand and held drawing. They are not moving perception. Experienced pace, fluidity, synchronization, shimmer, and subtle temporal defects remain unverified. Distinct frame hashes do not prove meaningful motion. Reference recognition in the comparison material was disclosed rather than treated as new independent evidence. No isolation test was performed. The result does not approve moving quality, overall source quality, native Resolve delivery, or pixel isolation. It does not establish superiority over another version, and no winner is selected. --- # Observatory portrait sequence ## Result **No supported material gap within the inspected static evidence under the previously fixed mechanism criteria.** This carries forward the completed review; no new viewing, decoding, or ranking was performed for this document. The blue-clothed illustrated subject and night observatory-like setting remain recognizable across surrounding context, a closer body view, and a face-led close view. The left-to-right one-panel, two-panel, three-panel build preserves the observed reference timing and fixed-window layout. Common edges, light gutters, and outer space remain coherent. The raised face, hair, scarf, and held diagram are readable in the inspected images. The night palette, setting, character, and geometric rendering differ from the photographic reference. Those artwork choices alone do not establish a failure of the fixed framing and readability criteria. No conspicuous deformation, smear, or panel seam is supported by the inspected detail views. This is not a claim that every fine detail in every frame is defect-free. ## Evidence scale The existing review contains seven still-image views of this content: - Four contact sheets cover all 59 native frame positions at 320 by 180 pixels per frame, a 1:6 linear reduction from 1920 by 1080. - One full native-raster image shows frame 29. - One unscaled right-panel crop revisits frame 29 at 552 by 824 pixels. - One side-by-side endpoint image contains unscaled left-panel crops from frames 0 and 58, each 552 by 824 pixels. The complete annotated image is 1104 by 848. Thus three time positions have some native-raster evidence; the other 56 have contact-scale evidence only. The 59 clock positions contain 56 distinct decoded RGB images. Repeated images, crops, and endpoint revisits do not earn extra unique content credit. The display may fit images, so physical one-to-one display sampling is not certified. The actual file's native clock was verified at 24000/1001 frames per second with 1001-tick frames on a 1/24000 time base. Contact observations place the additions at frames 11 and 19 and the final hold through frame 58. Fine native-scale detail at the addition boundaries was not separately inspected for this content. ## Unresolved questions Endpoint stills preserve the basic framing and pose while showing small changes in the hand and held diagram. They are not moving perception. Experienced pace, fluidity, synchronization, shimmer, and subtle temporal defects remain unverified. Distinct frame hashes do not prove meaningful motion. Reference recognition in the comparison material was disclosed rather than treated as new independent evidence. No isolation test was performed. The result does not approve moving quality, overall source quality, native Resolve delivery, or pixel isolation. It does not establish superiority over another version, and no winner is selected.