Process the fixture with Auto and the default settings only, with no case-specific options, and keep the automatic classification, route, and output grid identical to the approved baseline. The target comparison additionally measures how far the output still is from the fixed target image.
convert-deterministic-auto-palette
metexplicitTarget qualitymet · Time 20.29ms · Grid confidence 1.0000
Keep the image at its original 32 x 32 pixel dimensions and preserve fully transparent pixels while reducing its 947 opaque input colors to an automatically selected eight-color palette with full-strength Ordered dithering.
restore-soft-edged-sprite-to-34x47
metexplicitTarget qualitymet · Time 1056.19ms · Grid confidence 1.0000
Restore a 1254 x 1254 AI-generated pixel-art-style knight, whose cell boundaries carry antialiasing instead of hard edges, to the hand-given 34 x 47 logical grid, keeping the sword tip, the helmet slits, and the cross on the shield readable as single cells. This output is the target that the Auto case for the same fixture is measured against.
restore-blocky-sprite-to-20x18
metexplicitTarget qualitymet · Time 1106.34ms · Grid confidence 1.0000
Restore a 1254 x 1254 AI-generated slime drawn in 39.7 px blocks to the hand-given 20 x 18 logical grid, keeping the single-cell eye highlight and the one-cell-thick mouth line intact. Reconstruction error alone also accepts a roughly 5x finer reading of the same image, so this output records which of the two scales is the intended one.
restore-thin-features-and-alpha-coverage
metexplicitTarget qualitymet · Time 5.38ms · Grid confidence 1.0000
Restore area-coverage alpha from enlarged artwork containing thin lines and highlights, while selecting an input RGB without mixing in hidden colors from transparent pixels.
convert-continuous-tone-balanced
metexplicitTarget qualitymet · Time 148.05ms · Grid confidence 1.0000
Route a 48 x 32 continuous-tone image through Auto to the Convert pipeline, derive three candidate resolutions from its aspect ratio and information density, and emit the balanced candidate at 24 x 16 with edge-aware resampling instead of restoring an original grid.
convert-illustration-detailed
metexplicitTarget qualitymet · Time 44.67ms · Grid confidence 1.0000
Convert a 72 x 48 illustration with transparent margins and thin high-contrast lines to the detailed candidate at 54 x 36, keeping fully transparent pixels transparent and never picking the RGB of a nearly transparent pixel as a cell's representative color.
guide-recipe1-knight-sprite
metexplicitTarget qualitymet · Time 14975.72ms · Grid confidence 0.5468
Process the 2048 x 2048 AI-generated knight published in the recipes guide, whose cell boundaries are antialiased over a magenta background, with the Best Match preset the page tells the reader to select, and reproduce the published 60 x 85 result pixel for pixel. A mismatch means the published image no longer follows from the published steps.
guide-recipe2-potion-icon
metexplicitTarget qualitymet · Time 20952.49ms · Grid confidence 0.5570
Process the 2752 x 1536 AI-generated potion bottle published in the recipes guide, drawn over a chroma-key green background, with the Transparent Icon preset the page tells the reader to select, and reproduce the published 16 x 23 result pixel for pixel, including the transparent background.
guide-recipe3-dragon-sprite
metexplicitTarget qualitymet · Time 31702.54ms · Grid confidence 0.5362
Process the 2816 x 1536 AI-generated green dragon published in the recipes guide, drawn over a red background, with the Retro Game Style preset the page tells the reader to select, and reproduce the published 44 x 47 result pixel for pixel, including the four-tone palette the conversion replaces the generated colors with.
guide-recipe4-landscape
metexplicitTarget qualitymet · Time 18615.17ms · Grid confidence 0.5272
Process the 2752 x 1536 AI-generated mountain landscape published in the recipes guide, whose pixel cells are softened by generation, with the Background Artwork preset the page tells the reader to select, and reproduce the published 256 x 144 result pixel for pixel without making any part of the full-frame scenery transparent.
guide-recipe5-chibi-character
metexplicitTarget qualitymet · Time 2647.35ms · Grid confidence 1.0000
Process the 2816 x 1536 AI-generated chibi character published in the recipes guide, an ordinary flat-colored illustration rather than pixel art, with the Convert Illustration to Pixel Art preset the page tells the reader to select, and reproduce the published 60 x 81 result pixel for pixel, including the transparent background and the thick outlines carried over from the illustration.
auto-auto-grid-detection
metautoTarget qualitymet · Time 19802.35ms · Grid confidence 0.4961
Process the fixture with Auto and the default settings only, with no case-specific options, and keep the automatic classification, route, and output grid identical to the approved baseline. The target comparison additionally measures how far the output still is from the fixed target image.
auto-high-resolution
metautoTarget qualitymet · Time 18554.98ms · Grid confidence 0.4691
Process the fixture with Auto and the default settings only, with no case-specific options, and keep the automatic classification, route, and output grid identical to the approved baseline. The target comparison additionally measures how far the output still is from the fixed target image.
Process the fixture with Auto and the default settings only, with no case-specific options, and keep the automatic classification, route, and output grid identical to the approved baseline. The target comparison additionally measures how far the output still is from the fixed target image.
auto-no-trimming
metautoTarget qualitymet · Time 14684.09ms · Grid confidence 0.5285
Process the fixture with Auto and the default settings only, with no case-specific options, and keep the automatic classification, route, and output grid identical to the approved baseline. The target comparison additionally measures how far the output still is from the fixed target image.
auto-quality-prf400-ambiguous-grid-scale
metautoTarget qualitymet · Time 15152.71ms · Grid confidence 0.5819
Process the fixture with Auto and the default settings only, with no case-specific options, and keep the automatic classification, route, and output grid identical to the approved baseline. The target comparison additionally measures how far the output still is from the fixed target image.
auto-resize-and-remove-bg
metautoTarget qualitymet · Time 6141.45ms · Grid confidence 0.6268
Process the fixture with Auto and the default settings only, with no case-specific options, and keep the automatic classification, route, and output grid identical to the approved baseline. The target comparison additionally measures how far the output still is from the fixed target image.
auto-resize-with-trimming
metautoTarget qualitymet · Time 2008.57ms · Grid confidence 0.4538
Process the fixture with Auto and the default settings only, with no case-specific options, and keep the automatic classification, route, and output grid identical to the approved baseline. The target comparison additionally measures how far the output still is from the fixed target image.