jiaxinyang1104/y2k-pixel-doodle

Y2K White Background Pixel Hand-Drawing

Using crooked mouse pixel lines, loose coloring, small colorful stickers, and large areas of white canvas, draw something you can recognize but that is "just off" in a Y2K personal homepage doodle. Let the low-res feel come from shape, line, color fill, and detail reduction simultaneously. Decide the content anew from the current input each time.

Extract Content from Input

• Confirm the role of each image. The original image determines the subject, pose, perspective, cropping, and spatial context. Style images provide line style, color blocks, pixel scale, white space usage, and decoration approach. Interfaces, logos, people, and props in reference screenshots are only included when specified by the user.
• Extract two to four anchor points sufficient to identify the subject, such as hairstyle and carried items, ear-tail direction, and key openings of objects. Retain action intent, object relations, and scene distance, allowing visible deviations in shoulder lines, clothing, accessories, outer contours, or non-critical proportions. Pets may have slightly uneven ears or paws; objects may have off-center structures. These local errors serve the awkwardness effect.
• For the face, prioritize preserving gaze direction, eye opening and eyelid shape, eyebrow angle, mouth opening and corner direction, maintaining the original emotion. When the user specifies a particular image's face, extract these expression relationships and redraw the features using the unified drawing style of the whole image. Facial references determine the expression; line rules determine the actual strokes. Eye, eyebrow, mouth, nose, and blush strokes should all share the same pixel scale and degree of imperfection as the body contours. Do not uniformly replace the original expression with dot eyes, default smiles, big-eyed cute faces, or added blush.
• Read the original image's pixel width and height. By default, deliver the same dimensions and aspect ratio, and retain the subject's proportion, position, lens, and cropping within the frame. Only change when the user explicitly specifies a new size or composition. Awkward deformation of local shapes occurs within the original composition; top and side white space are allocated as in the original. The original image serves as recognition and action reference; redraw directly from it. Using prompts that require precise tracing or perfectly symmetrical local shapes will weaken this skill's drawing style.
• Check dimensions both during generation and final export. If the tool output differs from the original image's width and height, apply proportional export adaptation, preserving the entire frame and pixel corners. Only fill extremely narrow background color edges if necessary. Avoid changing original content via non-proportional stretching or cropping. Verify the final file's pixel dimensions before delivery.
• Select a few main colors from the original image, then choose two or three bright contrasting colors for surrounding stickers. The subject's colors can be slightly off; clothing patterns, floral prints, and small accessories are condensed into rough representative symbols. When the user specifies a color change, use the new color scheme.

Redraw with Unified Pixel Drawing Style

• The whole image uses one pen. Use a single pixel grid, the same rough mouse-drawn pencil, consistent stroke weight and wobble. First, determine the virtual pixel grid, base stroke width, step corner scale, and jitter amplitude for the entire image. Redraw facial features, body, objects, background structure lines, and small colored stickers according to this unified setting. The base stroke width is typically one to two grid cells on the virtual grid, with local natural fluctuations kept within the same range. The background recedes mainly by reducing line segments, lowering color contrast, and leaving white space; lines still maintain the same pixel grain and turning habits.
• Facial features are homogeneous with surroundings. Eye sockets, pupil edges, eyebrow arcs, lips, and nose tips should also have step-like, slightly non-circular, and stroke errors of corresponding scale. Small features are summarized with fewer pixels, preserving expression relationships, avoiding switching to higher-resolution fine-drawing for small features. Highlights in the original image's eyes can be converted to small white blocks on the same grid; mouth fills and blush also use consistent hard-edged brush strokes. Prevent a drawing style break where smooth, perfectly round eyes, delicate eyelashes, or clean lip lines are embedded within rough outer contours. Human figures, animals, and cartoon characters all follow this rule.
• Shape and lines. Use awkward amateur MS Paint doodle, imperfect memory sketch, wonky pixel contours. The outer contour should have visible offsets, turning mistakes, re-strokes, broken lines, and uneven thickness. Local arcs may suddenly kink; accessories may be slightly misaligned; facial features are simplified according to expression anchors. For ordinary compositions, use a virtual draft of approximately 160–220 pixels wide to describe the low-res degree, adjust as needed for detail, then apply a nearest-neighbor enlargement feel to preserve corners. Actual output resolution follows the tool settings.
• Color fills. Use scribbly pixel brush fills, misregistered color, uneven bucket fill. Retain short strokes with inconsistent directions, missed white spots, and a few overlapping color patches inside the color block. Local areas may overflow the line or be one or two grid cells away from the contour. Let the color and outline be slightly mismatched; several roughly merged small color blocks replace fine textures. Keep the rest of the paper clean. Avoid uniform noise, overall blur, smooth gradients, and neatly closed vector fills.
• When there is a human, animal, or object subject: The subject has the clearest contour, highest line contrast, and main color blocks. The background surface defaults to white space. Select a few structural lines from the original that let people notice the space, such as a corner of a wall, a segment of a window frame, or a ground turn, maintaining approximate positions and occlusion. Lines are drawn curved, slightly slanted, intermittent, and paler in color, sharing the base stroke width and pixel grid, allowing many line segments to naturally remain blank. Window frames and partitions may only need a few key strokes; repeated divisions and grid lines are greatly simplified according to readability needs. Only retain background objects involved in the subject's action or narrative. Keep background attention behind the subject.
• When the subject is a pure landscape or environment: Individually analyze the visual center, skyline, foreground/midground/background relations, major building or terrain contours of the image, then decide which structural lines and color blocks carry the picture. Omit secondary textures with white space, retaining the skeleton that allows reading the location and spatial layers. Lines are equally crooked and loose; line density varies with the importance of each layer. When the user specifies a solid color background, use that color to host the picture.
• Small stickers: By default, scatter a few tiny colored pixel doodles in the white space gaps, usually three to six. The longest side of each pattern is about 3%–5% of the canvas's shorter side, and should be noticeably smaller than the character's head or the main object. Choose from hollow hearts, crooked stars, X marks, small flowers, waves, arrows, or casual circles each time, using small amounts of bright pink, acid green, cobalt blue, or yellow. Sizes, angles, and spacing are inconsistent; positions are slightly unexpected, while avoiding facial features and key background structures. Small patterns are also drawn with mis-struck mouse pixel lines, giving a childish, cheap personal homepage sticker feel. Omit when the user explicitly requests no decorations.

For example, a full-body character is recognized by hairstyle, action, original expression, and a few clothing colors; shoulder lines and garment folds can be drawn awkwardly. A pet is recognized by ear-tail posture, gaze, and main fur colors; the outer contour can be slightly mishandled. An object is established by key contours and connection relationships, with minor parts simplified from memory. A pure landscape organizes space according to its visual center.

Deliver as Requested by the User

• Prompt generation or reverse-engineering. Deliver a directly copyable prompt. First define the original image's subject, composition, and reference image roles, then specify pixel lines, color blocks, background, and a few exclusions. Each generation instruction explicitly requires that facial features, subject contours, background, and stickers are drawn by the same mouse pixel pen on the same grid, forbidding individual refinement of the face. When the user wants a general template, write the subject, pose, color palette, scene, and frame size as replaceable items. When the user provides specific images, directly replace with the observed details.
• Image generation or conversion. When the user requests image generation, use the currently available built-in image generation tool, input the correct original image and style image as references, and annotate their roles in the instruction. Multiple independent original images default to being generated separately; combine them only when the user requests composition. Use a previously generated image as the editing base only if explicitly required to continue from a finished work.
• Post-generation check. Compare the original and output images' width, height, subject proportion, and cropping. Confirm that gaze and mouth emotion have been preserved. At the same magnification, compare eyes, mouth, face contour, body contour, and small stickers' stroke width, pixel steps, and jitter. If facial features appear significantly rounder, finer, or as if pasted on, first re-simplify them according to the whole-image grid while maintaining the original expression. When there is a clear subject, the background must recede to a secondary position, with structural lines sufficient to convey space and with distortion. For pure landscapes, check visual center and spatial layers. Decorations remain tiny; crooked contours and loose color fills should be visibly present.