浓色马克笔人物画

Dense Color Marker Portrait Painting | Turning a real person into an interesting piece of work

Creation Task

Using the photo uploaded by the user as the sole reference for the character, create a marker portrait painting with a sense of contemporary art.

Retain the identity, expression, posture, and recognizability of the person in the photo, but completely reorganize the colors, brushstrokes, and pictorial space. It is not applying a painting filter to the photo, but rather like an observant painter with a sense of humor who, after seeing this person, uses a few broad-tip markers with rich colors to redraw this moment.

The picture should be bold, thick, relaxed, and have distinct handmade traces. At first glance, one is attracted by the large color blocks; at second glance, one recognizes the person in the photo; at third glance, one discovers that slightly funny handwritten text next to them.

I. Artistic Direction

Color and shape organization references the distinct pure color relationships, simplified character silhouettes, and the rhythm between color blocks and negative space in Henri Matisse's paper cut-out works. Translate these inspirations into marker drawings rather than making paper cut-out collages. (Tate Gallery)

Text and humor reference the concise drawings, handwritten short sentences, and observations of the absurdity of daily life in David Shrigley's works. Only borrow this humorous relationship between text and image; this creation should be warmer, not adopting a dark, cynical tone, nor copying his specific works or sentences. (NGV)

Unify into a clear visual language: dense colors, broad strokes, simplified shapes, true expressions, a small number of handwritten side notes.

Do not mechanically mix multiple painting styles. The character is not a fixed role of a certain artist and must still be the person in the photo.

II. Recognize the person first, then simplify the person

Before drawing, first identify the most recognizable features in the photo: facial contour, hairstyle and hairline, the relationship between the eyes, the position of the nose and mouth, head angle, shoulder direction, posture, and memorable clothing or accessories.

Organize the picture around these features instead of depicting all details equally.

Retain the character's original sense of age, skin tone depth relationships, gender presentation, and bodily features. Retain obvious moles, freckles, glasses, beards, and a small number of facial lines that help identify identity.

Strictly retain the original expression and movement logic: closed eyes cannot be changed to open eyes, closed mouth cannot be changed to a toothy smile, profile face cannot be changed to a full face, and a back view cannot have a face added without authorization. Which direction the hair spreads and where the body tilts should continue from the original image.

Can simplify anatomical details, but the character cannot be reshaped into another face through beautifying, face thinning, enlarging eyes, or shrinking the nose.

Humor mainly comes from text and observation, not from drawing characters ugly.

III. Medium: Thick Opaque Pigment Markers

Set the drawing medium to opaque pigment markers, leaning towards an acrylic marker texture, using a wide, flat tip to lay down colors, a medium tip to establish shapes, and finally a small number of fine pens to complete necessary details and text.

Large color blocks should be full, solid, and have a sense of coverage, but there should be no three-dimensional piling up like oil paints.

Let people see how the pen passes across the paper surface: strips left by broad brush sweeps, slightly stiff corners when turning, seams of repeated overlapping, incompletely neat edges where strokes end, and small gaps formed by a few pen tips skipping over the paper texture.

Most color surfaces remain intact, showing these traces only locally. Do not replace real brushstrokes with granular noise spread across the entire picture.

Draw color blocks first, then add a small number of lines. Do not use uniform black lines to outline all contours first and then fill them in like a coloring book.

Allow local coverage, slight misalignment, and incompletely closed boundaries. Retain the drawing process, rather than polishing every stroke to perfection.

IV. Establish Characters with Large Color Blocks

Summarize the hair, face, neck, top, arms, and necessary accessories into a few heavy shapes. The relationship between shapes is more important than fine textures.

The face is mainly composed of a basic skin tone surface, two or three structural color blocks, and a small number of facial feature strokes. Cancel pores, photo micro-textures, fine reflections, and complex light-and-dark gradients, but retain the facial structure that supports identity recognition.

Eyes are completed with a small number of short strokes and dark color dots, and the direction of the gaze must be accurate. Do not universally draw them as large round eyes, and do not add exaggerated highlights and eyelashes.

Eyebrows are drawn with short strokes of varying thickness, retaining their original direction and distance.

The nose can be mainly prompted by a warm color block and one or two dark strokes, without precise three-dimensional modeling.

The mouth retains its original open/closed state, corner direction, and emotion. Draw teeth only when the original picture indeed shows teeth, and summarize them into simple shapes.

For hair, draw the overall outline first, then add a small number of strokes that explain the direction. For dark hair, deep brown, ink blue, or near-black color blocks can be used; light or gray-white hair retains its recognizability and is not uniformly changed to brown hair. Important stray hairs can be like suddenly drawn fine lines, loosening the character's edges.

衣服保留款式、领口、层次和关键色彩关系。褶皱只留下解释姿势所必需的几笔。 Clothing retains the style, neckline, layering, and key color relationships. Folds leave only a few strokes necessary to explain the posture.

When the gesture is the focus of the photo, the action must be retained; when the hand is not the focus, details can be reduced, but the position cannot be changed arbitrarily.

V. Color: Thick, Not Dull; Bright, Not Noisy

Abandon the uniform light pastel, cream skin tone, and pale blue background settings in the original prompt.

Based on the clothing, skin tone, and emotion in the photo, select a limited but powerful color palette. Usually use four to six main colors, with one dominant color establishing the overall atmosphere, one contrasting color creating tension, and then a small amount of accent color lighting up the picture.

Candidate colors include cobalt blue, ultramarine, vermilion, tomato red, ocher yellow, deep grass green, peacock green, eggplant purple, wine red, and warm ivory white. Select only a few suitable ones for each painting, do not put all colors in at the same time.

Let large areas of color truly participate in the composition, instead of just painting the background brightly while the character still retains photographic realism.

Skin tone allows artistic translation; color surfaces such as ocher, terracotta, apricot, or warm brown can be used, and cool colors that echo the background can be introduced locally. Retain the deep or light characteristics of the original character's skin tone, do not uniformly bleach it, and do not cut the face into messy rainbow fragments.

Use the warmth/coolness and brightness difference of adjacent color blocks to establish volume, without relying on soft-focus gradients, airbrush shadows, or cinematic lighting effects.

VI. Composition and Background

Extract the main characters based on the original image, deleting unrelated passersby, miscellaneous objects, and complex environments. If it is a group photo with no obvious primary or secondary characters, retain the relationship between the core characters and do not delete any of them arbitrarily.

By default, continue the horizontal or vertical orientation and key actions of the original image. Moderate cropping, enlarging, and adjusting of the main subject's position are allowed, but meaningful full-body actions should not be uniformly cropped into headshots.

Adopt a relaxed yet accurate composition. The character can be slightly biased to one side to leave a natural position for text; do not mechanically center it for the sake of symmetry.

The background is composed of a solid block of dominant color or two large color areas, with a small amount of warm white paper surface allowed to remain. Create space through the relationships between the edges of color blocks, character contours, and blank space, without recreating the full photographic environment.

Retain an object only when it directly constitutes the fun of the photo or the character's traits, such as a cup in hand, a hat blown by the wind, or a pet in arms. Paint it likewise as a simple color block.

Do not pointlessly add plants, stars, hearts, stickers, or geometric decorations. What the picture needs is selection and omission, not being piled high with design elements.

VII. Handwritten Text: Like a Painter Looking at This Person and Couldn't Help Writing a Line

First understand the actions, expressions, clothing, or objects visibly present in the photo, and then create corresponding humorous short sentences.

The text should make people feel: "This sentence was thought of only after looking at this photo."

Adopt slight contrasts, deadpan jokes, personified minor complaints, or small absurdities in daily life. Do not directly state what is in the photo, and do not use inspirational chicken soup or popular slogans unrelated to the picture.

Mainly in English, a shorter Chinese sentence can be added as appropriate; sentence-by-sentence bilingual translation is unnecessary. The main sentence should be controlled within three to eight English words, and an additional small side note of one to three words can be added. Except for the date, the text can form two groups at most.

The following only demonstrates the humor mechanism, it is not fixed copywriting, do not write all of it into the picture:

When the hair is obviously messy, you can write:
"MY HAIR HAS OTHER PLANS."
When the character is seriously holding a small cup of drink, you can write:
"SMALL CUP. BIG RESPONSIBILITIES."
When the character is clearly looking into the distance, appearing in a spaced-out posture, you can write:
"CURRENTLY BETWEEN THOUGHTS."
It must be re-judged and created based on the actual photo. Do not use these sentences if there are no corresponding visual clues, and do not add props to characters just to match the sentences.

Humor targets the situation, not deprecating the character's appearance or identity. When the photo's atmosphere is solemn, make the text lighter and more subtle, without forcing a punchline.

The text must look like handwriting: strokes have variations in pressure, letters vary slightly in size, the baseline wavers slightly, and spacing is naturally uneven, but spelling is correct and clearly legible.

Use a dark pen to write on light areas, or use a warm white pigment pen to write on dark areas. The text shares the same handmade texture with the character, unlike computer fonts overlaid in post-processing.

There can be a short hand-drawn underline or a small arrow that genuinely points to something. Do not let it turn into a page full of graffiti.

VIII. Timestamp Signature

After completing the painting and side notes, leave the actual creation date in smaller, lighter handwriting in the bottom right corner or the blank space below the picture.

The date uses the 
http://
YYYY.MM.DD format; when the specific time can be confirmed, the 
http://
YYYY.MM.DD · HH:mm format can be adopted.

Use the date and time verifiable when generating the work, giving priority to following the user-specified time zone. When the hour and minute cannot be confirmed, write only the date, do not invent an exact time. Do not forge the generation date as the photo capture date, and do not draw format placeholders directly into the work.

The signature is like a stroke casually left by the painter after finishing, separating its hierarchy from the main copywriting. Do not add fictional artist signatures, brand logos, or watermarks.

IX. Final Control

Avoid realistic photo textures, semi-realistic AI oil paintings, soft airbrush gradients, transparent watercolor smudges, pastel candy colors, plastic skin, anime big eyes, three-dimensional cartoons, vector contours, full-screen black outlines, and overly detailed hair and fabric textures.

Also avoid the other extreme: do not understand "artistic feel" as randomly destroying facial features, "handmade feel" as messy and dirty, or "humor" as piling up jokes.

The final picture should look like a marker portrait piece that can be hung in a small gallery: from afar, the color blocks are distinct and orderly; up close, the brushstrokes have pauses, overlaps, and a human touch; looking again, the character is still the person in the photo, and a small line of text next to them suddenly gives character to the whole moment.

Draw fewer irrelevant details, make the colors carry more weight, and paint the person to look more like themselves.

浓色马克笔人物画｜把一个真实的人，画成一件有趣的作品

创作任务

以用户上传的照片为唯一人物参考，创作一幅具有当代艺术感的马克笔人物画。

保留照片中人物的身份、神态、姿势和辨识度，但彻底重新组织颜色、笔触和画面空间。不是给照片套上绘画滤镜，而是像一位有观察力、有幽默感的画家，看过这个人之后，用几支颜色浓郁的宽头马克笔，把这个瞬间重新画出来。

画面应当大胆、浓厚、松弛，有鲜明的手工痕迹。第一眼被大色块吸引，第二眼认出照片中的人，第三眼发现旁边那句有点好笑的手写文字。

一、艺术方向

色彩与形状组织参考 Henri Matisse（亨利·马蒂斯）剪纸作品中鲜明的纯色关系、简化的人物轮廓，以及色块与留白之间的节奏。把这些启发转译为马克笔绘画，而不是制作剪纸拼贴。(泰特艺术馆)

文字与幽默参考 David Shrigley（大卫·史瑞格里）作品中简练的绘画、手写短句，以及对日常生活荒诞之处的观察。只借鉴这种图文之间的幽默关系；本次创作应更温暖，不沿用黑暗、尖刻的语气，也不复制他的具体作品或句子。(NGV)

统一成一种明确的视觉语言：浓色、宽笔、简化造型、真实神态、少量手写旁注。

不要机械混合多种画风。人物不是某位艺术家的固定角色，必须仍然是照片里的那个人。

二、先认出这个人，再简化这个人

绘画前，先识别照片中最有辨识度的几个特征：脸部轮廓、发型与发际线、眼睛之间的关系、鼻子与嘴的位置、头部角度、肩膀方向、姿势，以及有记忆点的服装或配饰。

围绕这些特征组织画面，而不是平均描绘所有细节。

保留人物原本的年龄感、肤色深浅关系、性别呈现与身体特征。保留明显的痣、雀斑、眼镜、胡须，以及帮助辨认身份的少量面部线条。

严格保留原始表情与动作逻辑：闭眼不能改成睁眼，闭嘴不能改成露齿微笑，侧脸不能改成正脸，背影不能擅自补出面孔。头发朝哪个方向散开，身体朝哪里倾斜，都应延续原图。

可以简化解剖细节，但不能通过美化、瘦脸、放大双眼、缩小鼻子，把人物重塑成另一张脸。

幽默主要来自文字和观察，不靠把人物画丑。

三、媒介：浓厚的不透明颜料马克笔

将绘画媒介设定为不透明颜料马克笔，偏丙烯马克笔质感，以宽扁笔头铺色，以中等笔头建立形状，最后用少量细笔完成必要细节和文字。

大色块应饱满、扎实、具有覆盖感，但不要出现油画颜料般的立体堆积。

让人能看见笔是怎样经过纸面的：宽笔横扫留下的条带，转弯时略生硬的折角，重复覆盖的接缝，收笔处不完全齐整的边缘，以及少量笔尖跳过纸纹形成的细小缺口。

大部分色面保持完整，只在局部露出这些痕迹。不要用遍布全图的颗粒噪点代替真实笔触。

先画色块，再补少量线条。不要先用均匀黑线勾出全部轮廓，再像填色本一样填满。

允许局部覆盖、轻微错位和不完全闭合的边界。保留画过的过程，不把每一笔修得完美。

四、用大色块建立人物

把头发、脸、颈部、上衣、手臂和必要配饰，概括成少数几块有重量的形状。形状之间的关系比细碎纹理更重要。

脸部主要由一个基础肤色色面、两三处结构性色块和少量五官笔画构成。取消毛孔、照片微纹理、细碎反光和复杂明暗渐变，但保留支撑身份辨识的面部结构。

眼睛用少量短笔和深色色点完成，眼神方向必须准确。不要统一画成大圆眼，不添加夸张的高光和睫毛。

眉毛用有粗细变化的短笔画出，保留原本的走向和距离。

鼻子可以主要由一块暖色与一两笔暗色提示，不进行精密的立体塑形。

嘴部保留原始开合状态、嘴角方向与情绪。只有原图确实露出牙齿时才画牙齿，并将其概括为简单形状。

头发先画整体轮廓，再添加少量能说明方向的笔画。深色头发可使用深褐、墨蓝或近黑的色块；浅色、灰白色头发保留其辨识度，不一律改成棕发。重要的碎发可以像突然划出的细线，松动人物边缘。

衣服保留款式、领口、层次和关键色彩关系。褶皱只留下解释姿势所必需的几笔。

手势是照片重点时，必须保留动作；手不是重点时，可以减少细节，但不能随意改变位置。

五、色彩：浓重，不沉闷；鲜明，不嘈杂

放弃原提示词中统一的浅粉彩、奶油肤色和淡蓝背景设定。

根据照片里的服装、肤色和情绪，选择一组有限而有力度的配色。通常使用四至六种主要颜色，以一种主导色建立整体气氛，一种对照色制造张力，再用少量强调色点亮画面。

候选色包括钴蓝、群青、朱红、番茄红、赭黄、深草绿、孔雀绿、茄紫、酒红和暖象牙白。每张画只选择适合它的几种，不要把所有颜色同时放进去。

让大面积颜色真正参与构图，而不是只把背景涂得鲜艳、人物仍然保留照片式写实。

肤色允许艺术化转译，可使用赭石、陶土、杏桃或暖棕等色面，并在局部引入与背景呼应的冷色。保留原人物肤色的深浅特征，不统一漂白，也不要把脸切成杂乱的彩虹碎片。

用相邻色块的冷暖与明度差建立体积，不依赖柔焦渐变、喷枪阴影或电影光效。

六、构图与背景

根据原图提取主要人物，删除无关路人、杂物和复杂环境。若是没有明显主次的合影，保留核心人物之间的关系，不任意删掉其中一人。

默认延续原图的横竖方向与关键动作。可以适度裁切、放大和调整主体位置，但不要把有意义的全身动作一律裁成头像。

采用松弛而准确的构图。人物可以略微偏向一侧，为文字留出自然的位置；不要为了对称而机械居中。

背景以一整块主色或两块大的颜色区域构成，可以保留少量暖白纸面。用色块边缘、人物轮廓与空白之间的关系创造空间，不还原完整的摄影环境。

只有当某个物件直接构成照片的趣味或人物特征时，才保留它，例如手中的杯子、被风吹起的帽子、怀里的宠物。把它同样画成简洁色块。

不要无意义地添加植物、星星、爱心、贴纸或几何装饰。画面需要的是取舍，不是堆满设计元素。

七、手写文字：像画家看着这个人，忍不住写了一句

先理解照片中真实可见的动作、表情、衣物或物件，再创作与之对应的幽默短句。

文字应当让人觉得：“这句话是看过这张照片才想到的。”

采用轻微反差、一本正经的玩笑、拟人化的小抱怨，或日常生活中的微小荒诞。不要直接说明照片里有什么，也不要使用励志鸡汤和与画面无关的流行口号。

英文为主，可酌情加入一句更短的中文；不必逐句双语翻译。主句控制在三至八个英文单词，另可添加一个一至三个单词的小旁注。除日期外，文字最多形成两组。

以下仅示范幽默机制，不是固定文案，不要全部写入画面：

头发明显凌乱时，可以写：
“MY HAIR HAS OTHER PLANS.”
人物认真拿着一小杯饮料时，可以写：
“SMALL CUP. BIG RESPONSIBILITIES.”

人物明显望向远处、呈放空姿态时，可以写：
“CURRENTLY BETWEEN THOUGHTS.”
必须根据实际照片重新判断和创作。没有对应的画面线索就不要使用这些句子，也不要为配合句子给人物添加道具。

幽默针对情境，不贬损人物外貌或身份。照片气氛庄重时，让文字更轻、更含蓄，不强行制造笑点。
文字必须像手写：笔画有压力变化，字母略有大小差异，基线轻微起伏，间距自然不均匀，但拼写正确、清晰可读。

使用深色笔写在浅色区域，或用暖白色颜料笔写在深色区域。文字与人物共享同一种手工质感，不像后期叠加的电脑字体。

可以有一条短短的手绘下划线，或一根确有指向作用的小箭头。不要变成满页涂鸦。

八、时间戳落款

完成绘画与旁注后，在画面右下角或下方留白处，以更小、更轻的手写字落下实际创作日期。

日期采用 
http://
YYYY.MM.DD 格式；能够确认具体时间时，可采用 
http://
YYYY.MM.DD · HH:mm 格式。

使用生成作品时可核实的日期与时间，优先遵循用户指定的时区。无法确认时分时，只写日期，不编造精确时间。不要把生成日期冒充照片拍摄日期，也不要把格式占位符直接画进作品。

落款像画家画完后随手留下的一笔，与主文案拉开层级。不添加虚构的艺术家签名、品牌标志或水印。

九、最终控制

避免写实照片质感、半写实 AI 油画、柔软喷枪渐变、透明水彩晕染、粉彩糖果色、塑料皮肤、动漫大眼、三维卡通、矢量轮廓、满画面黑色描边，以及过度精细的头发和织物纹理。

也避免另一种极端：不要把“艺术感”理解成随意破坏五官，把“手工感”理解成脏乱，把“幽默”理解成堆砌段子。

最终画面应当像一幅可以挂在小型画廊里的马克笔人物作品：远看，色块鲜明而有秩序；近看，笔触有停顿、有覆盖、有人的手感；再看，人物仍然是照片里的那个人，而旁边的一句小字，让整个瞬间突然有了性格。

少画一些无关细节，把颜色画得更有分量，把人画得更像他自己。