HUVA academy
Միանալ դասընթացին
Անվճար ուղեցույցներ
Gen AI · Prompt

Ինչպես քո նկարներից ստանալ վիզուալներ, որոնք անկկատ չեն մնա սոց ցանցերում

AM Ani Mkrtchyan 5 minutes Updated June 2026

AI ով կարող ես քո կոնտենտը դարձնել այնպիսին, որ նման չլինի ոչ մեկի։

Վերցնում ես քո նկարը։ GPT ին ասում ես վրան ավելացնել գրաֆիկ կամ նկարազարդում. ինչ որ այդ կադրին է կպչում։ Հետո Seedance 2.5 ում նկարը մնում է սառած, շարժվում է միայն այդ շերտը։

Մաս 1

Երեք քայլ

  1. Քո նկարը. Ինքդ ես նկարել։ Դեմքը, հագուստը, վայրը մնում են ինչպես կան։
  2. GPT (image). Կցիր նկարը ու խնդրիր վրան ավելացնել գրաֆիկ կամ նկարազարդում. Instagram carousel ի տեսք, նկարը մնում է առաջին պլանում, ավելացվածը հենց այդ կադրի համար է։
  3. Seedance 2.5. Կցիր GPT ի արդյունքը։ Նկարը սառած է։ Շարժվում է միայն գրաֆիկ շերտը. կարծես նույն ձևը նորից ես գծում անկայուն ձեռքով, մոտ 6 անգամ վայրկյանում։

GPT prompt (կցիր նկարը)

GPT graphics on photo
Use the ATTACHED PHOTO as the base image and keep the original photo as unchanged as possible.

Do not alter the person, face, body, pose, clothing, hair, proportions, lighting, background, architecture, landscape, perspective, framing, orientation, or aspect ratio. Do not redraw the photo or turn it into an illustration.

Add playful hand-drawn graphic doodles on top of the photo, inspired by a modern fashion / lifestyle Instagram carousel aesthetic.

First, analyze the specific photo:

the location and environment
the mood
the person’s pose and movement
the outfit and its colors
the empty areas of the composition
the dominant background colors
the visual contrast of different areas

Then choose graphic motifs that make sense specifically for THIS photo.

Do not use the same default elements on every image.
Do not automatically add suns, flowers, stars, hearts, or generic decorative symbols.

For example, a seaside image might suggest abstract waves, birds, shells, starfish, breeze lines, coastal plants, or other context-appropriate motifs. A city image might suggest graphic ornamental shapes, architectural-inspired doodles, abstract lines, or stylized botanical accents. A nature image might suggest leaves, wind, organic lines, plants, or other forms that fit the scene.

These are only examples. Choose the actual elements based on the image itself.

STYLE OF THE GRAPHICS:

simple, playful, naive hand-drawn illustration
visibly handmade, as if drawn with a marker, brush, gouache, crayon, or soft pastel
slightly uneven, organic lines
deliberately imperfect shapes
subtle handmade texture
bold enough to read clearly
medium-to-thick outlines where outlines are used

Do NOT make the graphics look like clean vector art.
Do NOT make them too perfect, symmetrical, geometric, polished, or sticker-like.
Do NOT use very thin outlines.
Do NOT add heavy grain or excessive texture.

COMPOSITION:
Use a medium amount of graphics.

The result should feel neither too minimal nor overloaded.

As a general guide:

around 3–6 noticeable graphic elements or grouped motifs
1–2 larger visual elements
several medium supporting elements
only a few small accent marks if they are truly needed

Prefer fewer larger elements over many tiny decorative details.

The graphics should remain clearly visible when the image is viewed on a phone.

Do not fill every empty area with doodles.

LAYERING AND DEPTH:
Integrate the graphics spatially into the photo.

Some graphic elements should sit BEHIND the person or behind parts of the body.

Use masking / occlusion so that an element can:

begin on one side of the person
disappear behind the body, arm, skirt, hair, or legs
reappear on the other side

Other elements can sit beside the person or occasionally in front.

The result should feel like the doodles exist as separate visual layers inside the scene, not simply pasted flat on top.

Do not cover the face.
Do not obscure important outfit details.

COLOR:
The graphic elements must have enough contrast against the exact area where they are placed.

Do not use colors that blend too much into the background.

For example:

on a blue background, do not make all doodles light blue
on a warm sandy background, use colors that separate clearly from the sand
on a dark background, use brighter or lighter tones

Use a small, harmonious palette of roughly 2–4 main colors.

The colors should relate to the outfit and environment, but still remain clearly visible.

Keep them playful, editorial, and fresh rather than neon or overly saturated.

OVERALL FEEL:
fashion travel editorial
playful hand-drawn doodles
casual handmade imperfection
fresh and artistic
Instagram carousel aesthetic
youthful but tasteful
visually balanced

The graphics should enhance the story and mood of the photo instead of looking like random stickers.

DO NOT ADD:

text
captions
letters
logos
borders
random symbols
a sun just because the photo was taken during the day
repetitive flower patterns
many tiny decorative details
thin hard-to-read outlines
overly polished vector graphics
excessive texture
too many elements

MAIN RULE:
photo first → understand the context → then add a small set of strong, contrasting, hand-drawn elements created specifically for that image.

The final result should look like an original fashion or travel photograph that has been thoughtfully enhanced with custom handmade graphic doodles designed specifically for that exact photo.

Seedance 2.5 prompt (կցիր GPT ի արդյունքը որպես հիմնական նկար)

Seedance 2.5 graphic overlay
Live-action photographic footage, shot on a real camera. is the sole visual master and defines the entire frame. Reproduce it exactly: add nothing, remove nothing, reposition nothing, recolor nothing, restyle nothing.

The photographic part of — the real photograph of the location and every person in it — stays photographic from the first frame to the last. It keeps real camera detail, real skin and fabric texture, real depth of field, the original colors and contrast. Nothing is applied to it: no outline, contour line, stroke or border is ever added around a person, a body, limbs, hair, clothing, footwear, rocks, plants or the horizon; no cartoon, anime or comic look; no cel shading; no painterly or brush-stroke rendering; no illustration, poster, vector or flat-color treatment; no skin smoothing; no added saturation; no regrading. Every silhouette in the photograph stays a soft photographic edge and is never traced.

The photographic part is also completely frozen. Carry it forward unchanged from frame to frame instead of re-rendering it: every pixel stays identical for the full duration. The camera does not move, zoom, pan, tilt, shake or breathe; no parallax, no reframing, no crop drift, no exposure shift. Nothing in it is alive: hair, fabric, skin, water, waves, clouds, grass and foliage do not move, sway, ripple or drift by even one pixel. No texture shimmer, no crawling detail on grass, gravel, sand, stone, hair or fabric, no flicker along edges or silhouettes, no grain drift. Silhouettes never wobble, stretch or reshape.

Separate from all of that, also carries a graphic overlay: simple marker shapes sitting on top of the photograph. These shapes are the only thing in the entire video that changes. Six times per second, each shape is replaced by a clearly different version of itself, held for about four frames before the next version appears. The versions never blend, morph or interpolate into each other — each one simply cuts to the next.

The difference between consecutive versions must be obvious at a glance, as if the same shape were redrawn from scratch each time by an unsteady hand. Measure every offset in stroke widths, not in pixels. From one version to the next: the outline shifts by one to two stroke widths along its length; the far ends of rays, petals, tails and limbs swing by up to three stroke widths, always more than the middle of the shape; the stroke swells and thins visibly along its length; corners and line ends land in noticeably different spots and their angles change by a few degrees; the shape as a whole nudges one or two stroke widths in a random direction, tilts by two to four degrees either way, and changes size by a few percent; flat color fills slide inside their outlines and overshoot or fall short of the edge in different places each time. Colors, opacity and marker texture of the overlay stay exactly as in .

All of this jitter happens around a fixed anchor and always returns to it: no shape ever drifts away, walks across the frame, grows, shrinks over time, or ends up anywhere other than where it sits in . Nothing travels, orbits, fades, sparkles or glows. Their number never changes. They are already complete in the first frame — no progressive draw-on, no reveal. All of this happens strictly inside the overlay shapes and never reaches the photograph beneath or around them.

Final look: an unmodified still photograph, with a set of marker shapes visibly jittering on top of it. The photograph itself is never redrawn, never outlined, never illustrated.

Pure video, no subtitles, no background music, no cuts, no camera movement, no added text.
Մաս 2

Որտեղ օգտագործել

  • Instagram carousel. մեկ նկար, մեկ գրաֆիկ շերտ, հետո կարճ շարժում։
  • Stories cover. դեմքը մաքուր, գրաֆիկը կողքին։
  • Ապրանքի նկար. գրաֆիկը ապրանքի շուրջ, ոչ դեմքի վրա։
  • Ճամփորդության կամ հագուստի հրապարակում. վայրին համապատասխան մոտիվներ, ոչ պատահական ծաղիկներ։
  • Բրենդային տեսք. նույն ձեռքի գրաֆիկ ոճը մի քանի հրապարակման մեջ. ճանաչելի, ոչ պատրաստի AI։

Մեկ որակյալ տարբերակ ավելի լավ է, քան տասը նույնը։

Կարճ ամփոփում

  • Նկար → GPT գրաֆիկ / նկարազարդում → Seedance 2.5 (շարժվում է միայն այդ շերտը)։
  • Պրոմպտերը վերևում են. պատճենիր ինչպես կա։
  • Նպատակը. յուրահատուկ տեսք, ոչ ծավալ։